Pfeiffer, JellaHaas, AlexanderHeßler, Pascal OliverPascal OliverHeßler2025-03-122025-03-122025-03-10https://jlupub.ub.uni-giessen.de/handle/jlupub/20251https://doi.org/10.22029/jlupub-19606This dissertation examines how anthropomorphized conversational agents (CAs) influence user behavior in prosocial decision-making contexts. It starts with the growing significance of CAs, particularly following the release of ChatGPT, and addresses the challenge of algorithm aversion—people’s tendency to distrust algorithms. Based on experimental studies, two key research questions are explored: (1) Does algorithm acceptance differ between prosocial and for-profit contexts? (2) How do anthropomorphized CAs affect user behavior in prosocial situations? The findings reveal that algorithm aversion is more pronounced in prosocial contexts but can be mitigated through the targeted design of CAs, such as the use of anthropomorphic features. Notably, perceived competence plays a more critical role than expected warmth in reducing aversion. This work contributes to a deeper understanding of human-CA interactions and offers practical insights for designing trustworthy, user-centered CAs.enAttribution-NonCommercial-NoDerivatives 4.0 Internationalddc:004ddc:650