Can a Hallucinating Model help in Reducing Human "Hallucination"?

TL;DR

This paper investigates whether large language models can reduce human hallucinations and misinformation by acting as personalized debunking agents, comparing their rationality to humans and proposing methods based on psychological persuasion theories.

Contribution

It introduces a novel approach to utilize LLMs for countering misinformation, grounded in psychological models, and explores their potential as debunking tools.

Findings

LLMs outperform humans in detecting logical fallacies.

Proposed methodologies leverage psychological theories for effective misinformation correction.

LLMs show promise as personalized debunking agents.

Abstract

The prevalence of unwarranted beliefs, spanning pseudoscience, logical fallacies, and conspiracy theories, presents substantial societal hurdles and the risk of disseminating misinformation. Utilizing established psychometric assessments, this study explores the capabilities of large language models (LLMs) vis-a-vis the average human in detecting prevalent logical pitfalls. We undertake a philosophical inquiry, juxtaposing the rationality of humans against that of LLMs. Furthermore, we propose methodologies for harnessing LLMs to counter misconceptions, drawing upon psychological models of persuasion such as cognitive dissonance theory and elaboration likelihood theory. Through this endeavor, we highlight the potential of LLMs as personalized misinformation debunking agents.