AI POLICY
Britain has a genuine opportunity in AI. I’m just not convinced we’re making the most of it.
The rhetoric is excellent. “AI superpower.” “Pro-innovation regulation.” “Compute sovereignty.” These are the right things to say, and the people saying them clearly understand the strategic stakes. But rhetoric and reality have a way of diverging, and the gap between the UK’s stated AI ambitions and its actual trajectory deserves an honest look.
What’s Going Well
The research base is genuinely world-class. The universities — Oxford, Cambridge, UCL, Edinburgh, Imperial — produce AI talent at a level that punches well above the UK’s economic weight. DeepMind remains one of the most important research organisations in the world. The financial services sector has been relatively sophisticated in deploying AI, and the regulatory sandbox approach has allowed experimentation that more cautious jurisdictions have blocked.
What Isn’t
Compute is the constraint that keeps coming up in every honest conversation I have with people in the sector. The UK simply doesn’t have enough sovereign compute capacity, and the government’s commitments in this area have been slower and smaller than the stated ambition requires.
“The UK produces world-class AI talent. It then watches a significant proportion of that talent relocate to San Francisco.”
Talent retention is the other elephant in the room. The UK produces world-class AI talent. It then watches a significant proportion of that talent relocate to San Francisco, where the salaries are higher, the equity is more valuable, and the concentration of frontier AI work is unmatched. A visa policy that makes it easier to hire a plumber from abroad than an AI researcher from a non-EU country is not an AI superpower strategy.
The Honest Assessment
The UK is likely to be a meaningful player in AI — but “meaningful player” and “superpower” are different things. Closing the gap requires decisions, investment, and trade-offs that the current political environment makes genuinely difficult. Acknowledging that isn’t defeatism. It’s the starting point for a more honest conversation about what’s actually achievable.
Tags: Artificial Intelligence • Opinion • Technology & Society • 192.168.1.22/