Healthcare experts are warning that growing reliance on artificial intelligence could weaken clinicians’ independent diagnostic and decision-making skills, a phenomenon known as AI-driven deskilling. In this Healio report, physicians and clinical informatics experts highlight concerns that clinicians may struggle when AI tools become unavailable or produce incorrect recommendations.
One example cited is a 2025 observational study published in The Lancet Gastroenterology & Hepatology, which found that adenoma detection rates during colonoscopies performed without AI declined from 28.4% in the three months before AI implementation to 22.4% in the three months afterward. Experts caution that performance may deteriorate when clinicians become accustomed to AI assistance, although the findings do not establish that AI use alone caused the decline.
The report identifies endoscopy, radiology, pathology and other pattern-recognition-heavy specialties as particularly vulnerable. Recommended safeguards include periodic assessments of clinicians’ unassisted performance, selected procedures conducted without AI, monitoring for automation bias, and clinician participation in AI governance. Medical education should also ensure trainees demonstrate core clinical competencies independently before relying on AI, alongside developing AI literacy.