The UK government’s ambition to release £45 billion in annual public sector savings through artificial intelligence and digitisation is facing growing scepticism from MPs and analysts, who question both the methodology underpinning the figure and the absence of any concrete workforce strategy required to convert projected efficiency gains into real cash savings.
The headline figure breaks down into three components: £36 billion attributed to simplifying and automating public sector service delivery, £4 billion from shifting service processing online, and £6 billion from reducing fraud and error.
Internally, government analysis identified an upper bound of £72 billion in potential annual productivity savings, with £36 billion selected as a conservative lower bound incorporated into the £45 billion total. Despite the precision implied by these figures, officials have stopped short of attaching any specific timeframe for their realisation, describing the savings as a long-term opportunity rather than a near-term commitment.
The methodology has attracted particular scrutiny. The core calculation scales findings from Central Digital and Data Office analysis of central government, the NHS, and police across the entire public sector, extrapolating from limited-scope data to far broader and more complex systems.
The £45 billion figure extrapolates from narrow datasets across a far broader and more complex public sector landscape.
Projections rest on the assumption that 100% of routine tasks and 10% of non-routine tasks can be automated, and that baseline public sector expenditure can be reduced by 15 to 30% through process simplification and AI-driven automation. The government’s own documentation concedes that quantifying AI-driven productivity savings represents an inherently complex research question carrying substantial uncertainty.
Pilot programmes have done little to resolve the credibility gap. Tools such as Microsoft 365 Copilot are projected to save civil servants nearly two working weeks per year on routine office tasks, with pilots estimating around £50 million in value from time savings.
However, those same pilots report no direct cash savings, because workforce numbers and budgets have remained unchanged. Time freed by automation has not translated into headcount reductions or budget releases, a distinction that critics argue exposes the fundamental weakness in the government’s framing.
Official communications have consistently emphasised productivity improvements, backlog reduction, and service quality rather than explicit civil service headcount reductions. Analysts argue this framing sidesteps the central challenge: achieving £45 billion in cash savings requires an integrated workforce management strategy covering roles, headcount, redeployment pathways, and skills development, all aligned with AI deployment timelines.
No such detailed civil service workforce plan has been published alongside the savings target, which critics say materially undermines the claim that large-scale financial savings are achievable within any credible horizon. The government is simultaneously exploring partnerships with companies such as IBM and Palantir to advance AI projects across public services, though critics note that vendor engagement alone does not substitute for the structural workforce planning the savings target demands.
The government has politically framed the £45 billion figure as a “jackpot” revealed through technology-driven reform of the state. Whether that prize proves realisable depends less on the technology itself and more on whether the organisational and workforce conditions necessary to capture savings are ever put in place.
