Between map and maze: reframing trust in healthcare AI
Authors:
- Hannah S. PiehlHPHannah S. Piehl
- Ricky JanssenRJRicky Janssen
- Bart PendersBPBart Penders
- Rianne R. R. FijtenRFRianne R. R. Fijten
Open Access
- 26 March 2026
- journal article
- Published by Springer Nature in AI & SOCIETY
- Vol. 41 (7) , 6925-6935
Abstract
Artificial intelligence (AI) is often presented as a transformative technology for healthcare, promising to augment clinical decision-making, streamline workflows, and enhance diagnostic precision. Yet its integration into healthcare practice is shaped by the complex and often ambiguous notion of trust . While trust in AI has become a recurring theme across disciplines, there has been little systematic analysis of different conceptualizations of the term. This paper addresses this gap through an interdisciplinary scoping review that examines how trust, trustworthiness, distrust, and mistrust are articulated in literature on AI in healthcare. Drawing on 82 publications (2015–2025) retrieved from six databases (Web of Science, Scopus, PubMed, PhilPapers, SocINDEX, and ACM Digital Library), it maps how trust is defined, measured, or problematized, and identifies the conceptual, epistemological, and disciplinary boundaries that shape these approaches. The analysis puts forward five different conceptualizations: trust as a set of designable principles , trust as an attitude or belief , trust as a binary between cognition/affect or human/technology, trust as a structural mechanism within uncertainty, and trust as a relational process embedded in socio-technical systems. These framings illuminate how trust is treated as an attribute to be engineered, a behavior to be calibrated, or a relation to be cultivated—often removed from clinical practice. Yet, across literature, trust disruptions which manifest as distrust, mistrust, overtrust, and undertrust are undertheorized and are typically framed as obstacles to overcome rather than phenomena warranting analysis in their own right. By introducing these diverging conceptualizations, the paper argues that trust in healthcare AI is less a stable condition than a dynamic negotiation that reveals power relations, uncertainties, and institutional dependencies of AI. Recognizing distrust and mistrust as analytically productive allows for a more reflexive understanding of how trust is enacted, distributed, and contested within healthcare. This review contributes to an emerging body of work that reframes trust not as a prerequisite for technological adoption but as a contested and dynamic concept.Keywords
Funding Information
- Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NGF.1607.22.040)
This publication has 70 references indexed in Scilit:
- Artificial intelligence and society: mapping the research through a systematic reviewAI & SOCIETY, 2025
- Challenges in Defining and Measuring Trust and Distrust in Science2025
- Trust and Mistrust in Science: Beyond the Binary2025
- AI Ethics beyond Principles: Strengthening the Life-world PerspectiveScience and Engineering Ethics, 2025
- Why we should talk about institutional (dis)trustworthiness and medical machine learningMedicine, Health Care and Philosophy, 2024
- Trust in Machine Learning Driven Clinical Decision Support Tools Among OtolaryngologistsThe Laryngoscope, 2024
- Trustworthy artificial intelligence in Alzheimer’s disease: state of the art, opportunities, and challengesArtificial Intelligence Review, 2023
- Trust in medical artificial intelligence: a discretionary accountEthics and Information Technology, 2022
- Implementing Ethics in Healthcare AI-Based Applications: A Scoping ReviewScience and Engineering Ethics, 2021
- Implicit trust in clinical decision-making by multidisciplinary teamsSynthese, 2017