Machine heuristic in algorithm aversion: Perceived creativity and effort of output created by or with artificial intelligence
Sigurd Birk Hansen, Anders Hauge Wien, Tarje Gaustad
Abstract
Products and services labeled as AI-created are perceived differently than identical ones attributed to humans, reflecting the phenomenon of algorithm aversion. This perception gap is partly explained by individuals’ tendency to attribute lower effort and creativity to work produced by AI algorithms. Through two pre-registered experiments, we investigate how individuals evaluate creative content when it is labeled as created by AI, humans, or human-AI collaboration. While previous research has established algorithm aversion across various domains, our findings reveal nuanced patterns in creative contexts. Individuals consistently rate content labeled as AI-authored as less creative and less favorable compared to content labeled as human-authored. This effect is driven by a sequential mediation: perceived effort influences perceived creativity, which in turn shapes overall attitudes. Importantly, content labeled as human-AI collaboration performs significantly better than AI-only content, suggesting a potential mitigation strategy. Our main theoretical contribution is the identification of the machine heuristic (MH), individuals' preexisting beliefs about AI capabilities, as a key moderator. Individuals with stronger MH beliefs attribute more effort to AI-generated work, which positively influences perceptions of creativity along with behavioral intentions. Using Netflix show descriptions as stimuli, we show these effects extend beyond attitudes to impact behavioral outcomes like willingness to watch. These results expand algorithm aversion theory by establishing MH as an important boundary condition and highlighting human-AI collaboration as an effective mitigation approach, offering valuable insights for creative industries integrating AI while maintaining positive perceptions among users. • Examines AI-human collaboration perceptions, showing hybrid approaches mitigate negative reactions to AI involvement. • Discovered machine heuristic as a key moderator. Stronger machine beliefs lead to more favorable of AI-created work evaluation. • Revealed how authorship labelling influences attitudes via perceived effort and creativity, affecting viewing intentions. • Offers theoretical insights and practical implications for consumer responses to AI integration in creative industries.