AI Auto Politicians is run by Stuart Thompson. He was granted patents in Europe and America for a technology which he recalls the UK Government initially said was impossible. He used them to pioneer one of the world’s first AI-driven markets, now forecast to pass $100 billion by 2035. He has already done once, at scale, what this programme asks of public decisions: take the problem everyone says cannot be solved, solve it, and publish the proof.

[His sentence on the origin of the programme, in his words. Suggested: I kept seeing groups of people who had never met arrive at the same answer on their own. It looked like luck. It was the majority, unmeasured.]

A woman was attacked near his road in Kent. He asked his neighbours whether they wanted a camera. They said no. They did not want to be watched on their own street.

So he asked a different question. Would you accept a camera you controlled, that nobody ever watched? They said yes. That condition became the design.

The industry gave him the reasons it would not work. It took two years to crack and two more to make it. He recalls that the Government first called the idea impossible, and then funded it through its own SMART award. The patents were granted in Europe and America, and a European examiner upheld one of them against a professional challenge.

He has described what he was told at the time as “the millions of reasons why it was a stupid idea”.

Building it was the smaller half. In 2003 he sat down with officers at the Metropolitan Police station in Bexleyheath, backed by the head of the CCTV User Group, and asked the only question that mattered: if I build this, will you use it? He recalls the answer as yes. Nothing was signed. Nothing needed to be. Luton Borough Council then tested the camera against its own manned cameras and put the result on the record.

That is the pattern. Take the thing everyone says cannot be done, do it, and come back with a measurement instead of an opinion.

One rule came out of all of it: measure it, and publish the measurement. AI Auto Politicians applies it to the one thing no institution measures: what a place agrees on, and what its representative did about it.