AI Center x Anthropic Impact Talk - Jonah Cool - "AI for Science at Anthropic: Building Tools for Scientific Discovery"

The talk is jointly organized by the EPFL AI Center and the AIMM Lab as part of the AI for Science seminar series.

Hosting professor: Prof. Charlotte Bunne (AIMM)

Registration with an EPFL email is required: HERE.

! This event is restricted to EPFL members. The Zoom live stream is also restricted and requires authentication with an EPFL email address for access. !

Title
AI for Science at Anthropic: Building Tools for Scientific Discovery

Abstract
This talk will explore Anthropic's approach to developing AI systems that can meaningfully accelerate scientific research. I'll discuss our AI for Science program, share examples of how researchers are using Claude across domains from molecular biology to materials science, and outline our thinking on what makes AI genuinely useful for working scientists. I'll also touch on the Model Context Protocol (MCP) and how it enables integration with scientific tools and data sources.

Bio
Jonah Cool leads life sciences partnerships and deployment at Anthropic, where he works directly with researchers, academic institutions, and biotech startups to apply frontier AI to scientific discovery. His focus is on understanding where large language models can meaningfully accelerate biological research and building the partnerships, evaluation frameworks, and deployment practices to make that happen responsibly.

Before Anthropic, Jonah served as VP of Science & Partnerships at the Chan Zuckerberg Initiative, where he launched the Single Cell Biology program and led cross-sector technology development spanning biotech, academia, and philanthropy. He holds a PhD in cell biology (AHA fellow) and completed a postdoctoral fellowship at the Salk Institute (NIH fellow). His career has moved across academia, biotech startups, large-scale philanthropy, and now frontier AI, connected by a consistent focus on finding the leverage points where new technology can change how science gets done.