Are you a domain scientist with hard tasks that current AI agents can't solve? Do you want to become an author on an AI benchmarking paper lead by other Stanford scientists? Join the Terminal Bench Science community today!
https://www.tbench.ai/news/tb-science-announcement
Step 1: Submit a proposal and join the discord. Your task must be (1) scientifically relevant (2) hard for current AI agents (aiming for a 15% pass rate) (3) programmatically verifiable (you have to write the tests) and (4) possible for a human expert (you have to write the oracle solution)
Step 2: Once the proposal is approved submit a PR to our github https://github.com/harbor-framework/terminal-bench-science and keep iterating until you pass three reviews.
Step 3: Get lots of citations on your Google Scholar!
Please do not message this poster about other commercial services.