Inaugural Frontiers of AI Summit Focuses on the Foundational Research Behind AI’s Rapid Progress

By Grace Stanley

On May 27, nearly 300 researchers, industry leaders, and nonprofit innovators gathered at Cornell Tech for the inaugural Frontiers of AI Summit, hosted by Cornell Tech and the Jacobs Technion-Cornell Institute, to discuss the foundational advances shaping the future of artificial intelligence.

Participants from institutions including Cornell University, Princeton University, Columbia University, and New York University mingled with representatives from nonprofit organizations such as the Simons Foundation and Biohub, and startups gaining traction in the AI space, such as Cursor, Gimlet Labs, Makora, and Radical AI. The mix reflected the wide reach of AI today, spanning everything from the design of core systems to their growing use in science, medicine, and engineering.

“We created the Frontiers of AI Summit to build a space in New York City where the people pushing the boundaries of AI — in academia, science, and industry — can come together to think deeply rather than just move fast. The inaugural event showed us there’s real hunger for that kind of dialogue, and the extent of AI innovation and research in New York,” said event organizer Yoav Artzi, associate professor of computer science at Cornell Tech and the Cornell Ann S. Bowers College of Computing and Information Science.

In an opening keynote, Cornell computer science professor Kilian Q. Weinberger argued that today’s dominant approach to AI — training models to predict the next word — is starting to reach its limits. He said these systems struggle to store and retrieve knowledge reliably. “Neurons are actually a terrible way to store data and facts,” he noted, pointing to persistent accuracy issues. Instead, he described pairing broader reasoning with external knowledge sources to produce results that are more reliable, accurate, and energy efficient.

Across a series of lightning talks and research presentations, that theme continued to unfold. Speakers pointed to a growing interest in AI systems that can work across different types of information — text, images, and video — and better understand the world around them.

Biohub researcher Roshan Rao described a newly released “world model” of protein biology trained on vast datasets of protein sequences and structures. With their new model, Biohub generated an atlas of more than one billion predicted proteins, which can uncover patterns beyond those observed in the lab.

“If you can learn to read and write the language of proteins,” Rao said, “then you could learn to modulate, modify, and really understand biology and create new biology.” By enabling scientists to generate and test large numbers of hypotheses before running experiments, these tools could significantly accelerate discovery in fields where traditional approaches are slow and resource-intensive.

Elsewhere, startup founders and industry engineers focused on how to turn these breakthroughs into systems that work at scale. Cursor presented its newly released agentic AI model, Composer 2.5, which demonstrated significant efficiency gains without sacrificing performance, while presentations from Gimlet Labs and Makora (co-founded by assistant professor Mohamed Abdelfattah) explored the technical challenges of building faster, more efficient AI infrastructure. Another startup, Radical AI, highlighted how these advances could scale scientific discovery.

Together, these efforts point to both the promise and the complexity of the field’s next phase. As AI systems become more powerful and more independent, questions about how they perform, how they are managed, and how they scale are becoming harder to answer. The discussion turned to the societal dimensions of that shift in a keynote conversation, “Human Extinction Is Not the Worst That Could Happen,” in which Helen Nissenbaum, the Andrew H. and Ann R. Tisch Professor at Cornell Tech, joined associate professor Yoav Artzi in conversation.

Similar questions carried into the day’s closing keynote conversation, where speakers stepped back to examine the broader impact of AI’s rapid rise. Cornell alumnus Andrew Ross Sorkin ’99 — award-winning journalist, author, co-anchor of CNBC’s Squawkbox, and member of the Cornell Tech council — joined Cornell Provost Kavita Bala to examine the current wave of investment and enthusiasm around AI, and whether it signals lasting transformation or a potential bubble.

“Anytime you get into one of these speculative moments, there’s a lot of good stuff that’s being financed, but a lot of silly stuff,” Sorkin said at the summit. “When the tide goes out, we will find out where the problems lie.”

His remarks echoed a broader undercurrent throughout the summit: that while AI is advancing at remarkable speed, its long-term impact will depend on sustained progress in the underlying science, and careful thought about how these systems are built and deployed.

The summit was followed up by a two-day academic symposium that brought together leading researchers from around the country. The symposium focused on two fundamental questions: what form will the next generation of AI models take? And how AI will revolutionize science, and enable the next wave of big discoveries?

The summit and symposium were made possible with generous support from the Secunda Family Foundation. Following this inaugural event, organizers plan to make the summit an ongoing forum for those conversations, rooted in the idea that the most important advances in AI often happen behind the scenes.

Grace Stanley is the staff writer-editor for Cornell Tech.