Paper Type
Complete
Paper Number
1292
Description
AI algorithms have been widely integrated into the decision-making process on various business platforms. However, the black-box nature of algorithmic decision-making poses challenges to user acceptance. AI transparency is seen as a potential solution. Drawing upon the cognitive dissonance theory and the cognitive load theory, this study explores how AI transparency affects algorithmic acceptance. The questionnaire survey was conducted both online and offline, with 646 participants. The empirical results indicated that AI transparency has a dual impact on users' acceptance of the algorithm: a negative pathway through cognitive dissonance and a positive pathway through cognitive load. Additionally, a test of multiple serial-mediation examines the pathway of AI transparency-cognitive load-cognitive dissonance-algorithmic acceptance. We further introduce a novel user characteristics variable: the desirability of control, moderating the effect of AI transparency on algorithmic acceptance. These findings contribute to human-computer interaction literature and the development of transparent AI designs.
Recommended Citation
Chen, Jiamin; Jiang, Yi; and Zheng, TianQi, "Unraveling the Double-Edged Sword Effect of AI Transparency on Algorithmic Acceptance" (2024). PACIS 2024 Proceedings. 5.
https://aisel.aisnet.org/pacis2024/track13_hcinteract/track13_hcinteract/5
Unraveling the Double-Edged Sword Effect of AI Transparency on Algorithmic Acceptance
AI algorithms have been widely integrated into the decision-making process on various business platforms. However, the black-box nature of algorithmic decision-making poses challenges to user acceptance. AI transparency is seen as a potential solution. Drawing upon the cognitive dissonance theory and the cognitive load theory, this study explores how AI transparency affects algorithmic acceptance. The questionnaire survey was conducted both online and offline, with 646 participants. The empirical results indicated that AI transparency has a dual impact on users' acceptance of the algorithm: a negative pathway through cognitive dissonance and a positive pathway through cognitive load. Additionally, a test of multiple serial-mediation examines the pathway of AI transparency-cognitive load-cognitive dissonance-algorithmic acceptance. We further introduce a novel user characteristics variable: the desirability of control, moderating the effect of AI transparency on algorithmic acceptance. These findings contribute to human-computer interaction literature and the development of transparent AI designs.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
Interaction