University of California, San Diego — AI Persuasiveness Evaluation

[Navigating Transformative AI] Open Philanthropy recommended a grant of $470,731 to the University of California, San Diego to support research led by Professor Benjamin Bergen on the persuasiveness of large language models (LLMs). ​​The study will use game-like evaluations to assess LLMs' ability to change minds, foster cooperation or defection in classic games, and negotiate for hidden goals in resource-allocation scenarios, all through freeform conversation. This falls within our focus area of potential risks from advanced artificial intelligence.