Artificial intelligence may soon outperform human physicians in key areas of patient care, raising concerns about workflow, liability, and the future of medical practice. A recent analysis in JAMA examines these challenges.
AI could surpass physicians—and their AI-assisted counterparts
A perspective essay published August 14 states that autonomous AI systems may soon deliver better outcomes than either human doctors working alone or those using AI as a tool. The authors suggest this shift could occur by 2030 for some medical tasks.
“Data from medicine and other fields show that when AI alone performs consistently better than humans, it also outperforms human-AI hybrids,” the authors wrote. The finding challenges the assumption that physician oversight improves AI-driven care. The analysis indicates that once AI surpasses human performance, adding a doctor to review its work may reduce accuracy.
This contradicts the current messaging from many AI vendors, who present physician sign-off as a necessary safeguard. The authors advise medical practices to reconsider whether that model will remain practical as AI advances.
Where AI already leads—and where it falls short
The essay identifies five areas where AI has shown an advantage over human physicians:
- Gathering patient information
- Diagnosing conditions
- Selecting tests to pinpoint diagnoses, while staying on budget
- Recommending treatments
- Managing chronic diseases, at least for hyperlipidemia, osteoarthritis, diabetes and breast cancer
The financial benefits are significant. In one study, a Microsoft diagnostic tool working within an $8,000 test-ordering budget reached the correct diagnosis on 56 complex cases roughly four times as often as physicians working without access to colleagues, textbooks or the internet—and did it at 19.1% lower cost per case ($2,396 vs. $2,963). Similarly, Cedars-Sinai’s AI system produced optimal treatment recommendations in 77.1% of 461 real patient cases, compared with 67.1% for physicians.
At scale, gaps like that in diagnostic accuracy and test utilization translate into fewer repeat visits, less unnecessary testing and less exposure to claims denied for lack of medical necessity.
Obstacles to adoption
Two forces work against smooth AI adoption in a practice, according to the article. One is algorithm aversion, a documented tendency, found in about 75% of studies on the topic.
The second barrier is the risk of deskilling. As clinicians depend on AI for tasks like documentation or diagnosis, their own skills in those areas may weaken. The authors note that this dynamic has already been observed in other fields.
Liability remains unresolved. If an autonomous AI tool misdiagnoses a patient, responsibility could—
The authors do not expect AI to replace physicians entirely. They argue the profession must prepare for a future where AI handles certain tasks more effectively—and where human oversight may not always enhance results. The challenge lies in adapting quickly without compromising patient care.
As practices integrate these tools, policy violations could create unexpected hurdles.
One hospital system already encountered this issue when a user was blocked with an error message after inputting incorrect billing codes into an AI-assisted system.
