The Argument
At the 2026 Aspen Ideas Health conference, LinkedIn co-founder Reid Hoffman made a pointed case: doctors should treat AI as a second opinion tool, not as a threat to their profession. His argument rests on three pillars:
- AI catches what humans miss — Studies show AI-assisted diagnosis catches 12% more cancers in radiology screenings
- AI doesn't replace judgment — The final decision always rests with the physician, who understands context AI cannot
- The best outcomes come from human-AI teams — "Centaur medicine" outperforms both human-alone and AI-alone
Where AI-Assisted Diagnosis Works
The evidence for AI in healthcare is growing in specific domains:
- Radiology — AI flags 23% more early-stage tumors than unassisted radiologists
- Dermatology — AI matches board-certified dermatologists on 97% of cases
- Pathology — AI-assisted biopsy analysis reduces false negatives by 34%
- Drug interactions — AI catches 89% of dangerous drug combinations vs pharmacists' 76%
// Editor's Take
Hoffman's framing is exactly right. The debate between "AI will replace doctors" and "AI can't do medicine" is a false dichotomy. The real question is: how do we design systems where AI augments human expertise? The answer is second-opinion workflows — AI flags, doctors decide. It works in radiology, it works in drug interaction checking, and it will work in more domains as the evidence accumulates.