When Doctors Should Tell You About AI: New Study Reveals Trust Depends on What the Algorithm Does
Chinese research shows patients accept AI for scheduling but resist it for treatment decisions—especially when stakes are high.

Illustration: AI-generated
When physicians use artificial intelligence in patient care, should they always disclose it? A new study suggests the answer is more nuanced than a simple yes or no, finding that patient trust hinges critically on what type of AI is being used and how serious the medical condition is.
Research published in npj Digital Medicine examined how 1,574 adults in mainland China responded to different AI disclosure scenarios across varying medical contexts. The findings reveal a clear hierarchy in patient comfort: administrative AI applications received significantly more favorable ratings than diagnostic or therapeutic AI tools, with high-severity conditions amplifying negative reactions specifically to therapeutic AI use.
The AI Trust Hierarchy
The factorial experiment tested patient responses across five disclosure conditions—administrative AI, diagnostic AI, therapeutic AI, explicit non-AI disclosure, and no disclosure—combined with two condition severity levels. Administrative AI, used for tasks like scheduling and documentation, consistently received the most positive ratings across measures of physician perception and care-seeking intention.
In contrast, therapeutic AI—systems that help determine treatment plans—faced the steepest trust barriers. This gap widened dramatically when patients considered high-severity medical conditions, suggesting that as clinical stakes rise, patients become increasingly wary of algorithmic involvement in treatment decisions.
Exploratory mediation analyses identified competence, trust, and empathy as key mechanisms explaining these different response patterns. The study also included a supplementary sample of 128 participants from Macao, where main effects were directionally consistent with the mainland findings.
The Paradox of Transparency
One of the study's most striking findings involves what researchers didn't expect: explicit disclosure that AI was not being used was associated with greater appointment willingness in high-severity conditions. This paradoxical effect suggests that in high-stakes clinical scenarios, patients may find reassurance in knowing their physician is making decisions without algorithmic assistance.
The research, funded by the Macau Science and Technology Development Fund, points to a fundamental tension in healthcare AI implementation. While transparency is generally considered an ethical imperative, blanket disclosure policies may inadvertently undermine patient confidence in precisely those situations where advanced decision support could be most valuable.
Implications for Healthcare Delivery
For healthcare providers and AI developers, these findings suggest that one-size-fits-all disclosure approaches may be counterproductive. The study indicates that healthcare organizations should consider implementing tiered transparency frameworks that distinguish between behind-the-scenes administrative automation and clinical decision-support systems.
Physicians treating serious conditions may need different communication strategies than those managing routine care. Health AI companies, meanwhile, face a market reality where administrative tools encounter fewer adoption barriers than clinical decision-making applications, potentially requiring differentiated deployment approaches based on condition severity.
As digital medicine continues to expand, understanding these nuanced patient preferences becomes critical for maintaining trust while advancing technological capabilities in healthcare delivery.
Sources
Follow Health AI Central
Keep up with health AI where you already read
On LinkedIn
The day’s strongest health AI stories in your feed, one or two each weekday.
On Google
Choose Health AI Central as a preferred source, and Google is more likely to show our reporting in Top Stories.
Reporting & accountability
How we maintain accuracy
Health AI Central separates reported facts, analysis, and public-source data. Readers can review our standards, request a correction, or consult the relevant research methodology before relying on a claim.