AI for Scientific Discovery

Over the past century, artificial intelligence (AI) has evolved from computational theory to everyday conversational technologies, capturing the attention and interest of the public and the media. It has also caught the attention of the scientific community, where it has provided a new tool to support inquiry and exploration. While AI in the context of scientific investigation has existed for decades, advances in computational technology and sensing in the physical world have created opportunities to integrate AI into science in unexpected ways, with capabilities that are rapidly accelerating. As a result, AI has been leveraged by an expanding collection of disciplines in the physical and biological sciences, as well as engineering domains. While the opportunities for AI in scientific discovery seem endless, there are numerous questions about what makes for trustworthy and reliable discovery, whether such investigation should be performed without human oversight or intervention, and how best to prioritize the research agenda and allocation of resources without magnifying disparities for individuals and nations alike.In recognition of the timeliness and sizable implications of AI in our world, the National Academies of Sciences, Engineering, and Medicine hosted AI for Scientific Discovery - A Workshop on October 12-23, 2023. Leaders from across the globe in the field of AI, preeminent researchers in various science and engineering disciplines, and experts in ethics, law, and social sciences met to appraise the state of the field and provide guidance on the opportunities, as well as the challenges, that lay ahead. Presentations and discussion explored the future of AI in terms of its role as an autonomous researcher carrying out discovery and considered ethical aspects of AI used for independent scientific discovery.