Announcement
METR says it will conduct a limited alignment assessment of OpenAI’s largest RL run
METR says it will publish its findings and assessment process in one or more reports.
TLDR
Under an agreement with OpenAI, METR says it will investigate the alignment properties of the company’s largest reinforcement-learning run. METR describes the assessment as a limited pilot and says it will publish one or more reports sharing its findings and explaining its process.
Combined views
412
3 Sources, first seen ago
Comments