Our Future of Britain initiative sets out a policy agenda for governing in the age of AI. This series focuses on how to deliver radical-yet-practical solutions for this new era of invention and innovation – concrete plans to reimagine the state for the 21st century, with technology as the driving force.
Chapter 1
The public sector in the United Kingdom is a major employer with nearly 6 million employees, or 5 million full-time equivalents,[_] and had an annual wage bill of £240 billion in 2022–23[_] or almost 10 per cent of GDP.
Public-sector productivity has been flat for more than a quarter of a century,[_] but there is now scope for artificial intelligence to radically reshape the way that the public sector operates and to deliver large improvements. This is particularly the case for administrative, back-office and policy functions that largely involve cognitive tasks. Our analysis in this paper examines the impact AI could have on public-sector productivity by estimating how much time AI could save workers and the associated cost savings. We then examine how quickly AI could be rolled out across the public sector and the associated costs, leading to a comprehensive estimate of the net savings expected from the use of AI in the public sector. Our results indicate that:
More than 40 per cent of tasks performed by public-sector workers could be partly automated by a combination of AI-based software, for example machine-learning models and large-language models, and AI-enabled hardware, ranging from AI-enabled sensors to advanced robotics.
These efficiency gains could help save a fifth of public-sector workers’ time in aggregate. These potential time savings could lead to significant fiscal savings if the government chose to bank them by reducing the size of the public-sector workforce accordingly.
However, not all these time savings will lead to a lower wage bill and fewer public-sector workers. Many public-sector professions already face severe staff shortages, with workers doing significant amounts of unpaid overtime to keep the system afloat. We assume that the benefits of AI in these stretched professions (which account for 2.3 million workers and include maths, science and language teachers, doctors, nurses and care workers[_]) do not lead to job cuts, but instead allow frontline workers to work fewer unpaid overtime hours and deliver better outcomes. Excluding these cases from the analysis reduces the overall potential time savings from AI from one-fifth to one-sixth, but still implies savings of £41 billion a year to the public-sector wage bill (1.5 per cent[_] of GDP) if AI were used to its fullest possible extent.
To achieve these gains, the government will need to invest in AI technology, upgrade its data systems, train its workforce to use the new tools and cover any redundancy costs associated with early exits from the workforce. Under an ambitious rollout scheme, we estimate these costs equate to £4 billion per year on average over this parliamentary term (0.15 per cent of GDP) and £7 billion per year over the next (0.24 per cent of GDP). In the long run, we estimate annual running costs of about £4 billion in today’s terms. This implies the net savings from fully utilising AI in the public sector to be nearly 1.3 per cent of GDP each year, equivalent to £37 billion a year in today’s terms. This equates to a benefit-cost ratio of 9:1 in aggregate and, since the setup costs of the programme are relatively small, the net benefit is positive almost straight away. Indeed, even after five years we estimate the programme could cumulatively save 0.5 per cent of annual GDP (or £15 billion in today’s terms), implying a benefit-cost ratio of 1.8:1 is possible if the technology is rolled out quickly.
The speed with which these gains are realised is within the government’s gift to determine. Government IT projects typically take nearly four years to complete, but these projects tend to be more targeted. Deploying AI across the entire public sector would present a bigger delivery challenge. We set out an ambitious scenario whereby the rollout of AI across the public sector is largely completed within two parliamentary terms. TBI’s recent paper, Governing in the Age of AI: A New Model to Transform the State, sets out a plan for the new government to meet this timetable.
The government will have a choice on how to spend any dividend from AI-enabled efficiency. It could choose to reinvest the savings in the public sector and boost the number of frontline public-sector workers. For example, a saving of 1 per cent of GDP would be enough to boost the size of the NHS workforce by around a third.[_] Alternatively, the government could choose to shrink the UK’s public-sector workforce by around a sixth (equivalent to around a million workers) and bank the fiscal savings. If it were to take this approach, it could lower the explosive path for debt shown in the paper The Economic Case for Reimagining the State as a share of GDP by 27 percentage points by 2050. Overall, the impact on annual public-sector net borrowing would be 2.4 per cent of GDP by 2050, including savings from lower debt interest.
All the above figures are based on a snapshot of AI’s potential capabilities as they are today, but the technology is advancing quickly. We therefore also explore a scenario where AI’s capabilities continue to advance. If AI advances such that it becomes possible to save a further 1 per cent of public-sector workers’ time each year, then we estimate that the annual net cost savings could reach 1.9 per cent of GDP by 2050 under similar rollout-speed assumptions. The size of the prize could grow further in the years to come.
Read the full paper here. This is one of four companion papers to The Economic Case for Reimagining the State.