Join UCSF researchers for a practical look at new generative AI models, tools, platforms and workflows designed to facilitate and expedite scientific research! Title: "Benchmarking Automated Scientific Enquiry: When AI co-scientists outpace peer review, what becomes the arbiter of truth?" Featured Speaker: Luca Foschini, PhD President & CEO, Sage Bionetworks Seminar 3-4pm, Networking Reception 4-5pm Share the Flyer! Bio: Luca Foschini, PhD, is President and CEO of Sage Bionetworks, a nonprofit that builds open infrastructure for biomedical data sharing, including the Synapse platform and community challenges such as DREAM. Before Sage he co-founded Evidation Health, where as Chief Data Scientist he spent over a decade leading research on wearable and real-world data for health. He holds a PhD in Computer Science from UC Santa Barbara, has published more than 70 peer-reviewed papers, and in a previous life represented Italy at the International Olympiad in Informatics. Abstract: In under two years, AI systems went from grade-school math to research-level problems, and the length of tasks an agent can complete reliably keeps doubling every few months. More than twenty AI co-scientists now promise to search the literature, generate hypotheses, analyze data and write up results. Do they reason like scientists? Recent evidence says not yet: agents often gather evidence and then ignore it, and better scaffolds barely help. The deeper problem is verification. AI moves fastest where answers can be checked instantly, as in code or proofs, while biology checks its claims at the pace of the wet lab and of peer review. I will walk through what works today, where these tools fail, and how benchmarks run as living community efforts, kept fresh and re-run on every new model, can keep us honest about what is still true.