Work

Ethical Uncertainty in Human-AI Collaborative Decision-Making: A Study of AI-Assisted Radiology

Type: article
Abstract: Decision-making in professional work is often ethically charged, and underlying ethical principles are rarely unambiguous. This ambiguity extends to human-AI collaboration, yet the literature says little about how professionals navigate different ethical principles (e.g., tacit and formal) when working with AI. We address this gap through an empirical study of AI-assisted clinical decision-making, examining how healthcare professionals perceive and exercise professional discretion under ethical uncertainty. We find that professionals are reluctant to override an AI recommendation under ethical uncertainty, and situational pressures increase the contextual appeal of adherence to the AI recommendations. Our analysis further identifies two distinct forms of AI adherence: precautionary adherence, by which professionals suppress their uncertainty in fully accepting an AI recommendation, and signalling adherence, by which professionals adhere to AI recommendations while signalling their reservations in communicating the decisions. These two types of adherence produce varied effects on downstream decision-making tasks.
Language: English
Field: Medicine
Open Access status: green
OpenAlex ID: W7222909981
Cites: 53
Cited by:
Related to: 10