IEEE Spectrum IEEE Spectrum
-
Barrier Breaker Shapes Aerospace Engineering's Future
by Willie D. Jones on 18. September 2024. at 12:00
Wesley L. Harris’s life is a testament to the power of mentorship and determination. Harris, born in 1941 in Richmond, Virginia, grew up during the tumultuous years of the Civil Rights Movement and faced an environment fraught with challenges. His parents, both of whom only had a third-grade education, walked to Richmond from rural Virginia counties when the Great Depression left the region’s farming communities destitute. They found work as laborers in the city’s tobacco factories but pushed their son to pursue higher education so he could live a better life.
Today, Harris is a professor of aeronautics and astronautics at MIT and heads the school’s Hypersonic Research Laboratory. More importantly, he is committed to fostering the next generation of engineers, particularly students of color.
“I’ve been keeping my head down, working with students of color—especially at the Ph.D. level—to produce more scholars,” Harris says. “I do feel good about that.”
From physics to aerospace engineering
Harris’s journey into the world of science began under the guidance of his physics teacher at the all-Black Armstrong High School, in Richmond. The instructor taught Harris how to build a cloud chamber to investigate the collision of alpha particles with water droplets. The chamber made it possible to visualize the passage of ionizing radiation emitted by radium 226, which Harris sourced from a wristwatch that used the substance to make the watch hands glow in the dark.
The project won first prize at Virginia’s statewide Black high school science fair, and he took the bold step of signing up for a separate science fair held for the state’s White students. Harris’s project received the third-place prize in physics at that event.
Those awards and his teacher’s unwavering belief in Harris’s potential pushed him to aim higher. He says that he wanted nothing more than to become a physicist like her. Ironically, it was also her influence that led him to shift his career path from physics to aeronautical engineering.
When discussing which college he should attend, she spoke to him as though he were a soldier getting his marching orders. “Wesley, you will go to the University of Virginia [in Charlottesville],” she proclaimed.
Harris applied, knowing full well that the school did not allow Black students in the 1960s to pursue degrees in mathematics, physics, chemistry, English, economics, or political science.
The only available point of entry for him was the university’s School of Engineering. He chose aerospace as his focus—the only engineering discipline that interested him. Harris became one of only seven Black students on a campus with 4,000 undergrads and the first Black student to join the prestigious Jefferson Society literary and debate club. He graduated in 1964 with a bachelor’s degree in aerospace engineering. He went on to earn his master’s and doctoral degrees in aerospace engineering from Princeton in 1966 and 1968, respectively.
Harris’s Ph.D. thesis advisor at Princeton reinforced the values of mentorship and leadership instilled by his high school teacher, urging Harris to focus not only on his research but on how he could uplift others.
Harris began his teaching career by breaking down barriers at the University of Virginia in 1968. He was the first Black person in the school’s history to be offered a tenured faculty position. He was also the university’s first Black engineering professor. In 1972, he joined MIT as a professor of aeronautics and astronautics.
Harris’s dedication to supporting underrepresented minority groups at MIT began early in his tenure. In 1975, he founded the Office of Minority Education, where he pioneered innovative teaching methods such as videotaping and replaying lectures, which helped countless students succeed. “Some of those old videotapes may still be around,” he says, laughing.
“I’ve been keeping my head down, working with students of color—especially at the Ph.D. level—to produce more scholars. I do feel good about that.”
Over the years, he has periodically stepped away from MIT to take on other roles, including Program Manager in the Fluid and Thermal Physics Office and as manager of Computational Methods at NASA’s headquarters in Washington, D.C., from 1979 to 1980. He returned to NASA in 1993 and served as Associate Administrator for Aeronautics, overseeing personnel, programs, and facilities until 1995.
He also served as Chief Administrative Officer and Vice President at the University of Tennessee Space Institute in Knoxville from 1990 to 1993 and as Dean of Engineering at the University of Connecticut, in Storrs, from 1985 to 1990.
He was selected for membership in an oversight group convened by the U.S. House of Representatives Science Subcommittee on Research and Technology to monitor the funding activities of the National Science Foundation. He has also been a member and chair of the U.S. Army Science Board.
Solving problems with aircraft
Harris is a respected aeronautical innovator. Near the end of the Vietnam War, the U.S. Army approached MIT to help it solve a problem. Helicopters were being shot down by the enemy, who had learned to distinguish attack helicopters from those used for performing reconnaissance or transporting personnel and cargo by the noise they made. The Army needed a solution that would reduce the helicopters’ acoustic signatures without compromising performance. Harris and his aeronautics team at MIT delivered that technology. In January 1978, they presented a lab report detailing their findings to the U.S. Department of Defense. “Experimental and Theoretical Studies on Model Helicopter Rotor Noise” was subsequently published in The Journal of Sound and Vibration. A year later, Harris and his colleagues at the Fluid Dynamic Research Laboratory wrote another lab report on the topic, “Parametric Studies of Model Helicopter Blade Slap and Rotational Noise.”
Harris has also heightened scientists’ understanding of the climate-altering effects of shock waves propagating upward from aircraft flying at supersonic speeds. He discovered that these high-speed airflows trigger chemical reactions among the carbon, oxides, nitrides, and sulfides in the atmosphere.
For these and other contributions to aerospace engineering, Harris, a member of the American Institute of Aeronautics and Astronautics, was elected in 1995 to the National Academy of Engineering. In 2022, he was named the academy’s vice president.
A model of educational leadership
Despite his technical achievements, Harris says his greatest fulfillment comes from mentoring students. He takes immense pride in the four students who recently earned doctorates in hypersonics under his guidance, especially a Black woman who graduated this year.
Harris’s commitment to nurturing young talent extends beyond his graduate students. For more than two decades, he has served as a housemaster at MIT’s New House residence hall, where he helps first-year undergraduate students successfully transition to campus life.
“You must provide an environment that fosters the total development of the student, not just mastery of physics, chemistry, math, and economics,” Harris says.
He takes great satisfaction in watching his students grow and succeed, knowing that he helped prepare them to make a positive impact on the world.
Reflecting on his career, Harris acknowledges the profound impact of the mentors who guided him. Their lessons continue to influence his work and his unwavering commitment to mentoring the next generation.
“I’ve always wanted to be like my high school teacher—a physicist who not only had deep knowledge of the scientific fundamentals but also compassion and love for Black folks,” he says.
Through his work, Harris has not only advanced the field of aerospace engineering but has also paved the way for future generations to soar.
-
ICRA@40 Conference Celebrates 40 Years of IEEE Robotics
by Evan Ackerman on 18. September 2024. at 11:30
Four decades after the first IEEE International Conference on Robotics and Automation (ICRA) in Atlanta, robotics is bigger than ever. Next week in Rotterdam is the IEEE ICRA@40 conference, “a celebration of 40 years of pioneering research and technological advancements in robotics and automation.” There’s an ICRA every year, of course. Arguably the largest robotics research conference in the world, the 2024 edition was held in Yokohama, Japan back in May.
ICRA@40 is not just a second ICRA conference in 2024. Next week’s conference is a single track that promises “a journey through the evolution of robotics and automation,” through four days of short keynotes from prominent roboticists from across the entire field. You can see for yourself, the speaker list is nuts. There are also debates and panels tackling big ideas, like: “What progress has been made in different areas of robotics and automation over the past decades, and what key challenges remain?” Personally, I’d say “lots” and “most of them,” but that’s probably why I’m not going to be up on stage.
There will also be interactive research presentations, live demos, an expo, and more—the conference schedule is online now, and the abstracts are online as well. I’ll be there to cover it all, but if you can make it in person, it’ll be worth it.
Forty years ago is a long time, but it’s not that long, so just for fun, I had a look at the proceedings of ICRA 1984 which are available on IEEE Xplore, if you’re curious. Here’s an excerpt of the forward from the organizers, which included folks from International Business Machines and Bell Labs:
The proceedings of the first IEEE Computer Society International Conference on Robotics contains papers covering practically all aspects of robotics. The response to our call for papers has been overwhelming, and the number of papers submitted by authors outside the United States indicates the strong international interest in robotics.
The Conference program includes papers on: computer vision; touch and other local sensing; manipulator kinematics, dynamics, control and simulation; robot programming languages, operating systems, representation, planning, man-machine interfaces; multiple and mobile robot systems.
The technical level of the Conference is high with papers being presented by leading researchers in robotics. We believe that this conference, the first of a series to be sponsored by the IEEE, will provide a forum for the dissemination of fundamental research results in this fast developing field.Technically, this was “ICR,” not “ICRA,” and it was put on by the IEEE Computer Society’s Technical Committee on Robotics, since there was no IEEE Robotics and Automation Society at that time; RAS didn’t get off the ground until 1987.
1984 ICR(A) had two tracks, and featured about 75 papers presented over three days. Looking through the proceedings, you’ll find lots of familiar names: Harry Asada, Ruzena Bajcsy, Ken Salisbury, Paolo Dario, Matt Mason, Toshio Fukuda, Ron Fearing, and Marc Raibert. Many of these folks will be at ICRA@40, so if you see them, make sure and thank them for helping to start it all, because 40 years of robotics is definitely something to celebrate. -
Glass Antenna Turns Windows Into 5G Base Stations
by Tim Hornyak on 18. September 2024. at 11:00
Since 5G began its rollout in 2018 or 2019, fifth-generation wireless networks have spread across the globe to cover hundreds of millions of users. But while it offers lower latency than precursor networks, 5G also requires more base stations. To avoid installing unsightly equipment on more and more shared spaces, Japanese companies are developing transparent glass antennas that allow windows to serve as base stations that can be shared by several carriers.
Because 5G networks include spectrum comprising higher frequencies than 4G, base stations for 5G networks serve a smaller coverage footprint. Which means more base stations are needed compared to 4G. Due to a lack of installation spots and the high cost of rolling out 5G networks, carriers in Japan have been sharing mobile infrastructure.
Last month the Tokyo-based communications company JTower announced the deployment of the new glass antenna, created in part by glassmaker AGC (one of the world’s largest) and the mobile carrier NTT Docomo. The first was installed on a window in Tokyo’s Shinjuku district.
The product is “the world’s first antenna that turns a window into a base station that can be attached to a building window inside and turn the outdoors into a service area without spoiling the cityscape or the exterior appearance of the building,” says Shota Ochiai, a marketing manager at AGC.
NTT Docomo reports that it uses transparent conductive materials as the basis for its antenna, sandwiching the conductive material along with a transparent resin, the kind used in laminated windshields, in between two sheets of glass.
“I don’t think the idea for using transparent conductive materials as an antenna existed before,” said AGC’s Kentaro Oka in a company statement. “The durability of the antenna was significantly increased by placing the conductive materials between glass.”
The transparent antenna can be engineered according to the thickness of the glass to reduce the attenuation and reflection of the radio signals being absorbed and emitted by the window-sized device. “The glass antenna uses our proprietary technology to smooth out the disruption in the direction of radio waves when they pass through a window,” says Ochia.
A brief history of the window antenna
Branded WAVEANTENNA, the antenna is installed on the interior surface of windows. Apart perhaps from its cabling, the WAVEANTENNA is an otherwise inconspicuous piece of equipment that is often tucked out of sight, placed near the top or otherwise at the edges of a window.
It is compatible with frequencies in the 5G Sub6 band—meaning signals that are less than 6 gigahertz (GHz). Sub6 antennas represent critical portions of a 5G deployment, as their lower frequency ranges penetrate barriers like walls and buildings better than the substantially higher-bandwidth millimeter-wave portions of the 5G spectrum.
An earlier version of the product was launched in 2020, while a version that could handle sharing by multiple cell networks was introduced last year, according to AGC. The company says its antenna is optimized for frequencies between 3.7 and 4.5 GHz bands, which still allows for substantial bandwidth—albeit not comparable with what an ideal millimeter-wave 5G deployment could reach. (Millimeter waves can deliver typically between 10 and 50 GHz of bandwidth.)
The glass antenna can help expand 5G coverage as infrastructure sharing will become more important to carriers, AGC says. Besides increasing the number of locations for base stations, the device makes it easier to select the appropriate installation height, according to Ochiai.
AGC has also applied 5G glass antennas to automobiles, where they can help reduce dropped signals. The company reports that users include Halo.Car, an on-demand EV rental service in Las Vegas that relies on high-speed networks for remote drivers to deliver cars to customers.
-
Engineering Students Innovate Accessibility Technology
by Ashley Moran on 17. September 2024. at 18:00
More than 15 percent of the world’s population—greater than 1 billion people—live with disabilities including hearing loss, vision problems, mental health challenges, and lack of mobility. EPICS in IEEE has engaged students’ ingenuity worldwide to address accessibility issues through adaptive services, redesigned technology, and new assistive technologies during its 2023 Access and Abilities Competition.
The competition challenged university students around the world to use their engineering skills to help with accessibility issues. The EPICS in IEEE Committee received 58 proposals and selected 23 projects, which were funded in early 2023.
EPICS is a grant-based program for IEEE Educational Activities that funds service learning projects for university and high school students.
The teams, which include faculty members and IEEE members, create and execute engineering projects in partnership with organizations to improve their communities.
“Some gamers with arm or hand deficiencies play with their feet, nose, mouth, or elbows, or they use devices not intended for that purpose and are forced to adapt. I realized that if there was a dedicated device designed for such individuals, they’d be able to play and experience the joy of gaming.” —John McCauley.
The four EPICS in IEEE pillars are access and abilities; environment; education and outreach; and human services. In the Access and Abilities Competition, student teams received between US $1,000 and $10,000. Each team had 12 months to build a prototype or solution in collaboration with its community partners. The projects, which involved more than 350 students and 149 IEEE volunteers, aimed to help an estimated 8,000 people in the first year of deployment.
The teams included participants from IEEE student branches, IEEE Women in Engineering groups, IEEE–Eta Kappa Nu honor society chapters, and IEEE sections.
Projects included a sound-detection device and a self-navigating robotic walking aid.
The competition was funded by the Taenzer Memorial Fund in 2019, with $90,000 allocated by the IEEE Foundation. The fund was established with a bequest from the estate of Jon C. Taenzer, an IEEE life senior member.
The student teams submitted their final reports this year.
Here are highlights from four of the projects:
Adaptive mouse for gaming
Members of the adaptive mouse EPICS in IEEE team at the University of Florida in Gainesville designed a device that contains keyboard functions and can be used with just one hand.EPICS in IEEE
A team of 10 biomedical engineering students at the University of Florida in Gainesville designed their project to help people whose hands or arms have an abnormality, so they could more easily play games.
The team built five adaptive mouse devices and plans to deliver them this year to five recipients involved with Hands to Love, a Florida-based organization that supports children with upper limb abnormalities.
The team incorporated the keyboard elements of gaming into a mouse, allowing gaming gestures and movements with just one hand. The 3D-printed mouse combines existing gaming technology, including the internal mechanisms of keyboards, a Logitech mouse, and Microsoft Xbox controller emulations. It allows the player to move and aim while gaming with just a mouse.
Gaming enthusiast John McCauley, a junior in the university’s biomedical engineering program, was behind the project’s conception.
“Some gamers with arm or hand deficiencies play with their feet, nose, mouth, or elbows, or they use devices not intended for that purpose and are forced to adapt,” McCauley says. “I realized that if there was a dedicated device designed for such individuals, they’d be able to play and experience the joy of gaming.”
The team used its $1,000 EPICS in IEEE grant to purchase the prototype’s components.
Making campus more accessible
Universidad Tecnólogica de Panamá students test their microcontroller-based prototype, designed to help make their school more accessible.EPICS in IEEE
A team of 15 undergraduate students from the Universidad Tecnológica de Panamá in Panama City and 24 students from four high schools in Chiriquí, Panama, created several projects focused on people with visual or physical disabilities. The team’s goal was to make their campus and community more accessible to those with different abilities. The projects enhanced their classmates’ autonomy and improved their quality of life.
The team made braille signs using a 3D printer, and they designed and built a personalized wheelchair. The students also automated the doors within the engineering department to provide better access to classrooms and corridors for those with disabilities.
“This project will be very useful, especially [in Panama], where buildings have not been adapted for people with disabilities,” said team member Gael Villarreal, a high school junior.
While working together on the project, team members honed their technical and interpersonal skills. They came to appreciate the importance of collaboration and communication.
“I learned that you need to have new experiences, be sociable, meet and get along with new people, and work as a team to be successful,” high school junior Gianny Rodriguez said.
The team used its $8,100 EPICS grant to purchase materials and train the community on using the new tools.
Helping children with hearing impairments
A team of students from the SRM Institute of Science and Technology student branch, in Chennai, India, worked with the Dr. MGR Home and Higher Secondary School for the Speech and Hearing Impaired, also in Chennai, to build a device to help children with hearing aids and cochlear implants learn Tamil, the local language. In rural areas, young children often do not have access to specialized speech and hearing health care providers to learn critical language skills. The team’s assistive device supports native language skill development, helping parents and trainers support the children in language and sound acquisition.
The project is designed to provide access to aural rehabilitation, including identifying hearing loss and therapies for children far from hospitals and rehabilitation centers.
The kiosklike device resembles an ATM and includes surround-sound speakers and touchscreens. It uses a touch monitor and microphones to access tasks and tests that help young children learn Tamil.
The team worked with 150 pupils at the school between the ages of 5 and 8 to develop the prototype. The built-in app includes tasks that focus on improving auditory awareness, auditory discrimination (the ability to recognize, compare, and distinguish between distinct sounds), and language acquisition (how people perceive and comprehend language).
The device tests the pupil’s hearing range based on sounds with visual cues, sounds at low intensity, sounds in the presence of noise, and sound direction.
The speakers emulate real-life situations and are used to relay the teacher’s instructions.
The team received a $1,605 grant to execute the project.
This video spotlights the challenges youngsters with hearing disabilities in Chenni, India, face and how the assistive technology will help them.
Self-navigating robotic walking aid
Students from the IEEE Swinburne Sarawak student branch in Malaysia brought a prototype of their walking aid to Trinity Eldercare, their community partner.EPICS in IEEE
To help senior citizens with mobility issues, a team of students from the IEEE Swinburne Sarawak student branch at the Swinburne University of Technology, in Malaysia, created a self-navigating walking aid.
The team wanted to improve existing walkers on the market, so they surveyed residents at Trinity Eldercare to find out what features would be useful to them.
The students’ prototype, based on a commercial walker, includes a wearable haptic belt that detects obstacles and alerts the user. Pressure sensors in the hand grips sense which direction the user wants to go. One of the senior citizens’ most requested features was the ability to locate a misplaced walker. The team was able to address the issue using sensors.
“I gained substantial knowledge in robotics programming and artificial intelligence and deep learning integration for person tracking and autonomous navigation,” one of the team members said. “Additionally, presenting our smart walker prototype at the International Invention, Innovation, Technolgy Competition and Exhibition in Malaysia enhanced my presentation skills, as I successfully articulated its viability and usefulness to the judges.”
The project received a $1,900 grant.
Join the EPICS in IEEE mailing list to learn more about all the Access and Abilities Competition projects and other impactful efforts made possible by donations to the IEEE Foundation. To learn more, check out the video of the competition:
The EPICS in IEEE program is celebrating its 15th year of supporting and facilitating service-learning projects and impacting students and communities worldwide
-
How and Why Gary Marcus Became AI's Leading Critic
by Eliza Strickland on 17. September 2024. at 12:00
Maybe you’ve read about Gary Marcus’s testimony before the Senate in May of 2023, when he sat next to Sam Altman and called for strict regulation of Altman’s company, OpenAI, as well as the other tech companies that were suddenly all-in on generative AI. Maybe you’ve caught some of his arguments on Twitter with Geoffrey Hinton and Yann LeCun, two of the so-called “godfathers of AI.” One way or another, most people who are paying attention to artificial intelligence today know Gary Marcus’s name, and know that he is not happy with the current state of AI.
He lays out his concerns in full in his new book, Taming Silicon Valley: How We Can Ensure That AI Works for Us, which was published today by MIT Press. Marcus goes through the immediate dangers posed by generative AI, which include things like mass-produced disinformation, the easy creation of deepfake pornography, and the theft of creative intellectual property to train new models (he doesn’t include an AI apocalypse as a danger, he’s not a doomer). He also takes issue with how Silicon Valley has manipulated public opinion and government policy, and explains his ideas for regulating AI companies.
Marcus studied cognitive science under the legendary Steven Pinker, was a professor at New York University for many years, and co-founded two AI companies, Geometric Intelligence and Robust.AI. He spoke with IEEE Spectrum about his path to this point.
What was your first introduction to AI?
Gary MarcusBen Wong
Gary Marcus: Well, I started coding when I was eight years old. One of the reasons I was able to skip the last two years of high school was because I wrote a Latin-to-English translator in the programming language Logo on my Commodore 64. So I was already, by the time I was 16, in college and working on AI and cognitive science.
So you were already interested in AI, but you studied cognitive science both in undergrad and for your Ph.D. at MIT.
Marcus: Part of why I went into cognitive science is I thought maybe if I understood how people think, it might lead to new approaches to AI. I suspect we need to take a broad view of how the human mind works if we’re to build really advanced AI. As a scientist and a philosopher, I would say it’s still unknown how we will build artificial general intelligence or even just trustworthy general AI. But we have not been able to do that with these big statistical models, and we have given them a huge chance. There’s basically been $75 billion spent on generative AI, another $100 billion on driverless cars. And neither of them has really yielded stable AI that we can trust. We don’t know for sure what we need to do, but we have very good reason to think that merely scaling things up will not work. The current approach keeps coming up against the same problems over and over again.
What do you see as the main problems it keeps coming up against?
Marcus: Number one is hallucinations. These systems smear together a lot of words, and they come up with things that are true sometimes and not others. Like saying that I have a pet chicken named Henrietta is just not true. And they do this a lot. We’ve seen this play out, for example, in lawyers writing briefs with made-up cases.
Second, their reasoning is very poor. My favorite examples lately are these river-crossing word problems where you have a man and a cabbage and a wolf and a goat that have to get across. The system has a lot of memorized examples, but it doesn’t really understand what’s going on. If you give it a simpler problem, like one Doug Hofstadter sent to me, like: “A man and a woman have a boat and want to get across the river. What do they do?” It comes up with this crazy solution where the man goes across the river, leaves the boat there, swims back, something or other happens.
Sometimes he brings a cabbage along, just for fun.
Marcus: So those are boneheaded errors of reasoning where there’s something obviously amiss. Every time we point these errors out somebody says, “Yeah, but we’ll get more data. We’ll get it fixed.” Well, I’ve been hearing that for almost 30 years. And although there is some progress, the core problems have not changed.
Let’s go back to 2014 when you founded your first AI company, Geometric Intelligence. At that time, I imagine you were feeling more bullish on AI?
Marcus: Yeah, I was a lot more bullish. I was not only more bullish on the technical side. I was also more bullish about people using AI for good. AI used to feel like a small research community of people that really wanted to help the world.
So when did the disillusionment and doubt creep in?
Marcus: In 2018 I already thought deep learning was getting overhyped. That year I wrote this piece called “Deep Learning, a Critical Appraisal,” which Yann LeCun really hated at the time. I already wasn’t happy with this approach and I didn’t think it was likely to succeed. But that’s not the same as being disillusioned, right?
Then when large language models became popular [around 2019], I immediately thought they were a bad idea. I just thought this is the wrong way to pursue AI from a philosophical and technical perspective. And it became clear that the media and some people in machine learning were getting seduced by hype. That bothered me. So I was writing pieces about GPT-3 [an early version of OpenAI's large language model] being a bullshit artist in 2020. As a scientist, I was pretty disappointed in the field at that point. And then things got much worse when ChatGPT came out in 2022, and most of the world lost all perspective. I began to get more and more concerned about misinformation and how large language models were going to potentiate that.
You’ve been concerned not just about the startups, but also the big entrenched tech companies that jumped on the generative AI bandwagon, right? Like Microsoft, which has partnered with OpenAI?
Marcus: The last straw that made me move from doing research in AI to working on policy was when it became clear that Microsoft was going to race ahead no matter what. That was very different from 2016 when they released [an early chatbot named] Tay. It was bad, they took it off the market 12 hours later, and then Brad Smith wrote a book about responsible AI and what they had learned. But by the end of the month of February 2023, it was clear that Microsoft had really changed how they were thinking about this. And then they had this ridiculous “Sparks of AGI” paper, which I think was the ultimate in hype. And they didn’t take down Sydney after the crazy Kevin Roose conversation where [the chatbot] Sydney told him to get a divorce and all this stuff. It just became clear to me that the mood and the values of Silicon Valley had really changed, and not in a good way.
I also became disillusioned with the U.S. government. I think the Biden administration did a good job with its executive order. But it became clear that the Senate was not going to take the action that it needed. I spoke at the Senate in May 2023. At the time, I felt like both parties recognized that we can’t just leave all this to self-regulation. And then I became disillusioned [with Congress] over the course of the last year, and that’s what led to writing this book.
You talk a lot about the risks inherent in today’s generative AI technology. But then you also say, “It doesn’t work very well.” Are those two views coherent?
Marcus: There was a headline: “Gary Marcus Used to Call AI Stupid, Now He Calls It Dangerous.” The implication was that those two things can’t coexist. But in fact, they do coexist. I still think gen AI is stupid, and certainly cannot be trusted or counted on. And yet it is dangerous. And some of the danger actually stems from its stupidity. So for example, it’s not well-grounded in the world, so it’s easy for a bad actor to manipulate it into saying all kinds of garbage. Now, there might be a future AI that might be dangerous for a different reason, because it’s so smart and wily that it outfoxes the humans. But that’s not the current state of affairs.
You’ve said that generative AI is a bubble that will soon burst. Why do you think that?
Marcus: Let’s clarify: I don’t think generative AI is going to disappear. For some purposes, it is a fine method. You want to build autocomplete, it is the best method ever invented. But there’s a financial bubble because people are valuing AI companies as if they’re going to solve artificial general intelligence. In my view, it’s not realistic. I don’t think we’re anywhere near AGI. So then you’re left with, “Okay, what can you do with generative AI?”
Last year, because Sam Altman was such a good salesman, everybody fantasized that we were about to have AGI and that you could use this tool in every aspect of every corporation. And a whole bunch of companies spent a bunch of money testing generative AI out on all kinds of different things. So they spent 2023 doing that. And then what you’ve seen in 2024 are reports where researchers go to the users of Microsoft’s Copilot—not the coding tool, but the more general AI tool—and they’re like, “Yeah, it doesn’t really work that well.” There’s been a lot of reviews like that this last year.
The reality is, right now, the gen AI companies are actually losing money. OpenAI had an operating loss of something like $5 billion last year. Maybe you can sell $2 billion worth of gen AI to people who are experimenting. But unless they adopt it on a permanent basis and pay you a lot more money, it’s not going to work. I started calling OpenAI the possible WeWork of AI after it was valued at $86 billion. The math just didn’t make sense to me.
What would it take to convince you that you’re wrong? What would be the head-spinning moment?
Marcus: Well, I’ve made a lot of different claims, and all of them could be wrong. On the technical side, if someone could get a pure large language model to not hallucinate and to reason reliably all the time, I would be wrong about that very core claim that I have made about how these things work. So that would be one way of refuting me. It hasn’t happened yet, but it’s at least logically possible.
On the financial side, I could easily be wrong. But the thing about bubbles is that they’re mostly a function of psychology. Do I think the market is rational? No. So even if the stuff doesn’t make money for the next five years, people could keep pouring money into it.
The place that I’d like to prove me wrong is the U.S. Senate. They could get their act together, right? I’m running around saying, “They’re not moving fast enough,” but I would love to be proven wrong on that. In the book, I have a list of the 12 biggest risks of generative AI. If the Senate passed something that actually addressed all 12, then my cynicism would have been mislaid. I would feel like I’d wasted a year writing the book, and I would be very, very happy.
-
Challengers Are Coming for Nvidia’s Crown
by Matthew S. Smith on 16. September 2024. at 14:00
It’s hard to overstate Nvidia’s AI dominance. Founded in 1993, Nvidia first made its mark in the then-new field of graphics processing units (GPUs) for personal computers. But it’s the company’s AI chips, not PC graphics hardware, that vaulted Nvidia into the ranks of the world’s most valuable companies. It turns out that Nvidia’s GPUs are also excellent for AI. As a result, its stock is more than 15 times as valuable as it was at the start of 2020; revenues have ballooned from roughly US $12 billion in its 2019 fiscal year to $60 billion in 2024; and the AI powerhouse’s leading-edge chips are as scarce, and desired, as water in a desert.
Access to GPUs “has become so much of a worry for AI researchers, that the researchers think about this on a day-to-day basis. Because otherwise they can’t have fun, even if they have the best model,” says Jennifer Prendki, head of AI data at Google DeepMind. Prendki is less reliant on Nvidia than most, as Google has its own homespun AI infrastructure. But other tech giants, like Microsoft and Amazon, are among Nvidia’s biggest customers, and continue to buy its GPUs as quickly as they’re produced. Exactly who gets them and why is the subject of an antitrust investigation by the U.S. Department of Justice, according to press reports.
Nvidia’s AI dominance, like the explosion of machine learning itself, is a recent turn of events. But it’s rooted in the company’s decades-long effort to establish GPUs as general computing hardware that’s useful for many tasks besides rendering graphics. That effort spans not only the company’s GPU architecture, which evolved to include “tensor cores” adept at accelerating AI workloads, but also, critically, its software platform, called Cuda, to help developers take advantage of the hardware.
“They made sure every computer-science major coming out of university is trained up and knows how to program CUDA,” says Matt Kimball, principal data-center analyst at Moor Insights & Strategy. “They provide the tooling and the training, and they spend a lot of money on research.”
Released in 2006, CUDA helps developers use an Nvidia GPU’s many cores. That’s proved essential for accelerating highly parallelized compute tasks, including modern generative AI. Nvidia’s success in building the CUDA ecosystem makes its hardware the path of least resistance for AI development. Nvidia chips might be in short supply, but the only thing more difficult to find than AI hardware is experienced AI developers—and many are familiar with CUDA.
That gives Nvidia a deep, broad moat with which to defend its business, but that doesn’t mean it lacks competitors ready to storm the castle, and their tactics vary widely. While decades-old companies like Advanced Micro Devices (AMD) and Intel are looking to use their own GPUs to rival Nvidia, upstarts like Cerebras and SambaNova have developed radical chip architectures that drastically improve the efficiency of generative AI training and inference. These are the competitors most likely to challenge Nvidia.
AMD: The other GPU maker
Pro: AMD GPUs are convincing Nvidia alternatives
Con: Software ecosystem can’t rival Nvidia’s CUDA
AMD has battled Nvidia in the graphics-chip arena for nearly two decades. It’s been, at times, a lopsided fight. When it comes to graphics, AMD’s GPUs have rarely beaten Nvidia’s in sales or mindshare. Still, AMD’s hardware has its strengths. The company’s broad GPU portfolio extends from integrated graphics for laptops to AI-focused data-center GPUs with over 150 billion transistors. The company was also an early supporter and adopter of high-bandwidth memory (HBM), a form of memory that’s now essential to the world’s most advanced GPUs.
“If you look at the hardware…it stacks up favorably” to Nvidia, says Kimball, referring to AMD’s Instinct MI325X, a competitor of Nvidia’s H100. “AMD did a fantastic job laying that chip out.”
The MI325X, slated to launch by the end of the year, has over 150 billion transistors and 288 gigabytes of high-bandwidth memory, though real-world results remain to be seen. The MI325X’s predecessor, the MI300X, earned praise from Microsoft, which deploys AMD hardware, including the MI300X, to handle some ChatGPT 3.5 and 4 services. Meta and Dell have also deployed the MI300X, and Meta used the chips in parts of the development of its latest large language model, Llama 3.1.
There’s still a hurdle for AMD to leap: software. AMD offers an open-source platform, ROCm, to help developers program its GPUs, but it’s less popular than CUDA. AMD is aware of this weakness, and in July 2024, it agreed to buy Europe’s largest private AI lab, Silo AI, which has experience doing large-scale AI training using ROCm and AMD hardware. AMD has also plans to purchase ZT Systems, a company with expertise in data-center infrastructure, to help the company serve customers looking to deploy its hardware at scale. Building a rival to CUDA is no small feat, but AMD is certainly trying.
Intel: Software success
Pro: Gaudi 3 AI accelerator shows strong performance
Con: Next big AI chip doesn’t arrive until late 2025
Intel’s challenge is the opposite of AMD’s.
While Intel lacks an exact match for Nvidia’s CUDA and AMD’s ROCm, it launched an open-source unified programming platform, OneAPI, in 2018. Unlike CUDA and ROCm, OneAPI spans multiple categories of hardware, including CPUs, GPUs, and FPGAs. So it can help developers accelerate AI tasks (and many others) on any Intel hardware. “Intel’s got a heck of a software ecosystem it can turn on pretty easily,” says Kimball.
Hardware, on the other hand, is a weakness, at least when compared to Nvidia and AMD. Intel’s Gaudi AI accelerators, the fruit of Intel’s 2019 acquisition of AI hardware startup Habana Labs, have made headway, and the latest, Gaudi 3, offers performance that’s competitive with Nvidia’s H100.
However, it’s unclear precisely what Intel’s next hardware release will look like, which has caused some concern. “Gaudi 3 is very capable,” says Patrick Moorhead, founder of Moor Insights & Strategy. But as of July 2024 “there is no Gaudi 4,” he says.
Intel instead plans to pivot to an ambitious chip, code-named Falcon Shores, with a tile-based modular architecture that combines Intel x86 CPU cores and Xe GPU cores; the latter are part of Intel’s recent push into graphics hardware. Intel has yet to reveal details about Falcon Shores’ architecture and performance, though, and it’s not slated for release until late 2025.
Cerebras: Bigger is better
Pro: Wafer-scale chips offer strong performance and memory per chip
Con: Applications are niche due to size and cost
Make no mistake: AMD and Intel are by far the most credible challengers to Nvidia. They share a history of designing successful chips and building programming platforms to go alongside them. But among the smaller, less proven players, one stands out: Cerebras.
The company, which specializes in AI for supercomputers, made waves in 2019 with the Wafer Scale Engine, a gigantic, wafer-size piece of silicon packed with 1.2 trillion transistors. The most recent iteration, Wafer Scale Engine 3, ups the ante to 4 trillion transistors. For comparison, Nvidia’s largest and newest GPU, the B200, has “just” 208 billion transistors. The computer built around this wafer-scale monster, Cerebras’s CS-3, is at the heart of the Condor Galaxy 3, which will be an 8-exaflop AI supercomputer made up of 64 CS-3s. G42, an Abu Dhabi–based conglomerate that hopes to train tomorrow’s leading-edge large language models, will own the system.
“It’s a little more niche, not as general purpose,” says Stacy Rasgon, senior analyst at Bernstein Research. “Not everyone is going to buy [these computers]. But they’ve got customers, like the [United States] Department of Defense, and [the Condor Galaxy 3] supercomputer.”
Cerebras’s WSC-3 isn’t going to challenge Nvidia, AMD, or Intel hardware in most situations; it’s too large, too costly, and too specialized. But it could give Cerebras a unique edge in supercomputers, because no other company designs chips on the scale of the WSE.
SambaNova: A transformer for transformers
Pro: Configurable architecture helps developers squeeze efficiency from AI models
Con: Hardware still has to prove relevance to mass market
SambaNova, founded in 2017, is another chip-design company tackling AI training with an unconventional chip architecture. Its flagship, the SN40L, has what the company calls a “reconfigurable dataflow architecture” composed of tiles of memory and compute resources. The links between these tiles can be altered on the fly to facilitate the quick movement of data for large neural networks.
Prendki believes such customizable silicon could prove useful for training large language models, because AI developers can optimize the hardware for different models. No other company offers that capability, she says.
SambaNova is also scoring wins with SambaFlow, the software stack used alongside the SN40L. “At the infrastructure level, SambaNova is doing a good job with the platform,” says Moorhead. SambaFlow can analyze machine learning models and help developers reconfigure the SN40L to accelerate the model’s performance. SambaNova still has a lot to prove, but its customers include SoftBank and Analog Devices.
Groq: Form for function
Pro: Excellent AI inference performance
Con: Application currently limited to inference
Yet another company with a unique spin on AI hardware is Groq. Groq’s approach is focused on tightly pairing memory and compute resources to accelerate the speed with which a large language model can respond to prompts.
“Their architecture is very memory based. The memory is tightly coupled to the processor. You need more nodes, but the price per token and the performance is nuts,” says Moorhead. The “token” is the basic unit of data a model processes; in an LLM, it’s typically a word or portion of a word. Groq’s performance is even more impressive, he says, given that its chip, called the Language Processing Unit Inference Engine, is made using GlobalFoundries’ 14-nanometer technology, several generations behind the TSMC technology that makes the Nvidia H100.
In July, Groq posted a demonstration of its chip’s inference speed, which can exceed 1,250 tokens per second running Meta’s Llama 3 8-billion parameter LLM. That beats even SambaNova’s demo, which can exceed 1,000 tokens per second.
Qualcomm: Power is everything
Pro: Broad range of chips with AI capabilities
Con: Lacks large, leading-edge chips for AI training
Qualcomm, well known for the Snapdragon system-on-a-chip that powers popular Android phones like the Samsung Galaxy S24 Ultra and OnePlus 12, is a giant that can stand toe-to-toe with AMD, Intel, and Nvidia.
But unlike those peers, the company is focusing its AI strategy more on AI inference and energy efficiency for specific tasks. Anton Lokhmotov, a founding member of the AI benchmarking organization MLCommons and CEO of Krai, a company that specializes in AI optimization, says Qualcomm has significantly improved the inference of the Qualcomm Cloud AI 100 servers in an important benchmark test. The servers’ performance increased from 180 to 240 samples-per-watt in ResNet-50, an image-classification benchmark, using “essentially the same server hardware,” Lokhmotov notes.
Efficient AI inference is also a boon on devices that need to handle AI tasks locally without reaching out to the cloud, says Lokhmotov. Case in point: Microsoft’s Copilot Plus PCs. Microsoft and Qualcomm partnered with laptop makers, including Dell, HP, and Lenovo, and the first Copilot Plus laptops with Qualcomm chips hit store shelves in July. Qualcomm also has a strong presence in smartphones and tablets, where its Snapdragon chips power devices from Samsung, OnePlus, and Motorola, among others.
Qualcomm is an important player in AI for driver assist and self-driving platforms, too. In early 2024, Hyundai’s Mobius division announced a partnership to use the Snapdragon Ride platform, a rival to Nvidia’s Drive platform, for advanced driver-assist systems.
The Hyperscalers: Custom brains for brawn
Pros: Vertical integration focuses design
Cons: Hyperscalers may prioritize their own needs and uses first
Hyperscalers—cloud-computing giants that deploy hardware at vast scales—are synonymous with Big Tech. Amazon, Apple, Google, Meta, and Microsoft all want to deploy AI hardware as quickly as possible, both for their own use and for their cloud-computing customers. To accelerate that, they’re all designing chips in-house.
Google began investing in AI processors much earlier than its competitors: The search giant’s Tensor Processing Units, first announced in 2015, now power most of its AI infrastructure. The sixth generation of TPUs, Trillium, was announced in May and is part of Google’s AI Hypercomputer, a cloud-based service for companies looking to handle AI tasks.
Prendki says Google’s TPUs give the company an advantage in pursuing AI opportunities. “I’m lucky that I don’t have to think too hard about where I get my chips,” she says. Access to TPUs doesn’t entirely eliminate the supply crunch, though, as different Google divisions still need to share resources.
And Google is no longer alone. Amazon has two in-house chips, Trainium and Inferentia, for training and inference, respectively. Microsoft has Maia, Meta has MTIA, and Apple is supposedly developing silicon to handle AI tasks in its cloud infrastructure.
None of these compete directly with Nvidia, as hyperscalers don’t sell hardware to customers. But they do sell access to their hardware through cloud services, like Google’s AI Hypercomputer, Amazon’s AWS, and Microsoft’s Azure. In many cases, hyperscalers offer services running on their own in-house hardware as an option right alongside services running on hardware from Nvidia, AMD, and Intel; Microsoft is thought to be Nvidia’s largest customer.
David Plunkert
Chinese chips: An opaque future
Another category of competitor is born not of technical needs but of geopolitical realities. The United States has imposed restrictions on the export of AI hardware that prevents chipmakers from selling their latest, most-capable chips to Chinese companies. In response, Chinese companies are designing homegrown AI chips.
Huawei is a leader. The company’s Ascend 910B AI accelerator, designed as an alternative to Nvidia’s H100, is in production at Semiconductor Manufacturing International Corp., a Shanghai-based foundry partially owned by the Chinese government. However, yield issues at SMIC have reportedly constrained supply. Huawei is also selling an “AI-in-a-box” solution, meant for Chinese companies looking to build their own AI infrastructure on-premises.
To get around the U.S. export control rules, Chinese industry could turn to alternative technologies. For example, Chinese researchers have made headway in photonic chips that use light, instead of electric charge, to perform calculations. “The advantage of a beam of light is you can cross one [beam with] another,” says Prendki. “So it reduces constraints you’d normally have on a silicon chip, where you can’t cross paths. You can make the circuits more complex, for less money.” It’s still very early days for photonic chips, but Chinese investment in the area could accelerate its development.
Room for more
It’s clear that Nvidia has no shortage of competitors. It’s equally clear that none of them will challenge—never mind defeat—Nvidia in the next few years. Everyone interviewed for this article agreed that Nvidia’s dominance is currently unparalleled, but that doesn’t mean it will crowd out competitors forever.
“Listen, the market wants choice,” says Moorhead. “I can’t imagine AMD not having 10 or 20 percent market share, Intel the same, if we go to 2026. Typically, the market likes three, and there we have three reasonable competitors.” Kimball says the hyperscalers, meanwhile, could challenge Nvidia as they transition more AI services to in-house hardware.
And then there’s the wild cards. Cerebras, SambaNova, and Groq are the leaders in a very long list of startups looking to nibble away at Nvidia with novel solutions. They’re joined by dozens of others, including d-Matrix, Untether, Tenstorrent, and Etched, all pinning their hopes on new chip architectures optimized for generative AI. It’s likely many of these startups will falter, but perhaps the next Nvidia will emerge from the survivors.
-
In 1926, TV Was Mechanical
by Allison Marsh on 16. September 2024. at 13:00
Scottish inventor John Logie Baird had a lot of ingenious ideas, not all of which caught on. His phonovision was an early attempt at video recording, with the signals preserved on phonograph records. His noctovision used infrared light to see objects in the dark, which some experts claim was a precursor to radar.
But Baird earned his spot in history with the televisor. On 26 January 1926, select members of the Royal Institution gathered at Baird’s lab in London’s Soho neighborhood to witness the broadcast of a small but clearly defined image of a ventriloquist dummy’s face, sent from the televisor’s electromechanical transmitter to its receiver. He also demonstrated the televisor with a human subject, who observers could see speaking and moving on the screen. For this, Baird is often credited with the first public demonstration of television.
John Logie Baird [shown here] used the heads of ventriloquist dummies in early experiments because they didn’t mind the heat and bright lights of his televisor. Science History Images/Alamy
How the Nipkow Disk Led to Baird’s Televisor
To be clear, Baird didn’t invent television. Television is one of those inventions that benefited from many contributors, collaborators, and competitors. Baird’s starting point was an idea for an “electric telescope,” patented in 1885 by German engineer Paul Nipkow.
Nipkow’s apparatus captured a picture by dividing it into a vertical sequence of lines, using a spinning disk with perforated holes around the edge. The perforations were offset in a spiral so that each hole captured one slice of the image in turn—known today as scan lines. Each line would be encoded as an electrical signal. A receiving apparatus converted the signals into light, to reconstruct the image. Nipkow never commercialized his electric telescope, though, and after 15 years the patent expired.
The inset on the left shows how the televisor split an image (in this case, a person’s face) into vertical lines. Bettmann/Getty Images
The system that Baird demonstrated in 1926 used two Nipkow disks, one in the transmitting apparatus and the other in the receiving apparatus. Each disk had 30 holes. He fitted the disk with glass lenses that focused the reflected light onto a photoelectric cell. As the transmitting disk rotated, the photoelectric cell detected the change in brightness coming through the individual lenses and converted the light into an electrical signal.
This signal was then sent to the receiving system. (Part of the receiving apparatus, housed at the Science Museum in London, is shown at top.) There the process was reversed, with the electrical signal first being amplified and then modulating a neon gas–discharge lamp. The light passed through a rectangular slot to focus it onto the receiving Nipkow disk, which was turning at the same speed as the transmitter. The image could be seen on a ground glass plate.
Early experiments used a dummy because the many incandescent lights needed to provide sufficient illumination made it too hot and bright for a person. Each hole in the disk captured only a small bit of the overall image, but as long as the disk spun fast enough, the brain could piece together the complete image, a phenomenon known as persistence of vision. (In a 2022 Hands On column, Markus Mierse explains how to build a modern Nipkow-disk electromechanical TV using a 3D printer, an LED module, and an Arduino Mega microcontroller.)
John Logie Baird and “True Television”
Regular readers of this column know the challenge of documenting historical “firsts”—the first radio, the first telegraph, the first high-tech prosthetic arm. Baird’s claim to the first public broadcast of television is no different. To complicate matters, the actual first demonstration of his televisor wasn’t on 26 January 1926 in front of those esteemed members of the Royal Institution; rather, it occurred in March 1925 in front of curious shoppers at a Selfridges department store.
As Donald F. McLean recounts in his excellent June 2022 article “Before ‘True Television’: Investigating John Logie Baird’s 1925 Original Television Apparatus,” Baird used a similar device for the Selfridges demo, but it had only 16 holes, organized as two groups of eight, hence its nickname the Double-8. The resolution was about as far from high definition as you could get, showing shadowy silhouettes in motion. Baird didn’t consider this “true television,” as McLean notes in his Proceedings of the IEEE piece.
In 1926, Baird loaned part of the televisor he used in his Selfridges demo to the Science Museum in London.PA Images/Getty Images
Writing in December 1926 in Experimental Wireless & The Wireless Engineer, Baird defined true television as “the transmission of the image of an object with all gradations of light, shade, and detail, so that it is seen on the receiving screen as it appears to the eye of an actual observer.” Consider the Selfridges demo a beta test and the one for the Royal Institution the official unveiling. (In 2017, the IEEE chose to mark the latter and not the former with a Milestone.)
The 1926 demonstration was a turning point in Baird’s career. In 1927 he established the Baird Television Development Co., and a year later he made the first transatlantic television transmission, from London to Hartsdale, N.Y. In 1929, the BBC decided to give Baird’s system a try, performing some experimental broadcasts outside of normal hours. After that, mechanical television took off in Great Britain and a few other European countries.
The BBC used various versions of Baird’s mechanical system from 1929 to 1937, starting with the 30-line system and upgrading to a 240-line system. But eventually the BBC switched to the all-electronic system developed by Marconi-EMI. Baird then switched to working on one of the earliest electronic color television systems, called the Telechrome. (Baird had already demonstrated a successful mechanical color television system in 1928, but it never caught on.) Meanwhile, in the United States, Columbia Broadcasting System (CBS) attempted to develop a mechanical color television system based on Baird’s original idea of a color wheel but finally ceded to an electronic standard in 1953.
Baird also experimented with stereoscopic or three-dimensional television and a 1,000-line display, similar to today’s high-definition television. Unfortunately, he died in 1946 before he could persuade anyone to take up that technology.
In a 1969 interview in TV Times, John’s widow, Margaret Baird, reflected on some of the developments in television that would have made her husband happy. He would enjoy the massive amounts of sports coverage available, she said. (Baird had done the first live broadcast of the Epsom Derby in 1931.) He would be thrilled with current affairs programs. And, my personal favorite, she thought he would love the annual broadcasting of the Eurovision song contest.
Other TV Inventors: Philo Farnsworth, Vladimir Zworykin
But as I said, television is an invention that’s had many contributors. Across the Atlantic, Philo Farnsworth was experimenting with an all-electrical system that he had first envisioned as a high school student in 1922. By 1926, Farnsworth had secured enough financial backing to work full time on his idea.
One of his main inventions was the image dissector, also known as a dissector tube. This video camera tube creates a temporary electron image that can be converted into an electrical signal. On 7 September 1927, Farnsworth and his team successfully transmitted a single black line, followed by other images of simple shapes. But the system could only handle silhouettes, not three-dimensional objects.
Meanwhile, Vladimir Zworykin was also experimenting with electronic television. In 1923, he applied for a patent for a video tube called the iconoscope. But it wasn’t until 1931, after he joined RCA, that his team developed a working version, which suspiciously came after Zworykin visited Farnsworth’s lab in California. The iconoscope overcame some of the dissector tube’s deficiencies, especially the storage capacity. It was also more sensitive and easier to manufacture. But one major drawback of both the image dissector and the iconoscope was that, like Baird’s original televisor, they required very bright lights.
Everyone was working to develop a better tube, but Farnsworth claimed that he’d invented both the concept of an electronic image moving through a vacuum tube as well as the idea of a storage-type camera tube. The iconoscope and any future improvements all depended on these progenitor patents. RCA knew this and offered to buy Farnsworth’s patents, but Farnsworth refused to sell. A multiyear patent-interference case ensued, finally finding for Farnsworth in 1935.
While the case was being litigated, Farnsworth made the first public demonstration of an all-electric television system on 25 August 1934 at the Franklin Institute in Philadelphia. And in 1939, RCA finally agreed to pay royalties to Farnsworth to use his patented technologies. But Farnsworth was never able to compete commercially with RCA and its all-electric television system, which went on to dominate the U.S. television market.
Eventually, Harold Law, Paul Weimer, and Russell Law developed a better tube at their Princeton labs, the image orthicon. Designed for TV-guided missiles for the U.S. military, it was 100 to 1,000 times as sensitive as the iconoscope. After World War II, RCA quickly adopted the tube for its TV cameras. The image orthicon became the industry standard by 1947, remaining so until 1968 and the move to color TV.
The Path to Television Was Not Obvious
My Greek teacher hated the word “television.” He considered it an abomination that combined the Greek prefix telos (far off) with a Latin base, videre (to see). But early television was a bit of an abomination—no one really knew what it was going to be. As Chris Horrocks lays out in his delightfully titled book, The Joy of Sets (2017), television was developed in relation to the media that came before—telegraph, telephone, radio, and film.
Was television going to be like a telegraph, with communication between two points and an image slowly reassembled? Was it going to be like a telephone, with direct and immediate dialog between both ends? Was it going to be like film, with prerecorded images played back to a wide audience? Or would it be more like radio, which at the time was largely live broadcasts? At the beginning, people didn’t even know they wanted a television; manufacturers had to convince them.
And technically, there were many competing visions—Baird’s, Farnsworth’s, Zworykin’s, and others. It’s no wonder that television took many years, with lots of false starts and dead ends, before it finally took hold.
Part of a continuing series looking at historical artifacts that embrace the boundless potential of technology.
An abridged version of this article appears in the September 2024 print issue as “The Mechanical TV.”
-
Amazon's Secret Weapon in Chip Design Is Amazon
by Samuel K. Moore on 15. September 2024. at 13:00
Big-name makers of processors, especially those geared toward cloud-based AI, such as AMD and Nvidia, have been showing signs of wanting to own more of the business of computing, purchasing makers of software, interconnects, and servers. The hope is that control of the “full stack” will give them an edge in designing what their customers want.
Amazon Web Services (AWS) got there ahead of most of the competition, when they purchased chip designer Annapurna Labs in 2015 and proceeded to design CPUs, AI accelerators, servers, and data centers as a vertically-integrated operation. Ali Saidi, the technical lead for the Graviton series of CPUs, and Rami Sinno, director of engineering at Annapurna Labs, explained the advantage of vertically-integrated design and Amazon-scale and showed IEEE Spectrum around the company’s hardware testing labs in Austin, Tex., on 27 August.
- What kind of engineer a vertically-integrated cloud company needs
- How the Graviton series of CPUs evolved
- How chip design is different at AWS
- How the lab location speeds design
What brought you to Amazon Web Services, Rami?
Rami SinnoAWS
Rami Sinno: Amazon is my first vertically integrated company. And that was on purpose. I was working at Arm, and I was looking for the next adventure, looking at where the industry is heading and what I want my legacy to be. I looked at two things:
One is vertically integrated companies, because this is where most of the innovation is—the interesting stuff is happening when you control the full hardware and software stack and deliver directly to customers.
And the second thing is, I realized that machine learning, AI in general, is going to be very, very big. I didn’t know exactly which direction it was going to take, but I knew that there is something that is going to be generational, and I wanted to be part of that. I already had that experience prior when I was part of the group that was building the chips that go into the Blackberries; that was a fundamental shift in the industry. That feeling was incredible, to be part of something so big, so fundamental. And I thought, “Okay, I have another chance to be part of something fundamental.”
Does working at a vertically-integrated company require a different kind of chip design engineer?
Sinno: Absolutely. When I hire people, the interview process is going after people that have that mindset. Let me give you a specific example: Say I need a signal integrity engineer. (Signal integrity makes sure a signal going from point A to point B, wherever it is in the system, makes it there correctly.) Typically, you hire signal integrity engineers that have a lot of experience in analysis for signal integrity, that understand layout impacts, can do measurements in the lab. Well, this is not sufficient for our group, because we want our signal integrity engineers also to be coders. We want them to be able to take a workload or a test that will run at the system level and be able to modify it or build a new one from scratch in order to look at the signal integrity impact at the system level under workload. This is where being trained to be flexible, to think outside of the little box has paid off huge dividends in the way that we do development and the way we serve our customers.
“By the time that we get the silicon back, the software’s done” —Ali Saidi, Annapurna Labs
At the end of the day, our responsibility is to deliver complete servers in the data center directly for our customers. And if you think from that perspective, you’ll be able to optimize and innovate across the full stack. A design engineer or a test engineer should be able to look at the full picture because that’s his or her job, deliver the complete server to the data center and look where best to do optimization. It might not be at the transistor level or at the substrate level or at the board level. It could be something completely different. It could be purely software. And having that knowledge, having that visibility, will allow the engineers to be significantly more productive and delivery to the customer significantly faster. We’re not going to bang our head against the wall to optimize the transistor where three lines of code downstream will solve these problems, right?
Do you feel like people are trained in that way these days?
Sinno: We’ve had very good luck with recent college grads. Recent college grads, especially the past couple of years, have been absolutely phenomenal. I’m very, very pleased with the way that the education system is graduating the engineers and the computer scientists that are interested in the type of jobs that we have for them.
The other place that we have been super successful in finding the right people is at startups. They know what it takes, because at a startup, by definition, you have to do so many different things. People who’ve done startups before completely understand the culture and the mindset that we have at Amazon.
What brought you to AWS, Ali?
Ali SaidiAWS
Ali Saidi: I’ve been here about seven and a half years. When I joined AWS, I joined a secret project at the time. I was told: “We’re going to build some Arm servers. Tell no one.”
We started with Graviton 1. Graviton 1 was really the vehicle for us to prove that we could offer the same experience in AWS with a different architecture.
The cloud gave us an ability for a customer to try it in a very low-cost, low barrier of entry way and say, “Does it work for my workload?” So Graviton 1 was really just the vehicle demonstrate that we could do this, and to start signaling to the world that we want software around ARM servers to grow and that they’re going to be more relevant.
Graviton 2—announced in 2019—was kind of our first… what we think is a market-leading device that’s targeting general-purpose workloads, web servers, and those types of things.
It’s done very well. We have people running databases, web servers, key-value stores, lots of applications... When customers adopt Graviton, they bring one workload, and they see the benefits of bringing that one workload. And then the next question they ask is, “Well, I want to bring some more workloads. What should I bring?” There were some where it wasn’t powerful enough effectively, particularly around things like media encoding, taking videos and encoding them or re-encoding them or encoding them to multiple streams. It’s a very math-heavy operation and required more [single-instruction multiple data] bandwidth. We need cores that could do more math.
We also wanted to enable the [high-performance computing] market. So we have an instance type called HPC 7G where we’ve got customers like Formula One. They do computational fluid dynamics of how this car is going to disturb the air and how that affects following cars. It’s really just expanding the portfolio of applications. We did the same thing when we went to Graviton 4, which has 96 cores versus Graviton 3’s 64.
How do you know what to improve from one generation to the next?
Saidi: Far and wide, most customers find great success when they adopt Graviton. Occasionally, they see performance that isn’t the same level as their other migrations. They might say “I moved these three apps, and I got 20 percent higher performance; that’s great. But I moved this app over here, and I didn’t get any performance improvement. Why?” It’s really great to see the 20 percent. But for me, in the kind of weird way I am, the 0 percent is actually more interesting, because it gives us something to go and explore with them.
Most of our customers are very open to those kinds of engagements. So we can understand what their application is and build some kind of proxy for it. Or if it’s an internal workload, then we could just use the original software. And then we can use that to kind of close the loop and work on what the next generation of Graviton will have and how we’re going to enable better performance there.
What’s different about designing chips at AWS?
Saidi: In chip design, there are many different competing optimization points. You have all of these conflicting requirements, you have cost, you have scheduling, you’ve got power consumption, you’ve got size, what DRAM technologies are available and when you’re going to intersect them… It ends up being this fun, multifaceted optimization problem to figure out what’s the best thing that you can build in a timeframe. And you need to get it right.
One thing that we’ve done very well is taken our initial silicon to production.
How?
Saidi: This might sound weird, but I’ve seen other places where the software and the hardware people effectively don’t talk. The hardware and software people in Annapurna and AWS work together from day one. The software people are writing the software that will ultimately be the production software and firmware while the hardware is being developed in cooperation with the hardware engineers. By working together, we’re closing that iteration loop. When you are carrying the piece of hardware over to the software engineer’s desk your iteration loop is years and years. Here, we are iterating constantly. We’re running virtual machines in our emulators before we have the silicon ready. We are taking an emulation of [a complete system] and running most of the software we’re going to run.
So by the time that we get to the silicon back [from the foundry], the software’s done. And we’ve seen most of the software work at this point. So we have very high confidence that it’s going to work.
The other piece of it, I think, is just being absolutely laser-focused on what we are going to deliver. You get a lot of ideas, but your design resources are approximately fixed. No matter how many ideas I put in the bucket, I’m not going to be able to hire that many more people, and my budget’s probably fixed. So every idea I throw in the bucket is going to use some resources. And if that feature isn’t really important to the success of the project, I’m risking the rest of the project. And I think that’s a mistake that people frequently make.
Are those decisions easier in a vertically integrated situation?
Saidi: Certainly. We know we’re going to build a motherboard and a server and put it in a rack, and we know what that looks like… So we know the features we need. We’re not trying to build a superset product that could allow us to go into multiple markets. We’re laser-focused into one.
What else is unique about the AWS chip design environment?
Saidi: One thing that’s very interesting for AWS is that we’re the cloud and we’re also developing these chips in the cloud. We were the first company to really push on running [electronic design automation (EDA)] in the cloud. We changed the model from “I’ve got 80 servers and this is what I use for EDA” to “Today, I have 80 servers. If I want, tomorrow I can have 300. The next day, I can have 1,000.”
We can compress some of the time by varying the resources that we use. At the beginning of the project, we don’t need as many resources. We can turn a lot of stuff off and not pay for it effectively. As we get to the end of the project, now we need many more resources. And instead of saying, “Well, I can’t iterate this fast, because I’ve got this one machine, and it’s busy.” I can change that and instead say, “Well, I don’t want one machine; I’ll have 10 machines today.”
Instead of my iteration cycle being two days for a big design like this, instead of being even one day, with these 10 machines I can bring it down to three or four hours. That’s huge.
How important is Amazon.com as a customer?
Saidi: They have a wealth of workloads, and we obviously are the same company, so we have access to some of those workloads in ways that with third parties, we don’t. But we also have very close relationships with other external customers.
So last Prime Day, we said that 2,600 Amazon.com services were running on Graviton processors. This Prime Day, that number more than doubled to 5,800 services running on Graviton. And the retail side of Amazon used over 250,000 Graviton CPUs in support of the retail website and the services around that for Prime Day.
The AI accelerator team is colocated with the labs that test everything from chips through racks of servers. Why?
Sinno: So Annapurna Labs has multiple labs in multiple locations as well. This location here is in Austin… is one of the smaller labs. But what’s so interesting about the lab here in Austin is that you have all of the hardware and many software development engineers for machine learning servers and for Trainium and Inferentia [AWS’s AI chips] effectively co-located on this floor. For hardware developers, engineers, having the labs co-located on the same floor has been very, very effective. It speeds execution and iteration for delivery to the customers. This lab is set up to be self-sufficient with anything that we need to do, at the chip level, at the server level, at the board level. Because again, as I convey to our teams, our job is not the chip; our job is not the board; our job is the full server to the customer.
How does vertical integration help you design and test chips for data-center-scale deployment?
Sinno: It’s relatively easy to create a bar-raising server. Something that’s very high-performance, very low-power. If we create 10 of them, 100 of them, maybe 1,000 of them, it’s easy. You can cherry pick this, you can fix this, you can fix that. But the scale that the AWS is at is significantly higher. We need to train models that require 100,000 of these chips. 100,000! And for training, it’s not run in five minutes. It’s run in hours or days or weeks even. Those 100,000 chips have to be up for the duration. Everything that we do here is to get to that point.
We start from a “what are all the things that can go wrong?” mindset. And we implement all the things that we know. But when you were talking about cloud scale, there are always things that you have not thought of that come up. These are the 0.001-percent type issues.
In this case, we do the debug first in the fleet. And in certain cases, we have to do debugs in the lab to find the root cause. And if we can fix it immediately, we fix it immediately. Being vertically integrated, in many cases we can do a software fix for it. We use our agility to rush a fix while at the same time making sure that the next generation has it already figured out from the get go.
-
Conference To Spotlight Harm Caused by Online Platforms
by Joanna Goodrich on 14. September 2024. at 18:00
This year’s IEEE Conference on Digital Platforms and Societal Harms is scheduled to be held on 14 and 15 October in a hybrid format, with both in-person and virtual keynote panel sessions. The in-person events are to take place at American University, in Washington, D.C.
The annual conference focuses on how social media and similar platforms amplify hate speech, extremism, exploitation, misinformation, and disinformation, as well as what measures are being taken to protect people.
With the popularity of social media and the rise of artificial intelligence, content can be more easily created and shared online by individuals and bots, says Andre Oboler, the general chair of IEEE DPSH. The IEEE senior member is CEO of the Online Hate Prevention Institute, which is based in Sydney. Oboler cautions that a lot of content online is fabricated, so some people are making economic, political, social, and health care decisions based on inaccurate information.
“Addressing the creation, propagation, and engagement of harmful digital information is a complex problem. It requires broad collaboration among various stakeholders including technologists; lawmakers and policymakers; nonprofit organizations; private sectors; and end users.”
Misinformation (which is false) and disinformation (which is intentionally false) also can propagate hate speech, discrimination, violent extremism, and child sexual abuse, he says, and can create hostile online environments, damaging people’s confidence in information and endangering their lives.
To help prevent harm, he says, cutting-edge technical solutions and changes in public policy are needed. At the conference, academic researchers and leaders from industry, government, and not-for-profit organizations are gathering to discuss steps being taken to protect individuals online.
Experts to explore challenges and solutions
The event includes panel discussions and Q&A sessions with experts from a variety of technology fields and organizations. Scheduled speakers include Paul Giannasi from the U.K. National Police Chiefs’ Council; Skip Gilmour of the Global Internet Forum to Counter Terrorism; and Maike Luiken, chair of IEEE’s Planet Positive 2030 initiative.
“Addressing the creation, propagation, and engagement of harmful digital information is a complex problem,” Oboler says. “It requires broad collaboration among various stakeholders including technologists; lawmakers and policymakers; nonprofit organizations; private sectors; and end users.
“There is an emerging need for these stakeholders and researchers from multiple disciplines to have a joint forum to understand the challenges, exchange ideas, and explore possible solutions.”
To register for in-person and online conference attendance, visit the event’s website. Those who want to attend only the keynote panels can register for free access to the discussions. Attendees who register by 22 September and use the code 25off2we receive a 25 percent discount.
Check out highlights from the 2023 IEEE Conference on Digital Platforms and Societal Harms.
-
Ultrasonic Chips Could Cut Interference in the IoT
by Tim Hornyak on 14. September 2024. at 13:00
The proliferation of IoT technology has made chatterboxes out of everyday hardware and new gadgets too, but it comes with a downside: the more devices sharing the airwaves the more trouble they have communicating. The nearly 30 billion connected devices expected by 2030 will be operating using different wireless standards while sharing the same frequency bands, potentially interfering with one another. To overcome this, researchers in Japan say they have developed a way to shrink the devices that filter out interfering signals. Instead of many individual filters, the technology would combine them onto single chips.
For smartphones to work with different communications standards and in different countries, they need dozens of filters to keep out unwanted signals. But these filters can be expensive and collectively take up a relatively large amount of real estate in the phone. With increasingly crowded electromagnetic spectrum , engineers will have to cram even more filters into phones and other gadgets, meaning further miniaturization will be necessary. Researchers at Japanese telecom NTT and Okayama University say they’ve developed technology that could shrink all those filters down to a single device they describe as an ultrasonic circuit that can steer signals without unintentionally scattering them.
The ultrasonic circuit incorporates filters that are similar to surface acoustic wave (SAW) filters used in smartphones. SAW filters convert an electronic RF signal into a mechanical wave on the surface of a substrate and back again, filtering out particular frequencies in the process. Because the mechanical wave is thousands of times shorter than the RF wave that creates it, SAW filters can be compact.
Today’s filters screen out unwanted RF signals by converting them to ultrasonic signals and back again. New research could lead to a way to integrate many such filters onto a single chip.NTT Corporation
“In the future IoT society, communication bandwidth and methods will increase, so we will need hundreds of ultrasonic filters in smartphones, but we cannot allocate a large area to them,” because the battery, display, processor and other components need room too, says Daiki Hatanaka a senior research scientist in the Nanomechanics Research Group at NTT. “Our technology allows us to confine ultrasound in a very narrow channel on a micrometer scale then guide the signal as we want. Based on this ultrasonic circuit, we can integrate many filters on just one chip.”
Valley Pseudospin-dependent Transport
Guiding ultrasonic waves along a path that changes direction can cause backscattering, degrading the signal quality. To counter this, Hatanaka and colleagues tapped Okayama University’s research into acoustic topological structures. Topology is mathematics concerned with how different shapes can be thought of as equivalent if they satisfy certain conditions—the classic example is a donut and a coffee mug being equivalent because they each have just one hole. But as highlighted by the 2016 Nobel Prize in Physics, it’s also used to explore exotic states of matter including superconductivity.
In their experiments, the researchers in Japan fashioned a waveguide made up of arrays of periodic holes with three-fold rotational symmetry. Where two arrays with holes that were rotated 10 degrees apart from each other met, a topological property called valley pseudospin arises. At this edge, tiny ultrasonic vortexes “pseudospin” in opposite directions, generating a unique ultrasonic wave known as valley pseudospin-dependent transport. This propagates a 0.5 GHz signal in only one direction even if there is a sharp bend in the waveguide, according to NTT. So the signal can’t suffer backscattering.
“The direction of the polarization of the valley states of ultrasound automatically forces it to propagate in only one direction, and backscattering is prohibited,” says Hatanaka. “
NTT says the gigahertz topological circuit is the first of its kind. The research team is now trying to fabricate a waveguide that connects 5 to 10 filters on a single chip. The initial chip will be about 1 square centimeter, but the researchers hope to shrink it to a few hundred square micrometers. In the second stage of research, they will try to dynamically control the ultrasound, amplify the signal, convert its frequency, and integrate these functions into one system.
The company will consider plans for commercialization as the research proceeds over the next two years. If the research becomes a commercial product the impact on future smartphones and IoT systems could be important, says Hatanaka. He estimates that future high-end smartphones could be equipped with up to around 20 ultrasonic circuits.
“We could use the space saved for a better user experience, so by using ultrasonic filters or other analog signal components we can improve the display or battery or other important components for the user experience,” he says.
-
From Punch Cards to Python
by Joanna Goodrich on 13. September 2024. at 18:00
In today’s digital world, it’s easy for just about anyone to create a mobile app or write software, thanks to Java, JavaScript, Python, and other programming languages.
But that wasn’t always the case. Because the primary language of computers is binary code, early programmers used punch cards to instruct computers what tasks to complete. Each hole represented a single binary digit.
That changed in 1952 with the A-0 compiler, a series of specifications that automatically translates high-level languages such as English into machine-readable binary code.
The compiler, now an IEEE Milestone, was developed by Grace Hopper, who worked as a senior mathematician at the Eckert-Mauchly Computer Corp., now part of Unisys, in Philadelphia.
IEEE Fellow’s innovation allowed programmers to write code faster and easier using English commands. For her, however, the most important outcome was the influence it had on the development of modern programming languages, making writing code more accessible to everyone, according to a Penn Engineering Today article.
The dedication of the A-0 compiler as an IEEE Milestone was held in Philadelphia on 7 May at the University of Pennsylvania. That’s where the Eckert-Mauchly Computer Corp. got its start.
“This milestone celebrates the first step of applying computers to automate the tedious portions of their own programming,” André DeHon, professor of electrical systems, engineering, and computer science, said at the dedication ceremony.
Eliminating the punch-card system
To program a computer, early technicians wrote out tasks in assembly language—a human-readable way to write machine code, which is made up of binary numbers. They then manually translated the assembly language into machine code and punched holes representing the binary digits into cards, according to a Medium article on the method. The cards were fed into a machine that read the holes and input the data into the computer.
The punch-card system was laborious; it could take days to complete a task. The cards couldn’t be used with even a slight defect such as a bent corner. The method also had a high risk of human error.
After leading the development of the Electronic Numerical Integrator and Computer (ENIAC) at Penn, computer scientists J. Presper Eckert and John W. Mauchly set about creating a replacement for punch cards. ENIAC was built to improve the accuracy of U.S. artillery during World War II, but the two men wanted to develop computers for commercial applications, according to a Pennsylvania Center for the Book article.
The machine they designed was the first known large-scale electronic computer, the Universal Automatic, or UNIVAC I. Hopper was on its development team.
UNIVAC I used 6,103 vacuum tubes and took up a 33-square-meter room. The machine had a memory unit. Instead of punch cards, the computer used magnetic tape to input data. The tapes, which could hold audio, video, and written data, were up to 457 meters long. Unlike previous computers, the UNIVAC I had a keyboard so an operator could input commands, according to the Pennsylvania Center for the Book article.
“This milestone celebrates the first step of applying computers to automate the tedious portions of their own programming.” —André DeHon
Technicians still had to manually feed instructions into the computer, however, to run any new program.
That time-consuming process led to errors because “programmers are lousy copyists,” Hopper said in a speech for the Association for Computing Machinery. “It was amazing how many times a 4 would turn into a delta, which was our space symbol, or into an A. Even B’s turned into 13s.”
According to a Hidden Heroes article, Hopper had an idea for simplifying programming: Have the computer translate English to machine code.
She was inspired by computer scientist Betty Holberton’s sort/merge generator and Mauchly’s Short Code. Holberton is one of six women who programmed the ENIAC to calculate artillery trajectories in seconds, and she worked alongside Hopper on the UNIVAC I. Her sort/merge program, invented in 1951 for the UNIVAC I, handled the large data files stored on magnetic tapes. Hopper defined the sort/merge program as the first version of virtual memory because it made use of overlays automatically without being directed to by the programmer, according to a Stanford presentation about programming languages. The Short Code, which was developed in the 1940s, allowed technicians to write programs using brief sequences of English words corresponding directly to machine code instructions. It bridged the gap between human-readable code and machine-executable instructions.
“I think the first step to tell us that we could actually use a computer to write programs was the sort/merge generator,” Hopper said in the presentation. “And Short Code was the first step in moving toward something which gave a programmer the actual power to write a program in a language which bore no resemblance whatsoever to the original machine code.”
IEEE Fellow Grace Hopper inputting call numbers into the Universal Automatic (UNIVAC I), which allows the computer to find the correct instructions to complete. The A-0 compiler translates the English instructions into machine-readable binary code.Computer History Museum
Easier, faster, and more accurate programming
Hopper, who figured computers should speak human-like languages, rather than requiring humans to speak computer languages, began thinking about how to allow programmers to call up specific codes using English, according to an IT Professional profile.
But she needed a library of frequently used instructions for the computer to reference and a system to translate English to machine code. That way, the computer could understand what task to complete.
Such a library didn’t exist, so Hopper built her own. It included tapes that held frequently used instructions for tasks that she called subroutines. Each tape stored one subroutine, which was assigned a three-number call sign so that the UNIVAC I could locate the correct tape. The numbers represented sets of three memory addresses: one for the memory location of the subroutine, another for the memory location of the data, and the third for the output location, according to the Stanford presentation.
“All I had to do was to write down a set of call numbers, let the computer find them on the tape, and do the additions,” she said in a Centre for Computing History article. “This was the first compiler.”
The system was dubbed the A-0 compiler because code was written in one language, which was then “compiled” into a machine language.
What previously had taken a month of manual coding could now be done in five minutes, according to a Cockroach Labs article.
Hopper presented the A-0 to Eckert-Mauchly Computer executives. Instead of being excited, though, they said they didn’t believe a computer could write its own programs, according to the article.
“I had a running compiler, and nobody would touch it, because they carefully told me computers could only do arithmetic; they could not do programs,” Hopper said. “It was a selling job to get people to try it. I think with any new idea, because people are allergic to change, you have to get out and sell the idea.”
It took two years for the company’s leadership to accept the A-0.
In 1954, Hopper was promoted to director of automatic programming for the UNIVAC division. She went on to create the first compiler-based programming languages including Flow-Matic, the first English language data-processing compiler. It was used to program UNIVAC I and II machines.
Hopper also was involved in developing COBOL, one of the earliest standardized computer languages. It enabled computers to respond to words in addition to numbers, and it is still used in business, finance, and administrative systems. Hopper’s Flow-Matic formed the foundation of COBOL, whose first specifications were made available in 1959.
A plaque recognizing the A-0 is now displayed at the University of Pennsylvania. It reads:
During 1951–1952, Grace Hopper invented the A-0 Compiler, a series of specifications that functioned as a linker/loader. It was a pioneering achievement of automatic programming as well as a pioneering utility program for the management of subroutines. The A-0 Compiler influenced the development of arithmetic and business programming languages. This led to COBOL (Common Business-Oriented Language), becoming the dominant high-level language for business applications.
The IEEE Philadelphia Section sponsored the nomination.
Administered by the IEEE History Center and supported by donors, the Milestone program recognizes outstanding technical developments worldwide.
About Grace Hopper
Hopper didn’t start as a computer programmer. She was a mathematician at heart, earning bachelor’s degrees in mathematics and physics in 1928 from Vassar College, in Poughkeepsie, N.Y. She then received master’s and doctoral degrees in mathematics and mathematical physics from Yale in 1930 and 1934, respectively.
She taught math at Vassar, but after the bombing of Pearl Harbor and the U.S. entry into World War II, Hopper joined the war effort. She took a leave of absence from Vassar to join the U.S. Naval Reserve (Women’s Reserve) in December 1943. She was assigned to the Bureau of Ships Computation Project at Harvard, where she worked for mathematician Howard Aiken. She was part of Aiken’s team that developed the Mark I, one of the earliest electromechanical computers. Hopper was the third person and the first woman to program the machine.
After the war ended, she became a research fellow at the Harvard Computation Laboratory. In 1946 she joined the Eckert-Mauchly Computer Corp., where she worked until her retirement in 1971. During 1959 she was an adjunct lecturer at Penn’s Moore School of Electrical Engineering.
Her work in programming earned her the nickname “Amazing Grace,” according to an entry about her on the Engineering and Technology History Wiki.
Hopper remained a member of the Naval Reserve and, in 1967, was recalled to active duty. She led the effort to standardize programming languages for the military, according to the ETHW entry. She was eventually promoted to rear admiral. When she retired from the Navy at the age of 79 in 1989, she was the oldest serving officer in all the U.S. armed forces.
Among her many honors was the 1991 U.S. National Medal of Technology and Innovation “for her pioneering accomplishments in the development of computer programming languages that simplified computer technology and opened the door to a significantly larger universe of users.”
She received 40 honorary doctorates from universities, and the Navy named a warship in her honor.
-
Driving Middle East’s Innovation in Robotics and Future of Automation
by Khalifa University on 13. September 2024. at 16:29
This is a sponsored article brought to you by Khalifa University of Science and Technology.
Abu Dhabi-based Khalifa University of Science and Technology in the United Arab Emirates (UAE) will be hosting the 36th edition of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2024) to highlight the Middle East and North Africa (MENA) region’s rapidly advancing capabilities in the robotics and intelligent transport systems.
Themed “Robotics for Sustainable Development,” the IROS 2024 will be held from 14-18 October 2024 at the Abu Dhabi National Exhibition Center (ADNEC) in the UAE’s capital city. It will offer a platform for universities and research institutions to display their research and innovation activities and initiatives in robotics, gathering researchers, academics, leading corporate majors, and industry professionals from around the globe.
A total of 13 forums, nine global-level competitions and challenges covering various aspects of robotics and AI, an IROS Expo, as well as an exclusive Career Fair will also be part of IROS 2024. The challenges and competitions will focus on physical or athletic intelligence of robots, remote robot navigation, robot manipulation, underwater robotics, as well as perception and sensing.
Delegates for the event will represent sectors including manufacturing, healthcare, logistics, agriculture, defense, security, and mining sectors with 60 percent of the talent pool having over six years of experience in robotics. A major component of the conference will be the poster sessions, keynotes, panel discussions by researchers and scientists, and networking events.
Khalifa University will be hosting IROS 2024 to highlight the Middle East and North Africa (MENA) region’s rapidly advancing capabilities in the robotics and intelligent transport systems.Khalifa University
Abu Dhabi ranks first on the world’s safest cities list in 2024, according to online database Numbeo, out of 329 global cities in the 2024 standings, holding the title for eight consecutive years since 2017, reflecting the emirate’s ongoing efforts to ensure a good quality of life for citizens and residents.
With a multicultural community, Abu Dhabi is home to people from more than 200 nationalities and draws a large number of tourists to some of the top art galleries in the city such as Louvre Abu Dhabi and the Guggenheim Abu Dhabi, as well as other destinations such as Ferrari World Abu Dhabi and Warner Bros. World Abu Dhabi.
The UAE and Abu Dhabi have increasingly become a center for creative skillsets, human capital and advanced technologies, attracting several international and regional events such as the global COP28 UAE climate summit, in which more than 160 countries participated.
Abu Dhabi city itself has hosted a number of association conventions such as the 34th International Nursing Research Congress and is set to host the UNCTAD World Investment Forum, the 13th World Trade Organization (WTO) Ministerial Conference (MC13), the 12th World Environment Education Congress in 2024, and the IUCN World Conservation Congress in 2025.
Khalifa University’s Center for Robotics and Autonomous Systems (KU-CARS) includes a vibrant multidisciplinary environment for conducting robotics and autonomous vehicle-related research and innovation.Khalifa University
Dr. Jorge Dias, IROS 2024 General Chair, said: “Khalifa University is delighted to bring the Intelligent Robots and Systems 2024 to Abu Dhabi in the UAE and highlight the innovations in line with the theme Robotics for Sustainable Development. As the region’s rapidly advancing capabilities in robotics and intelligent transport systems gain momentum, this event serves as a platform to incubate ideas, exchange knowledge, foster collaboration, and showcase our research and innovation activities. By hosting IROS 2024, Khalifa University aims to reaffirm the UAE’s status as a global innovation hub and destination for all industry stakeholders to collaborate on cutting-edge research and explore opportunities for growth within the UAE’s innovation ecosystem.”
“This event serves as a platform to incubate ideas, exchange knowledge, foster collaboration, and showcase our research and innovation activities” —Dr. Jorge Dias, IROS 2024 General Chair
Dr. Dias added: “The organizing committee of IROS 2024 has received over 4000 submissions representing 60 countries, with China leading with 1,029 papers, followed by the U.S. (777), Germany (302), and Japan (253), as well as the U.K. and South Korea (173 each). The UAE with a total of 68 papers comes atop the Arab region.”
Driving innovation at Khalifa University is the Center for Robotics and Autonomous Systems (KU-CARS) with around 50 researchers and state-of-the-art laboratory facilities, including a vibrant multidisciplinary environment for conducting robotics and autonomous vehicle-related research and innovation.
IROS 2024 is sponsored by IEEE Robotics and Automation Society, Abu Dhabi Convention and Exhibition Bureau, the Robotics Society of Japan (RSJ), the Society of Instrument and Control Engineers (SICE), the New Technology Foundation, and the IEEE Industrial Electronics Society (IES).
More information at https://iros2024-abudhabi.org/
-
Video Friday: Jumping Robot Leg, Walking Robot Table
by Evan Ackerman on 13. September 2024. at 15:35
Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.
ICRA@40: 23–26 September 2024, ROTTERDAM, NETHERLANDS
IROS 2024: 14–18 October 2024, ABU DHABI, UAE
ICSR 2024: 23–26 October 2024, ODENSE, DENMARK
Cybathlon 2024: 25–27 October 2024, ZURICH
Enjoy today’s videos!
Researchers at the Max Planck Institute for Intelligent Systems and ETH Zurich have developed a robotic leg with artificial muscles. Inspired by living creatures, it jumps across different terrains in an agile and energy-efficient manner.
Thanks, Toshi!
ETH Zurich researchers have now developed a fast robotic printing process for earth-based materials that does not require cement. In what is known as “impact printing,” a robot shoots material from above, gradually building a wall. On impact, the parts bond together, and very minimal additives are required.
[ ETH Zurich ]
How could you not be excited to see this happen for real?
[ arXiv paper ]
Can we all agree that sanding, grinding, deburring, and polishing tasks are really best done by robots, for the most part?
Thanks, David!
Using doors is a longstanding challenge in robotics and is of significant practical interest in giving robots greater access to human-centric spaces. The task is challenging due to the need for online adaptation to varying door properties and precise control in manipulating the door panel and navigating through the confined doorway. To address this, we propose a learning-based controller for a legged manipulator to open and traverse through doors.
[ arXiv paper ]
Isaac is the first robot assistant that’s built for the home. And we’re shipping it in fall of 2025.
Fall of 2025 is a long enough time from now that I’m not even going to speculate about it.
[ Weave Robotics ]
By patterning liquid metal paste onto a soft sheet of silicone or acrylic foam tape, we developed stretchable versions of conventional rigid circuits (like Arduinos). Our soft circuits can be stretched to over 300% strain (over 4x their length) and are integrated into active soft robots.
[ Science Robotics ] via [ Yale ]
NASA’s Curiosity rover is exploring a scientifically exciting area on Mars, but communicating with the mission team on Earth has recently been a challenge due to both the current season and the surrounding terrain. In this Mars Report, Curiosity engineer Reidar Larsen takes you inside the uplink room where the team talks to the rover.
[ NASA ]
I love this and want to burn it with fire.
[ Carpentopod ]
Very often, people ask us what Reachy 2 is capable of, which is why we’re showing you the manipulation possibilities (through teleoperation) of our technology. The robot shown in this video is the Beta version of Reachy 2, our new robot coming very soon!
[ Pollen Robotics ]
The Scalable Autonomous Robots (ScalAR) Lab is an interdisciplinary lab focused on fundamental research problems in robotics that lie at the intersection of robotics, nonlinear dynamical systems theory, and uncertainty.
[ ScalAR Lab ]
Astorino is a 6-axis educational robot created for practical and affordable teaching of robotics in schools and beyond. It has been created with 3D printing, so it allows for experimentation and the possible addition of parts. With its design and programming, it replicates the actions of #KawasakiRobotics industrial robots, giving students the necessary skills for future work.
[ Astorino ]
I guess fish-fillet-shaping robots need to exist because otherwise customers will freak out if all their fish fillets are not identical, or something?
[ Flexiv ]
Watch the second episode of the ExoMars Rosalind Franklin rover mission—Europe’s ambitious exploration journey to search for past and present signs of life on Mars. The rover will dig, collect, and investigate the chemical composition of material collected by a drill. Rosalind Franklin will be the first rover to reach a depth of up to two meters below the surface, acquiring samples that have been protected from surface radiation and extreme temperatures.
[ ESA ]
-
The Next Frontier for EV Batteries: Nanoscale Coatings
by Willie D. Jones on 12. September 2024. at 11:00
Over the past 25 years, the longest driving range of an electric vehicle on a single charge has gone from about 260 kilometers to slightly over 800 km. Increasingly, these advanced battery packs have also begun storing energy from the grid or renewable sources to power homes or businesses. No wonder, then, that the global automotive battery market has surpassed US $50 billion a year and there is increasing pressure to produce greater numbers of even better batteries.
Now, several companies are applying a well-established chemical technique called atomic layer deposition (ALD) to coat battery electrodes with metal oxides or nitrides, which they claim improves both the energy capacity and the lifespan of lithium-ion batteries. The companies include Thornton, Colo.–based Forge Nano, Picosun (a wholly-owned subsidiary of Santa Clara, Calif.–based Applied Materials), and Beneq, in Espoo, Finland; they are leveraging the technique, which was originally developed in the 1960s. After years of refining their respective processes, these companies now hope to gain a toehold in markets for EV and smartphone batteries dominated by such giants as CATL, Panasonic, and Samsung.
Of the three, Forge Nano appears to have the most developed technology. It recently announced that its subsidiary, Forge Battery, has begun sending samples of a prototype battery cell made with ALD-coated materials to customers for testing. The company says its proprietary ALD formulation, which it calls Atomic Armor, makes batteries’ electrodes better at storing energy and helps them last longer.
What Goes Into a Lithium-Ion Battery?
The batteries found in today’s electric vehicles and smartphones consist of three main components. The anode, or negative electrode, usually made of graphite, is where lithium ions are stored during the charging process. The cathode (positive electrode) is made of a lithium-metal oxide such as lithium cobalt oxide or lithium-iron phosphate. Then there’s the electrolyte, which is a lithium salt dissolved in an organic solvent that allows lithium ions to move between the anode and cathode. Also important is the separator, a semi-porous material that allows the movement of ions between the cathode and anode during charging and discharging but blocks the flow of electrons directly between the two, which would quickly short out the battery.
A cathode coating is deposited for R&D battery cells by Forge Nano.Forge Nano
Coating the materials that make up the anode, cathode, and separator at the molecular level, these companies say, boosts batteries’ the performance and durability without an appreciable increase in their weight or volume.
. The films are formed by a chemical reaction between two gaseous precursor substances, which are introduced to the substrate by turns. The first one reacts with the substrate surface at active sites, the points on the precursor molecules and on the surface of the substrate where the two materials chemically bond. Then, after all the non-reacted precursor gas is pumped away, the next precursor is introduced and bonds with the first precursor at their respective active sites. ALD technology is self-terminating, meaning that when all active sites are filled, the reaction stops. The film forms one atomic layer at a time, so its thickness can be set with precision as fine as a few tenths of a nanometer simply by cutting off exposure of the substrate to the precursors once the desired coating thickness is reached.
In a conventional lithium-ion battery, with a graphite anode, silicon (and sometimes other materials) is added to the graphite to improve the anode’s ability to store ions. The practice boosts energy density, but silicon is much more prone to side reactions with the electrolyte and to expansion and contraction during charging and discharging, which weakens the electrode. Eventually, the mechanical degradation diminishes the battery’s storage capacity. ALD technology, by coating anode molecules with a protective layer, enables a higher proportion of silicon in the anode while also inhibiting the expansion-contraction cycles and therefore, slowing the mechanical degradation. The result is a lighter, more energy-dense battery that is more durable than conventional lithium-ion batteries.
Picosun says its ALD technology has been used to create coated nickel oxide anodes with more than twice the energy storage capacity and three times the energy density of those relying on traditional graphite.How big is the benefit? Forge Nano says that although the third-party testing and validation are underway, it’s too soon to make definitive statements about the coating-enhanced batteries’ lifespans. But a company spokesperson told IEEE Spectrum the data it has received thus far indicates that specific energy is improved by 15 percent compared with comparable batteries currently on the market.
The company has made a big bet that the players all along the battery production chain—from fabricators of anodes and cathodes to Tier 1 battery suppliers, and even electric vehicle manufacturers—will view its take on ALD as a must-have step in battery manufacturing. Forge Battery is building a 25,700 square meter gigafactory in North Carolina that it says will turn out 1 gigawatt-hour of its Atomic Armor–enhanced lithium-ion cells and finished batteries when it becomes operational in 2026.
-
Transistor-like Qubits Hit Key Benchmark
by Dina Genkina on 11. September 2024. at 12:00
A team in Australia has recently demonstrated a key advance in metal-oxide-semiconductor-based (or MOS-based) quantum computers. They showed that their two-qubit gates—logical operations that involve more than one quantum bit, or qubit—perform without errors 99 percent of the time. This number is important, because it is the baseline necessary to perform error correction, which is believed to be necessary to build a large-scale quantum computer. What’s more, these MOS-based quantum computers are compatible with existing CMOS technology, which will make it more straightforward to manufacture a large number of qubits on a single chip than with other techniques.
“Getting over 99 percent is significant because that is considered by many to be the error correction threshold, in the sense that if your fidelity is lower than 99 percent, it doesn’t really matter what you’re going to do in error correction,” says Yuval Boger, CCO of quantum computing company QuEra and who wasn’t involved in the work. “You’re never going to fix errors faster than they accumulate.”
There are many contending platforms in the race to build a useful quantum computer. IBM, Google and others are building their machines out of superconducting qubits. Quantinuum and IonQ use individual trapped ions. QuEra and Atom Computing use neutrally-charged atoms. Xanadu and PsiQuantum are betting on photons. The list goes on.
In the new result, a collaboration between the University of New South Wales (UNSW) and Sydney-based startup Diraq, with contributors from Japan, Germany, Canada, and the U.S., has taken yet another approach: trapping single electrons in MOS devices. “What we are trying to do is we are trying to make qubits that are as close to traditional transistors as they can be,” says Tuomo Tanttu, a research fellow at UNSW who led the effort.
Qubits That Act Like Transistors
These qubits are indeed very similar to a regular transistor, gated in such a way as to have only a single electron in the channel. The biggest advantage of this approach is that it can be manufactured using traditional CMOS technologies, making it theoretically possible to scale to millions of qubits on a single chip. Another advantage is that MOS qubits can be integrated on-chip with standard transistors for simplified input, output, and control, says Diraq CEO Andrew Dzurak.
The drawback of this approach, however, is that MOS qubits have historically suffered from device-to-device variability, causing significant noise on the qubits.
“The sensitivity in [MOS] qubits is going to be more than in transistors, because in transistors, you still have 20, 30, 40 electrons carrying the current. In a qubit device, you’re really down to a single electron,” says Ravi Pillarisetty, a senior device engineer for Intel quantum hardware who wasn’t involved in the work.
The team’s result not only demonstrated the 99 percent accurate functionality on two-qubit gates of the test devices, but also helped better understand the sources of device-to-device variability. The team tested three devices with three qubits each. In addition to measuring the error rate, they also performed comprehensive studies to glean the underlying physical mechanisms that contribute to noise.
The researchers found that one of the sources of noise was isotopic impurities in the silicon layer, which, when controlled, greatly reduced the circuit complexity necessary to run the device. The next leading cause of noise was small variations in electric fields, likely due to imperfections in the oxide layer of the device. Tanttu says this is likely to improve by transitioning from a laboratory clean room to a foundry environment.
“It’s a great result and great progress. And I think it’s setting the right direction for the community in terms of thinking less about one individual device, or demonstrating something on an individual device, versus thinking more longer term about the scaling path,” Pillarisetty says.
Now, the challenge will be to scale up these devices to more qubits. One difficulty with scaling is the number of input/output channels required. The quantum team at Intel, who are pursuing a similar technology, has recently pioneered a chip they call Pando Tree to try to address this issue. Pando Tree will be on the same plane as the quantum processor, enabling faster inputs and outputs to the qubits. The Intel team hopes to use it to scale to thousands of qubits. “A lot of our approach is thinking about, how do we make our qubit processor look more like a modern CPU?” says Pillarisetty.
Similarly, Diraq CEO Dzurak says his team plan to scale their technology to thousands of qubits in the near future through a recently announced partnership with Global Foundries. “With Global Foundries, we designed a chip that will have thousands of these [MOS qubits]. And these will be interconnected by using classical transistor circuitry that we designed. This is unprecedented in the quantum computing world,” Dzurak says.
-
Where VR Gaming Took a Wrong Turn
by Marcus Carter on 10. September 2024. at 13:00
In 2017 Mark Zuckerberg stated a bold goal: He wanted one billion people to try virtual reality (VR) by 2027. While he still has a few years to pull it off, the target remains impossibly farfetched. The most recent estimates place total worldwide VR headset sales at only 34 million.
VR Gaming was expected to lead this uptake, but why hasn’t it? We believe that VR gaming has been held back by game developers who are committed to a fantasy. In this fantasy, VR games align with the values of “hardcore” gamer culture, with advanced graphics and wholly immersive play. Aspirational attempts to reach this flawed fantasy have squashed the true potential of VR for gaming.
VR Gaming’s Contemporary Emergence
The 1990s and 2000s saw several ill-fated attempts to launch VR gaming systems—including the Sega VR system, which the company promoted breathlessly but then never released, because it gave players motion sickness and headaches. But VR gaming’s contemporary emergence really began in August 2009, when then-17-year-old Palmer Luckey began posting on a VR enthusiasts forum about his plan to make a head-mounted VR gaming device. One early reader of Luckey’s posts was John Carmack, lead programmer for several of the most influential first-person shooter games, including Doom and Wolfenstein.
Palmer Luckey, shown here in 2013 at the age of 20, holds an early Oculus Rift virtual reality head-mounted display.Allen J. Schaben/Los Angeles Times/Getty Images
While working on the remaster of Doom 3—which included support for 3D displays—Carmack was experimenting with different VR headsets that were available at the time. The two connected through their forum posts, and Luckey sent one of his prototype VR headsets to Carmack. When Carmack took the prototype to the major gaming expo E3 in 2012, it catalyzed an avalanche of interest in the project.
Carmack’s involvement put Luckey’s newly formed company, Oculus VR, on a trajectory towards a particular kind of gaming: the hyper-violent games with high-fidelity graphics that hardcore gamers revere. Carmack, far beyond anyone else, pioneered the genre of hardcore games with his first-person shooter games.
Sega Visions Magazine promoted Sega VR in its August/September 1993 issue. Sega
Here’s how the gaming scholar Shira Chess sums up the genre: “Traditionally, ‘hardcore’ describes games that are difficult to learn, expensive, and unforgiving of mistakes and that must be played over longer periods of time. Conversely, casual games can be learned quickly, are forgiving of mistakes and cheap or free, and can be played for either longer or shorter periods of time, depending on one’s schedule.”
Oculus’s Kickstarter campaign in 2012 was proudly “designed for gamers, by gamers.” Soon after, Meta (then Facebook) acquired Oculus for US $3 billion in March 2014. The acquisition enraged many of those in the gaming community and those who had backed the original Kickstarter. Facebook was already an unpopular platform with the tech-enthusiast community, associated more closely with data collection and surveillance than gaming. If Facebook was associated with gaming, it was with casual social media games like Farmville and Bejeweled. But as it turns out, Meta went on to invest billions in VR, a level of investment highly unlikely if Oculus had remained independent.
The Three Wrong Assumptions of VR Gaming
VR’s origin in hardcore gaming culture resulted in VR game development being underpinned by three false assumptions about the types of experiences that would (or could) make VR gaming successful. These assumptions were that gamers wanted graphical realism and fast-paced violence, and that they didn’t want casual play experiences.
Over the past three decades, “AAA” game development—a term used in the games industry to signify high-budget games distributed by large publishers—has driven the massive expansion of computing power in consumer gaming devices. Particularly in PC gaming, part of what made a game hardcore was the computing power needed to run it at “maximum settings,” with the most detailed and textured graphics available.
The enormous advances in game graphics over the past 30 years contributed to significant improvements in player experience. This graphical realism became closely entwined with the concept of immersion.
For VR—which sold itself as “ truly immersive”—this meant that hardcore gamers expected graphically real VR experiences. But VR environments need to be rendered smoothly in order to not cause motion sickness, something made harder by a commitment to graphical realism. This aspiration saddling VR games with a nearly impossible compute burden.
One game that sidesteps this issue—and has subsequently become one of the most celebrated VR games—is Superhot VR, an action puzzle with basic graphics in which enemy avatars and their bullets only move when the player moves their body.
The video game Superhot VR remains one of the top-selling VR games years after its release due to its unique experience of time manipulation through body movements.Superhot VR
Play begins with the player surrounded by attacking enemies, with death immediately returning the player to the starting moment. Play thus involves discovering what sequence of movements and attacks can get the player out of this perilous situation. It’s a learning curve reminiscent of the 2014 science-fiction film Edge of Tomorrow, in which a hapless soldier (played by Tom Cruise) quicky becomes an elite, superhuman soldier while stuck in a time loop.
The attention in Superhot’s gameplay is not to visual fidelity or sensory immersion, but what genuinely makes VR distinct: embodiment. The effect of its conceit is a superhuman-like control of time manipulation, with players deftly contorting their bodies to evade slow moving bullets while dispatching enemies with an empowering ease. Superhot VR provides an experience worth donning a headset for, and it consequently remains one of VR gaming’s top selling titles eight years after its release.
When Immersion Is Too Much
John Carmack’ Doom and Wolfenstein, on which VR’s gaming fantasy was based, are first-person shooters that closely map to hardcore gaming ideals. They’re hyperviolent, fast-paced, and difficult; they have a limited focus on story; and they feature some of the goriest scenes in games. In the same way that VR gaming has been detrimentally entwined with the pursuit of photorealism, VR gaming has been co-opted by these hardcore values that ultimately limit the medium. They lack mainstream appeal and valorise experiences that simply aren’t as appealing in VR as it is in a flat screen.
In a discussion around the design of Half Life: Alyx—one of the only high-budget VR-only games—designers Greg Coomer and Robin Walker explain that VR changes the way that people interact with virtual environments. As Coomer says, “people are slower to traverse space, and they want to slow down and be more interactive with more things in each environment. It has affected, on a fundamental level, how we’ve constructed environments and put things together.” Walker adds that the changes aren’t “because of some constraint around how they move through the world, it’s just because they pay so much more attention to things and poke at things.” Environments in VR games are much denser; on PC they feel small, but in VR they feel big.
This in part explains why few games originally designed for flat screens and “ported” to VR have been successful. The rapidly paced hyperviolence best characterized by Doom is simply sensory overload in VR, and the “intensity of being there”—one of Carmack’s aspirations—is unappealing. In VR, unrelenting games are unpleasurable: Most of us aren’t that coordinated, and we can’t play for extended periods of time in VR. It’s physically exhausting.
Casual Virtual Reality?
Beat Saber is a prime example of a game that might be derided as casual, if it weren’t the bestselling VR game of all time. Beat Saber is a music rhythm-matching game, a hybrid of Dance Dance Revolution, Guitar Hero, and Fruit Ninja. In time with electronic music, a playlist of red or blue boxes streams towards the player. Armed with two neon swords—commonly described as light sabers—the player must strike these boxes in the correct direction, denoted by a subtle white arrow.
Striking a box releases a note in the accompanying song, resulting in an experience that is half playing an instrument, and half dance. Well patterned songs create sweeping movements and rhythms reminiscent of the exaggerated gestures used by Nintendo Wii players.
Beat Saber youtube
Beat Saber’s appeal is immersion-through-embodiment, also achieved by disregarding VR’s gaming fantasy of hardcore experiences. With each song being, well, song length, Beat Saber supports a shorter, casual mode of engagement that isn’t pleasurable because it is difficult or competitive, but simply because playing a song feels good.
Gaming in VR has been subjected to a vicious self-reinforcing cycle wherein VR developers create hardcore games, which appeal to a certain kind of hardcore gamer user, whose purchasing habits in turn drive further development of those kinds of games, and not others. Attempts to penetrate this feedback loop have been met with the hostility of VR’s online gaming culture, appropriated from gamer culture at large.
As a result, the scope of VR games remains narrow, and oblivious to the kinds of games that might take VR to its billionth user. Maybe then, the one thing that could save VR gaming is the one possibility that VR enthusiasts decried the most when Facebook purchased Oculus in 2014: Farmville VR.
-
Meet the Teens Whose Tech Reduces Drownings and Fights Air Pollution
by Elizabeth Fuscaldo on 09. September 2024. at 18:00
Drowning is the third leading cause of accidental deaths globally, according to the World Health Organization. The deaths disproportionately impact low- and middle-income communities, whose beaches tend to lack lifeguards because of limited funds. Last year 104 drownings of the 120 reported in the United States occurred on unstaffed beaches.
That fueled Angelina Kim’s drone project. Kim is a senior at the Bishop’s School in La Jolla, Calif. Her Autonomous Unmanned Aerial Vehicle (UAV) System for Ocean Hazard Recognition and Rescue: Scout and Rescue UAV Prototype project was showcased in May at Regeneron’s International Science and Engineering Fair (ISEF) in Los Angeles.
Kim project took first place in this year’s IEEE Presidents’ Scholarship competition: a prize of US $10,000 payable over four years of undergraduate university study. The IEEE Foundation established the award to acknowledge a deserving student whose project demonstrates an understanding of electrical or electronics engineering, computer science, or other IEEE field of interest. The scholarship is administered by IEEE Educational Activities.
Kim has long been motivated to help those in need, especially after her mother’s illness when Kim was a young child.
“I’ve been determined to find ways to create a safer community using technology my whole life,” she says. “I realized as I got more into robotics that technology can be used to protect those in my community.”
Kim and the second- and third-place scholarship winners also received a complimentary IEEE student membership.
In addition, to mark the scholarship’s quarter century, IEEE established a 25th anniversary award to honor a project that is likely to make a difference. It also was given out at ISEF.
Drones that can save swimmers’ lives
Angelina Kim’s autonomous drone system, consisting of a scout and rescue drone, aims to prevent drownings on beaches that have no lifeguards.Lynn Bowlby
The autonomous UAV lifeguard system consists of two types of drones: a scout craft and a rescue craft. The scout drone surveys approximately 1 kilometer of shoreline, taking photographs and analyzing them for rip currents, which can be deadly to swimmers.
The scout drone “implements a new version of differential frame displacement and a new depth-risk model,” Kim says. The displacement compares the previous image to the next one and notes in what direction a wave is moving. The algorithm detects rip currents and, using the depth-risk model, focuses on strong currents in the deep end—which are more dangerous. If the scout drone detects a swimmer caught in such a current, it summons the rescue drone. That drone drops a flotation device outfitted with a heaving rope and can pull the endangered swimmer to shore, Kim says.
The rescue drones operate with roll-and-pitch tilt rotors, allowing them to fly in whatever direction and orientation they need to.
Kim says she was shocked and ecstatic to receive the award: “It felt like a dream come true because IEEE has been the organization I’ve always wanted to be a part of.”
She presented her project at two IEEE gatherings this year: the IEEE International Conference on Control and Automation and the IEEE International Conference on Automatic Control and Intelligent Systems.
Kim says some of the roadblocks she encountered with her project have made her a better engineer.
“If everything goes perfectly, then there will not be much to learn,” she says. “But if you fail—like accidentally flying your drone into your parents’ house, like I did—you get opportunities to learn.”
She plans to study electrical or mechanical engineering in college. Eventually, she says, she would like to build and mass-produce technologies that help communities at a low cost.
Spotting carbon dioxide in soil
Sahiti Bulusu’s Carboflux device measures and tracks carbon levels in soil to collect data for climatologists. Lynn Bowlby
Excessive carbon dioxide is a leading cause of global warming and climate change, according to NASA. Burning fossil fuels is a prominent source of carbon dioxide, but it also can be released from the ground.
“People don’t account for the large amounts of CO2 that come from soil,” second-place winner Sahiti Bulusu says. “Carbon flux is the rate at which CO2 is exchanged between the soil and the atmosphere, and can account for 80 percent of net ecosystem carbon exchange.”
Bulusu is a senior at the Basis Independent high school in Fremont, California.
Ecosystem carbon exchange is the transfer of carbon dioxide between the atmosphere and the physical environment. Because the exchange is not often accounted for, there’s a gap in data. To provide additional information, Bulusu created the Carboflux Network.
The sensor node is designed to measure enhanced flux CO2 and add to the Global Ecosystem Monitoring Network. The system contains three main subunits: a flux chamber, an under-soil sensor array, and a microcontroller with a data transmission unit. The enclosed flux chamber houses sensors. When CO2 accumulates in the chamber, the linear increase in carbon dioxide concentration is used to calculate carbon flux. The second subunit uses the gradient method, which measures flux using Fick’s law, looking at the CO2 gradient in the soil, and multiplying it with the soil diffusivity coefficient. Sensors are put in the ground at different depths to measure the gradient.
“Carboflux uses the gas concentration between the depths to predict what the soil carbon flux would be,” Bulusu says.
The system is automated and provides real-time data through microcontrollers and a modem. The device studies soil flux, both above and beneath the ground, she says.
“It is able to be scaled locally and globally,” she says, “helping to pinpoint local carbon sources and carbon sinks.” The data also can give scientists a large-scale global perspective.
Bulusu came up with her idea with help from mentors and professors Helen Dahlke and Elad Levintal.
“They told me about the lack of global carbon flux data, and the idea for a network of CO2 flux sensors,” Bulusu says. “This data is needed for climate modeling and mitigation strategies.
“I didn’t think I would be so passionate about a project. I love science, but I never thought I would be someone who would sit in the rain and cold for hours trying to figure out a problem. I was so invested in this project, and it has come so far.”
Bulusu plans to pursue a degree in computer science or environmental science. Whatever field she choses, she says, she wants to improve the environment with the technology she creates.
She was awarded a $600 scholarship for her project.
Accessible communication for ALS patients
Gaze Link, developed by Xiangzhou Sun, uses a mobile app, camera, and artificial intelligence to help those with ALS communicate through their eye movements. Lynn Bowlby
Xiangzhou “Jonas” Sun has been volunteering to help people with amyotrophic lateral sclerosis, also known as Lou Gehrig’s disease. After Sun and his family spent time helping ALS patients and assisting caretakers in Hong Kong and the United States, he was inspired to create a mobile app to assist them. He is a senior at the Webb School of California, in Claremont.
While volunteering, he noticed that ALS patients had trouble communicating because of the disease.
“ALS damages neurons in the body, and patients gradually lose the ability to walk, move their hands, and speak,” he says. “My objective with Gaze Link was to build a mobile application that could help ALS patients to type sentences with only their eyes and without external assistance.”
The low-cost smartphone app uses eye-gesture recognition, AI sentence generation, and text-to-speech capabilities to allow people with disabilities linked to ALS to use their phone’s front-end camera to communicate.
Gaze Link now works with three languages: English, Mandarin, and Spanish.
Users’ eye gestures are mapped to words and pronunciations, and an AI next-word-prediction feature is incorporated.
Sun’s passion for his project is palpable.
“My favorite moment was when I brought a prototype of Gaze Link to a caretaker from the ALS Association, and they were overjoyed,” he says. “That moment was really emotional for me because I was motivated by these patients. It gave me a huge sense of achievement that I could finally help them.”
Sun received a $400 scholarship for placing third.
He says he plans to use his engineering know-how to help people with disabilities. He is passionate about design and visual arts as well, and says he hopes to combine them with his engineering skills.
Gaze Link is available through Google Play.
An inclusive way to code
Abhisek Shah’s AuralStudio provides a way for programmers with visual impairments to code using the Rattle language he wrote, which can be read aloud.Lynn Bowlby
The 25th anniversary award, in the amount of $1,000, was given to Abishek Shah, a senior at Green Level High School, in Cary, N.C.
Last year Shah had a temporary vision issue, which made looking at a computer screen nearly impossible, even for a short time. After that, while visiting family back home in India, he got a chance to interact with some students at a school for blind girls.
During his interactions with the students, he says, he realized they were determined and driven and wanted to be financially independent. None had even considered a career in computer programming, however, believing it to be off-limits because of their visual impairment, he says.
He wondered if there was a solution that could help blind people code.
“How can we redesign and rethink the way we code today?” he asked himself before devising his AuralStudio. He realized he could put his passion for coding to good use and created a software application for those whose vision impairments are permanent.
His AuralStudio development environment allows programmers with a visual disability to write, build, run, and test prototypes.
“All this was built toward helping those with disabilities learn to code,” Shah says.
It eliminates the need for a keyboard and mouse in favor of a custom control pad. It includes a voice-only option for those who cannot use their hands.
The testbed uses Rattle, a programming language that Shah created to be read aloud, both from the computer and the programmer. AuralStudio also uses acyclic digraphs to render code, making it easier and more intuitive. Shah wrote a two-part autocorrect algorithm to prevent homophones and homonyms from causing errors. It was done by integrating AI into the application.
Shah used the programming languages Rust, C, JavaScript, and Python. Raspberry Pi was the primary hardware.
He worked with visually impaired students from the Governor Morehead School of Raleigh, N.C., teaching them programming basics.
“The students were so eager to learn,” he says. “Getting to see their eagerness, grit, and hard-working nature was so heartwarming and probably the best part of this project.”
Shah says he plans to study computer science and business.
“Anything I build will have the end goal of improving the lives of others,” he says.
-
Will the "AI Scientist" Bring Anything to Science?
by Eliza Strickland on 09. September 2024. at 14:41
When an international team of researchers set out to create an “AI scientist” to handle the whole scientific process, they didn’t know how far they’d get. Would the system they created really be capable of generating interesting hypotheses, running experiments, evaluating the results, and writing up papers?
What they ended up with, says researcher Cong Lu, was an AI tool that they judged equivalent to an early Ph.D. student. It had “some surprisingly creative ideas,” he says, but those good ideas were vastly outnumbered by bad ones. It struggled to write up its results coherently, and sometimes misunderstood its results: “It’s not that far from a Ph.D. student taking a wild guess at why something worked,” Lu says. And, perhaps like an early Ph.D. student who doesn’t yet understand ethics, it sometimes made things up in its papers, despite the researchers’ best efforts to keep it honest.
Lu, a postdoctoral research fellow at the University of British Columbia, collaborated on the project with several other academics, as well as with researchers from the buzzy Tokyo-based startup Sakana AI. The team recently posted a preprint about the work on the ArXiv server. And while the preprint includes a discussion of limitations and ethical considerations, it also contains some rather grandiose language, billing the AI scientist as “the beginning of a new era in scientific discovery,” and “the first comprehensive framework for fully automatic scientific discovery, enabling frontier large language models (LLMs) to perform research independently and communicate their findings.”
The AI scientist seems to capture the zeitgeist. It’s riding the wave of enthusiasm for AI for science, but some critics think that wave will toss nothing of value onto the beach.
The “AI for Science” Craze
This research is part of a broader trend of AI for science. Google DeepMind arguably started the craze back in 2020 when it unveiled AlphaFold, an AI system that amazed biologists by predicting the 3D structures of proteins with unprecedented accuracy. Since generative AI came on the scene, many more big corporate players have gotten involved. Tarek Besold, a SonyAI senior research scientist who leads the company’s AI for scientific discovery program, says that AI for science is “a goal behind which the AI community can rally in an effort to advance the underlying technology but—even more importantly—also to help humanity in addressing some of the most pressing issues of our times.”
Yet the movement has its critics. Shortly after a 2023 Google DeepMind paper came out claiming the discovery of 2.2 million new crystal structures (“equivalent to nearly 800 years’ worth of knowledge”), two materials scientists analyzed a random sampling of the proposed structures and said that they found “scant evidence for compounds that fulfill the trifecta of novelty, credibility, and utility.” In other words, AI can generate a lot of results quickly, but those results may not actually be useful.
How the AI Scientist Works
In the case of the AI scientist, Lu and his collaborators tested their system only on computer science, asking it to investigate topics relating to large language models, which power chatbots like ChatGPT and also the AI scientist itself, and the diffusion models that power image generators like DALL-E.
The AI scientist’s first step is hypothesis generation. Given the code for the model it’s investigating, it freely generates ideas for experiments it could run to improve the model’s performance, and scores each idea on interestingness, novelty, and feasibility. It can iterate at this step, generating variations on the ideas with the highest scores. Then it runs a check in Semantic Scholar to see if its proposals are too similar to existing work. It next uses a coding assistant called Aider to run its code and take notes on the results in the format of an experiment journal. It can use those results to generate ideas for follow-up experiments.
The AI scientist is an end-to-end scientific discovery tool powered by large language models. University of British Columbia
The next step is for the AI scientist to write up its results in a paper using a template based on conference guidelines. But, says Lu, the system has difficulty writing a coherent nine-page paper that explains its results—”the writing stage may be just as hard to get right as the experiment stage,” he says. So the researchers broke the process down into many steps: The AI scientist wrote one section at a time, and checked each section against the others to weed out both duplicated and contradictory information. It also goes through Semantic Scholar again to find citations and build a bibliography.
But then there’s the problem of hallucinations—the technical term for an AI making stuff up. Lu says that although they instructed the AI scientist to only use numbers from its experimental journal, “sometimes it still will disobey.” Lu says the model disobeyed less than 10 percent of the time, but “we think 10 percent is probably unacceptable.” He says they’re investigating a solution, such as instructing the system to link each number in its paper to the place it appeared in the experimental log. But the system also made less obvious errors of reasoning and comprehension, which seem harder to fix.
And in a twist that you may not have seen coming, the AI scientist even contains a peer review module to evaluate the papers it has produced. “We always knew that we wanted some kind of automated [evaluation] just so we wouldn’t have to pour over all the manuscripts for hours,” Lu says. And while he notes that “there was always the concern that we’re grading our own homework,” he says they modeled their evaluator after the reviewer guidelines for the leading AI conference NeurIPS and found it to be harsher overall than human evaluators. Theoretically, the peer review function could be used to guide the next round of experiments.
Critiques of the AI Scientist
While the researchers confined their AI scientist to machine learning experiments, Lu says the team has had a few interesting conversations with scientists in other fields. In theory, he says, the AI scientist could help in any field where experiments can be run in simulation. “Some biologists have said there’s a lot of things that they can do in silico,” he says, also mentioning quantum computing and materials science as possible fields of endeavor.
Some critics of the AI for science movement might take issue with that broad optimism. Earlier this year, Jennifer Listgarten, a professor of computational biology at UC Berkeley, published a paper in Nature Biotechnology arguing that AI is not about to produce breakthroughs in multiple scientific domains. Unlike the AI fields of natural language processing and computer vision, she wrote, most scientific fields don’t have the vast quantities of publicly available data required to train models.
Two other researchers who study the practice of science, anthropologist Lisa Messeri of Yale University and psychologist M.J. Crockett of Princeton University, published a 2024 paper in Nature that sought to puncture the hype surrounding AI for science. When asked for a comment about this AI scientist, the two reiterated their concerns over treating “AI products as autonomous researchers.” They argue that doing so risks narrowing the scope of research to questions that are suited for AI, and losing out on the diversity of perspectives that fuels real innovation. “While the productivity promised by ‘the AI Scientist’ may sound appealing to some,” they tell IEEE Spectrum, “producing papers and producing knowledge are not the same, and forgetting this distinction risks that we produce more while understanding less.”
But others see the AI scientist as a step in the right direction. SonyAI’s Besold says he believes it’s a great example of how today’s AI can support scientific research when applied to the right domain and tasks. “This may become one of a handful of early prototypes that can help people conceptualize what is possible when AI is applied to the world of scientific discovery,” he says.
What’s Next for the AI Scientist
Lu says that the team plans to keep developing the AI scientist, and he says there’s plenty of low-hanging fruit as they seek to improve its performance. As for whether such AI tools will end up playing an important role in the scientific process, “I think time will tell what these models are good for,” Lu says. It might be, he says, that such tools are useful for the early scoping stages of a research project, when an investigator is trying to get a sense of the many possible research directions—although critics add that we’ll have to wait for future studies to see if these tools are really comprehensive and unbiased enough to be helpful.
Or, Lu says, if the models can be improved to the point that they match the performance of “a solid third-year Ph.D. student,” they could be a force multiplier for anyone trying to pursue an idea (at least, as long as the idea is in an AI-suitable domain). “At that point, anyone can be a professor and carry out a research agenda,” says Lu. “That’s the exciting prospect that I’m looking forward to.”
-
Greener Steel Production Requires More Electrochemical Engineers
by Dan Steingart on 08. September 2024. at 13:00
In the 1800s, aluminum was considered more valuable than gold or silver because it was so expensive to produce the metal in any quantity. Thanks to the Hall-Héroult smelting process, which pioneered the electrochemical reduction of aluminum oxide in 1886, electrochemistry advancements made aluminum more available and affordable, rapidly transforming it into a core material used in the manufacturing of aircraft, power lines, food-storage containers and more.
As society mobilizes against the pressing climate crisis we face today, we find ourselves seeking transformative solutions to tackle environmental challenges. Much as electrochemistry modernized aluminum production, science holds the key to revolutionizing steel and iron manufacturing.
Electrochemistry can help save the planet
As the world embraces clean energy solutions such as wind turbines, electric vehicles, and solar panels to address the climate crisis, changing how we approach manufacturing becomes critical. Traditional steel production—which requires a significant amount of energy to burn fossil fuels at temperatures exceeding 1,600 °C to convert ore into iron—currently accounts for about 10 percent of the planet’s annual CO2 emissions. Continuing with conventional methods risks undermining progress toward environmental goals.
Scientists already are applying electrochemistry—which provides direct electrical control of oxidation-reduction reactions—to convert ore into iron. The conversion is an essential step in steel production and the most emissions-spewing part. Electrochemical engineers can drive the shift toward a cleaner steel and iron industry by rethinking and reprioritizing optimizations.
When I first studied engineering thermodynamics in 1998, electricity—which was five times the price per joule of heat—was considered a premium form of energy to be used only when absolutely required.
Since then the price of electricity has steadily decreased. But emissions are now known to be much more harmful and costly.
Engineers today need to adjust currently accepted practices to develop new solutions that prioritize mass efficiency over energy efficiency.
In addition to electrochemical engineers working toward a cleaner steel and iron industry, advancements in technology and cheaper renewables have put us in an “electrochemical moment” that promises change across multiple sectors.
The plummeting cost of photovoltaic panels and wind turbines, for example, has led to more affordable renewable electricity. Advances in electrical distribution systems that were designed for electric vehicles can be repurposed for modular electrochemical reactors.
Electrochemistry holds the potential to support the development of clean, green infrastructure beyond batteries, electrolyzers, and fuel cells. Electrochemical processes and methods can be scaled to produce metals, ceramics, composites, and even polymers at scales previously reserved for thermochemical processes. With enough effort and thought, electrochemical production can lead to billions of tons of metal, concrete, and plastic. And because electrochemistry directly accesses the electron transfer fundamental to chemistry, the same materials can be recycled using renewable energy.
As renewables are expected to account for more than 90 percent of global electricity expansion during the next five years, scientists and engineers focused on electrochemistry must figure out how best to utilize low-cost wind and solar energy.
The core components of electrochemical systems, including complex oxides, corrosion-resistant metals, and high-power precision power converters, are now an exciting set of tools for the next evolution of electrochemical engineering.
The scientists who came before have created a stable set of building blocks; the next generation of electrochemical engineers needs to use them to create elegant, reliable reactors and other systems to produce the processes of the future.
Three decades ago, electrochemical engineering courses were, for the most part, electives and graduate-level. Now almost every institutional top-ranked R&D center has full tracks of electrochemical engineering. Students interested in the field should take both electroanalytical chemistry and electrochemical methods classes and electrochemical energy storage and materials processing coursework.
Although scaled electrochemical production is possible, it is not inevitable. It will require the combined efforts of the next generation of engineers to reach its potential scale.
Just as scientists found a way to unlock the potential of the abundant, once-unattainable aluminum, engineers now have the opportunity to shape a cleaner, more sustainable future. Electrochemistry has the power to flip the switch to clean energy, paving the way for a world in which environmental harmony and industrial progress go hand in hand.
-
Get to Know the IEEE Board of Directors
by IEEE on 06. September 2024. at 18:00
The IEEE Board of Directors shapes the future direction of IEEE and is committed to ensuring IEEE remains a strong and vibrant organization—serving the needs of its members and the engineering and technology community worldwide—while fulfilling the IEEE mission of advancing technology for the benefit of humanity.
This article features IEEE Board of Directors members A. Matt Francis, Tom Murad, and Christopher Root.
IEEE Senior Member A. Matt Francis
Director, IEEE Region 5: Southwestern U.S.
Moriah Hargrove Anders
Francis’s primary technology focus is extreme environment and high-temperature integrated circuits. His groundbreaking work has pushed the boundaries of electronics, leading to computers operating in low Earth orbit for more than a year on the International Space Station and on jet engines. Francis and his team have designed and built some of the world’s most rugged semiconductors and systems.
He is currently helping explore new computing frontiers in supersonic and hypersonic flight, geothermal energy exploration, and molten salt reactors. Well versed in shifting technology from idea to commercial application, Francis has secured and led projects with the U.S. Air Force, DARPA, NASA, the National Science Foundation, the U.S. Department of Energy, and private-sector customers.
Francis’s influence extends beyond his own ventures. He is a member of the IEEE Aerospace and Electronic Systems, IEEE Computer, and IEEE Electronics Packaging societies, demonstrating his commitment to industry and continuous learning.
He attended the University of Arkansas in Fayetteville for both his undergraduate and graduate degrees. He joined IEEE while at the university and was president of the IEEE–Eta Kappa Nu honor society’s Gamma Phi chapter. Francis’s other past volunteer roles include serving as chair of the IEEE Ozark Section, which covers Northwest Arkansas, and also as a member of the IEEE-USA Entrepreneurship Policy Innovation Committee.
His deep-rooted belief in the power of collaboration is evident in his willingness to share knowledge and support aspiring entrepreneurs. Francis is proud to have helped found a robotics club (an IEEE MGA Local Group) in his rural Elkins, Ark., community and to have served on steering committees for programs including IEEE TryEngineering and IEEE-USA’s Innovation, Research, and Workforce Conferences. He serves as an elected city council member for his town, and has cofounded two non-profits, supporting his community and the state of Arkansas.
Francis’s journey from entrepreneur to industry leader is a testament to his determination and innovative mindset. He has received numerous awards including the IEEE-USA Entrepreneur Achievement Award for Leadership in Entrepreneurial Spirit, IEEE Region 5 Directors Award, and IEEE Region 5 Outstanding Individual Member Achievement Award.
IEEE Senior Member Tom Murad
Director, IEEE Region 7: Canada
Siemens Canada
Murad is a respected technology leader, award-winning educator, and distinguished speaker on engineering, skills development, and education. Recently retired, he has 40 years of experience in professional engineering and technical operations executive management, including more than 10 years of academic and R&D work in industrial controls and automation.
He received his doctorate (Ph.D.) degree in power electronics and industrial controls from Loughborough University of Technology in the U.K.
Murad has held high-level positions in several international engineering and industrial organizations, and he contributed to many global industrial projects. His work on projects in power utilities, nuclear power, oil and gas, mining, automotive, and infrastructure industries has directly impacted society and positively contributed to the economy. He is a strong advocate of innovation and creativity, particularly in the areas of digitalization, smart infrastructure, and Industry 4.0. He continues his academic career as an adjunct professor at University of Guelph in Ontario, Canada.
His dedication to enhancing the capabilities of new generations of engineers is a source of hope and optimism. His work in significantly improving the quality and relevance of engineering and technical education in Canada is a testament to his commitment to the future of the engineering profession and community. For that he has been assigned by the Ontario Government to be a member of the board of directors of the Post Secondary Education Quality Assessment Board (PEQAB).
Murad is a member of the IEEE Technology and Engineering Management, IEEE Education, IEEE Intelligent Transportation Systems, and IEEE Vehicular Technology societies, the IEEE-Eta Kappa Nu honor society, and the Editorial Advisory Board Chair for the IEEE Canadian Review Magazine. His accomplishments show his passion for the engineering profession and community.
He is a member of the Order of Honor of the Professional Engineers of Ontario, Canada, Fellow of Engineers Canada, Fellow of Engineering Institutes of Canada (EIC), and received the IEEE Canada J.M. Ham Outstanding Engineering Educator Award, among other recognitions highlighting his impact on the field.
IEEE Senior Member Christopher Root
Director, Division VII
Vermont Electric Power Company and Shana Louiselle
Root has been in the electric utility industry for more than 40 years and is an expert in power system operations, engineering, and emergency response. He has vast experience in the operations, construction, and maintenance of transmission and distribution utilities, including all phases of the engineering and design of power systems. He has shared his expertise through numerous technical presentations on utility topics worldwide.
Currently an industry advisor and consultant, Root focuses on the crucial task of decarbonizing electricity production. He is engaged in addressing the challenges of balancing an increasing electrical market and dependence on renewable energy with the need to provide low-cost, reliable electricity on demand.
Root’s journey with IEEE began in 1983 when he attended his first meeting as a graduate student at Rensselaer Polytechnic Institute, in Troy, N.Y. Since then, he has served in leadership roles such as treasurer, secretary, and member-at-large of the IEEE Power & Energy Society (PES). His commitment to the IEEE mission and vision is evident in his efforts to revitalize the dormant IEEE PES Boston Chapter in 2007 and his instrumental role in establishing the IEEE PES Green Mountain Section in Vermont in 2015. He also is a member of the editorial board of the IEEE Power & Energy Magazine and the IEEE–Eta Kappa Nu honor society.
Root’s contributions and leadership in the electric utility industry have been recognized with the IEEE PES Leadership in Power Award and the PES Meritorious Service Award. -
Video Friday: HAND to Take on Robotic Hands
by Evan Ackerman on 06. September 2024. at 15:53
Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion.
ICRA@40: 23–26 September 2024, ROTTERDAM, NETHERLANDS
IROS 2024: 14–18 October 2024, ABU DHABI, UAE
ICSR 2024: 23–26 October 2024, ODENSE, DENMARK
Cybathlon 2024: 25–27 October 2024, ZURICH
Enjoy today’s videos!
The National Science Foundation Human AugmentatioN via Dexterity Engineering Research Center (HAND ERC) was announced in August 2024. Funded for up to 10 years and $52 million, the HAND ERC is led by Northwestern University, with core members Texas A&M, Florida A&M, Carnegie Mellon, and MIT, and support from Wisconsin-Madison, Syracuse, and an innovation ecosystem consisting of companies, national labs, and civic and advocacy organizations. HAND will develop versatile, easy-to-use dexterous robot end effectors (hands).
[ HAND ]
The Environmental Robotics Lab at ETH Zurich, in partnership with Wilderness International (and some help from DJI and Audi), is using drones to sample DNA from the tops of trees in the Peruvian rainforest. Somehow, the treetops are where 60 to 90 percent of biodiversity is found, and these drones can help researchers determine what the heck is going on up there.
[ ERL ]
Thanks, Steffen!
1X introduces NEO Beta, “the pre-production build of our home humanoid.”
“Our priority is safety,” said Bernt Børnich, CEO at 1X. “Safety is the cornerstone that allows us to confidently introduce NEO Beta into homes, where it will gather essential feedback and demonstrate its capabilities in real-world settings. This year, we are deploying a limited number of NEO units in selected homes for research and development purposes. Doing so means we are taking another step toward achieving our mission.”
[ 1X ]
We love MangDang’s fun and affordable approach to robotics with Mini Pupper. The next generation of the little legged robot has just launched on Kickstarter, featuring new and updated robots that make it easy to explore embodied AI.
The Kickstarter is already fully funded after just a day or two, but there are still plenty of robots up for grabs.
[ Kickstarter ]
Quadrupeds in space can use their legs to reorient themselves. Or, if you throw one off a roof, it can learn to land on its feet.
To be presented at CoRL 2024.
[ ARL ]
HEBI Robotics, which apparently was once headquartered inside a Pittsburgh public bus, has imbued a table with actuators and a mind of its own.
[ HEBI Robotics ]
Carcinization is a concept in evolutionary biology where a crustacean that isn’t a crab eventually becomes a crab. So why not do the same thing with robots? Crab robots solve all problems!
[ KAIST ]
Waymo is smart, but also humans are really, really dumb sometimes.
[ Waymo ]
The Robotics Department of the University of Michigan created an interactive community art project. The group that led the creation believed that while roboticists typically take on critical and impactful problems in transportation, medicine, mobility, logistics, and manufacturing, there are many opportunities to find play and amusement. The final piece is a grid of art boxes, produced by different members of our robotics community, which offer an eight-inch-square view into their own work with robotics.
I appreciate that UBTECH’s humanoid is doing an actual job, but why would you use a humanoid for this?
[ UBTECH ]
I’m sure most actuators go through some form of life-cycle testing. But if you really want to test an electric motor, put it into a BattleBot and see what happens.
Yes, but have you tried fighting a BattleBot?
[ AgileX ]
In this video, we present collaboration aerial grasping and transportation using multiple quadrotors with cable-suspended payloads. Grasping using a suspended gripper requires accurate tracking of the electromagnet to ensure a successful grasp while switching between different slack and taut modes. In this work, we grasp the payload using a hybrid control approach that switches between a quadrotor position control and a payload position control based on cable slackness. Finally, we use two quadrotors with suspended electromagnet systems to collaboratively grasp and pick up a larger payload for transportation.
[ Hybrid Robotics ]
I had not realized that the floretizing of broccoli was so violent.
[ Oxipital ]
While the RoboCup was held over a month ago, we still wanted to make a small summary of our results, the most memorable moments, and of course an homage to everyone who is involved with the B-Human team: the team members, the sponsors, and the fans at home. Thank you so much for making B-Human the team it is!
[ B-Human ]
-
When "AI for Good" Goes Wrong
by Payal Arora on 05. September 2024. at 14:59
This guest article is adapted from the author’s new book From Pessimism to Promise: Lessons from the Global South on Designing Inclusive Tech, published by MIT Press.
What do AI-enabled rhino collars in South Africa, computer-vision pest-detection drones in the Punjab farmlands, and wearable health devices in rural Malawi have in common?
These initiatives are all part of the AI for Good movement, which aligns AI technologies with the United Nations sustainable development goals to find solutions for global challenges like poverty, health, education, and environmental sustainability.
MIT Press
The hunger for AI-based solutions is understandable. In 2023, 499 rhinos were killed by poachers in South Africa, an increase of more than 10 percent from 2022. Several farmers in Punjab lost about 90 percent of their cotton yield to the pink bollworm; if the pest had been detected in time, they could have saved their crops. As for healthcare, despite decades of effort to boost the numbers of healthcare practitioners in rural areas, they continue to migrate to cities.
What makes AI “good,” though? Why do we need to preface AI applications in the Global South with morality and charity? And will noble intent translate to making AI tools work for the majority of the world?
A Changed Reality
The fact is, the Global South of decades ago does not exist.
Today the countries in the Global South are more confident, more entrepreneurial, and are taking leadership to pioneer locally appropriate AI tools that work for their people. Startups understand that the success of new tech is contingent on leveraging local knowledge for meaningful adoption and scaling.
The old formula of “innovate in the West and disseminate to the rest” is out of sync with this new reality. While the West holds onto its old missionary zeal, the South-South collaboration continues to grow, sharing new tech and building AI governance. What’s more, some tech altruism initiatives have come under scrutiny as they obfuscate their data extraction activities, making them more transactional than charitable.
The Market for Tech Altruism
In August, the European Union’s legal framework on AI, the AI Act, entered into force. Its measures are meant to help citizens and stakeholders optimize these tools while mitigating the risks. There is little mention of AI for good in their documents; it’s simply the default. Yet, as we shift from the Global North to the South, morality kicks in.
Tech altruism underlines this shift. Many of the AI for Good initiatives are funded by tech philanthropists in partnership with global aid agencies. Doing good manifests in piloting tech solutions, with the Global South as a live laboratory. A running joke with development workers is that their field suffers from “pilotitis,” an acute syndrome of pilot projects that never scale. The Global South is typically viewed as a recipient, a market, a beneficiary for techno-solutionism.
Take AI collars for rhinos. The Conservation Collar initiative in South Africa, for example, detects abnormal behavior, and these signals are sent to an AI system that computes the probability of risk. If it determines that the animal is at urgent risk, the rangers can hypothetically act immediately to stop the poaching. But when my team investigated the ground realities, we found that rangers face numerous obstacles to fast action, including dirt roads, old vehicles, and long distances. Many rangers had not been paid in months, and their motivation was low. And to top all this, they faced an armed militia protecting a multibillion rhino trade business.
In Punjab, drones with computer vision can guide farmers to detect pests before they destroy crops. The Global Alliance for Climate-Smart Agriculture funds projects involving many such AI-enabled technologies as farmers face the vagaries of the climate crisis. However, detection is just one part of a larger problem. Farmers struggle with poor quality and unaffordable pesticides, loan sharks, the vulnerabilities brought on by monocropping, and water scarcity. Agricultural innovators complain that there are few early adopters of their tech, however good their tools may be. Afterall, young people in the Global South increasingly don’t see their future in farming.
Meanwhile, we’ve seen philanthropies such as the Bill & Melinda Gates Foundation launch grand challenges for AI to help alleviate burdens on African healthcare systems. This has resulted in winners such as IntelSurv in Malawi, an intelligent disease surveillance data feedback system that computes data from smart wearables. Yet, even with hundreds of patents for such devices being registered every year, they’re not yet capable of consistently capturing high-quality data. In places like Malawi, these devices may become the single source of training data for healthcare AI, amplifying errors in their healthcare system
The fact is, we can’t really solve problems with AI without accompanying social reforms. Building proper roads or paying your rangers on time is not an innovation, it’s common sense. Likewise, whether it’s in the healthcare or the agricultural sector, people need social incentives to adopt these technologies. Otherwise these AI tools will remain in the wild, and won’t be domesticated.
Data Is Currency
Tech altruism has increasingly become suspect as AI companies are now facing an acute data shortage. They’re scrambling for data in the Global South, where majority of tech users live. Take, for instance, the case of Worldcoin, co-founded by OpenAI CEO Sam Altman. It plans to become “the world’s largest privacy-preserving human identity and financial network, giving ownership to everyone.” Worldcoin started as a nonprofit in 2019 by collecting biometric data, mostly in Global South countries, through its “orb” device and in exchange for cryptocurrency. Today, it’s a for-profit entity and is under investigation by many countries for its dubious data-collection methods.
The German nonprofit Heinrich-Böll-Stiftung recently reported on the aggressive growth of digital agricultural platforms across Africa which promise farmers precision agriculture and increased yields via AI-enabled apps. Yet these apps often provide corporations with free access to data about seeds, soil, crops, fertilizers, and weather from the farms where they’re used. Corporations can use AI analytics to weaponize this information, perhaps creating discriminatory agricultural insurance policies or micro-targeting ads for seeds and fertilizers. Similarly, in the health care sector, the Center for Digital Health at Brown University has reported on the selling of personal health data to third-party advertisers without user consent.
The problem is that, unlike private companies that are compelled to follow the law, altruistic initiatives often succeed in circumventing regulations due to their “charitable” intent. Almost a decade ago, Facebook launched Free Basics, which provided access to limited internet services in the Global South by violating net neutrality principles. When India blocked Free Basics in 2015, Mark Zuckerberg appeared shocked and remarked, “Who could possibly be against this?”
Today we ask, who could possibly get on board?
From Paternalism to Partnerships
As of 2024, according to one estimate, the Global South contributes 80 percent to global economic growth. Close to 90 percent of the world’s young population reside in these regions. And it has become a vital space for innovation. In 2018, China entered the global innovation index rankings as one of the top twenty most innovative countries in the world. India’s government has set up its “tech stack,” the largest open source, interoperable, and public digital infrastructure in the world. This stack is enabling entrepreneurs to build their products and services away from the Apple and Google duopoly that constrains competition and choice.
Despite the Global South demonstrating its innovative prowess, the imitator label remains sticky. This perception often translates to Western organizations treating Global South countries as beneficiaries, and not as partners and leaders in global innovation.
It is time we stop underestimating the Global South. Instead, Western organizations should channel their energies by looking at how different consumers can help to rethink opportunity, safeguards, and digital futures for the world’s majority. Inclusion is not an altruistic act. It is an essential element to generating solutions for the wicked problems that humanity faces today.
In designing new tech, we need to shift away from morality-driven design with grandiose visions of doing good. Instead, we should strive for design that focuses on the relationships between people, contexts, and policies.
Designers, programmers, and funders can benefit from listening to what users and entrepreneurs in the Global South have to say about AI intervening in their lives. And policymakers should bury the term “AI for Good.”
Media outlets must stop debating whether tech alone can solve the world’s problems. The real contextual intelligence we need won’t come from AI, but from human beings.
-
How Region Realignment Will Impact IEEE Elections
by IEEE Member and Geographic Activities on 04. September 2024. at 18:00
The work of restructuring IEEE’s geographic regions is well underway. Six U.S. regions will be consolidated into five, joining together the current IEEE Region 1 (Northeastern U.S.) and Region 2 (Eastern U.S.) to form Region 2 (Eastern and Northeastern U.S.). IEEE Region 10 (Asia and Pacific) will be split into two to form Region 10 (North Asia) and Region 11 (South Asia and Pacific).
The restructuring of IEEE’s 10 geographic regions will provide a more equitable representation across its global membership, as outlined in The Institute’s February 2023 article, “IEEE is Realigning Its Geographic Regions.”
The realignment will impact this year’s annual IEEE election process, which runs through 1 October.
In this year’s IEEE annual election, eligible voting members residing in Region 2 will be electing the IEEE Region 2 director-elect for the 2025—2026 term (serving as director in 2027—2028). The elected officer will first serve as director of the current Region 2 in 2027, and then in 2028 as director of the new Region 2.
The eligible voting members residing in Region 10 will be electing the IEEE Region 10 director-elect for the term 2025—2026 (serving as Director in 2027—2028). The elected officer will first serve as director of the current Region 10 in 2027 and then in 2028 as Director of the new Region 10 .
IEEE Member and Geographic Activities is continuing to coordinate the transition in tandem with the other IEEE organizational units in preparation for the realignment, which takes effect in January 2028.
This article appears in the September 2004 print issue.
-
How the Designer of the First Hydrogen Bomb Got the Gig
by Glenn Zorpette on 02. September 2024. at 14:00
By any measure, Richard Garwin is one of the most decorated and successful engineers of the 20th century. The IEEE Life Fellow has won the Presidential Medal of Freedom, the National Medal of Science, France’s La Grande Médaille de l’Académie des Sciences, and is one of just a handful of people elected to all three U.S. National Academies: Engineering, Science, and Medicine. At IBM, where he worked from 1952 to 1993, Garwin was a key contributor or a facilitator on some of the most important products and breakthroughs of his era, including magnetic resonance imaging, touchscreen monitors, laser printers, and the Cooley-Tukey fast Fourier transform algorithm.
And all that was after he did the thing for which he is most famous. At age 23 and at the behest of Edward Teller, Garwin designed the very first working hydrogen bomb, which was referred to as “the Sausage.” It was detonated in a test code-named Ivy Mike at Enewetak Atoll in November 1952, yielding 10.4 megatons of TNT. (The largest detonation before Ivy Mike was of a bomb code-named George, which yielded a mere 225 kilotons.)
Richard Garwin
Richard Garwin is an IBM Fellow Emeritus, an IEEE Life Fellow, and the designer of the first working hydrogen bomb.
Not until 2001—50 years after Garwin’s work on the bomb—did his pivotal role become publicly known. The definitive history of the hydrogen bomb, Richard Rhodes’s Dark Sun: The Making of the Hydrogen Bomb, published in 1995, has barely a page about Garwin. However, in 1979, after suffering a heart attack and contemplating his mortality, Teller sat down with the physicist George A. Keyworth II to record an oral testimony about the project. Teller’s verbal reckoning was kept secret for 22 years, until 2001, at which time a transcript was obtained by The New York Times.
In the transcript, Teller discounts the role of the mathematician Stanislaw Ulam, who was thought to have been Teller’s partner in what is still called the Teller-Ulam configuration. This “configuration” was actually a theory-based framework that envisioned a two-stage thermonuclear device based in part on a fission bomb (the first stage) that would generate the enormous temperatures and pressures needed to trigger a runaway fusion reaction (in the second stage). In the same transcript, Teller lavishes praise on Garwin’s design and declares, “that first design was made by Dick Garwin.” Because of the enduring secrecy around that first thermonuclear bomb, Garwin’s role had been largely unknown outside of a small circle of Los Alamos physicists, mathematicians, and engineers who were involved with the project—notably Teller, Enrico Fermi, Hans Bethe, and Ulam. Teller died in 2003.
Starting in the early 1950s and continuing in parallel with his career at IBM, Garwin also served as an advisor or consultant to U.S. government agencies on some of the most vital tech-related issues, and some of the most prestigious panels, of his times. That work continues to this day with his service as a member of the Jason group, the elite panel that offers technical and scientific advice, often classified, to the U.S. Defense Department and other agencies. Garwin, who has served in advisory roles under every U.S. president from Dwight Eisenhower to Barak Obama, has also been known for his writing and speaking on issues related to nuclear proliferation and arms control.
IEEE Spectrum spoke via videoconference with Garwin, now 96, who was at his home in Westchester County, New York.
Richard Garwin on:
- Returning to Los Alamos to work on the hydrogen bomb
- What it’s like to hold plutonium
- Designing “the Sausage”
- His proudest contributions during his time at IBM
- His friendship with Enrico Fermi
- Whether he considers himself an engineer or a physicist
- His efforts in the nuclear arms-control movement
- The future of nuclear and renewable energy
Garwin arrived at Los Alamos for the second time to work as a physicist in May of 1951. In the interview, he spoke early on, and without prompting, about Edward Teller’s ideas at the time about how a thermonuclear (fusion) bomb would work. Teller had not had much success translating his ideas into a working bomb, in part, Garwin says, because Teller did not understand that the deuterium fuel would “burn” (react) when it was very highly compressed, as it would be in the basic, Teller-Ulam conception of a hydrogen bomb.
Garwin: When I got to Los Alamos for the second time, in 1951, I had already known Edward Teller. He was on the physics faculty of the University of Chicago. And I went to Edward and I said, “What is the progress on your ideas for burning deuterium?” And he told me that he had met with the mathematician, Stanislaw Ulam, who worked for him. Ulam was in his small group. Teller was allowed only about four people in his group, much to his distress. And he resented that. But it was the right choice because you would need an atomic bomb, according to the Ulam-Teller concept. And there was no sense diluting the effort on working on the atomic bomb.
But Edward had had for many years a wrong theorem which he had never written down. He confesses this in his 1979 paper [ Editor’s note: This is the statement dictated to Keyworth after Teller’s heart attack] in which he gives me credit for the hydrogen bomb. But his theorem was that compression wouldn’t help. That if you couldn’t burn deuterium at normal liquid density—I think it’s 0.19 grams per cubic centimeter—you can’t burn it at 100-fold or 1,000-fold density. Everything would just happen faster, 100 times faster, or 1,000 times faster. This was a wrong theorem. He had never written it down, and it was wrong. And when he told Stan Ulam, he said in his 1979 effort, that he had been wasting a lot of time talking to Stan.
And so Edward decided that I would write it [a detailed engineering design for a working hydrogen bomb] up and give him a fair shot. And Ulam’s idea, according to this still-secret document in the Los Alamos report library, was given away by the title of the report. The title of the report that is, and always has been, unclassified. The first part of the title was: “Hydrodynamic Lenses.” The second part of the title was: “and Radiation Mirrors.” [ Editor’s Note: The paper, published in secret in March 1951, is titled, “On Heterocatalytic Detonations I: Hydrodynamic Lenses and Radiation Mirrors,” and it is the paper that contains the first description of the Teller-Ulam configuration.]
That was the option that Teller thought was best. So I went to Teller in his office at Los Alamos, and I asked him what had happened. He said that he had written up the meeting he had had recently with Stan Ulam and that Ulam had proposed acoustic lenses, of which we had 32 on the original implosion weapon [detonated at the Trinity Test near Alamogordo, N.M.]. So you could get 32 segments of the sphere. They had fast and slow explosives. And so most of the mass of the explosive—of the 8 tons of the weapon, probably 4 tons was the lenses, which didn’t count in accelerating the plutonium.
And so that was the Nagasaki bomb and the one that was tested in Alamogordo on July 16, 1945. [Teller] told me about his report, and that was the end of the conversation, except that he said what he really needed was a small experiment to prove to the most skeptical physicists that this was the way to build the atomic bomb and the hydrogen bomb. And I took that as a challenge. I started and tried to make a 20-kiloton experiment, but I couldn’t make one that was sufficiently convincing and decided to make it full scale. And so that’s what I did. I published my report of the Sausage based on the concepts current at the time. I wrote that up and published it in the classified report library, also on July 25th, 1951.
And it was detonated, as Teller says later, “exactly as Dick Garwin had devised it,” on November 1, 1952, so just 16 months afterwards. And it could never have been done faster. And the only way it got done that fast was because I wrote the paper all by myself. I was sitting in the office with Enrico Fermi. I had two offices: One was with Fermi in the theoretical division, and the other was in the physics division, where I was working on developing a means for accelerating deuterons and protons to 100 kilovolts.
Ivy Mike, detonated in the Marshall Islands on 1 November, 1952, was the first successful test of a full-scale thermonuclear device. Richard Garwin designed the bomb used in the test.Los Alamos National Laboratory/AP
Garwin took exception to my suggestion that Teller “entrusted” him with the design of the first thermonuclear bomb. He also revealed poignant details about the daily routine in his office, which he shared with Enrico Fermi.
Garwin: [Teller] challenged me. He didn’t entrust me. He didn’t know that it could be done. But he said, “I’d like a small experiment that would persuade the most skeptical and that this is the way to do it.” And it persuaded the person who counted—It persuaded [Los Alamos Director] Norris Bradbury, and Norris Bradbury [then allocated more resources for continued work on hydrogen bombs], without asking anybody else, because that’s how things worked then. Truman had said, “We’re going to build a hydrogen bomb,” and nobody knew how to build it. But Truman didn’t say that. People thought that Edward Teller probably knew how to build it. But he had been working on it since 1939, and he didn’t know how to build it, either. He continually complained that he didn’t have enough people. But any time, he could have written down his theorem and found out that it was wrong. But he found out it was wrong when he wrote down what he and Ulam had talked about.
When I sat in the room, it was a very small room, which had two desks, my desk faced Fermi’s desk. I could see him face-to-face. He taught me a lot the first year and the second year. But only the first year did I share an office with him. And that year, he worked with Stan Ulam in the mornings. The coders would come in. And they would deliver the code, the results of their work. They had been following a spreadsheet that Fermi had started. The first few lines he had actually calculated and sat next to the coder and calculated the first few lines of the spreadsheet, which were various zones along the axis of this infinitely long cylinder. And it started at one end, which was enriched with tritium. And so it reacted about 100 times as fast as deuterium itself. And so then, the second line across the spreadsheet would be the second set of zones along the axis. And the third line would be the third set of zones, and so on. And so they would come in with the whole thing, 100 zones, perhaps. And Fermi would discuss that with the coder, and then he would think of what to calculate next with Stan Ulam.
What’s it like to hold plutonium in your hands? Garwin is among the few people on the planet who can tell you.
Garwin: You can put plutonium in your pocket if it’s coated with nickel, as were the original plutonium hemispheres for the atomic bomb. I’ve held it in my two hands. It was a very dangerous thing to do. But at Los Alamos, you could be admitted to the sanctum and hold the nickel-plated plutonium in your hands. It’s warm, like a rabbit. And of course, if you isolate it, it gets even warmer.
Garwin spoke about the basic design of the Sausage, the first thermonuclear bomb. He disclosed how the device persuaded Hans Bethe, a star of the Manhattan Project and later, like Garwin, active in arms-control causes, about the viability of thermonuclear weapons. In a touching aside, he remembered his wife, Lois Garwin, to whom he was married from 1947 until her death in 2018.
Garwin: In order to get the Sausage to work, you needed to have a different way of getting equal forces on all sides. And that was the design of the full-size bomb. We used a normal atomic bomb at one end, and then, as has been revealed since then, a cylinder containing deuterium and surrounding that, a cylinder containing hydrogen. And beyond that, the very heavy container. All of that was at liquid-hydrogen temperature or at liquid-deuterium temperature. I can’t go into more detail at present, even now. So I made a full-size weapon. And it was very big. But I argued with Hans Bethe, who was head of various committees for building the atomic bomb, the hydrogen bomb, even though he didn’t want to build it. He wanted to prove that it couldn’t be built. But he was an honest man and excellent physicist. And so he accepted that it could be built.
But I never saw a weapons test, not even in Nevada. Never saw a weapons test. But I traveled to Hawaii a couple of times during the Ivy series and the George series in order to talk to people who came from the test site back to Hawaii to talk to me and others.
I want to mention my wife, Lois Garwin. I could not have done any of this without her. She died in 2018, February 4. And she was the one who took care of the children, except for waking up and diapering them or feeding them a bottle at night, because I could wake up and go to sleep much faster than she.
Garwin also weighed in on one of the most enduring controversies of nuclear-weapons history, which was the relative contributions of Teller and Ulam to the Teller-Ulam configuration. Garwin was asked, was it really the two of them working on this, or was it all Edward Teller?
Garwin: It was really all Edward Teller. I have volunteered that in various interviews. Ulam was a very good mathematician. But he was interested in things whether they were useful or not. He reminded me of Samuel Eilenberg, a mathematician at Columbia with whom I used to have lunch, together with I.I. Rabi and other Nobel Prize winners. Eilenberg used to say, “It’s like a tailor. Sometimes, you make a suit which has three sleeves, sometimes two sleeves, whatever looks best. Sometimes it’s useful, sometimes it’s not. And that’s mathematics for you.”
Why did Teller choose Garwin, a 23-year-old newly minted physicist, over the many staff physicists at Los Alamos to design that first hydrogen bomb?
Garwin: Well, he probably was influenced by something that I found out only in 1981. And that was in an article in Science magazine. Fermi had told people very publicly that I was the only true genius he had ever met. And it was too late to ask Fermi, who died in 1954. He had said that at a meeting at Fuller Lodge at lunchtime. That was the school for boys at Los Alamos that was taken over at the beginning of the atomic bomb program. I was not at that lunch. And he had said, “I’ve met the only true genius I have ever met.” And the people started preening themselves and so on, expecting to be named by him. And then Fermi said, “His name’s Dick Garwin.” And I suppose I had been working, at that time, on the hydrogen-bomb paper. Anyhow, so that’s all I know. Those people were very disappointed.
Garwin was asked which, of the many things he invented or helped invent during his career at IBM, he was most proud of. He did not hesitate in answering.
Garwin: Really the Cooley-Tukey algorithm because I was the midwife for that. I didn’t invent it. I just sat next to John Tukey. I sat next to Tukey so I could eat his dried prunes, with his permission. I had worked with him in 1953 to ’54 on the intelligence project of the Killian Committee. There were 67 consultants who were members of the Killian Committee. It was a very well-received report. And I worked for six months with NSA under William O Baker, who was the Vice President for research at Bell Labs. So I worked halftime there for six months. I can’t explain what we worked on. But I met all the people, Bill Friedman [legendary cryptanalyst William F. Friedman] and others.
I worked also with Jerry Wiesner. The first time I saw him, in the Lamp Light study, he said, “You know, Dick, you can either accomplish something or get credit for it, but not both.”
Richard Garwin, shown here in 1960, spent decades working at the IBM Research Laboratory in New York.AP
PSAC was the President’s Science Advisory Committee. If you look in Wikipedia, you’ll find many items for PSAC. It was formed in 1957 as part of the Killian Committee report. Eisenhower created PSAC, and Killian was the first head. [Editor’s note: James Rhyne Killian was the 10th president of MIT, from 1948 to 1959.] I had two terms on PSAC. One was with Kennedy, beginning January 21, 1961, and the other one was with Nixon, his second term.
When I came home from the PSAC meeting, a two-day meeting [in 1963], I wrote the person who was head of mathematics at Yorktown Heights. I was, at that time, head of the [IBM] Watson Scientific Laboratory, at Columbia University. And so I wrote to him and I said, “Can you find me a numerical analyst, somebody who can code this up and who will go to Princeton and talk with Tukey.” And he said, “Cooley is your man.”
Jim Cooley wasn’t enthusiastic about stopping what he was doing and going to Princeton. He needed additional influence. So I wrote him and I told him what it was that I wanted him to do. And in fact, the idea came not from Tukey, but it came from a colleague of his at [Government Communications Headquarters] in England [mathematician I. J. Good], a fellow whom we both knew from our days working on the Killian Committee. This was persuasive to Cooley, that his employment would be dependent upon his going to Princeton and talking to Tukey. So Cooley went to Princeton, and he talked to Tukey, and I don’t know for how long, but then he sat down and he wrote a Fortran program. And I then went to the IBM Science Advisory Committee, which was headed by Jerome Wiesner, and persuaded them to make this a free-for-everybody Fortran program rather than charging money for it. And so that was my additional contribution. And I then started distributing the program within a few months. I would send the name of the program and people could write in and get it. But mostly, they weren’t persuaded.
Garwin counted Enrico Fermi among his closest friends and associates. I asked him if there was something about Fermi that most people did not know, and that he wanted to share.
Garwin: He was a very ordinary-appearing person, but he had great round eyes. And in fact, I gave a talk at the IISS, the International Institute of Strategic Studies, meaning I was on the board for nine years while my daughter, Laura, was a graduate student at Oxford. And then at Cambridge, she was in the first batch of women Rhodes Scholars. And I saw somebody in the front row who looked very familiar. And I realized that it was his eyes. That he had Fermi’s eyes. It was Giulio Fermi, his son. I had met Giulio when he was 12 or 13 years old at their home in Chicago.
And I knew Nella, the older scion, who was a daughter. And I knew Laura, Laura Fermi, after whom our daughter, Laura, is named. She was a refined person. But Enrico was self-taught. There was an engineer, a friend of his father. His father, I think, worked for the railroad. The engineering friend would lend him books and Enrico would read them and learn them and learn the various languages involved. And he would solve the problems in the books, many of which were not easy, but nothing was too little for him or too big. He kept very good notes, in his laboratory notebook. And actually, he would write in my laboratory notebook at Los Alamos, and I would lock it up in the safe at night so that he wouldn’t have to do that.
There he would record the four shock equations and derive them and teach me how to do things like that. The degree to which he was self-taught… He organized the kids, the ragazzi, the kids of whatever street it was in Rome. And when he got to Rome from where he had gone to college and went to graduate school, he brought all kinds of people. Some of whom joined him in Los Alamos.
We were good friends of the Fermis in Los Alamos. And Lois and Jeffrey, my oldest boy, were with me in the summers. And then gradually, the other two children, Tom and Laura, joined us. So there were many years when we had rental homes in Los Alamos because people were always going away for the summer.
Garwin, an IEEE Life Fellow, is renowned for his ability to not only understand theory but also to put it to practical use. Nevertheless, his answer to the question of whether he considers himself an engineer or a physicist was surprising.
Garwin: I’m a physicist. I don’t think there’s a ranking. I just don’t know enough to qualify as an engineer.
All through the 41 years he was working for IBM, Garwin was also extraordinarily active on countless government committees and boards and also active in the nuclear arms–control movement. I asked him how he was able to maintain such an active and public professional life outside of IBM without raising eyebrows.
Garwin: My agreement with IBM was that they would not know what I was working on [outside of IBM]. They wouldn’t know what I testified about, and so on. And they signed an employment contract. Because otherwise, I knew that they would want to approve it. They would have lawyers saying, “Is this a good thing for us to do or not?” And so then when I started testifying in Congress and the testimony was public, I decided that I should tell IBM. I told them the same day. I gave them a copy of the testimony at the same time I gave it. I printed 100 copies, and we stood around the dining room table in our house, and whatever children were of suitable age would sort these things or unsort these things and staple them together and put them into a suitcase. One hundred copies of 10 pages of testimony is a lot to carry. And I would lug them to the airport for the early morning flight to Washington.
And IBM was as good as its word. I think they considered firing me a couple of times, but once I saw Manny Piori, who was the first director of research, and then he was head of the IBM Science Advisory Committee, and then he was various other things. Once I saw him furiously writing the head of IBM, who was, at that time, Thomas J. Watson Jr. And he was writing him to tell him that whatever they did, they shouldn’t fire me.
Not surprisingly, Garwin had strong opinions about the United States’ planned resumption of the production of plutonium, intended for a new missile warhead, the W87-1.
Garwin: That’s very bad, but it’s a matter of monkey see, monkey do. But it’s not that we need to do these things. It’s just that they don’t want to be caught short when the Russians resume testing or the Chinese resume testing. [The Chinese] have a lot to learn from their tests because they’ve had only 40-some total in history compared with the thousands or more, mostly underground. And underground tests up to 5 megatons, with the antiballistic missile warhead. So they don’t need to do this, but they don’t want to be caught short. And people say, “Look, the Russians are testing,” and the Russians have manufactured plutonium and you aren’t. They don’t really need new plutonium.
Somebody in Congress would say that the Russians are ahead of us, that they have these thousands of…they had 60,000 weapons at one time in 1962, I guess. And they could have those weapons that would destroy whole cities. And the United States doesn’t have a comparable number of weapons. They could destroy industrial centers. Anyhow, by appearance, the Americans would lose the race for appearance.
Not that the Russians could do anything useful with their weapons, but Russians aren’t constrained by logic. The Chinese, unfortunately, under Xi Jinping, have lost their way. Their way really is to make things and sell them to the world, even though the labor is transferring elsewhere as a matter of relative size and cost. But I had hoped that before Xi Jinping, beginning with Deng Xiaoping, I guess, that the Chinese would see the benefit of being supplier to the world. But now we have a kind of trade war with China. And I think that’s a big mistake for the United States. The United States ought to encourage China. And although we should set tariffs, the tariffs ought to be modest, in my opinion.
Garwin also had strong feelings about the surging funding for small, modular nuclear power reactors in the United States and elsewhere. I asked him if he thought these reactors were likely to succeed economically where larger reactors had not.
Garwin: The answer is no. I think that they won’t succeed because of economics. And the beginning of subsidy for those reactors, they want to subsidize the “valley of death.” But you can’t do that because there are many competing firms. You can subsidize all you want, but you can’t get out of the fact that they’re uneconomical. They’re not economical, and they will find out when they try to build them and when they exhaust the subsidy, and they can’t make it work. So no, I don’t think that they will work. I think it will put us in a large plutonium economy in order to breed plutonium and reprocess the material that accumulates in the reactors.
Does he think the only way to progress toward a carbon-neutral energy regime is with renewables?
Garwin: Yes. I think that using—I’m sorry to claim credit for something, but nobody has picked it up. A few years ago, I published a paper on green hydrogen and green ammonia. And the key is to use the steep trench up the west coast of the United States and Chile and the other countries in Latin America. Within 100 miles of the coast, there is this undersea trench where you can store hydrogen in ordinary weighted culverts. So it sits on the bottom. And it goes down to five kilometers or more. And all you need is one kilometer for 100 bar. You would have land-based solar and land-based wind turbines. They would be constantly electrolyzing either at surface or at depth, one way or another. So you would accumulate hydrogen. You would store it in a bladder which is held down by the negative buoyancy of the culverts. And it would just displace the water as you fill the bladder and as you empty the bladder during hydrogen usage, you would send the hydrogen back to shore. All that was worked out in my paper. And I even say how you would start by making green ammonia. You would get nitrogen from the atmosphere and then combine it with the hydrogen from the electrolysis. You can start small.
And you can just electrolyze and convert to ammonia and have it trucked away. And it would initially be sold at a high price, which would work because it would be sold for fertilizer. There’s a large market for fertilizer. And then when that gets saturated because of too much production, then you’d have to start using green ammonia for fuel anyhow.
An abridged version of this article appears in the September 2024 print issue as “5 Questions for Richard Garwin.”
-
IEEE President’s Note: Why Students Should Stay with IEEE
by Tom Coughlin on 01. September 2024. at 15:00
I would like our student members to know that IEEE is much more than just a club you join at school. It is an international community that can help students build and sustain successful careers as technical professionals after they graduate.
For more than 40 years, IEEE has been a great place to build my personal brand and to create a valuable professional network. I know it can do the same for the next generation of engineers.
By maintaining their membership after graduation and throughout their careers, students will have access to more resources—both professional and personal—that can help them advance within their field and discover new interests. They will also have the opportunity to build soft skills, raise their visibility, and make friendships that last a lifetime. All these resources and experiences, and more, will be of value in their life and work.
Lifelong professional home
The students of today likely will have many jobs across different organizations throughout their career. IEEE can be their lifelong professional home and help them meet their long-term technical and professional needs. The organization provides professional contacts and a community that offers support, advice, and mentoring, independent of where they work.
IEEE is here to support members at each stage of their professional journey. Within the organization, there are young people just starting to explore their professional passions. But there are also active engineers currently working in industry, government, and academia, as well as retired professionals who have a wealth of acquired knowledge and experience to share.
The students of today likely will have many jobs across different organizations throughout their career. IEEE can be their lifelong professional home and help them meet their long-term technical and professional needs.
Through its technical societies, IEEE also has a tremendous reach that spans areas including circuits, communications, computers, power and energy, semiconductors, and more. Whatever your focus is, there is a community within IEEE that will meet your needs.
For students and early-career practitioners, membership can help expand one’s professional network and elevate one’s professional image. It can lead to meaningful collaborative research opportunities that jump-start and advance one’s career, and it can provide professional-development pathways that refine skills through leadership opportunities.
Many of the key benefits of volunteering are also help with skills critical to continued professional success. Volunteering can advance your knowledge in all aspects of technology and science, provide opportunities to help guide the evolution of numerous fields, and network with others from around the world. Serving as a volunteer can also help empower members to champion their ideas and hone their communication and presentation skills, as well as management experience, which are important for professional development.
The future looks bright
There are great career opportunities in technology and engineering. The number of jobs in science, technology, engineering, and mathematics (STEM) fields is increasing, and many of them are lucrative, as they are essential components of today’s competitive global economy.
A career in these areas enables IEEE members to make a significant impact in the world, as STEM fields are imperative to solving some of the grand challenges facing society, such as climate change, cyberwarfare, and public health.
IEEE is committed to empowering and inspiring the next generation of engineering and technology leaders. Members, from students to retirees, will play a pivotal role in not only helping the organization continue to flourish but also in advancing technology for the benefit of humanity. They are the future of IEEE.
The best way to realize the true value of the organization is by engaging with IEEE colleagues. Get involved. Volunteer with your local section or engage with a society or technical community or work on an IEEE standard. Become an active participant—whether it be with affinity groups, humanitarian efforts, continuing education, or a standards working group and make IEEE your professional home.
—Tom Coughlin
IEEE president and CEO
This article appears in the September 2024 print issue as “Why Students Should Stay with IEEE.”
-
AI Has Created a Battle Over Web Crawling
by Eliza Strickland on 31. August 2024. at 13:00
Most people assume that generative AI will keep getting better and better; after all, that’s been the trend so far. And it may do so. But what some people don’t realize is that generative AI models are only as good as the ginormous data sets they’re trained on, and those data sets aren’t constructed from proprietary data owned by leading AI companies like OpenAI and Anthropic. Instead, they’re made up of public data that was created by all of us—anyone who’s ever written a blog post, posted a video, commented on a Reddit thread, or done basically anything else online.
A new report from the Data Provenance Initiative, a volunteer collective of AI researchers, shines a light on what’s happening with all that data. The report, “Consent in Crisis: The Rapid Decline of the AI Data Commons,” notes that a significant number of organizations that feel threatened by generative AI are taking measures to wall off their data. IEEE Spectrum spoke with Shayne Longpre, a lead researcher with the Data Provenance Initiative, about the report and its implications for AI companies.
Shayne Longpre on:
- How websites keep out web crawlers, and why
- Disappearing data and what it means for AI companies
- Synthetic data, peak data, and what happens next
The technology that websites use to keep out web crawlers isn’t new—the robot exclusion protocol was introduced in 1995. Can you explain what it is and why it suddenly became so relevant in the age of generative AI?
Shayne Longpre
Shayne Longpre: Robots.txt is a machine-readable file that crawlers—bots that navigate the web and record what they see—use to determine whether or not to crawl certain parts of a website. It became the de facto standard in the age where websites used it primarily for directing web search. So think of Bing or Google Search; they wanted to record this information so they could improve the experience of navigating users around the web. This was a very symbiotic relationship because web search operates by sending traffic to websites and websites want that. Generally speaking, most websites played well with most crawlers.
Let me next talk about a chain of claims that’s important to understand this. General-purpose AI models and their very impressive capabilities rely on the scale of data and compute that have been used to train them. Scale and data really matter, and there are very few sources that provide public scale like the web does. So many of the foundation models were trained on [data sets composed of] crawls of the web. Under these popular and important data sets are essentially just websites and the crawling infrastructure used to collect and package and process that data. Our study looks at not just the data sets, but the preference signals from the underlying websites. It’s the supply chain of the data itself.
But in the last year, a lot of websites have started using robots.txt to restrict bots, especially websites that are monetized with advertising and paywalls—so think news and artists. They’re particularly fearful, and maybe rightly so, that generative AI might impinge on their livelihoods. So they’re taking measures to protect their data.
When a site puts up robots.txt restrictions, it’s like putting up a no trespassing sign, right? It’s not enforceable. You have to trust that the crawlers will respect it.
Longpre: The tragedy of this is that robots.txt is machine-readable but does not appear to be legally enforceable. Whereas the terms of service may be legally enforceable but are not machine-readable. In the terms of service, they can articulate in natural language what the preferences are for the use of the data. So they can say things like, “You can use this data, but not commercially.” But in a robots.txt, you have to individually specify crawlers and then say which parts of the website you allow or disallow for them. This puts an undue burden on websites to figure out, among thousands of different crawlers, which ones correspond to uses they would like and which ones they wouldn’t like.
Do we know if crawlers generally do respect the restrictions in robots.txt?
Longpre: Many of the major companies have documentation that explicitly says what their rules or procedures are. In the case, for example, of Anthropic, they do say that they respect the robots.txt for ClaudeBot. However, many of these companies have also been in the news lately because they’ve been accused of not respecting robots.txt and crawling websites anyway. It isn’t clear from the outside why there’s a discrepancy between what AI companies say they do and what they’re being accused of doing. But a lot of the pro-social groups that use crawling—smaller startups, academics, nonprofits, journalists—they tend to respect robots.txt. They’re not the intended target of these restrictions, but they get blocked by them.
In the report, you looked at three training data sets that are often used to train generative AI systems, which were all created from web crawls in years past. You found that from 2023 to 2024, there was a very significant rise in the number of crawled domains that had since been restricted. Can you talk about those findings?
Longpre: What we found is that if you look at a particular data set, let’s take C4, which is very popular, created in 2019—in less than a year, about 5 percent of its data has been revoked if you respect or adhere to the preferences of the underlying websites. Now 5 percent doesn’t sound like a ton, but it is when you realize that this portion of the data mainly corresponds to the highest quality, most well-maintained, and freshest data. When we looked at the top 2,000 websites in this C4 data set—these are the top 2,000 by size, and they’re mostly news, large academic sites, social media, and well-curated high-quality websites—25 percent of the data in that top 2,000 has since been revoked. What this means is that the distribution of training data for models that respect robots.txt is rapidly shifting away from high-quality news, academic websites, forums, and social media to more organization and personal websites as well as e-commerce and blogs.
That seems like it could be a problem if we’re asking some future version of ChatGPT or Perplexity to answer complicated questions, and it’s taking the information from personal blogs and shopping sites.
Longpre: Exactly. It’s difficult to measure how this will affect models, but we suspect there will be a gap between the performance of models that respect robots.txt and the performance of models that have already secured this data and are willing to train on it anyway.
But the older data sets are still intact. Can AI companies just use the older data sets? What’s the downside of that?
Longpre: Well, continuous data freshness really matters. It also isn’t clear whether robots.txt can apply retroactively. Publishers would likely argue they do. So it depends on your appetite for lawsuits or where you also think that trends might go, especially in the U.S., with the ongoing lawsuits surrounding fair use of data. The prime example is obviously The New York Times against OpenAI and Microsoft, but there are now many variants. There’s a lot of uncertainty as to which way it will go.
The report is called “Consent in Crisis.” Why do you consider it a crisis?
Longpre: I think that it’s a crisis for data creators, because of the difficulty in expressing what they want with existing protocols. And also for some developers that are non-commercial and maybe not even related to AI—academics and researchers are finding that this data is becoming harder to access. And I think it’s also a crisis because it’s such a mess. The infrastructure was not designed to accommodate all of these different use cases at once. And it’s finally becoming a problem because of these huge industries colliding, with generative AI against news creators and others.
What can AI companies do if this continues, and more and more data is restricted? What would their moves be in order to keep training enormous models?
Longpre: The large companies will license it directly. It might not be a bad outcome for some of the large companies if a lot of this data is foreclosed or difficult to collect, it just creates a larger capital requirement for entry. I think big companies will invest more into the data collection pipeline and into gaining continuous access to valuable data sources that are user-generated, like YouTube and GitHub and Reddit. Acquiring exclusive access to those sites is probably an intelligent market play, but a problematic one from an antitrust perspective. I’m particularly concerned about the exclusive data acquisition relationships that might come out of this.
Do you think synthetic data can fill the gap?
Longpre: Big companies are already using synthetic data in large quantities. There are both fears and opportunities with synthetic data. On one hand, there have been a series of works that have demonstrated the potential for model collapse, which is the degradation of a model due to training on poor synthetic data that may appear more often on the web as more and more generative bots are let loose. However, I think it’s unlikely that large models will be hampered much because they have quality filters, so the poor quality or repetitive stuff can be siphoned out. And the opportunities of synthetic data are when it’s created in a lab environment to be very high quality, and it’s targeting particularly domains that are underdeveloped.
Do you give credence to the idea that we may be at peak data? Or do you feel like that’s an overblown concern?
Longpre: There is a lot of untapped data out there. But interestingly, a lot of it is hidden behind PDFs, so you need to do OCR [optical character recognition]. A lot of data is locked away in governments, in proprietary channels, in unstructured formats, or difficult to extract formats like PDFs. I think there’ll be a lot more investment in figuring out how to extract that data. I do think that in terms of easily available data, many companies are starting to hit walls and turning to synthetic data.
What’s the trend line here? Do you expect to see more websites putting up robots.txt restrictions in the coming years?
Longpre: We expect the restrictions to rise, both in robots.txt and in terms of service. Those trend lines are very clear from our work, but they could be affected by external factors such as legislation, companies themselves changing their policies, the outcome of lawsuits, as well as community pressure from writers’ guilds and things like that. And I expect that the increased commoditization of data is going to cause more of a battlefield in this space.
What would you like to see happen in terms of either standardization within the industry to making it easier for websites to express preferences about crawling?
Longpre: At the Data Province Initiative, we definitely hope that new standards will emerge and be adopted to allow creators to express their preferences in a more granular way around the uses of their data. That would make the burden much easier on them. I think that’s a no-brainer and a win-win. But it’s not clear whose job it is to create or enforce these standards. It would be amazing if the [AI] companies themselves could come to this conclusion and do it. But the designer of the standard will almost inevitably have some bias towards their own use, especially if it’s a corporate entity.
It’s also the case that preferences shouldn’t be respected in all cases. For instance, I don’t think that academics or journalists doing prosocial research should necessarily be foreclosed from accessing data with machines that is already public, on websites that anyone could go visit themselves. Not all data is created equal and not all uses are created equal.
-
Was an AI Image Generator Taken Down for Making Child Porn?
by David Evan Harris on 30. August 2024. at 20:04
Why are AI companies valued in the millions and billions of dollars creating and distributing tools that can make AI-generated child sexual abuse material (CSAM)?
An image generator called Stable Diffusion version 1.5, which was created by the AI company Runway with funding from Stability AI, has been particularly implicated in the production of CSAM. And popular platforms such as Hugging Face and Civitai have been hosting that model and others that may have been trained on real images of child sexual abuse. In some cases, companies may even be breaking laws by hosting synthetic CSAM material on their servers. And why are mainstream companies and investors like Amazon, Google, Nvidia, Intel, Salesforce, and Andreessen Horowitz pumping hundreds of millions of dollars into these companies? Their support amounts to subsidizing content for pedophiles.
As AI safety experts, we’ve been asking these questions to call out these companies and pressure them to take the corrective actions we outline below. And we’re happy today to report one major triumph: seemingly in response to our questions, Stable Diffusion version 1.5 has been removed from Hugging Face. But there’s much still to do, and meaningful progress may require legislation.
The Scope of the CSAM Problem
Child safety advocates began ringing the alarm bell last year: Researchers at Stanford’s Internet Observatory and the technology non-profit Thorn published a troubling report in June 2023. They found that broadly available and “open-source” AI image-generation tools were already being misused by malicious actors to make child sexual abuse material. In some cases, bad actors were making their own custom versions of these models (a process known as fine-tuning) with real child sexual abuse material to generate bespoke images of specific victims.
Last October, a report from the U.K. nonprofit Internet Watch Foundation (which collects reports of child sexual abuse material) detailed the ease with which malicious actors are now making photorealistic AI-generated child sexual abuse material, at scale. The researchers included a “snapshot” study of one dark web CSAM forum, analyzing more than 11,000 AI-generated images posted in a one-month period; of those, nearly 3,000 were judged severe enough to be classified as criminal. The report urged stronger regulatory oversight of generative AI models.
AI models can be used to create this material because they’ve seen examples before. Researchers at Stanford discovered last December that one of the most significant data sets used to train image-generation models included hundreds of pieces of CSAM. Many of the most popular downloadable open-source AI image generators, including the popular Stable Diffusion version 1.5 model, were trained using this data. While Runway created that version of Stable Diffusion, Stability AI paid for the computing power to produce the dataset and train the model, and Stability AI released the subsequent versions.
Runway did not respond to a request for comment. A Stability AI spokesperson emphasized that the company did not release or maintain Stable Diffusion version 1.5, and says the company has “implemented robust safeguards” against CSAM in subsequent models, including the use of filtered data sets for training.
Also last December, researchers at the social media analytics firm Graphika found a proliferation of dozens of “undressing” services, many based on open-source AI image generators, likely including Stable Diffusion. These services allow users to upload clothed pictures of people and produce what experts term nonconsensual intimate imagery (NCII) of both minors and adults, also sometimes referred to as deepfake pornography. Such websites can be easily found through Google searches, and users can pay for the services using credit cards online. Many of these services only work on women and girls, and these types of tools have been used to target female celebrities like Taylor Swift and politicians like U.S. representative Alexandria Ocasio-Cortez.
AI-generated CSAM has real effects. The child safety ecosystem is already overtaxed, with millions of files of suspected CSAM reported to hotlines annually. Anything that adds to that torrent of content—especially photorealistic abuse material—makes it more difficult to find children that are actively in harm’s way. Making matters worse, some bad actors are using existing CSAM to generate synthetic images of these survivors—a horrific re-violation of their rights. Others are using the readily available “nudifying” apps to create sexual content from benign imagery of real children, and then using that newly generated content in sexual extortion schemes.
One Victory Against AI-Generated CSAM
Based on the Stanford investigation from last December, it’s well-known in the AI community that Stable Diffusion 1.5 was trained on child sexual abuse material, as was every other model trained on the LAION-5B data set. These models are being actively misused by malicious actors to make AI-generated CSAM. And even when they’re used to generate more benign material, their use inherently revictimizes the children whose abuse images went into their training data. So we asked the popular AI hosting platforms Hugging Face and Civitai why they hosted Stable Diffusion 1.5 and derivative models, making them available for free download?
It’s worth noting that Jeff Allen, a data scientist at the Integrity Institute, found that Stable Diffusion 1.5 was downloaded from Hugging Face over 6 million times in the past month, making it the most popular AI image-generator on the platform.
When we asked Hugging Face why it has continued to host the model, company spokesperson Brigitte Tousignant did not directly answer the question, but instead stated that the company doesn’t tolerate CSAM on its platform, that it incorporates a variety of safety tools, and that it encourages the community to use the Safe Stable Diffusion model that identifies and suppresses inappropriate images.
Then, yesterday, we checked Hugging Face and found that Stable Diffusion 1.5 is no longer available. Tousignant told us that Hugging Face didn’t take it down, and suggested that we contact Runway—which we did, again, but we have not yet received a response.
It’s undoubtedly a success that this model is no longer available for download from Hugging Face. Unfortunately, it’s still available on Civitai, as are hundreds of derivative models. When we contacted Civitai, a spokesperson told us that they have no knowledge of what training data Stable Diffusion 1.5 used, and that they would only take it down if there was evidence of misuse.
Platforms should be getting nervous about their liability. This past week saw the arrest of Pavel Durov, CEO of the messaging app Telegram, as part of an investigation related to CSAM and other crimes.
What’s Being Done About AI-Generated CSAM
The steady drumbeat of disturbing reports and news about AI-generated CSAM and NCII hasn’t let up. While some companies are trying to improve their products’ safety with the help of the Tech Coalition, what progress have we seen on the broader issue?
In April, Thorn and All Tech Is Human announced an initiative to bring together mainstream tech companies, generative AI developers, model hosting platforms, and more to define and commit to Safety by Design principles, which put preventing child sexual abuse at the center of the product development process. Ten companies (including Amazon, Civitai, Google, Meta, Microsoft, OpenAI, and Stability AI) committed to these principles, and some also co-authored a related paper with more detailed recommended mitigations. The principles call on companies to develop, deploy, and maintain AI models that proactively address child safety risks; to build systems to ensure that any abuse material that does get produced is reliably detected; and to limit the distribution of the underlying models and services that are used to make this abuse material.
These kinds of voluntary commitments are a start. Rebecca Portnoff, Thorn’s head of data science, says the initiative seeks accountability by requiring companies to issue reports about their progress on the mitigation steps. It’s also collaborating with standard-setting institutions such as IEEE and NIST to integrate their efforts into new and existing standards, opening the door to third party audits that would “move past the honor system,” Portnoff says. Portnoff also notes that Thorn is engaging with policy makers to help them conceive legislation that would be both technically feasible and impactful. Indeed, many experts say it’s time to move beyond voluntary commitments.
We believe that there is a reckless race to the bottom currently underway in the AI industry. Companies are so furiously fighting to be technically in the lead that many of them are ignoring the ethical and possibly even legal consequences of their products. While some governments—including the European Union—are making headway on regulating AI, they haven’t gone far enough. If, for example, laws made it illegal to provide AI systems that can produce CSAM, tech companies might take notice.
The reality is that while some companies will abide by voluntary commitments, many will not. And of those that do, many will take action too slowly, either because they’re not ready or because they’re struggling to keep their competitive advantage. In the meantime, bad actors will gravitate to those services and wreak havoc. That outcome is unacceptable.
What Tech Companies Should Do About AI-Generated CSAM
Experts saw this problem coming from a mile away, and child safety advocates have recommended common-sense strategies to combat it. If we miss this opportunity to do something to fix the situation, we’ll all bear the responsibility. At a minimum, all companies, including those releasing open source models, should be legally required to follow the commitments laid out in Thorn’s Safety by Design principles:
- Detect, remove, and report CSAM from their training data sets before training their generative AI models.
- Incorporate robust watermarks and content provenance systems into their generative AI models so generated images can be linked to the models that created them, as would be required under a California bill that would create Digital Content Provenance Standards for companies that do business in the state. The bill will likely be up for signature by Governor Gavin Newson in the coming month.
- Remove from their platforms any generative AI models that are known to be trained on CSAM or that are capable of producing CSAM. Refuse to rehost these models unless they’ve been fully reconstituted with the CSAM removed.
- Identify models that have been intentionally fine-tuned on CSAM and permanently remove them from their platforms.
- Remove “nudifying” apps from app stores, block search results for these tools and services, and work with payment providers to block payments to their makers.
There is no reason why generative AI needs to aid and abet the horrific abuse of children. But we will need all tools at hand—voluntary commitments, regulation, and public pressure—to change course and stop the race to the bottom.
The authors thank Rebecca Portnoff of Thorn, David Thiel of the Stanford Internet Observatory, Jeff Allen of the Integrity Institute, Ravit Dotan of TechBetter, and the tech policy researcher Owen Doyle for their help with this article.
-
Andrew Ng: Unbiggen AI
by Eliza Strickland on 09. February 2022. at 15:31
Andrew Ng has serious street cred in artificial intelligence. He pioneered the use of graphics processing units (GPUs) to train deep learning models in the late 2000s with his students at Stanford University, cofounded Google Brain in 2011, and then served for three years as chief scientist for Baidu, where he helped build the Chinese tech giant’s AI group. So when he says he has identified the next big shift in artificial intelligence, people listen. And that’s what he told IEEE Spectrum in an exclusive Q&A.
Ng’s current efforts are focused on his company Landing AI, which built a platform called LandingLens to help manufacturers improve visual inspection with computer vision. He has also become something of an evangelist for what he calls the data-centric AI movement, which he says can yield “small data” solutions to big issues in AI, including model efficiency, accuracy, and bias.
Andrew Ng on...
- What’s next for really big models
- The career advice he didn’t listen to
- Defining the data-centric AI movement
- Synthetic data
- Why Landing AI asks its customers to do the work
The great advances in deep learning over the past decade or so have been powered by ever-bigger models crunching ever-bigger amounts of data. Some people argue that that’s an unsustainable trajectory. Do you agree that it can’t go on that way?
Andrew Ng: This is a big question. We’ve seen foundation models in NLP [natural language processing]. I’m excited about NLP models getting even bigger, and also about the potential of building foundation models in computer vision. I think there’s lots of signal to still be exploited in video: We have not been able to build foundation models yet for video because of compute bandwidth and the cost of processing video, as opposed to tokenized text. So I think that this engine of scaling up deep learning algorithms, which has been running for something like 15 years now, still has steam in it. Having said that, it only applies to certain problems, and there’s a set of other problems that need small data solutions.
When you say you want a foundation model for computer vision, what do you mean by that?
Ng: This is a term coined by Percy Liang and some of my friends at Stanford to refer to very large models, trained on very large data sets, that can be tuned for specific applications. For example, GPT-3 is an example of a foundation model [for NLP]. Foundation models offer a lot of promise as a new paradigm in developing machine learning applications, but also challenges in terms of making sure that they’re reasonably fair and free from bias, especially if many of us will be building on top of them.
What needs to happen for someone to build a foundation model for video?
Ng: I think there is a scalability problem. The compute power needed to process the large volume of images for video is significant, and I think that’s why foundation models have arisen first in NLP. Many researchers are working on this, and I think we’re seeing early signs of such models being developed in computer vision. But I’m confident that if a semiconductor maker gave us 10 times more processor power, we could easily find 10 times more video to build such models for vision.
Having said that, a lot of what’s happened over the past decade is that deep learning has happened in consumer-facing companies that have large user bases, sometimes billions of users, and therefore very large data sets. While that paradigm of machine learning has driven a lot of economic value in consumer software, I find that that recipe of scale doesn’t work for other industries.
It’s funny to hear you say that, because your early work was at a consumer-facing company with millions of users.
Ng: Over a decade ago, when I proposed starting the Google Brain project to use Google’s compute infrastructure to build very large neural networks, it was a controversial step. One very senior person pulled me aside and warned me that starting Google Brain would be bad for my career. I think he felt that the action couldn’t just be in scaling up, and that I should instead focus on architecture innovation.
“In many industries where giant data sets simply don’t exist, I think the focus has to shift from big data to good data. Having 50 thoughtfully engineered examples can be sufficient to explain to the neural network what you want it to learn.”
—Andrew Ng, CEO & Founder, Landing AII remember when my students and I published the first NeurIPS workshop paper advocating using CUDA, a platform for processing on GPUs, for deep learning—a different senior person in AI sat me down and said, “CUDA is really complicated to program. As a programming paradigm, this seems like too much work.” I did manage to convince him; the other person I did not convince.
I expect they’re both convinced now.
Ng: I think so, yes.
Over the past year as I’ve been speaking to people about the data-centric AI movement, I’ve been getting flashbacks to when I was speaking to people about deep learning and scalability 10 or 15 years ago. In the past year, I’ve been getting the same mix of “there’s nothing new here” and “this seems like the wrong direction.”
How do you define data-centric AI, and why do you consider it a movement?
Ng: Data-centric AI is the discipline of systematically engineering the data needed to successfully build an AI system. For an AI system, you have to implement some algorithm, say a neural network, in code and then train it on your data set. The dominant paradigm over the last decade was to download the data set while you focus on improving the code. Thanks to that paradigm, over the last decade deep learning networks have improved significantly, to the point where for a lot of applications the code—the neural network architecture—is basically a solved problem. So for many practical applications, it’s now more productive to hold the neural network architecture fixed, and instead find ways to improve the data.
When I started speaking about this, there were many practitioners who, completely appropriately, raised their hands and said, “Yes, we’ve been doing this for 20 years.” This is the time to take the things that some individuals have been doing intuitively and make it a systematic engineering discipline.
The data-centric AI movement is much bigger than one company or group of researchers. My collaborators and I organized a data-centric AI workshop at NeurIPS, and I was really delighted at the number of authors and presenters that showed up.
You often talk about companies or institutions that have only a small amount of data to work with. How can data-centric AI help them?
Ng: You hear a lot about vision systems built with millions of images—I once built a face recognition system using 350 million images. Architectures built for hundreds of millions of images don’t work with only 50 images. But it turns out, if you have 50 really good examples, you can build something valuable, like a defect-inspection system. In many industries where giant data sets simply don’t exist, I think the focus has to shift from big data to good data. Having 50 thoughtfully engineered examples can be sufficient to explain to the neural network what you want it to learn.
When you talk about training a model with just 50 images, does that really mean you’re taking an existing model that was trained on a very large data set and fine-tuning it? Or do you mean a brand new model that’s designed to learn only from that small data set?
Ng: Let me describe what Landing AI does. When doing visual inspection for manufacturers, we often use our own flavor of RetinaNet. It is a pretrained model. Having said that, the pretraining is a small piece of the puzzle. What’s a bigger piece of the puzzle is providing tools that enable the manufacturer to pick the right set of images [to use for fine-tuning] and label them in a consistent way. There’s a very practical problem we’ve seen spanning vision, NLP, and speech, where even human annotators don’t agree on the appropriate label. For big data applications, the common response has been: If the data is noisy, let’s just get a lot of data and the algorithm will average over it. But if you can develop tools that flag where the data’s inconsistent and give you a very targeted way to improve the consistency of the data, that turns out to be a more efficient way to get a high-performing system.
“Collecting more data often helps, but if you try to collect more data for everything, that can be a very expensive activity.”
—Andrew NgFor example, if you have 10,000 images where 30 images are of one class, and those 30 images are labeled inconsistently, one of the things we do is build tools to draw your attention to the subset of data that’s inconsistent. So you can very quickly relabel those images to be more consistent, and this leads to improvement in performance.
Could this focus on high-quality data help with bias in data sets? If you’re able to curate the data more before training?
Ng: Very much so. Many researchers have pointed out that biased data is one factor among many leading to biased systems. There have been many thoughtful efforts to engineer the data. At the NeurIPS workshop, Olga Russakovsky gave a really nice talk on this. At the main NeurIPS conference, I also really enjoyed Mary Gray’s presentation, which touched on how data-centric AI is one piece of the solution, but not the entire solution. New tools like Datasheets for Datasets also seem like an important piece of the puzzle.
One of the powerful tools that data-centric AI gives us is the ability to engineer a subset of the data. Imagine training a machine-learning system and finding that its performance is okay for most of the data set, but its performance is biased for just a subset of the data. If you try to change the whole neural network architecture to improve the performance on just that subset, it’s quite difficult. But if you can engineer a subset of the data you can address the problem in a much more targeted way.
When you talk about engineering the data, what do you mean exactly?
Ng: In AI, data cleaning is important, but the way the data has been cleaned has often been in very manual ways. In computer vision, someone may visualize images through a Jupyter notebook and maybe spot the problem, and maybe fix it. But I’m excited about tools that allow you to have a very large data set, tools that draw your attention quickly and efficiently to the subset of data where, say, the labels are noisy. Or to quickly bring your attention to the one class among 100 classes where it would benefit you to collect more data. Collecting more data often helps, but if you try to collect more data for everything, that can be a very expensive activity.
For example, I once figured out that a speech-recognition system was performing poorly when there was car noise in the background. Knowing that allowed me to collect more data with car noise in the background, rather than trying to collect more data for everything, which would have been expensive and slow.
What about using synthetic data, is that often a good solution?
Ng: I think synthetic data is an important tool in the tool chest of data-centric AI. At the NeurIPS workshop, Anima Anandkumar gave a great talk that touched on synthetic data. I think there are important uses of synthetic data that go beyond just being a preprocessing step for increasing the data set for a learning algorithm. I’d love to see more tools to let developers use synthetic data generation as part of the closed loop of iterative machine learning development.
Do you mean that synthetic data would allow you to try the model on more data sets?
Ng: Not really. Here’s an example. Let’s say you’re trying to detect defects in a smartphone casing. There are many different types of defects on smartphones. It could be a scratch, a dent, pit marks, discoloration of the material, other types of blemishes. If you train the model and then find through error analysis that it’s doing well overall but it’s performing poorly on pit marks, then synthetic data generation allows you to address the problem in a more targeted way. You could generate more data just for the pit-mark category.
“In the consumer software Internet, we could train a handful of machine-learning models to serve a billion users. In manufacturing, you might have 10,000 manufacturers building 10,000 custom AI models.”
—Andrew NgSynthetic data generation is a very powerful tool, but there are many simpler tools that I will often try first. Such as data augmentation, improving labeling consistency, or just asking a factory to collect more data.
To make these issues more concrete, can you walk me through an example? When a company approaches Landing AI and says it has a problem with visual inspection, how do you onboard them and work toward deployment?
Ng: When a customer approaches us we usually have a conversation about their inspection problem and look at a few images to verify that the problem is feasible with computer vision. Assuming it is, we ask them to upload the data to the LandingLens platform. We often advise them on the methodology of data-centric AI and help them label the data.
One of the foci of Landing AI is to empower manufacturing companies to do the machine learning work themselves. A lot of our work is making sure the software is fast and easy to use. Through the iterative process of machine learning development, we advise customers on things like how to train models on the platform, when and how to improve the labeling of data so the performance of the model improves. Our training and software supports them all the way through deploying the trained model to an edge device in the factory.
How do you deal with changing needs? If products change or lighting conditions change in the factory, can the model keep up?
Ng: It varies by manufacturer. There is data drift in many contexts. But there are some manufacturers that have been running the same manufacturing line for 20 years now with few changes, so they don’t expect changes in the next five years. Those stable environments make things easier. For other manufacturers, we provide tools to flag when there’s a significant data-drift issue. I find it really important to empower manufacturing customers to correct data, retrain, and update the model. Because if something changes and it’s 3 a.m. in the United States, I want them to be able to adapt their learning algorithm right away to maintain operations.
In the consumer software Internet, we could train a handful of machine-learning models to serve a billion users. In manufacturing, you might have 10,000 manufacturers building 10,000 custom AI models. The challenge is, how do you do that without Landing AI having to hire 10,000 machine learning specialists?
So you’re saying that to make it scale, you have to empower customers to do a lot of the training and other work.
Ng: Yes, exactly! This is an industry-wide problem in AI, not just in manufacturing. Look at health care. Every hospital has its own slightly different format for electronic health records. How can every hospital train its own custom AI model? Expecting every hospital’s IT personnel to invent new neural-network architectures is unrealistic. The only way out of this dilemma is to build tools that empower the customers to build their own models by giving them tools to engineer the data and express their domain knowledge. That’s what Landing AI is executing in computer vision, and the field of AI needs other teams to execute this in other domains.
Is there anything else you think it’s important for people to understand about the work you’re doing or the data-centric AI movement?
Ng: In the last decade, the biggest shift in AI was a shift to deep learning. I think it’s quite possible that in this decade the biggest shift will be to data-centric AI. With the maturity of today’s neural network architectures, I think for a lot of the practical applications the bottleneck will be whether we can efficiently get the data we need to develop systems that work well. The data-centric AI movement has tremendous energy and momentum across the whole community. I hope more researchers and developers will jump in and work on it.
This article appears in the April 2022 print issue as “Andrew Ng, AI Minimalist.”
-
How AI Will Change Chip Design
by Rina Diane Caballar on 08. February 2022. at 14:00
The end of Moore’s Law is looming. Engineers and designers can do only so much to miniaturize transistors and pack as many of them as possible into chips. So they’re turning to other approaches to chip design, incorporating technologies like AI into the process.
Samsung, for instance, is adding AI to its memory chips to enable processing in memory, thereby saving energy and speeding up machine learning. Speaking of speed, Google’s TPU V4 AI chip has doubled its processing power compared with that of its previous version.
But AI holds still more promise and potential for the semiconductor industry. To better understand how AI is set to revolutionize chip design, we spoke with Heather Gorr, senior product manager for MathWorks’ MATLAB platform.
How is AI currently being used to design the next generation of chips?
Heather Gorr: AI is such an important technology because it’s involved in most parts of the cycle, including the design and manufacturing process. There’s a lot of important applications here, even in the general process engineering where we want to optimize things. I think defect detection is a big one at all phases of the process, especially in manufacturing. But even thinking ahead in the design process, [AI now plays a significant role] when you’re designing the light and the sensors and all the different components. There’s a lot of anomaly detection and fault mitigation that you really want to consider.
Heather GorrMathWorks
Then, thinking about the logistical modeling that you see in any industry, there is always planned downtime that you want to mitigate; but you also end up having unplanned downtime. So, looking back at that historical data of when you’ve had those moments where maybe it took a bit longer than expected to manufacture something, you can take a look at all of that data and use AI to try to identify the proximate cause or to see something that might jump out even in the processing and design phases. We think of AI oftentimes as a predictive tool, or as a robot doing something, but a lot of times you get a lot of insight from the data through AI.
What are the benefits of using AI for chip design?
Gorr: Historically, we’ve seen a lot of physics-based modeling, which is a very intensive process. We want to do a reduced order model, where instead of solving such a computationally expensive and extensive model, we can do something a little cheaper. You could create a surrogate model, so to speak, of that physics-based model, use the data, and then do your parameter sweeps, your optimizations, your Monte Carlo simulations using the surrogate model. That takes a lot less time computationally than solving the physics-based equations directly. So, we’re seeing that benefit in many ways, including the efficiency and economy that are the results of iterating quickly on the experiments and the simulations that will really help in the design.
So it’s like having a digital twin in a sense?
Gorr: Exactly. That’s pretty much what people are doing, where you have the physical system model and the experimental data. Then, in conjunction, you have this other model that you could tweak and tune and try different parameters and experiments that let sweep through all of those different situations and come up with a better design in the end.
So, it’s going to be more efficient and, as you said, cheaper?
Gorr: Yeah, definitely. Especially in the experimentation and design phases, where you’re trying different things. That’s obviously going to yield dramatic cost savings if you’re actually manufacturing and producing [the chips]. You want to simulate, test, experiment as much as possible without making something using the actual process engineering.
We’ve talked about the benefits. How about the drawbacks?
Gorr: The [AI-based experimental models] tend to not be as accurate as physics-based models. Of course, that’s why you do many simulations and parameter sweeps. But that’s also the benefit of having that digital twin, where you can keep that in mind—it’s not going to be as accurate as that precise model that we’ve developed over the years.
Both chip design and manufacturing are system intensive; you have to consider every little part. And that can be really challenging. It’s a case where you might have models to predict something and different parts of it, but you still need to bring it all together.
One of the other things to think about too is that you need the data to build the models. You have to incorporate data from all sorts of different sensors and different sorts of teams, and so that heightens the challenge.
How can engineers use AI to better prepare and extract insights from hardware or sensor data?
Gorr: We always think about using AI to predict something or do some robot task, but you can use AI to come up with patterns and pick out things you might not have noticed before on your own. People will use AI when they have high-frequency data coming from many different sensors, and a lot of times it’s useful to explore the frequency domain and things like data synchronization or resampling. Those can be really challenging if you’re not sure where to start.
One of the things I would say is, use the tools that are available. There’s a vast community of people working on these things, and you can find lots of examples [of applications and techniques] on GitHub or MATLAB Central, where people have shared nice examples, even little apps they’ve created. I think many of us are buried in data and just not sure what to do with it, so definitely take advantage of what’s already out there in the community. You can explore and see what makes sense to you, and bring in that balance of domain knowledge and the insight you get from the tools and AI.
What should engineers and designers consider when using AI for chip design?
Gorr: Think through what problems you’re trying to solve or what insights you might hope to find, and try to be clear about that. Consider all of the different components, and document and test each of those different parts. Consider all of the people involved, and explain and hand off in a way that is sensible for the whole team.
How do you think AI will affect chip designers’ jobs?
Gorr: It’s going to free up a lot of human capital for more advanced tasks. We can use AI to reduce waste, to optimize the materials, to optimize the design, but then you still have that human involved whenever it comes to decision-making. I think it’s a great example of people and technology working hand in hand. It’s also an industry where all people involved—even on the manufacturing floor—need to have some level of understanding of what’s happening, so this is a great industry for advancing AI because of how we test things and how we think about them before we put them on the chip.
How do you envision the future of AI and chip design?
Gorr: It’s very much dependent on that human element—involving people in the process and having that interpretable model. We can do many things with the mathematical minutiae of modeling, but it comes down to how people are using it, how everybody in the process is understanding and applying it. Communication and involvement of people of all skill levels in the process are going to be really important. We’re going to see less of those superprecise predictions and more transparency of information, sharing, and that digital twin—not only using AI but also using our human knowledge and all of the work that many people have done over the years.
-
Atomically Thin Materials Significantly Shrink Qubits
by Dexter Johnson on 07. February 2022. at 16:12
Quantum computing is a devilishly complex technology, with many technical hurdles impacting its development. Of these challenges two critical issues stand out: miniaturization and qubit quality.
IBM has adopted the superconducting qubit road map of reaching a 1,121-qubit processor by 2023, leading to the expectation that 1,000 qubits with today’s qubit form factor is feasible. However, current approaches will require very large chips (50 millimeters on a side, or larger) at the scale of small wafers, or the use of chiplets on multichip modules. While this approach will work, the aim is to attain a better path toward scalability.
Now researchers at MIT have been able to both reduce the size of the qubits and done so in a way that reduces the interference that occurs between neighboring qubits. The MIT researchers have increased the number of superconducting qubits that can be added onto a device by a factor of 100.
“We are addressing both qubit miniaturization and quality,” said William Oliver, the director for the Center for Quantum Engineering at MIT. “Unlike conventional transistor scaling, where only the number really matters, for qubits, large numbers are not sufficient, they must also be high-performance. Sacrificing performance for qubit number is not a useful trade in quantum computing. They must go hand in hand.”
The key to this big increase in qubit density and reduction of interference comes down to the use of two-dimensional materials, in particular the 2D insulator hexagonal boron nitride (hBN). The MIT researchers demonstrated that a few atomic monolayers of hBN can be stacked to form the insulator in the capacitors of a superconducting qubit.
Just like other capacitors, the capacitors in these superconducting circuits take the form of a sandwich in which an insulator material is sandwiched between two metal plates. The big difference for these capacitors is that the superconducting circuits can operate only at extremely low temperatures—less than 0.02 degrees above absolute zero (-273.15 °C).
Superconducting qubits are measured at temperatures as low as 20 millikelvin in a dilution refrigerator.Nathan Fiske/MIT
In that environment, insulating materials that are available for the job, such as PE-CVD silicon oxide or silicon nitride, have quite a few defects that are too lossy for quantum computing applications. To get around these material shortcomings, most superconducting circuits use what are called coplanar capacitors. In these capacitors, the plates are positioned laterally to one another, rather than on top of one another.
As a result, the intrinsic silicon substrate below the plates and to a smaller degree the vacuum above the plates serve as the capacitor dielectric. Intrinsic silicon is chemically pure and therefore has few defects, and the large size dilutes the electric field at the plate interfaces, all of which leads to a low-loss capacitor. The lateral size of each plate in this open-face design ends up being quite large (typically 100 by 100 micrometers) in order to achieve the required capacitance.
In an effort to move away from the large lateral configuration, the MIT researchers embarked on a search for an insulator that has very few defects and is compatible with superconducting capacitor plates.
“We chose to study hBN because it is the most widely used insulator in 2D material research due to its cleanliness and chemical inertness,” said colead author Joel Wang, a research scientist in the Engineering Quantum Systems group of the MIT Research Laboratory for Electronics.
On either side of the hBN, the MIT researchers used the 2D superconducting material, niobium diselenide. One of the trickiest aspects of fabricating the capacitors was working with the niobium diselenide, which oxidizes in seconds when exposed to air, according to Wang. This necessitates that the assembly of the capacitor occur in a glove box filled with argon gas.
While this would seemingly complicate the scaling up of the production of these capacitors, Wang doesn’t regard this as a limiting factor.
“What determines the quality factor of the capacitor are the two interfaces between the two materials,” said Wang. “Once the sandwich is made, the two interfaces are “sealed” and we don’t see any noticeable degradation over time when exposed to the atmosphere.”
This lack of degradation is because around 90 percent of the electric field is contained within the sandwich structure, so the oxidation of the outer surface of the niobium diselenide does not play a significant role anymore. This ultimately makes the capacitor footprint much smaller, and it accounts for the reduction in cross talk between the neighboring qubits.
“The main challenge for scaling up the fabrication will be the wafer-scale growth of hBN and 2D superconductors like [niobium diselenide], and how one can do wafer-scale stacking of these films,” added Wang.
Wang believes that this research has shown 2D hBN to be a good insulator candidate for superconducting qubits. He says that the groundwork the MIT team has done will serve as a road map for using other hybrid 2D materials to build superconducting circuits.