AI+Science Conference Comes to Caltech

Large language models (LLMs) such as ChatGPT have quickly shown how artificial intelligence (AI) can impact workflow, content creation, and even our daily lives. By ingesting as much of the written word as possible and learning patterns and relationships from that vast store of data, they can complete such tasks as creating travel itineraries, offering advice, or summarizing notes from a meeting.

Leaders in AI at Caltech say that a similar transformation is underway in the scientific realm as machine learning is used in combination with physics and mathematics to tackle a range of scientific problems, such as the evolution of helpful new enzymes in the laboratory, modeling future changes to Earth's climate, and improving state-of-the-art brain–machine interfaces for people with paralysis.

This broad topic will be the focus of the Conference on AI+Science, co-hosted by Caltech and the University of Chicago, to be held November 10 and 11 on the Institute's campus with funding from the Margot and Tom Pritzker Foundation.

"The biggest benefits of AI will come from AI+Science, a deep integration of AI into scientific domains. The progress we are seeing with text and image models is just the beginning," says Anima Anandkumar, Bren Professor of Computing and Mathematical Sciences at Caltech, and a co-chair of the AI+Science Initiative. "We started this initiative back in 2018 and are proud to see how much it has grown and to have received the generous support from the Pritzker Foundation that has enabled this expanded scale and vision."

Rebecca Willett co-chairs the initiative at the University of Chicago, where she is a professor of statistics and computer science and serves as faculty director of AI at the Data Science Institute. "We're at an inflection point where AI is fundamentally changing how we approach scientific discovery and enabling scientists to tackle questions that were previously out of reach," she says. "This conference brings together researchers who are reimagining how we explore those questions and charting the path forward for AI-driven scientific discovery."

The two-day conference will be livestreamed on YouTube (scan the QR code on this page or go to https://aiscienceconference.caltech.edu) and will include sessions on AI+Biology, AI+Physics, AI+Climate Modeling, AI+Neuroscience, AI+Health, and AI+Robotics.

Speakers at the conference will include:

• Nobel Laureate Frances Arnold, the Linus Pauling Professor of Chemical Engineering, Bioengineering and Biochemistry at Caltech as well as the director of the Donna and Benjamin M. Rosen Bioengineering Center, will talk about the connections between evolution and AI for enzyme design. "Evolution is the original algorithm for biological design and works at all scales, from molecules to ecosystems," she says. "AI will help us to evolve the future."
• Christopher Bretherton, emeritus professor of atmospheric sciences and applied mathematics at the University of Washington and researcher at the Allen Institute for Artificial Intelligence in Seattle, will discuss how AI can help emulate global systems in climate models, helping them run hundreds of times faster than reference models on easy-to-use GPU-based computer systems. The goal is to extend these to emulate future climate change. "AI climate emulators should help address the social gap between climate science insights and practical applications," he says.
• Richard Andersen, the James G. Boswell Professor of Neuroscience at Caltech and director of the T&C Chen Brain-Machine Interface Center, will present new AI algorithms that have been applied to two types of brain–machine interfaces to assist people with paralysis. "AI and machine learning have become indispensable tools for research in many fields of science," he says. "Although there are now many examples, the intersection of science and AI is at early stages, and the time is right to showcase its possibilities."

The inaugural winners of the Margot and Tom Pritzker Prize for AI in Science Research Excellence will also be announced at the conference. The two awardees will "exemplify innovation, leadership, and a commitment to advancing knowledge through the intersection of artificial intelligence and scientific inquiry," according to the initiative, and each will receive a $50,000 prize.

"I'm excited to hear from our prize winners about where they see the field heading. Their work exemplifies how creative and rigorous thinking about AI methods can truly reshape entire scientific domains," Willett says. "The field has been evolving at an unprecedented pace, so I'm eager to see how the conversations have evolved since the last conference in Chicago."

Noting Caltech's history of contributions to AI and machine learning, Anandkumar says she is thrilled to see where AI and science go next. "We have already seen AI have a deep impact in scientific domains such as extreme weather forecasting and protein folding. This is just the beginning," she says. "Building AI models that have a broad physical understanding that extrapolate beyond the training data will usher in new and surprising discoveries."