AIHealthcare Analytics

Knowledge Wiki

Wiki › Popular AI Models Get Medication Doses Wrong, Some Dangerously: Doximity Study

Popular AI Models Get Medication Doses Wrong, Some Dangerously: Doximity Study

OVERVIEW
Rev 1 Oct 31, 2026 05:42 UTC 4 sources

Doximity released research showing that popular general-purpose AI models (ChatGPT, Gemini, Claude) make dangerous medication dosing errors, while specialized clinical AI tools demonstrate significantly higher accuracy. The study underscores the risks of deploying unvalidated AI models for clinical decision support. Key Findings: General-purpose LLMs produced dosing recommendations that deviated substantially from clinical guidelines, with some errors potentially harmful. Doximity's specialized dose verification tool outperformed general-use models by 40+ percentage points on accuracy metrics. The research challenges assumptions that general-purpose AI can safely substitute for domain-specific clinical tools. Clinical Practice Implications: Clinicians and health systems using general-purpose AI for medication recommendations face liability and patient safety risks. The study reinforces FDA's rationale for requiring clinical validation before AI tools enter clinical workflows. Healthcare organizations must establish governance frameworks to prevent unsanctioned use of general-purpose AI for clinical decision-making. ---

Sources: harvested
🔒

Subscribe to read full wiki analysis

Full wiki content — including source provenance, contradiction tracking, related pages, and analyst notes — is available to Subscriber members.

View Plans & Subscribe

New users get a 7-day trial — no credit card required.

← Back to Wiki Index