Policy
AI is no longer a promise about the future. In medicine it is already reading scans and helping diagnose patients. The question now is whether the state knows how to use it.
Marc has built AI that is in clinical use. He led the development of the first AI diagnostic for prostate cancer cleared by the FDA, and ran a portfolio of CE-marked AI products for cancer. Earlier in his career, with the Cabinet Office's Open Innovation Team, he worked on how AI was being introduced across government.
That experience makes him impatient with both sides of the current debate. The tools are not hype. But they are not magic either. A model that performs well on the day it is approved can drift as patients and practice change. It needs monitoring once it is in use, and a clear rule for when a human has to take over.
Marc wants rules that are strict enough to protect people and clear enough that good tools are not left waiting. Behind that, the state needs the basic infrastructure: health data that is usable rather than locked in silos, and procurement that can buy software without treating it like a building contract.
Used well, AI can free up clinicians, speed up diagnosis and make public services more productive. Used badly, it will waste money and erode trust. The difference comes down to governance, and governance is the job of government.