On May 23rd, I spoke at Anti-Aging Festival 2026 at Palatul Bragadiru in Bucharest. Ten minutes, one idea: AI is not going to enter medicine — it's already inside it.
The talk opens with ten seconds of silence and a heartbeat ticking once per second. Every beat is one person dying of cancer somewhere in the world. Roughly 9.7 million people per year. I wanted the room to feel the scale before any of the AI talk landed.
"Nothing I'm about to show you is the future. It's happening now. The question isn't if AI enters medicine — it's how fast."
One cancer, one study: MASAI
To make the abstract concrete, I anchored the talk on a single example: breast cancer screening. In Europe today, every mammogram is read by two radiologists — the second pair of eyes catches what the first might miss. The setup is robust, but radiologists are tired, and cancers still slip through.
Sweden asked the obvious question: what if one of the two readers were a machine? The MASAI trial at Lund University put it to the test — a real randomized clinical trial, the first of its kind in AI breast screening, with 105,934 women enrolled. Half were screened the usual way; the other half had an AI reading alongside the radiologists.
The numbers came back like this:
- +29% more cancers detected
- −12% fewer interval cancers (the ones that normally slip between screenings)
- −27% fewer aggressive cancers found late
- −44% less reading workload for radiologists
More cancers found. The dangerous ones found earlier. Less human work, not more. And for context: breast cancer caught early has a 5-year survival of around 99% — caught late, it drops to roughly 30%. Early detection isn't a nice-to-have. It's the difference between living and not.
But MASAI is just one example
From there, the talk moves into a fast montage. AI is already doing three different kinds of work inside medicine — and each one was true before I stepped on that stage.
AI reads.
- Lung CT — an AI from Google and Northwestern beat six radiologists at lung cancer detection
- Heart — Mayo Clinic's AI-ECG sees patterns in a routine ECG that the human eye can't
- Eye — IDx-DR, the first fully autonomous AI diagnostic ever approved by the FDA, for diabetic retinopathy
AI creates.
- AlphaFold solved the structures of around 200 million proteins — work that earned the 2024 Nobel Prize in Chemistry
- An MIT AI sifted through millions of molecules and found halicin and abaucin — entirely new structural classes of antibiotics, active against multi-resistant bacteria
AI predicts.
- Sybil, from MIT, predicts lung cancer risk up to six years before it appears
- Your biological age can now be read from blood, from a retina, and — with FaceAge — from a selfie
The closing
The talk ends where it started — with the heartbeat. Three more beats. Then ten seconds of silence, phone held to the microphone, no beats at all. That's the world I hope to open the same talk with in ten years.
Watch the talk
Thanks to the Anti-Aging Festival team for the invitation and the room, to Laude-Reut Educational Complex for partnering on the event, and to everyone who stayed quiet for that opening minute. If your event wants a talk that pulls AI out of the hype lane and into clinical reality, I'm happy to put one together.