Periodical tech commentary from the in-between.
This month, the UK Government published its 10 year health plan for England. Among its more futuristic promises was a vision where the Care Quality Commission (CQC), leads a more data-driven inspection model. Even more ambitious; an AI powered early warning system to detect issues in healthcare services before they spiral.
On paper, this sounds extremely promising. But one very basic question that nobody seems able to answer: what data is this AI going to be trained on?
Because at the time of writing, the evidence seems to suggest that the CQC doesn't have a data asset, but a data crisis.
Just last year, an independent review of the CQC's new £99 million system (partially funded by taxpayers) laid things bare. According to the report, digital transformation of the CQC, as well as the new IT system, were all failing to deliver. What's more, the data is in a dire state. It's not accurate, or standardised and the CQC needs to go back to the drawing board.
In the context of the plans, this is not a small glitch. AI systems need massive amounts of accurate, standardised and clean data to work. Without that, any predictions or alerts they generate are not just unreliable, they are dangerous.
The CQC is just one example of a broader trend: public institutions, under funding pressure, chasing “innovation” without first securing the fundamentals—Lead by people who lack technical understanding, control significant budgets, and are all too happy to blindly pan for gold.
I appreciate the appeal. As Arthur C. Clark once said, 'any sufficiently advanced technology is indistinguishable from magic'. However, I can promise you; AI really is not magic. It's closer to a Mechanical Turk: impressive on the surface, but limited by the human behind the curtain.
If we don't fix the data first, this won't be innovation. It will be expensive, esoteric theatre. And much like British theatre establishment in real life; we will be sitting here again in five years asking where all the money went.
Real innovation will have to start with the unsexy task of working on the underlying data and infrastructure. Not promising to automate existing failures.
