Big AI for Small Devices at the Inaugural Athena Summit

Industry, government and academic leaders came together to share the latest developments in artificial intelligence (AI) and edge computing at the inaugural summit of the U.S. National Science Foundation AI Institute for Edge Computing Leveraging Next Generation Networks (Athena).

Hosted by Duke University on May 5-6, this year’s summit centered on expanding industry-university collaborations. Representatives from industries ranging from Silicon Valley to telecommunications to financial management came to participate in panels, learn about the latest in academic research and observe demonstrations.

Funded by the National Science Foundation (NSF) and Department of Homeland Security (DHS), Athena brings together a multidisciplinary team of scientists and engineers from 11 universities, led by Duke, who collaborate on research, broadening participation in AI, and education and workforce development for the future.

The institute is focused on advancing edge computing, a concept that aims to bring AI closer to where data originates on local devices as opposed to cloud-based servers. Combining edge computing and AI also enhances the ability to gather data from the physical world since it can process data in real-time. Athena’s leaders often used the phrase “Big AI for Small Devices,” during the summit to describe their goals.

Yiran Chen, director of Athena and the John Cocke Distinguished Professor of Electrical and Computer Engineering (ECE) at Duke, highlighted the importance of working together to foster innovation in AI across industries.

“Athena is not just about research. It’s not just about publishing papers. It’s a comprehensive organization for research, industry collaboration and community building,” Chen said during his opening remarks.

Charlie Zhang, senior vice president at Samsung Research America, echoed Chen on the importance of cooperation as industries are changing due to AI.

“On behalf of Samsung and all these industry partners, we are committed to continuing, strengthening and expanding our collaboration with Athena in the AI area because I believe working together with academia, government and industry is the only way to bring these wonderful innovations to real life,” Zhang said.

Remarks from government officials including US Congresswoman Valerie Foushee and North Carolina Representative Zack Forde-Hawkins highlighted the importance of Athena’s work to the nation’s technological and cybersecurity future, as well as the state and local economy.

“Duke has long been a leader in shaping the future of technology, particularly in AI,” Foushee said. “The Athena Institute stands as a prime example of this leadership.”

​​”We believe North Carolina is the next frontier for AI in the United States of America,” Forde-Hawkins said. “The state of North Carolina and the government don’t want to get in the way. In fact, we want to be partners, and Athena is showing the way.”

Edge Computing Poised to Transform Industries

In various panels and talks, industry and academic leaders highlighted the potential benefits of on-device AI, including its energy efficiency and cost effectiveness.

“Even though we’ve made a lot of advancements on the AI front, we still see a large gap in the data and energy efficiency of AI compared to human usage,” said Jilei Hou, senior vice president of engineering and the head of AI research at Qualcomm Technologies, in his keynote address. “There are a lot of reasons to expect edge computing to play a very strong role.”

Theodore Rappaport, the David Lee/Ernst Weber Professor at New York University and founder of NYU WIRELESS, shared Hou’s sentiment on edge computing’s capacity to personalize user experiences.

“We’re going to see incredible new use cases for our mobile devices,” Rappaport said. “It will become your detector for air quality, for allergens in your food, your smoke detector. I see the smartphone becoming more and more an appliance that handles many aspects of our life.”

Demos Display Engineering Expertise

The summit dedicated space for Athena faculty labs to demonstrate their capabilities and engineering feats. The signature demo of the day showcased a large collaboration between the labs of Miroslav Pajic, professor of ECE at Duke; Maria Gorlatova, the Nortel Networks Assistant Professor of ECE at Duke; Suman Banerjee, professor of computer sciences at the University of Wisconsin, Madison; Lin Zhong, professor of computer science at Yale University; and other Athena principal investigators.

Picture this scenario: An unknown biohazard has gotten out in a lab or office building. Although there are human survivors and victims inside, first responders can’t rush in without putting themselves at risk. This is where engineers can play a role, leveraging the power of edge computing and AI to act quickly.

First, a flying drone enters and surveys the building to construct a virtual map. Next, a robotic dog with stronger capabilities to navigate doors and obstacles uses the map to search the building. Finally, a small flying drone launches off the robotic dog to search high and low. This isn’t just an imaginary scenario—real-time demos at the summit showed this is a realistic possibility in the not-too-distant future. The entire rescue process is supported by computing performed on a remote edge server that runs on AI and large language models. All of this occurs autonomously, but it’s also monitored by engineers, who can redirect the robots. In a life-or-death scenario like this, fast processing speeds made possible by edge computing are essential.

Preparing the Next Generation of AI Engineers

On day two, the workforce development and educational outreach arm of Athena welcomed about 40 middle and high schoolers to Duke’s campus. Middle schoolers from Lowe’s Grove and high schoolers from Cary Academy observed similar demos from the previous day, discussed AI topics with graduate students and coded cars to autonomously navigate through a maze.

One of the institute’s founding pillars is cultivating the next generation of scientists and engineers to work in AI. Through age-appropriate lessons, Athena has worked with more than 1200 students with more than 300 contact hours. These efforts are largely led by the LIFT Lab team including Shaundra Daily, Cue Family Professor of the Practice of ECE, Sandra Roach, associate in research of ECE, Karis Boyd-Sinkler, research scientist of ECE, and Whitney McCoy, research scientist at the Sanford School of Public Policy.

“Exposing young minds to STEM and AI isn’t just about teaching technical skills—it’s about igniting curiosity and showing students they can be creators of technology, not just consumers,” Daily said. “We’re not just sharing knowledge—we’re opening doors to futures many haven’t yet imagined for themselves. The excitement I see when a student successfully codes their first autonomous vehicle or grasps a complex AI concept tells me we’re not just building better technology—we’re building better technologists for tomorrow.”

Industry Collaborations and Entrepreneurship

While the educational outreach team worked with students, other summit attendees discussed opportunities in AI entrepreneurship and industry collaborations with Athena. The collaborations panel focused on how industry-funded research might evolve given the current headwinds in federal funding.

Moderated by Daniel Dardani, director of physical sciences and digital innovation licensing and corporate alliances at Duke’s Office of Translation and Commercialization, the panel included:

• Mario Aguilar-Simon, head of AI and fellow at Teledyne Scientific & Imaging
• Victor Bahl, technical fellow and chief technology officer (CTO) at Microsoft Azure for Operators
• Nakjung Choi, mobile network systems department head at Nokia Bell Labs
• Hai “Helen” Li, Marie Foote Reel E’46 Distinguished Professor and chair of ECE at Duke
• John Nicholson, distinguished researcher at Lenovo

The group acknowledged the traditional benefits that industry gets from university collaborations and sponsored research; namely an opportunity to look further into technology’s future than their companies can while also being introduced to talented students who could one day join their ranks. The conversation then shifted to what technologies the industry panelists were currently most interested in pursuing and how universities can demonstrate a clear value proposition in their proposals.

Bahl was quick to point out that to be truly valuable, a technology needs to be disruptive.

“The world is currently too dependent on GPUs that cost $44,000 each—it’s just crazy,” said Bahl, who also spoke on day one in a CTO fireside chat with Zhong. “And they’re incredibly power hungry, so our energy is going to run out. There’s an opportunity to develop new hardware that costs less and doesn’t consume as much electricity.”

Aguilar-Simon added that he saw developments in how humans and AI interact and team with one another as a potential avenue to commercial success.

“Creating well-oiled machines from teams that optimally pair humans and AI. Understanding what makes a team function well and harnessing the skills and abilities that each brings—that’s where I see opportunity,” Aguilar-Simon said.

The second panel, which was moderated by Tingjun Chen, assistant professor of ECE, focused on entrepreneurship and introduced a few new faces to the discussions.

• Daniel Dardani
• Jungsang Kim, co-founder of IonQ and the Schiciano Family Distinguished Professor of ECE at Duke
• Jing Li, former investment manager at DUMAC
• Caleb Miles, investment manager at DUMAC
• David Talpalar, investor at Conversion Capital
• Steven Xi, co-founder and managing partner at Eastlink Capital

The session started with observations about what opportunities are currently available for new businesses including vertical applications in AI, energy-saving AI, and AI aimed at disrupting traditional knowledge-based industries.

Xi was especially enthusiastic about Duke and the Research Triangle as a promising area for new start-ups to thrive. There is a lot of available, inexpensive land, the talent pool is dense, infrastructure like health care already exists, and entities like the Duke Law School and the Duke Fuqua School of Business are already well-known.

“Engineering is jumping up too. I won’t be surprised if Duke Engineering keeps going up into the top 10,” Xi said. “Just look at the playbook Stanford used to assume an important role in the development of Silicon Valley. Duke can play such a role, too.”

The panel also recognized that, because AI is already off and running, potential entrepreneurs might need to find a new niche to fill. Athena, they said, could help bridge the gap between public uncertainty and building trust in AI.

“Athena and Duke are well-positioned to do that, but you have to pick a battle you can win,” Kim said. “And you do that by coming up with an idea that nobody else can.”