Topic

Human Feedback Model Training

The practice of having people rate and compare model outputs against written guidelines so the scores can be used to tune later versions.

Current clusters

build2 publishers

Hundreds of contractors read ChatGPT chats behind a filter tuned for identifier fields

OpenAI's outside reviewers read real conversations with account names masked. The 404 Media documents show some review tasks also carry a summary of what ChatGPT has remembered about the user, including location.

Publishers:runtimewire.comthe-decoder.com

Reality

Evidence72
Adoption78
Hype gap+12
Incentives58
Confidence70