Will Douglas Heaven / MIT Technology Review:

OpenAI is testing training LLMs to produce “confessions”, or self-report how they carried out a task and own up to bad behavior, like appearing to lie or cheat  —  OpenAI is testing another new way to expose the complicated processes at work inside large language models.

Will Douglas Heaven / MIT Technology Review:

OpenAI is testing training LLMs to produce “confessions”, or self-report how they carried out a task and own up to bad behavior, like appearing to lie or cheat  —  OpenAI is testing another new way to expose the complicated processes at work inside large language models.

Source: TechMeme
Source Link: http://www.techmeme.com/251203/p51#a251203p51