Agent Evaluation Science
EvalScience’s one-day 2026 symposium will bring researchers and practitioners together to examine how AI agents should be evaluated, from benchmark reliability and evaluation infrastructure to real-world interactive settings. Speakers will include experts from Microsoft Research, Meta, Cornell Tech, Cohere Labs, Princeton, UIUC, J.P. Morgan, and IBM Research, alongside talks, research presentations, posters, and discussion.
Explore event











