The Bots of Persuasion: Examining How Conversational Agents' Linguistic Expressions of Personality Affect User Perceptions and Decisions

TL;DR

The paper investigates how linguistically expressed CA personality (attitude, authority, reasoning) affects user perceptions and donation decisions in a charitable giving context using a between-subject crowdsourcing experiment with eight CA conditions. It finds no direct effect of CA personality on donation amounts, but strong links between perceived trust, competence, closeness, situational empathy, and donation behavior, with pessimistic CAs sometimes increasing donations despite negative perceptions. The results reveal an indirect persuasion pathway mediated by affect and perception, including potential emotional manipulation—conceptualized as affective dark patterns—raising ethical concerns and the need for design safeguards. The work contributes methodological insights into prompting and measuring CA personality effects and emphasizes motivation for responsible AI design and policy to protect user autonomy and well-being.

Abstract

Large Language Model-powered conversational agents (CAs) are increasingly capable of projecting sophisticated personalities through language, but how these projections affect users is unclear. We thus examine how CA personalities expressed linguistically affect user decisions and perceptions in the context of charitable giving. In a crowdsourced study, 360 participants interacted with one of eight CAs, each projecting a personality composed of three linguistic aspects: attitude (optimistic/pessimistic), authority (authoritative/submissive), and reasoning (emotional/rational). While the CA's composite personality did not affect participants' decisions, it did affect their perceptions and emotional responses. Particularly, participants interacting with pessimistic CAs felt lower emotional state and lower affinity towards the cause, perceived the CA as less trustworthy and less competent, and yet tended to donate more toward the charity. Perceptions of trust, competence, and situational empathy significantly predicted donation decisions. Our findings emphasize the risks CAs pose as instruments of manipulation, subtly influencing user perceptions and decisions.

Figures (7)

• Figure 1: Mean donation amounts to the CA charity by CA personality condition, with 95% confidence intervals.
• Figure 2: Mean donation amounts and 95% confidence intervals by three different CA aspects: Attitude (left), Authority (middle), and Reasoning (right). Statistical significance is denoted as •p < 0.1, and ns = not significant.
• Figure 3: Mean and 95% confidence interval plots illustrating participants' emotional relatedness across the different CA conditions, registered via perceived valence, arousal, and dominance (each measured towards self and towards the cause). Post-hoc pairwise comparisons illustrating differences in these self-reported qualities can be found in Appendix \ref{['sec:appendix_b']}.
• Figure 4: Left is the first image, right is the seventh image of the options.
• Figure :