Humanity’s Last Choice

Humanity’s Last Choice is an open research community studying how AI chooses for people. We ask whether recommendations serve the person when the information a system reads is written to influence it. Through papers, shared evaluations, and contributions across domains, we build evidence that others can inspect, reproduce, and use.

Start here

• Website: humanitylastchoice.org
• Leaderboard: humanitylastchoice.org/leaderboard
• SafeGEO, our EMNLP 2026 paper: arXiv · dataset · code · explainer

Take part

Bring a research question, propose an experiment, evaluate a system, or contribute a domain. Write to info@humanitylastchoice.org.

About the data here

Datasets published under this organisation are public suites. Hidden test sets are never published, so they stay out of training data and the leaderboard keeps measuring what it claims to.