Years ago, when I was a minister at the Department of Culture (DCMS), I had a row with my opposite number at the Department of Trade (DTI). DCMS had over a few years accumulated a number of quangos working in the digital space. DTI had a few, too. Money was tight, and I suggested a review, to see if we could merge some, and abolish others. I was told in no uncertain terms where to get off.
The creation of the Sovereign AI fund a few weeks ago triggered this memory. The fund is a great idea, ably stewarded into being by the brilliant venture capitalist James Wise and run by the equally brilliant investor Josephine Kant. It will be housed in the Department of Science (DSIT). It will invest in AI startups that promise to give the UK some independence in the AI space.
But there are other government organisations that already have a lot of heft in investing in startups. Innovate UK has just announced its priorities, which include investing in AI startups. The British Business Bank has also set out a new strategy which includes, you guessed it, investing in early-stage AI companies. Both these institutions are well-led and do great work. So in announcing a Sovereign AI fund, wouldn’t the government have made everybody’s life easier by parking it with one of them? I don't want to confuse you by bringing up the National Wealth Fund, but that also does similar work to the Business Bank, but at a larger scale. Oh and I almost forgot ARIA — our version of DARPA, which invests in moonshot projects, a lot of which (I’m guessing here) involve AI.
Confused? Me too. That’s at least five institutions that invest, in theory and I guess in practice, in AI companies. What is a poor founder meant to do? I’m sure there are subtle differences in the nature and scale of investment, and some have broader remits that go beyond AI, but that’s not the point. It’s an increasingly confusing and complex funding landscape, and it seems it’s the customer’s job to navigate it. That creates unnecessary friction.
That’s just the funding side of things, and I am sure readers could come up with more examples. It’s equally confusing on the policy side of things. Ofcom has overall responsibility for regulating content on platforms, through the online safety act. Then we have the AI Security Institute, which has recently rebranded from the AI Safety Institute. I wonder how they both work with Alan Turing Institute, the UK Cyber Security Council, the Centre for Data Ethics, the Information Commissioner’s Office, the NHS AI Lab and Ofsted in education.
We know how this happens. Every year ministers think of what they can announce that will grab a headline and give the impression that they are taking action. And each initiative is often well meant. But execution is not the government’s forte, and they never ask — is there an existing body that could seamlessly take on this new initiative? So over the years we end up with a cumulative mess. And it needs sorting out.
I’m sure if I did more research I would work out how all of the above fits together. But I’m a busy person. And so are you. And that’s the point.
