Trillion Parameter Foundation Models as Discovery Accelerators - Lecture 10

Explore the potential of advanced AI methods in accelerating scientific discovery through this 33-minute lecture by Dr. Ian T. Foster at the University of Michigan's MICDE. Delve into the hypotheses surrounding the use of foundation models for harnessing vast knowledge, automating hypothesis generation, and operating complex tools. Examine the challenges and opportunities of building science-focused foundation models, including their enormous cost and the need for extensive collaboration. Consider how research infrastructure must evolve to enable effective collaboration between human scientists, AI, high-performance computing, and robotics. Gain insights into the future of scientific research and the role of trillion-parameter foundation models as discovery accelerators.

10. SciFM24 Ian T. Foster: Trillion Parameter Foundation Models as Discovery Accelerators