Why LLMs Agree With You Even When You’re Wrong

T2Covered by 1 source · first reported by The Pragmatic Engineer / ByteByteGo / Architecture Weekly at 06 Oct, 15:31 UTC

When LLMs sometimes agree with incorrect claims, it is mostly because their training rewards such behaviour. This reward system is built on several things at once, such as accuracy, helpfulness, politeness, and responses that people like.

ai

Sources

• T2Why LLMs Agree With You Even When You’re WrongThe Pragmatic Engineer / ByteByteGo / Architecture Weekly06 Oct, 15:31