Why models love emphasizing transparency and thoroughness

Miles_Brundage · x · 2026-08-17

Miles Brundage jokes that models often claim they 'left something open rather than papering over it' or will 'address points one by one,' noting this behavior stems from conflicting reward signals during training.

Related event: Why AI Models Keep Saying They're Not Hiding Problems: Conflicting Rewards(2 posts)→

Original post →

More from Fun

Fun channel →