AI-only care predicted to surpass doctors by 2030

The article's bold prediction stems from a partnership between Emanuel, a prominent medical ethicist, and Khosla, a long-time AI advocate. For years, Khosla argued AI would do 85% of what doctors do by 2035, a claim Emanuel once dismissed as "bullshit." But after reading Wachter's book on AI and medicine, Emanuel began to reconsider. The resulting paper is not just a prediction; it actively advocates for removing humans from the loop, citing evidence that human involvement can degrade AI performance.

The pushback highlights a critical tension in medical AI. While some studies show AI's potential, real-world hurdles remain. The Nature study's finding that patients often can't use LLMs effectively points to a practical gap. Yet even skeptics like Wachter acknowledge the argument's force, suggesting that the hybrid model may not be permanent. The debate extends beyond medicine, raising existential questions about the role of human expertise in an AI-driven world.

The future likely involves a gradual shift. AI may first gain permission to prescribe, then take on more complex tasks. The "doorman fallacy" offers some comfort, suggesting doctors may find new roles beyond diagnosis. But for now, the JAMA article sets a clear timeline: by 2030, AI alone could provide the best care, leaving doctors to ask what's left for them.