Paper Type

Complete

Abstract

As reliance on AI systems in professional settings continues to grow, understanding how individuals assign responsibility when human-AI collaboration produces flawed outcomes has become a critical concern. This study examined how perfect automation beliefs and self-reported AI expertise shape responsibility attribution between a human analyst and an AI system. Using vignettes designed to simulate a realistic human-AI collaborative work environment, participants (N = 48) completed measures of automation beliefs, AI expertise, and responsibility attribution in both pre- and post-error contexts. Perfect automation beliefs consistently and significantly affected responsibility redistribution across contexts, reducing blame attributed to the human analyst while increasing blame attributed to the AI system. Self-reported AI expertise showed no significant effects pre-error, but significantly predicted greater AI blame following failure. These findings extend the theory of automation bias into the domain of responsibility attribution and carry practical implications for designing accountability structures in AI-assisted organizational workflows.

Paper Number

1821

Comments

SIG OSRA

Share

COinS
 
Aug 15th, 12:00 AM

Who Is Responsible When Human-AI Collaboration Fails? The Role of Perfect Automation Beliefs and Self-Assessed Expertise in Responsibility Attribution

As reliance on AI systems in professional settings continues to grow, understanding how individuals assign responsibility when human-AI collaboration produces flawed outcomes has become a critical concern. This study examined how perfect automation beliefs and self-reported AI expertise shape responsibility attribution between a human analyst and an AI system. Using vignettes designed to simulate a realistic human-AI collaborative work environment, participants (N = 48) completed measures of automation beliefs, AI expertise, and responsibility attribution in both pre- and post-error contexts. Perfect automation beliefs consistently and significantly affected responsibility redistribution across contexts, reducing blame attributed to the human analyst while increasing blame attributed to the AI system. Self-reported AI expertise showed no significant effects pre-error, but significantly predicted greater AI blame following failure. These findings extend the theory of automation bias into the domain of responsibility attribution and carry practical implications for designing accountability structures in AI-assisted organizational workflows.

When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.