Topic

Model self-reports and introspection

Research into what language models say about their own nature, feelings or internal states, and whether such statements are reliable evidence about the model.

Current clusters

build1 publisher

Chat template tokens switch on the 'just an AI' disclaimer in eight open instruct models

Jędrzej Maczan's paper finds the chat template turns on the 'just an AI' disclaimer in all eight open instruct models he tested. For eval teams, a model's self-description now depends on a formatting step most developers never see.

Publishers:dev.to

Reality

Evidence45
Adoption: Insufficient
Hype gap+15
Incentives: Insufficient
Confidence40