x

LESSWRONG

LW

omni_allie — LessWrong

omni_allie

omni_allie

Message

https://anruigu.github.io/

I'm researching environment-behavior provenance and human alignment at Fleet AI; also increasingly interested in automating alignment research. Previously I worked on non-verifiable domains and user alignment at xAI.

3

1

21d

omni_allie

https://anruigu.github.io/

I'm researching environment-behavior provenance and human alignment at Fleet AI; also increasingly interested in automating alignment research. Previously I worked on non-verifiable domains and user alignment at xAI.

Behavior Theory-Building for Diverse Agents and Environments

The oversight problem consists of describing, eliciting, and understanding behavior. Prior work in black-box behavior investigation and interpretability seek to construct compact, faithful, and informative theories that explain model behavior. We propose training investigator agents and subject models to emit explicit theories, and using these theories as common currency for...