Paper Number
ICIS2025-1621
Paper Type
Complete
Abstract
In this paper, we explore whether and why the dominant finding of algorithm aversion in predictive AI (which focuses on forecasting-oriented tasks) may not apply to generative AI (which can produce new content). We propose that algorithm aversion may be reduced for generative (vs. predictive) AI and this reduced aversion may be mediated by lower reactance or higher trust. Through an experiment, we find that algorithm aversion is reversed for generative (vs. predictive) AI and trust mediates algorithm appreciation. Our findings contribute to algorithm aversion vs. appreciation literature by distinguishing between predictive and generative AI and identifying AI type as a boundary condition for algorithm aversion. While algorithm aversion in predictive AI can be reduced through automatic processes, we suggest trust plays a more critical role in attenuating aversion in the generative (vs. predictive) AI context. Methodologically, we extend the widely used weight-on-advice paradigm from predictive to generative AI context.
Recommended Citation
Lei, Zhanfei; Yin, Dezhi; and Zhang, Han, "From Predictive to Generative AI: Exploring Algorithm Aversion vs. Appreciation" (2025). ICIS 2025 Proceedings. 13.
https://aisel.aisnet.org/icis2025/gen_ai/gen_ai/13
From Predictive to Generative AI: Exploring Algorithm Aversion vs. Appreciation
In this paper, we explore whether and why the dominant finding of algorithm aversion in predictive AI (which focuses on forecasting-oriented tasks) may not apply to generative AI (which can produce new content). We propose that algorithm aversion may be reduced for generative (vs. predictive) AI and this reduced aversion may be mediated by lower reactance or higher trust. Through an experiment, we find that algorithm aversion is reversed for generative (vs. predictive) AI and trust mediates algorithm appreciation. Our findings contribute to algorithm aversion vs. appreciation literature by distinguishing between predictive and generative AI and identifying AI type as a boundary condition for algorithm aversion. While algorithm aversion in predictive AI can be reduced through automatic processes, we suggest trust plays a more critical role in attenuating aversion in the generative (vs. predictive) AI context. Methodologically, we extend the widely used weight-on-advice paradigm from predictive to generative AI context.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
12-GenAI