AI Overreliance May Erode Critical Thinking in Medicine

More than 80% of U.S. physicians now use artificial intelligence in their work—double the rate from 2023, according to an American Medical Association survey. While most clinicians believe AI improves efficiency and reduces burnout, emerging research suggests a troubling trade-off: the technology may be quietly eroding the cognitive skills essential to medical practice.

A 2025 MIT experiment found that people's brain activity decreased when writing essays with AI assistance. Another study analyzing 370,000 college admission essays observed a narrowing range of original ideas after generative AI chatbots became available, suggesting creativity itself may be at risk.

Why it matters

Physicians' ability to think critically, question diagnoses, and generate novel solutions directly impacts patient outcomes. If AI tools become cognitive crutches rather than supplements, the medical profession could lose the very capabilities that distinguish human judgment from algorithmic output—particularly in complex cases where AI performs poorly.

The critical thinking gap

Medical students increasingly turn to AI applications like OpenEvidence for diagnostic support. While the tool scored 100% on the U.S. Medical Licensing Examination, a December 2025 pilot study found it answered only about one-third of complex subspecialty questions correctly.

Youngjin Cho, a professor of medical education at the University of Georgia School of Medicine, has observed students immediately consulting AI during exercises rather than reasoning through problems independently. "I do worry that it cuts learners' essential skills related to reasoning on their own, knowing the right resources to find information, and judging the information," Cho told the Association of American Medical Colleges.

Fares Alahdab, who directs graduate studies in health informatics at the University of Missouri School of Medicine, notes students are "starting to off-load their cognitive skills" before recognizing the shift. His institution has redesigned curriculum to include exercises where students assess AI output and identify hallucinations.

De-skilling among experienced physicians

Nearly 75% of clinicians worry about de-skilling—the dulling or loss of previously learned abilities—according to a recent industry survey. A 2025 Lancet study provided concrete evidence: Polish gastroenterologists using an AI-assisted endoscopy tool saw their tumor detection rate drop six percentage points when the tool was removed after three months.

Robert Wachter, chair of medicine at the University of California, San Francisco, called the findings sobering but argues de-skilling isn't universally problematic. "I no longer know how to read a map, and that's fine," he said. However, since AI tools aren't always available, physicians must maintain sharp clinical skills.

Wachter suggests medicine adopt practices from aviation, where pilots face simulator tests every six months to prevent autopilot-related skill degradation. AI tools could be designed to periodically test users rather than simply provide answers.

Preserving creativity and innovation

Adam Green, a Georgetown University neuroscientist who studies creativity, warns that "creative intelligence and creative problem-solving also lead to innovation, especially in the areas of science, medicine, and technology." He notes that measures of intelligence and critical thinking are declining for the first time in the internet era.

Green advocates testing and rewarding medical students for originality beyond AI's homogenized solutions, not just performance outcomes. Arthur Lazarus, an adjunct professor of psychiatry at Temple University, recommends a simple principle: "Do the work first, consult AI second."

Despite the risks, Wachter argues against avoiding AI adoption. "The status quo is immoral," he said, pointing to quality gaps, clinician burnout, and unsustainable costs. "If we say AI is too scary, we're going to stick with the status quo. I think that's not a neutral decision, that's a negative decision."

These findings and recommendations were first reported by the Association of American Medical Colleges.