Reeves is betting big on AI, but does the public back her?
Rachel Reeves is betting the UK’s economic future on an AI-driven productivity surge, but there are serious questions about the plausibility of this bet given the UK’s weaknesses in implementation, scaling capacity, talent retention and public pessimism, says Paul Armstrong
Rachel Reeves has placed an enormous bet on AI rescuing the UK’s dwindling economy. Fraser Nelson argued in the Times that the Chancellor is effectively gambling that AI will trigger a productivity surge strong enough to lift tax receipts, stabilise debt and shift the country out of its long malaise. While many in government speak as if this outcome is inevitable, serious questions remain about what is plausible, what is fantasy and what needs to be confronted now before the gap between ambition and delivery becomes unmanageable.
Momentum inside UK firms is real enough. Almost a quarter of businesses already deploy AI in some form according to the latest ONS survey. The number alone is not a sign of transformation, yet the speed of uptake signals something deeper. AI no longer sits in innovation labs or side projects, it’s moving into everyday workflows in finance, media, logistics, research and operations. Early adopters are already bringing automation into the centre of their processes rather than leaving it at the edges.
Consultants continue to publish optimistic forecasts because the models look promising if adoption becomes systemic. KPMG suggests generative AI could add more than pounds of productivity to the UK over the medium term. A broader PwC estimate points to far larger gains once AI diffuses across entire value chains. These numbers depend on firms doing the tiny thing of totally redesigning how they operate. Right now they are sprinkling AI on problems rather than seeing if they can solve the problem at all. Reeves has given herself a huge cross to bear.
Where’s the money?
Investment signals political intent but also exposes gaps. One flagship example is the government backed Isambard AI supercomputing cluster, designed to give researchers serious compute power. Oxford’s expansion through the Ellison Institute shows willingness to build world class AI enabled science facilities, but other nations are already ahead. The University of Florida built Malachowsky Hall, a fully AI native research building, Duke University has an academic partnership with OpenAI that embeds machine reasoning across disciplines, Britain gets a data deal.
A deeper problem sits below these announcements. Britain has a pattern of world class ideas and weak implementation due to venture scaling capacity remaining shallow compared with the United States. Corporate adoption is slower than European peers, and AI research talent drains away because the best researchers still leave for better funded labs abroad. These weaknesses compound each other. Britain invents, and others industrialise. Reeves would do well to address these issues head on if serious about AI being an economic factor.
Risk is rising just as quickly as optimism thanks to the Bank of England being sniffy on how large models could introduce new fragilities into financial markets. Synthetic sentiment, automated trading and opaque model behaviour create conditions where mispricing becomes systemic. Public institutions face the opposite challenge with over enthusiastic adoption of AI before clear governance is in place. Moves like this erode trust in government, and those who misread synthetic signals could make policy on illusions rather than evidence. Not good for long or short term smart planning for companies.
Workforce disruption is already visible with uneven effects across regions and sectors. Highly skilled roles are being transformed while middle tier roles risk getting hollowed out. Low wage roles are vulnerable from automation, and that’ll add to everyone’s woes. Where productivity gaps go is a good question.
Tom Johnson, MD at Trajectory, a leading UK foresight agency, isn’t bullish for Reeves. According to their monthly Optimism Index barometer: “The UK public are stuck in a pessimistic mood. Since January 2022, there have been only four months in which the public have been broadly optimistic about the future.” On AI, the nation is split: “Only nine per cent of us are enthusiastic and optimistic about AI, while 45 per cent are actively hostile. This hostile group don’t see the benefits of AI, either to themselves or society, and aren’t keen to see business or government make greater use of AI technologies.” If Reeves wants to win with AI, she needs leaders to get the messaging right yesterday.
The UK has big steps to take
Firms need leadership that understands AI’s strategic impact, not just how it’s made. National compute must be broadly accessible rather than concentrated in elite institutions. Talent needs to be retained rather than incentivised to bolt to Silicon Valley. Public sector adoption must be responsible rather than reckless. None of which is easy, fast or cheap, but Britain has strengths in research, regulation and creative industries that aren’t worthless. Britain does have weaknesses in skills, scaling, infrastructure and long term industrial discipline. Not small hills to climb, and growth isn’t going to emerge automatically from Reeves and co spouting optimism and inking a few datacentre deals.
Boards need to begin modelling multiple potential outcomes rather than assume Whitehall’s optimism will carry them forward. One future features rapid AI adoption, productivity gains and aggressive competitive divergence. Another features stalled projects, employee resistance, regulatory friction and stranded capital. Both futures remain equally likely. Firms that assume the government is right risk walking into a strategic blind spot at speed.
Practical steps look deceptively simple
Work must be mapped at task level rather than job title level. AI should be deployed where it can remove friction, raise quality or unlock new propositions. Governance has to sit alongside deployment rather than act as a late stage apology. Risk teams must be embedded from the start, and procurement should shift from experimentation to measurable operational change. Workforce planning should move in tandem with automation so people know what is coming and why. If you want a case study look at Amazon ripping off the plaster earlier this year. Expect more of this to happen in 2026.
Whole sectors need realistic timelines. Finance will move quickly because incentives align. Logistics will follow because of tangible gains. Public sector organisations will move slowly because procurement is political and risk tolerance is fundamentally low. SMEs will need targeted support or risk being left behind. Universities and colleges will need to rethink curricula faster than they prefer. Multinational firms will need to navigate conflicting regulatory regimes and increasingly divergent safety standards. There’s a huge opportunity for anyone who is training or creating tools that help AI get into the veins of businesses.
Growth will only come from firms that treat AI as disciplined operational reconstruction rather than national theatre. Political ambition will never compensate for organisational unreadiness and the UK has a small window to convert aspiration into economic reality. Firms that take steps to shorten the distance between rhetoric and capability now will see the best results. Those who have ambition without architecture are just making noise. Leaders who recognise that difference before the market forces clarity upon them will dominate the next economic chapter.