UNDERSTANDING THE IMPACT OF PERCEIVED RISKS ON AI ADOPTION AMONG PHYSICIANS

Artificial intelligence (AI) is progressively transforming the medical field, offering new diagnostic and therapeutic possibilities. However, its adoption remains uneven among physicians, shaped by the tension between technological promises and perceived risks. This study explores the factors influencing AI adoption by combining the concepts of technological promises and the temporality of perceived risks.
We conducted 20 semi-structured interviews with physicians from various specialties, analyzed through a thematic approach using NVivo. Findings reveal a dual dynamic structuring attitudes toward AI. On the one hand, physicians view AI as a promising tool capable of improving diagnostic accuracy, streamlining patient management, and reducing administrative burden. On the other hand, four categories of perceived risks hinder adoption: (1) professional risks (loss of autonomy, algorithmic dependence), (2) risks for patients (dehumanization, standardized care), (3) algorithmic biases (inequities, diagnostic errors), and (4) data security concerns (breaches of confidentiality, control of sensitive information).
Perceptions are strongly shaped by temporality: immediate risks (such as errors or loss of expertise) are more salient, while delayed risks (such as long-term role transformation or evolving care standards) create persistent uncertainty. This temporal tension influences adoption decisions, particularly in specialties where human interaction is central.
The study underscores the need for proactive management of expectations and uncertainties to support the responsible integration of AI in medical practice.

DOI: https://doi.org/10.47743/rebs-2025-2-0002