When Machines Take Over: Lancet Study Warns Doctors Could Lose Critical Skills from Regular AI Use

As AI technologies become more embedded in clinical practices, their ability to transform medicine is being hailed globally. The promise: Accelerated diagnoses, reduced errors, and customized treatments. But hidden behind these shiny possibilities is a subtle danger: the deskilling of physicians who become overly dependent on AI assistance.

A landmark study in The Lancet included 1,443 colonoscopy procedures across several Polish healthcare facilities, with skilled endoscopists who perform these regularly. Researchers detected a significant reduction in physicians' capacity to spot precancerous growths without AI after AI support was implemented: adenoma detection rates reduced from 28.4% to 22.4% a 20% relative decrease within months of regular AI exposure. This indicates that during the time before AI existed, physicians detected fewer precancerous tumours than after they began working with machine assistants.

Why Does Skill Loss Occur?

Speed and accuracy of AI can produce a false sense of security, luring physicians into deferring to machines and eventually making less autonomous decisions. This "automation bias" makes clinicians doubt their own judgment; they may even defer to AI suggestions, opting for technical concurrence over professional judgment, a problem particularly dangerous when AI systems fail or become unavailable because of cyberattacks.


Apart from colonoscopy, specialists fear this wave may cascade through other areas of medicine. As medical dependence on AI increases, abilities sharpened by years of practice and training can become stale much like drivers who rely too heavily on autopilot are likely to be unprepared in case of emergencies. Deskilling could render physicians liable if technology malfunctions or needs close attention.

Striking a Safe Balance

The Lancet study authors and independent commentators urge the medical community not to rush headlong into full AI adoption. They call for more robust research into AI’s long-term impact on clinical skills and strategies for safeguarding human expertise. The key may lie in collaborating with AI as a partner, not a replacement, and in maintaining training requirements that ensure doctors can always revert to “manual mode” when technology falters.

While AI promises to deliver vast potential, the silent erosion of core clinical competencies must serve as a wake-up call. The finest medicine is a combination of innovation and unreplaceable human judgment a marriage of code and compassion