Blogs

I attended this year’s IEEE Vision, Innovation, and Challenges Summit and Honors Ceremony on 23 and 24 April at the Hilton Tokyo Odaiba hotel. The event celebrates pioneers in engineering who have developed technology that changes the way people connect and learn about the world. This year’s celebrants included the engineers behind the first digital cable set-top box modem chipset and the James Webb Space Telescope.The event included the inaugural IEEE Distinguished Young Professionals and Laureate Forum. Fifty young professionals attended the networking event with IEEE leaders, IEEE Medal of Honor laureates, and award recipients.Here are highlights of the sessions, which are available to watch in full on IEEE.tv.Networking opportunities for young professionalsBefore the VIC summit got underway on 23 April, the networking forum took place that morning. In her speech welcoming the attendees, Sophie Muirhead, IEEE executive director and chief operating officer, encouraged the young professionals to engage in IEEE’s mission of developing technology for the benefit of humanity.The participants heard from 2020 IEEE President Toshio Fukuda and award recipient Aishwarya Bandla about their careers and volunteer work. Bandla received this year’s IEEE Theodore W. Hissey Young Professionals Award for her “leadership in patient-centric health technology innovation and inspiring IEEE young professionals to drive meaningful change.” The award is sponsored by the IEEE Photonics and IEEE Power & Energy societies, as well as IEEE Young Professionals.She is an IEEE senior member and the clinical innovation manager at Paxman, a medical equipment manufacturer headquartered in Huddersfield, England. She is developing a wearable device that cools a person’s limbs. The Paxman limb “cryocompression” system helps prevent nerve damage associated with certain types of intravenous chemotherapy.As someone who follows the Japanese concept of ikigai—a sense of purpose—Bandla said her “passion and profession intersected not at technology in the lab but at bringing technology to the people.”She shows similar passion in her role as chair of IEEE Region 10’s Young […]

Take a look around the airport during your travels this summer and you might spot a string of new technologies at every touchpoint: from pre-arrival, bag drop, and security to the moment you board the plane. In this new world, your face is your boarding pass, your electronic luggage tag transforms itself for each new flight, and gate scanners catch line cutters trying to sneak onto the plane early. It isn’t the future—it’s now. Each of the technologies to follow is in use at airports around the world today, transforming your journey-before-the-journey. Virtual queuing speeds up airport security As you pack the night before your trip, you ponder the age-old travel question: What time should I get to the airport? The right answer requires predicting the length of the security line. But at some airports, you no longer have to guess; in fact, you don’t have to wait in line at all. Instead, you can book ahead and choose a specific time for your security screening—so you can arrive right before your reserved slot, confident that you’ll be whisked to the front of the line, thanks to Copenhagen Optimization’s Virtual Queuing system. Copenhagen Optimization’s machine learning models use linear regression, heuristic models, and other techniques to forecast the volume of passenger arrivals based on historical data. The system is integrated with airport programs to access flight schedules and passenger-flow data from boarding-pass scans, and it also takes in data from lidar sensors and cameras at security checkpoints, X-ray luggage scanners, and other areas. If a given day’s passenger volume ends up differing from historical projections, the platform can use real-time data from these inputs to adjust the Virtual Queuing time slots—and recommend that the airport make changes to security staffing and the number of open lanes. The Virtual Queuing system is constantly adjusting to flatten the passenger arrival curve, tactically redistributing demand across time slots to optimize resources and reduce congestion. While this system is doing the most, you as a passenger can do the least. Just book a time slot on your airport’s website or app, and […]

For those who enjoy rooting for the underdog, the latest MLPerf benchmark results will disappoint: Nvidia’s GPUs have dominated the competition yet again. This includes chart-topping performance on the latest and most demanding benchmark, pretraining the Llama 3.1 403B large language model. That said, the computers built around the newest AMD GPU, MI325X, matched the performance of Nvidia’s H200, Blackwell’s predecessor, on the most popular LLM fine-tuning benchmark. This suggests that AMD is one generation behind Nvidia. MLPerf training is one of the machine learning competitions run by the MLCommons consortium. “AI performance sometimes can be sort of the Wild West. MLPerf seeks to bring order to that chaos,” says Dave Salvator, director of accelerated computing products at Nvidia. “This is not an easy task.” The competition consists of six benchmarks, each probing a different industry-relevant machine learning task. The benchmarks are content recommendation, large language model pretraining, large language model fine-tuning, object detection for machine vision applications, image generation, and graph node classification for applications such as fraud detection and drug discovery. The large language model pretraining task is the most resource intensive, and this round it was updated to be even more so. The term “pretraining” is somewhat misleading—it might give the impression that it’s followed by a phase called “training.” It’s not. Pretraining is where most of the number crunching happens, and what follows is usually fine-tuning, which refines the model for specific tasks. In previous iterations, the pretraining was done on the GPT3 model. This iteration, it was replaced by Meta’s Llama 3.1 403B, which is more than twice the size of GPT3 and uses a four times larger context window. The context window is how much input text the model can process at once. This larger benchmark represents the industry trend for ever larger models, as well as including some architectural updates. Blackwell Tops the Charts, AMD on Its Tail For all six benchmarks, the fastest training time was on Nvidia’s Blackwell GPUs. Nvidia itself […]

This article is excerpted from Every American an Innovator: How Innovation Became a Way of Life, by Matthew Wisnioski (The MIT Press, 2025). Imagine a point-to-point transportation service in which two parties communicate at a distance. A passenger in need of a ride contacts the service via phone. A complex algorithm based on time, distance, and volume informs both passenger and driver of the journey’s cost before it begins. This novel business plan promises efficient service and lower costs. It has the potential to disrupt an overregulated taxi monopoly in cities across the country. Its enhanced transparency may even reduce racial discrimination by preestablishing pickups regardless of race. aspect_ratio Every American an Innovator: How Innovation Became a Way of Life, by Matthew Wisnioski.The MIT Press Sounds like Uber, but it’s not. Prototyped in 1975, this automated taxi-dispatch system was the brainchild of mechanical engineer Dwight Baumann and his students at Carnegie Mellon University. The dial-a-ride service was designed to resurrect a defunct cab company that had once served Pittsburgh’s African American neighborhoods. The ride service was one of 11 entrepreneurial ventures supported by the university’s Center for Entrepreneurial Development. Funded by a million-dollar grant from the National Science Foundation, the CED was envisioned as an innovation “hatchery,” intended to challenge the norms of research science and higher education, foster risk-taking, birth campus startups focused on market-based technological solutions to social problems, and remake American science to serve national needs. Today, university incubators like the CED are commonplace. Whether they’re seeking to nurture the next Uber, or social ventures like the dial-a-ride service, they all aim to transform ideas into businesses, discoveries into applications, classroom assignments into revenue, and faculty and students into entrepreneurs. Indeed, the idea that universities are engines of innovation is so ingrained that we take it for granted that it was always the case. So it’s instructive to look back to the time when the first innovation incubators […]

The robots that share our public spaces today are so demure. Social robots and service robots aim to avoid offense, erring toward polite airs, positive emotions, and obedience. In some ways, this makes sense—would you really want to have a yelling match with a delivery robot in a hotel? Probably not, even if you’re in New York City and trying to absorb the local culture. In other ways, this passive social robot design aligns with paternalistic standards that link assistance to subservience. Thoughtlessly following such outdated social norms in robot design may be ill-advised, since it can help to reinforce outdated or harmful ideas such as restricting people’s rights and reflecting only the needs of majority-identity users. In my robotics lab at Oregon State University, we work with a playful spirit and enjoy challenging the problematic norms that are entrenched within “polite” interactions and social roles. So we decided to experiment with robots that use foul language around humans. After all, many people are using foul language more than ever in 2025. Why not let robots have a chance, too? Why and How to Study Cursing Robots Societal standards in the United States suggest that cursing robots would likely rub people the wrong way in most contexts, as swearing has a predominantly negative connotation. Although some past research shows that cursing can enhance team cohesion and elicit humor, certain members of society (such as women) are often expected to avoid risking offense through profanity. We wondered whether cursing robots would be viewed negatively, or if they might perhaps offer benefits in certain situations. We decided to study cursing robots in the context of responding to mistakes. Past work in human-robot interaction has already shown that responding to error (rather than ignoring it) can help robots be perceived more positively in human-populated spaces, especially in the case of personal and service robots. And one study found that compared to other faux pas, foul language is more forgivable in a robot. With this past work in mind, we generated videos with three common types of robot failure: bumping into a table, dropping an […]