University of York-Led White Paper Calls for Urgent Reforms to Empower Clinicians with AI in Patient Care
(IN BRIEF) A new White Paper, developed by the Centre for Assuring Autonomy at the University of York, the MPS Foundation, and the Bradford Institute for Health Research, warns that the benefits of AI in patient care could be lost if these tools do not effectively support clinicians. Building on the Shared CAIRE project’s research, the report highlights the risk that clinicians might become “liability sinks” if they are forced to shoulder the legal repercussions of flawed AI-driven decisions. Emphasizing the importance of preserving clinician autonomy, the paper advocates for AI tools that supply information rather than prescribe treatment, urging that clinicians be actively involved in the design and development process. The report calls for urgent revisions to product liability laws and swift action from regulators and developers to prevent clinician burnout and ensure that AI can truly enhance patient care.
(PRESS RELEASE) YORK, 26-Mar-2025 — /EuropaWire/ — A newly published White Paper warns that the transformative potential of artificial intelligence in patient care may be squandered unless immediate action is taken to tailor these technologies to the needs of frontline clinicians. Developed through a collaborative effort involving the Centre for Assuring Autonomy at the University of York, the MPS Foundation, and the Improvement Academy at the Bradford Institute for Health Research, the document draws on findings from the Shared CAIRE (Shared Care AI Role Evaluation) research project. This project evaluated six AI decision-support tools, engaging experts across safety, medicine, AI, human-computer interaction, ethics, and law.
The report identifies the “off switch” phenomenon—where clinicians, perceiving AI systems as burdensome, ill-suited, or legally risky, reject their integration—as the most significant obstacle to AI adoption in healthcare. It warns that if clinicians are forced to absorb disproportionate legal liability for AI-influenced decisions, they may become “liability sinks,” which would ultimately compromise patient care and stifle technological innovation.
Different AI integration models were assessed, ranging from systems that simply provide data to those that directly offer treatment recommendations or interact with patients. Both clinicians and patients showed a clear preference for tools that augment clinical judgment by highlighting critical information without supplanting the decision-making process. The paper strongly advocates for the involvement of clinicians in every phase of AI tool design and development, and calls for urgent reform of current product liability laws to better accommodate the nuances of AI-assisted care.
Professor Ibrahim Habli of the University of York, who led the Shared CAIRE project, explained that the White Paper delivers clear, actionable recommendations aimed at ensuring the safe and effective use of AI in the NHS and broader healthcare environments. He emphasized that these insights, derived from real-world scenarios and informed by both clinician and patient feedback, are immediately applicable. Professor Tom Lawton, a consultant in Critical Care and Anaesthetics at Bradford Teaching Hospitals NHS Trust and the Clinical and AI lead on Shared CAIRE, cautioned that if AI systems are developed to serve the interests of their creators rather than the needs of clinicians and patients, they risk causing burnout and inefficiencies that could lead to a wholesale rejection of AI technologies.
In conclusion, the paper offers seven targeted recommendations to prevent the premature “switching off” of AI tools, urging government bodies, regulators, and AI developers to implement these measures swiftly. Professor Gozie Offiah, Chair of the MPS Foundation, stressed that while rapid technological advancements hold enormous promise for healthcare, it is imperative that clinicians remain empowered, informed users rather than passive recipients of technology.
———-
First published in this link of EuropaWIRE.
