In-Context Impersonation Reveals Large Language Models' Strengths and Biases

TL;DR

This paper investigates how large language models can adopt different personas in-context to reveal their strengths and biases across various tasks, including reasoning, exploration, and description.

Contribution

It demonstrates that in-context impersonation enables LLMs to exhibit human-like developmental behaviors, improve task performance, and uncover biases.

Findings

Impersonation reveals human-like developmental stages in exploration.

Domain-specific impersonation enhances reasoning performance.

Impersonation uncovers gender and expertise biases.

Abstract

In everyday conversations, humans can take on different roles and adapt their vocabulary to their chosen roles. We explore whether LLMs can take on, that is impersonate, different roles when they generate text in-context. We ask LLMs to assume different personas before solving vision and language tasks. We do this by prefixing the prompt with a persona that is associated either with a social identity or domain expertise. In a multi-armed bandit task, we find that LLMs pretending to be children of different ages recover human-like developmental stages of exploration. In a language-based reasoning task, we find that LLMs impersonating domain experts perform better than LLMs impersonating non-domain experts. Finally, we test whether LLMs' impersonations are complementary to visual information when describing different categories. We find that impersonation can improve performance: an LLM prompted to be a bird expert describes birds better than one prompted to be a car expert. However, impersonation can also uncover LLMs' biases: an LLM prompted to be a man describes cars better than one prompted to be a woman. These findings demonstrate that LLMs are capable of taking on diverse roles and that this in-context impersonation can be used to uncover their hidden strengths and biases.