Human judgment

…responsible evaluation infrastructure, human judgment in evaluation, and methods for assessing AI deployment in emerging markets. Use the ARA proposal template; four pages excluding appendices are encouraged. Decisions are expected in February 2027. Applications close on 4 November 2026 at 11:59pm Pacific Time.

…interacting agents, and the role of human judgment and verification. It also examines diversity in machine-generated ideas, benchmark and harness overfitting, persistent agent errors, continual learning, world models, and where information theory helps explain AI. Hosted by Ravid Shwartz Ziv and Allen Roush; the full conversation is available on the publisher's YouTube channe