Impact of Agent-Generated Rationales on Online Social Conformity

Le résumé fourni par la source

Social conformity, in which individuals adjust their opinions to align with the majority, is a widely established phenomenon.As digital agents are increasingly integrated into group decision-making, while also having shown to convincingly present misinformation, it is crucial to understand their impact on online social conformity.In this paper, we investigate the effects of a large language modelpowered agent on social conformity in an online multiple-choice quiz (𝑁 = 80).We present participants with an agent's rationale for both informative (objective) and normative (subjective) questions.We collected participants' judgements and confidence both prior to and following the presentation of both the agent's judgement and rationale, as well as a fabricated distribution of other people's judgements.Our results replicate majority conformance and show a significant influence of agent rationale on conformity.We discuss the implications of our results for the integration of LLM-based agents into people's everyday decision-making tasks.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.