Will you donate money to a chatbot? The effect of chatbot anthropomorphic features and persuasion strategies on willingness to donate (2412.19976v1)

Abstract: This work investigates the causal mechanism behind the effect of chatbot personification and persuasion strategies on users' perceptions and donation likelihood. In a 2 (personified vs. non-personified chatbot) x 2 (emotional vs. logical persuasion strategy) between-subjects experiment (N=76), participants engaged with a chatbot that represented a non-profit charitable organization. The results suggest that interaction with a personified chatbot evokes perceived anthropomorphism; however, it does not elicit greater willingness to donate. In fact, we found that commonly used anthropomorphic features, like name and narrative, led to negative attitudes toward an AI agent in the donation context. Our results showcase a preference for non-personified chatbots paired with logical persuasion appeal, emphasizing the significance of consistency in chatbot interaction, mirroring human-human engagement. We discuss the importance of moving from exploring the common scenario of a chatbot with machine identity vs. a chatbot with human identity in light of the recent regulations of AI systems.

• The paper finds that personified chatbots did not increase willingness to donate (WTD) and could provoke negative perceptions, challenging the idea that human-like features always enhance user experience.
• A key result shows that a logical persuasion strategy is more effective for non-personified chatbots, suggesting consistency between chatbot identity and persuasion approach is important.
• The study indicates that perceived anthropomorphism, whether mindless or mindful, did not significantly influence willingness to donate, separating user attribution of human traits from donation behavior.

Analysis of the Impact of Chatbot Personification and Persuasion Strategies on Donation Behavior

The paper titled "Will you donate money to a chatbot? The effect of chatbot anthropomorphic features and persuasion strategies on willingness to donate" offers a meticulous examination of how chatbot personification influences user perception and willingness to donate (WTD). The research aims to unravel the causal mechanisms behind these effects by experimenting with personified and non-personified chatbots utilizing either emotional or logical persuasion strategies.

Key Findings

The research yielded several insightful findings:

1. Impact of Personification: Contrary to prior assumptions, personified chatbots—those with human-like features, such as names and background stories—did not result in higher WTD. In fact, such anthropomorphic cues provoked negative perceptions towards the AI agents, challenging the notion that adding human-like attributes enhances user experience in donation contexts.
2. Persuasion Strategy Influence: The paper revealed that a logical persuasion strategy is more effective when implemented by non-personified chatbots. Logical appeals, which rely on data and statistics, appeared more congruent with the non-personified chatbots, thereby fostering more favorable perceptions. This aligns with the hypothesis that the consistency of cues between chatbot identity and persuasion strategies influences user attitudes.
3. Mediation of Perceived Anthropomorphism: Chatbot personification was found to evoke mindless rather than mindful anthropomorphism. However, neither type of anthropomorphism significantly influenced WTD, indicating that while users may unconsciously attribute human-like characteristics to chatbots, these perceptions do not translate into increased donation willingness.

Theoretical and Practical Implications

The paper offers significant implications for both the theoretical understanding of anthropomorphism in AI systems and practical applications in chatbot design. The findings suggest a reevaluation of the widely accepted use of anthropomorphic design in AI, especially in contexts requiring high trust and objectivity, such as charitable donations. Designing chatbots that leverage logical persuasion strategies without resorting to personification aligns better with user expectations and regulatory requirements emphasizing transparency.

Furthermore, the paper provides important insights into the psychological mechanisms of human-chatbot interaction, underscoring the tendency towards mindless anthropomorphism. It suggests that prevalent anthropomorphic design elements may need refinement to harness these mechanisms effectively.

Future Directions

The paper's limitations, including its small and homogenous sample size, point towards the need for broader research to generalize these findings across diverse demographic segments. Future work could explore a wider array of personification elements, assess long-term engagement patterns, and incorporate varied contexts beyond charitable donations to build comprehensive chatbot interaction models.

In conclusion, this research enriches the discourse on chatbot design by challenging existing paradigms about anthropomorphism and providing empirical evidence for the optimization of chatbot interactions in specific domains like fundraising. As AI continues to integrate into the non-profit sector, carefully calibrated designs will be crucial to align technological tools with organizational goals and user expectations.