AI agents for scientific discoveries and data science

AI agents — large language models equipped with tools and reasoning capabilities — are emerging as powerful research enablers. This talk will explore how agentic AI can accelerate scientific discoveries. I will first introduce the Virtual Biotech, a platform where tens of thousands of AI agents work together to advance drug discovery and development. As AI agents become capable of conducting end-to-end data analyses, they also introduce new challenges for scientific reliability. I will present Agentic Bootstrap, a framework that estimates the distribution of defensible analyses an AI agent could produce from the same data, and the m-value, a measure of how representative a reported finding is within that distribution.