Laure Zanna Serves on Advisory Committee for Agents’ Last Exam

Laure Zanna is serving on the advisory committee for Agents’ Last Exam (ALE), a new large-scale benchmark designed to evaluate whether AI agents can successfully perform real-world professional tasks. Developed through a collaboration led by UC Berkeley researchers and industry partners, ALE assesses AI systems on long-horizon workflows across dozens of industries, using tasks with verifiable outcomes rather than traditional question-answering benchmarks.

The project aims to provide a more realistic measure of progress toward AI systems capable of supporting economically valuable work. The benchmark currently includes more than 1,500 tasks spanning 55 industry domains and is expected to continue expanding as a community resource for evaluating next-generation AI agents.

Learn more about the benchmark and current results on the Agents’ Last Exam leaderboard.