Skip to main content
Event

Using AI to Accelerate Scientific Discovery — MIT seminar with Demis Hassabis

Dr. Omar Costilla Reyes hosted Demis Hassabis at MIT on April 5, 2022, for a seminar on using artificial intelligence to accelerate scientific discovery. The Center for Brains, Minds and Machines organized the seminar.

The seminar “Using AI to Accelerate Scientific Discovery” at MIT, April 5, 2022, hosted by Dr. Costilla Reyes. Credit: Image courtesy of Dr. Omar Costilla Reyes.

MIT seminar on AI for scientific discovery

MIT’s Center for Brains, Minds and Machines (CBMM) and the MIT Quest for Intelligence listed Demis Hassabis, founder and chief executive of DeepMind, for the seminar “Using AI to Accelerate Scientific Discovery” on April 5, 2022. The talk took place in room 26-100 of Building 26 at MIT as part of CBMM’s spring 2022 Brains, Minds and Machines seminar series, and was also livestreamed. Dr. Omar Costilla Reyes, then a research scientist at MIT CSAIL, the Computer Science and Artificial Intelligence Laboratory, wrote that day that he was glad to be hosting Hassabis at MIT and that the visit was giving him and his colleagues a view of the team’s efforts to advance science with artificial intelligence.

The question behind the talk

The seminar’s central question was how AI might give researchers new ways into scientific problems that have resisted familiar approaches. CBMM’s announcement named DeepMind’s AlphaFold as an example. AlphaFold is a deep-learning system that predicts the three-dimensional shape of a protein from the sequence of its amino acids, a problem biologists had worked on for decades because a protein’s shape largely determines what it does. The photograph from the seminar shows a slide titled “AlphaGo’s original strategies,” referring to DeepMind’s Go-playing system and a move labeled Move 37, Game 2.

Connection to his research

Machine learning for scientific discovery is also the subject of Dr. Costilla Reyes’s own research at MIT CSAIL, where he worked on neurosymbolic programming for science, the subject of the paper “Neurosymbolic programming for science.” Hassabis later shared the 2024 Nobel Prize in Chemistry for computational methods that predict the three-dimensional structures of proteins.

Topics