June 11-12, 2024
London, UK
Rapid advancements in the field of AI pose both opportunities and challenges to scientific research. The Forum explored the emergence of AI in support of scientific research; applications of AI in different scientific fields; new paradigms of doing science enabled by AI; the intersection of AI with open science and the issues of reproducibility and replicability; and ethical issues related to the use of AI in science.
Session 1 – The Emergence of AI Science to Support Scientific Research
Welcome Remarks:
Sir Adrian Smith, The Royal Society
Professor Alison Noble, University of Oxford & The Royal Society
Professor William Press, University of Texas at Austin & National Academy of Sciences
Session Chair:
Professor Andy Cooper, University of Liverpool
Speakers:
Professor Jeannette Wing, Columbia University
Dr. Frank Noé, Microsoft Research
Dr. Pushmeet Kohli, Google DeepMind
Professor Tom Mitchell, Carnegie Mellon University
This session laid foundations for the Forum, discussing the current state of AI as a discipline and how AI can be integrated into scientific research.
Session 2 – AI at the Frontier of Scientific Discovery: Transformative Applications Across Disciplines
Session Chair:
Dr. Tapio Schneider, California Institute of Technology
Speakers:
Professor Alexander Szalay, Johns Hopkins University
Professor Kristin Persson, University of California, Berkeley
Professor Kim Jelfs, Imperial College London
Dr. Rémi Lam, Google DeepMind
Professor Anil Madhavapeddy, University of Cambridge
This session explored pioneering case studies demonstrating the application of AI across scientific disciplines, including astronomy, materials science, and weather forecasting. We explored how AI is transforming the landscape of scientific discovery, highlighting its potential and examining its limitations.
Session 3 – Scientific Advances Powered by AI, and New Science Enabled by AI
Session Chair:
Dr. David Donoho, Stanford University
Speakers:
Dr. John Jumper, Director, Google DeepMind
Professor John Moult, University of Maryland
Dr. Jonathon Phillips, National Institute of Standards and Technology
Professor Caroline Uhler, Broad Institute of Massachusetts Institute of Technology & Harvard
Professor Andrew Zisserman, University of Oxford
Recent progress in protein folding has been dramatic. AlphaFold seized the scientific public’s imagination, winning recent CASP Protein Folding challenges. Challenges in biometrics, such as facial recognition, have also seen progress over time, where a ‘Moore’s Law’ for recognition performance has been observed across 30 years.
Scientists are finding diverse applications of AI in scientific practice across many fields. AI accelerates practice within individual disciplines, for example by enabling data to be processed with super-human capabilities. With impacts in chemistry and biology, AIs also are helping design future experiments and identify improved algorithms.
Session 4 – Enabling Open Science, Reproducibility, Replicability, and Privacy
Chair:
Professor Chris Holmes, University of Oxford
Speakers:
Professor Victoria Stodden, University of Southern California
Dr. Sasha Luccioni, Hugging Face
Professor Mark Kelson, University of Exeter
Dr. Xiaoxuan Liu, University of Birmingham & University Hospitals Birmingham
Dr. Rebecca Lawrence, Managing Director, F1000
This session discussed the issues and opportunities that arise when conducting open science with AI, as well as the unique challenges to ensuring reproducibility and replicability when using AI in science.
Session 5 – Panel on Recent Publications
Session Chair:
Professor Alison Noble, University of Oxford & The Royal Society
Speakers:
Professor Terence Tao, University of California, Los Angeles & Co-Lead, PCAST
Professor Maria Zuber, Massachusetts Institute of Technology & Co-Chair, PCAST
Areeq Chowdhury, The Royal Society
This panel covered two recent publications: President’s Council of Advisors on Science and Technology (PCAST) report “Supercharging Research: Harnessing Artificial Intelligence to Meet Global Challenges”, and the Royal Society report “Science in the Age of AI: How Artificial Intelligence is Changing the Nature and Method of Scientific Research”.
Session 6 – Responsible AI in Science
Session Chair:
Professor William Press, University of Texas at Austin & National Academy of Sciences
Speakers:
Dr. Yolanda Gil, University of South California
Dr. Stuart Feldman, Schmidt Sciences
Professor David Leslie, Queen Mary University of London & The Alan Turing Institute
Professor Shannon Vallor, University of Edinburgh
Lord Martin Rees, Astronomer Royal
This session addressed the care that must be taken to ensure AI in science is used responsibly and ethically, in ways that support science’s broad goal to better humanity.
Related Events
Event Disclaimer
It is essential to the National Academy of Sciences mission of providing evidence-based advice that participants in any of our meetings or events avoid political or partisan statements or commentary and maintain a culture of mutual respect. The statements and presentations during our meetings or events are solely those of the individual participants and do not necessarily represent the views of other participants or the National Academy of Sciences, which is a non-partisan, tax exempt organization.