Trust Me, I'm an Expert: Decoding and Steering Authority Bias in Large Language Models

TL;DR

This paper investigates how large language models are influenced by endorsements from perceived experts, revealing susceptibility to authority bias that affects accuracy and confidence, and proposing methods to mitigate this bias.

Contribution

It uncovers the presence of authority bias in language models and demonstrates how to steer models away from this bias to improve performance.

Findings

Models are more influenced by high-authority endorsements.

Authority bias leads to increased confidence in incorrect answers.

Bias can be mechanistically identified and mitigated.

Abstract

Prior research demonstrates that performance of language models on reasoning tasks can be influenced by suggestions, hints and endorsements. However, the influence of endorsement source credibility remains underexplored. We investigate whether language models exhibit systematic bias based on the perceived expertise of the provider of the endorsement. Across 4 datasets spanning mathematical, legal, and medical reasoning, we evaluate 11 models using personas representing four expertise levels per domain. Our results reveal that models are increasingly susceptible to incorrect/misleading endorsements as source expertise increases, with higher-authority sources inducing not only accuracy degradation but also increased confidence in wrong answers. We also show that this authority bias is mechanistically encoded within the model and a model can be steered away from the bias, thereby improving its performance even when an expert gives a misleading endorsement.