Request for Proposals: AI Interpretability

Schmidt Sciences invites proposals for a pilot program in AI interpretability. This RFP seeks new methods for detecting and mitigating deceptive behaviors from AI models, such as when models knowingly give misleading or harmful advice to users.

This RFP focuses on three directions:

1. Detecting deceptive behaviors from LLMs
2. Steering models to improve truthfulness
3. Applications of detection/ steering methods