Empathic Artificial Intelligence and Human Creativity : The Role of User Awareness in AI-Assisted Idea Generation
Desiree Lorenz · Digital Library of the University of Innsbruck (University of Innsbruck) · 2026
Modern artificial intelligence (AI) systems are increasingly designed to engage with users in emotionally responsive and empathic ways. While such systems can simulate empathy convincingly, they do not and cannot genuinely feel it, yet users often respond as if they did, with potential consequences for how AI outputs are evaluated and integrated into human decision-making. This thesis investigates whether making users explicitly aware of the simulated nature of AI empathy shifts their perception of an AI chatbot, and whether such a perceptual shift in turn influences the novelty of ideas generated during human-AI co-creation. A between-subjects online experiment (N = 100) was conducted in which participants completed a small-space student productivity ideation task with an AI chatbot. In the awareness prime condition, participants received a brief reminder before the interaction that the AI's empathic expressions are simulated rather than genuinely felt; in the control condition, no such information was provided. Perceived empathy was assessed via the Perceived Empathy of Technology Scale (Schmidmaier et al., 2024), and idea novelty was rated by independent experts using the Consensual Assessment Technique. Perceived empathy was reduced numerically in the prime condition (M = 66.75 vs. 71.14, d = 0.24), though the difference did not reach significance with the achieved sample size. Behavioral engagement data nonetheless revealed substantial negative associations between perceived empathy and indicators of critical engagement (task duration r = -.50, user messages r = -.38, idea edits r = -.28, all p ≤ .006), providing preliminary support for an engagement-mediated mechanism linking AI empathy perception to creative outcomes. The thesis additionally contributes a transparent, reproducible quality-screening procedure to address online sampling challenges introduced by AI-assisted survey responding.