Overview

Artificial Intelligence (AI) is poised to disrupt science by unlocking new approaches that enable scientific advances with far-reaching societal consequences. In addition, challenges unique to scientific problems also offer an opportunity to advance AI, leading to a virtuous cycle of fundamental AI research driving scientific discovery, and use-inspired challenges in scientific problems advancing the frontiers of AI. 

In March 2023, a workshop was held at NSF Headquarters to identify key challenges and steps that need to be taken to enable the next AI revolution in the Sciences. This workshop led the discussion on a new frontier in AI, where novel AI frameworks will drive scientific inquiry, suggest novel experiments, elucidate new theories, and thus revolutionize the traditional discovery process across multiple scientific disciplines. It identified limitations of the current state-of-the-art in AI for existing grand scientific challenge problems and discussed AI advances that are needed to catalyze synergistic research across scientific communities and how these advances can be incorporated into the practice of scientific discovery. Discussions at the workshop were summarized in this report .

The goal of this second workshop is to gather feedback on the initial report and revisit its recommendations in light of recent developments in AI. A major focus will be on understanding the power of new advances in areas such as generative AI for scientific discovery, their potential limitations (e.g., hallucinations), and ways to strengthen them (e.g., by leveraging scientific theories).