Paper Type
Complete
Paper Number
PACIS2025-1724
Description
Algorithm dysfunction, a phenomenon where algorithms temporarily fail to perform their tasks, is common in human-algorithm collaborations. Despite its widespread occurrence, the understanding of the effects of algorithm dysfunction remains limited. This paper empirically investigates the impact of algorithm dysfunction on human reliance and collaboration performance through a field experiment in the logistic industry. The results show that algorithm dysfunction reduces human reliance and undermines collaboration performance. Notably, this negative effect does not completely recover even after the algorithm is restored to normal functioning. The occurrence of algorithm dysfunction can encourage users with a higher initial level of reliance to overcome overreliance and invest more effort in evaluating and improving algorithmic recommendations. Our study provides empirical evidence on the effects of algorithm dysfunction, offering valuable insights for businesses in managing algorithm dysfunction strategically.
Recommended Citation
DU, Muxuan; Wu, Ding; Wang, Lingli; Guo, Xunhua; Deng, Hongshuyu; and Zhuang, Xiaotian, "The Impact of Algorithm Dysfunction on Human-algorithm Collaboration" (2025). PACIS 2025 Proceedings. 23.
https://aisel.aisnet.org/pacis2025/aiandml/aiandml/23
The Impact of Algorithm Dysfunction on Human-algorithm Collaboration
Algorithm dysfunction, a phenomenon where algorithms temporarily fail to perform their tasks, is common in human-algorithm collaborations. Despite its widespread occurrence, the understanding of the effects of algorithm dysfunction remains limited. This paper empirically investigates the impact of algorithm dysfunction on human reliance and collaboration performance through a field experiment in the logistic industry. The results show that algorithm dysfunction reduces human reliance and undermines collaboration performance. Notably, this negative effect does not completely recover even after the algorithm is restored to normal functioning. The occurrence of algorithm dysfunction can encourage users with a higher initial level of reliance to overcome overreliance and invest more effort in evaluating and improving algorithmic recommendations. Our study provides empirical evidence on the effects of algorithm dysfunction, offering valuable insights for businesses in managing algorithm dysfunction strategically.
Comments
AI ML