Will AI Really Cure Most Diseases Within a Decade?

It has become one of the signature promises of the AI boom. Dario Amodei of Anthropic has suggested that AI could help cure most human disease within five to ten years. Demis Hassabis of Google DeepMind and Sam Altman of OpenAI have made similar forecasts. The pitch is intoxicating, and it is doing real work in the industry's fundraising. The trouble is that the people who actually develop medicines are not buying the timeline.

Cancer researchers surveyed by The Information, including several who describe themselves as AI enthusiasts, pushed back on those cure-by-the-2030s predictions. Their objection is not that AI is useless in biology. It plainly is not. The objection is that intelligence was never the main bottleneck. Biology is slow and messy, and it resists shortcuts. A model that can read every paper ever written still cannot skip the years of trials that tell you whether a drug works and whether it harms people.

What the record actually shows

Here is the number that should anchor any honest conversation. As of late 2026, no drug designed or discovered by AI has received full approval from the US Food and Drug Administration. Not one. More than 200 AI-originated programs are working their way through clinical trials, and analysts give the first approval a reasonable chance in the next year or two. That would be a genuine achievement. It is also a long way from curing most disease.

AlphaFold, the DeepMind system that predicted the shapes of nearly every known protein, is held up as the showcase for AI in medicine. It is a remarkable tool, and it has not yet produced an approved drug. The gap between a brilliant prediction and a pill that survives a trial is where most of the hard, unglamorous work of medicine lives, and it is precisely the part that intelligence alone does not clear.

So is it a myth?

Not exactly. AI is already speeding up parts of drug discovery. It has helped hunt for new antibiotics and design protein binders, and tools that predict cancer recurrence are reaching clinics. The useful distinction is between help and miracle. The claim that AI will accelerate medicine is well supported. The claim that it will cure most diseases within a decade is a forecast made by people raising money, not a finding backed by evidence. Treat the first as news, and the second as a hope worth watching with your skepticism intact.