AMSTERDAM — Artificial intelligence (AI) integration into triage and clinical workflows is raising concerns about a growing “blind trust” among clinicians and patients. This trend involves deferring to algorithmic confidence over independent medical judgment, according to experts at the HLTH Europe 2026 conference. Panelists emphasized that a person’s information ecosystem, including social media consumption, podcasts, and AI interactions, increasingly determines health outcomes.
Clive Flashman, chief digital officer at Patient Safety Learning in Borehamwood, UK, stated that by 2030, health outcomes will be shaped more by information ecosystems than by proximity to hospitals or financial status. This shift highlights a critical change in determinants of health.
Automation bias, the human tendency to favor automated system suggestions even when incorrect, was a central discussion point. Flashman defined blind trust in this context as a clinician ceasing independent thought and judgment because an algorithm has provided information they believe or should act upon.
This bias carries significant clinical consequences. An Oxford study, cited by Flashman, examined how clinical users interacted with large language models (LLMs). The study found that clinical users employing LLMs were 1.5 times less likely to identify serious underlying conditions compared to control groups working without the software. Only 34% of the conditions suggested by the LLMs in the study were accurate.
A simulation involving subarachnoid hemorrhages further illustrated the risks. Entering identical symptoms with a single word variation—changing the timeline to include ‘sudden’—resulted in completely opposing advice. One user was told they could safely stay home, while another was instructed to call emergency services immediately. Flashman characterized this as a “life-and-death situation.”
The issue extends beyond clinical settings. Approximately one-third of US adults and 27% of Canadian patients reportedly use chatbots for health information. This trend positions AI as an unauthorized gatekeeper to care, influencing patient access and decisions.
Systemic access gaps are driving this reliance on AI. In Canada, an estimated 1 in 6 individuals lacks access to a traditional family physician. This void prompts patients to seek health information and guidance online, often through AI chatbots.
The increasing reliance on AI for health information and clinical decision support raises questions about accountability and regulatory oversight. Future developments will likely focus on establishing clear guidelines for AI use in healthcare and educating both clinicians and the public on its limitations. Monitoring the impact of information ecosystems on public health will be crucial as AI integration continues to expand.