Over 700,000 clinicians actively query medical AI engines for treatment decisions each month, rapidly outstripping formal clinical validation.
Physicians are integrating point-of-care search models into daily practice without waiting for randomized trials to prove patient safety.
Unvalidated Daily Practice
Between January 2024 and June 2026, over 950,000 U.S. clinicians submitted 271 million queries through OpenEvidence. Monthly active users surged from 355 to over 703,000 during that period.
More than half of these searches involved direct treatment decisions. Doctors frequently seek guidance outside their primary medical specialties.
That speed creates immediate risk.
Adoption Outpaces Patient Evidence
High search volume reflects convenience, not diagnostic accuracy. Retrospective usage stats confirm massive demand, but yield zero data on patient outcomes or error rates.
Using search models to manage unfamiliar conditions saves time. However, without controlled trials evaluating real-world efficacy, health systems are overseeing an unmonitored experiment at scale.
System leaders must prioritize rigorous outcome tracking rather than confusing rapid user growth with clinical safety.
