Official UK messaging about artificial intelligence has leaned for roughly two years on a headline estimate of £400 billion ($538 billion) in possible extra economic output. The number was meant to reassure people worried that AI could cost many jobs or trigger a downturn. Wired now reports that the estimate came mostly from one outside study, and that the government’s own figure is much smaller.
How the figure came to light
Last week, AI minister Kanishka Narayan responded to a question in Parliament and conceded that the £400 billion number sat well above what the government’s own forecasts show. He traced it to a 2023 study that Google paid for, carried out by Public First, a British consultancy that now trades as PF Global.
Also read: Google says schools can turn AI and Search on after backlash
According to Narayan, AI uptake and investment in data could add between £55 billion and £140 billion to the UK economy by 2030, with £120 billion as the middle estimate. He called even this newer calculation highly uncertain. Wired points out that his answer neither retracted the £400 billion figure nor stated that it was mistaken.
Why the two numbers differ
PF Global’s head of economics, Jonathan Dupont, told Wired that part of the gap reflects how slowly the government has adopted AI. In his view, the official numbers presume a slower take-up than his firm’s model, built in 2023, had anticipated. He added that AI changes quickly, so the assumptions behind any model need regular revision.
PF Global says its underlying view of the technology has not changed. Given the slower uptake it has observed, though, its central 2030 estimate now sits close to the government’s figure, with faster growth expected in the years after. The firm suggested that more top-down adoption and use of AI within civil service work could help. It also says its Google work continues and that its research is independent.
Political and expert reaction
Dame Chi Onwurah, Labour MP and chair of the Commons Science, Innovation and Technology Committee, warned Wired that exaggerating what technology can do weakens the argument for change. Her committee has identified exaggeration as a major obstacle to modernising public services. She called on ministers to be realistic about AI and open with the public about likely savings.
Lord Holmes of Richmond, a Conservative peer, had raised the source of the £400 billion claim with ministers in January 2025 and received no reply. He told Wired that the figure, repeated by government and then by other bodies, has now been substantially and properly revised, and that officials should have explained this earlier. He stressed that reliable data matters because public spending decisions rely on it.
Sumedha Deshmukh of the Ada Lovelace Institute welcomed the updated figures but said it was troubling that ministers had relied for nearly two years on research funded by Google. She argued that precise-looking AI numbers often contain hidden assumptions and motives, and that companies frequently pay for or shape the studies behind them. Wired also notes that former technology minister Peter Kyle cited the £400 billion number while promoting AI.
Timeline and what happens next
Ministers have been working from the lower estimate since at least November 2025, roughly a year before this week’s admission. Even so, Skills England, the public body that oversees apprenticeships, was still publishing the older number about four months after that.
Wired says several government departments and agencies relied on the Google-linked figure across two different administrations. The wider issue is how governments balance backing new technology with scrutinising the claims made about it.
Separately, Google has expanded its AI & Economy Research Program. On 18 September 2026 it announced new academic advisers, including Nobel laureate Philippe Aghion, who won the 2025 economics prize, and two directors, Anu Madgavkar and Daniel Rock, from the Wharton School. The programme aims to study how AI affects jobs and productivity. That announcement does not mention the 2023 report.
The sources do not say whether the government will formally withdraw the £400 billion figure. The Cabinet Office, which handles AI for government, and Google were asked for comment, but the sources do not include any replies.
Frequently asked questions
Why did the UK government's AI estimate change?
Wired reports that the £400 billion figure came from a 2023 Google-commissioned report. The government now uses a lower central estimate of £120 billion, which the AI minister called hugely uncertain. PF Global says the gap is partly due to slower AI adoption than its model assumed.
Who produced the original £400 billion figure?
AI minister Kanishka Narayan said the number came from a 2023 study that Google paid for, carried out by Public First, a consultancy now called PF Global.
Has the government withdrawn the £400 billion figure?
Wired reports that the parliamentary reply neither retracted the figure nor admitted it was incorrect. The sources do not say whether it will be formally withdrawn.
Why it matters
Figures about AI's economic effects can shape decisions on public spending, training and policy. This case shows how a single number can be repeated across government for years before its origins and assumptions are checked. Readers can use it as a reminder to ask who produced an estimate, what it assumes and who paid for it. Public money depends on reliable data, so the way governments test and explain such numbers matters to everyone.
Sources
Share this article
This article was prepared with AI assistance from the sources listed above and published after automatic quality checks. Our editor reviews published articles and corrects mistakes. How we work.
Spotted a mistake? See how we correct errors.
