A new study published in the Journal of the American Medical Association argues that artificial intelligence may soon outperform human doctors in five fundamental medical tasks. The authors, including medical ethicist Ezekiel Emanuel and venture capitalist Vinod Khosla, claim that AI could deliver the best outcomes in taking medical histories, establishing diagnoses, identifying necessary tests, prescribing treatment, and managing chronic diseases. They argue that clinicians should refrain from meddling, as human involvement may degrade AI performance. 'We’re making a prediction for four years from now,' Emanuel said. 'You don’t think that [AI] is going to be way more advanced than a doctor?'
The study reviewed all published articles on AI in medicine since January 1, 2024, and found that medicine is rapidly approaching a transition point where AI alone could exceed physicians and physician-AI hybrids. The authors predict that a 'superior autonomous AI' will likely surpass humans using AI by 2030. This prediction, they admit, is 'unsettling but seems probable.' The study also highlights that some of the research it surveyed consists of simulations, not blind experiments, and not all support the article’s conclusion.
The authors acknowledge that AI is now so good that clinicians should step back, but the American Medical Association’s CEO John Whyte expressed concerns about removing humans from the clinical loop. He argued that AI tools must be used in the context of a care plan governed by a physician. Meanwhile, physician Robert Wachter, who initially doubted the study’s premise, now concedes that the authors have a point. He warned that while AI is good, humans are still vital for tasks like delivering grim prognoses and guiding patients to the best treatment paths.
Source: wired
