UK auditors challenge AI savings case behind £45 billion efficiency plan

The UK National Audit Office has told government leaders to build AI and digital change into long-term workforce planning before relying on claimed public-sector savings of £45 billion a year. The warning matters beyond Whitehall because the government is treating automation, self-service and AI as a core productivity lever, while its own auditor says departments have not explained how those gains will be delivered.

The NAO said civil service leaders should account for major technology shifts when planning workforce size, job roles, grades, tasks and required skills. Its report says senior management needs to focus on those changes if the government is to convert AI and digital transformation into productivity gains.

The auditor framed the central issue as an operating-model problem, not a procurement slogan. “The major challenge for the government is reflecting the fundamental impact of AI and digital change in planning for how organisations will operate in future,” the NAO said, adding that this includes likely effects on staff numbers, the type of work being done and the capabilities needed to do it.

£45 billion claim remains thinly evidenced

The government has said digital transformation and AI could produce annual efficiencies worth £45 billion. The NAO did not endorse that figure. It pointed instead to gaps in how departments have calculated workforce-related savings.

Other public-sector observers have already questioned the number. Nick Davies, programme director at the Institute for Government, said the figure was a “huge amount” given that much public spending goes on salaries or infrastructure. Parliament’s Science, Innovation and Technology Committee later described the estimate as “worryingly optimistic” and said it was an example of “hype” that could hold back digital transformation.

The NAO said published efficiency plans do not give enough detail on how departments arrived at expected workforce savings. It warned that departments may be estimating those savings in inconsistent ways. The report also raised a broader risk: departments may be missing larger opportunities to redesign workforces around broad AI and digital adoption.

For technology suppliers selling into government, that distinction is material. A department buying AI tools for document review or case handling is a different market from a department that has redesigned workflows, roles and governance around automation. The NAO’s concern is that the government has made the productivity claim before showing enough of the workforce math behind it.

Automation use cases are named, but constraints remain

The report points to areas where reduced staffing needs and higher productivity are expected, including digital self-service for tax and benefits, and automation of manual tasks such as checking planning applications. Those examples are familiar categories for public-sector software vendors, but the NAO said the workforce plans need to be tied directly to departmental digital and AI plans.

The auditor also highlighted a skills constraint. The public sector lacks some of the technology and digital capabilities needed to secure the efficiencies being claimed from AI, according to the report. That gap raises questions about whether departments can implement and manage AI systems at the scale implied by the government’s savings target.

Cabinet Secretary Dame Antonia Romeo had already said current civil service workforce plans “do not sufficiently reflect how AI and technology are going to change the shape of the civil service,” the NAO noted. The auditor said departments should prepare digital and AI plans alongside strategic workforce plans rather than treating them as separate exercises.

The report also flags limits on deployment. Some AI uses may raise ethical concerns, the NAO said, leading organizations to restrict where the technology is used. That caveat is another reason the £45 billion figure remains exposed: savings from automation depend not only on technical capability, but on where government is willing and able to use it.

This story draws on original reporting from The Register.