"LLMs also demonstrate a strong confirmation bias when the external evidence contains some information that is consistent with their parametric memory"
Well, maybe they do act human. 🧐
Unraveling the Behavior of Large Language Models in Knowledge Clashes
https://arxiv.org/abs/2305.13300?utm_source=tldrai
Follow
@jeffjarvis Wait, what? It's an overgrown pattern-matching algorithm. Why are people surprised that its output matches patterns? (Because that's all "confirmation bias" really is in this case: pattern matching).
@LouisIngenthron
Bingo