Can AI help solve the UK’s public sector productivity puzzle?
Jeremy Hunt has made improving public sector productivity a core plank of his economic strategy.
“We need a more productive state not a bigger state,” he said in the Spring Budget last week. “It’s not fair to ask taxpayers to pay for more when public service productivity has fallen”.
In the Spring Budget, the Chancellor announced plans to invest £800m in public services – funds that are specifically earmarked for new technologies, including artificial intelligence. This came alongside a £3.4bn investment package for the NHS, of which £1bn will go towards data and AI.
According to the Office for National Statistics (ONS), public sector productivity increased by just 0.2 per cent per year between 1997 and 2019. Since the pandemic, it has actually fallen and is now about five per cent below pre-pandemic levels.
With an ageing population likely to place greater demand on public services, there is an urgent need to increase productivity otherwise a larger and larger share of national income will be swallowed up by public spending.
The potential benefits of boosting productivity should not be underestimated. Raising public sector productivity to pre-pandemic levels would be the equivalent of around £20bn extra in funding, the Office for Budget Responsibility has estimated.
Hunt’s ambition is not new. As Anne-Marie Morris, MP, noted to Hunt in a Treasury Committee session last year, improving productivity has been “nirvana for many of your predecessors”.
But Hunt is hopeful that new technologies, and artificial intelligence (AI) in particular, will enable him to succeed where previous Chancellors have failed.
A Silver Bullet?
Oliver Dowden, deputy Prime Minister, has argued that “AI is genuinely a silver bullet”, and could help solve the UK’s ‘productivity puzzle‘.
Likewise, Bart van Ark, managing director at think tank the Productivity Institute, told City A.M. that AI is a “game changer” in terms of addressing stagnant productivity.
Its easy to see why. Treasury estimates from last year suggest that new technologies, including AI, could help cut a quarter of a million policing hours every week.
But AI is not a silver bullet on its own. Structural reforms will need to take place while rolling out AI across many parts of the public sector. This will take time and, importantly, cost money, van Ark explained.
For a start, workers need to be educated about how to use new technologies to help them see where it can be useful. And it’s not clear that this work has been undertaken.
A 2023 survey by the FDA union, which represents senior civil servants, showed that 90 per cent of members had never been consulted on the roll-out of AI, while 70 per cent had received no training on AI use in the last two years.
“Our members show a clear willingness to take AI forward to improve productivity and deliver for the public,” FDA assistant general secretary, Lucille Thirlby, told City A.M. “However, the wider use of AI needs to be matched with additional resources and a proper workforce strategy.”
“The wider use of AI needs to be matched with additional resources and a proper workforce strategy.”
Lucille Thirlby, FDA assistant general secretary
The other issue is managerial. “Digital technologies are about communication and information,” Diane Coyle, Bennett Professor of public policy at Cambridge University, told City A.M. “This means organisations have to change their structure in order to use the data effectively,” she continued.
For example, police staff are already using AI to improve facial recognition technologies. But for real productivity gains, staff both need to have confidence in the results and have the ability to act on them without going through traditional hierarchies.
In particular, van Ark said the government would need to enhance the role of those providing oversight of the use of AI. “We cannot just throw it at the wall and assume that it will do well,” he said.
No quick wins
Putting in place proper structures is essential to getting the most out of any new technology, and this will take time.
“The benefits of electricity took 50 years to show up in the productivity statistics,” Coyle said.
“The benefits of electricity took 50 years to show up in the productivity statistics.”
Diane Coyle, Bennett Professor of public policy at Cambridge University
AI might show up more quickly, but there’s no guarantee that it will. Despite the hype, only six per cent of firms in the US private sector are using AI in a productive way, according to BCG.
“We’re in the UK, which has slower productivity growth, and looking in the public sector, where it is more challenging,” van Ark said. “You can only get results if you have a longer term strategy”.
And since Brexit, the Conservatives have hardly been a shining example of strategic thinking. “The government changes its mind virtually every day,” Coyle lamented.
And doing this properly costs money.
Hunt’s current investment package is worth about £840m a year, or 0.03 per cent of GDP.
Tom Smith, director of economic policy at the Tony Blair Institute, said the Chancellor’s plans were a “good down payment” but warned that the investment “will not be enough on its own to turn around the UK’s fiscal fortunes”.
Hunt has effectively pencilled in steep spending cuts for unprotected government departments after the election, and it is unclear how this will impact this public sector productivity drive.
These departments, like justice and local government, are hardly flush with cash at the moment. Slashing budgets can deliver some increases in productivity, but continual budget cuts cannot continue delivering improvements if workers go without the necessary equipment.
Capital investment meanwhile will be frozen in cash terms, amounting to steep cut in real terms worth £26bn. The Resolution Foundation noted “it is hard to think of a more anti-growth policy choice”.
Thirlby, from the FDA union, warned that “inadequate investment” would stand in the way of the civil service’s ability to get the most out of AI.