Government plans to make the UK a world leader in artificial intelligence could worsen existing problems around productivity
Labour’s ambitious plans to use artificial intelligence to transform the UK economy and public services risk making existing inequalities worse, according to research by experts from The University of Manchester and The University of Edinburgh.
The study argues that the Government’s approach risks treating AI as a technological solution to complex economic and social problems, while underestimating the practical obstacles to its widespread adoption.
Key findings
AI adoption is heavily concentrated in London, the South East and larger businesses, meaning its economic benefits could reinforce existing regional and productivity inequalities
The country lacks the computing and energy infrastructure needed to deliver the Government’s ambitions, which could consume more than a quarter of the UK’s electricity
Poor-quality, fragmented data and outdated IT systems could hamper AI adoption across government
Skills shortages are holding back AI adoption, particularly among smaller businesses and in the public sector, despite years of government investment in digital skills
AI is not a ‘magic bullet’ for Britain’s economic and public-service problems, and rapid expansion could create new environmental, social and governance risks
Unequal benefits and risks
Researchers Nathan Critch and Darcy Luke say years of short-term and fragmented policymaking by successive governments have left the UK with significant infrastructure and capacity gaps which could undermine Labour’s plans.
The Government’s AI Opportunities Action Plan includes plans to increase UK AI computing capacity twentyfold by 2030, develop AI Growth Zones, establish a National Data Library and expand the use of AI across government and the wider economy.
The researchers warn that the benefits of AI are unlikely to be shared evenly across the UK without policies specifically designed to encourage wider adoption.
Only 15% of UK businesses had adopted at least one AI technology in 2022, with adoption particularly concentrated among larger companies. AI adoption was highest in IT and telecommunications and legal services, while it was much lower in hospitality, health and retail - sectors which employ large numbers of people.
AI firms and businesses using AI are also heavily concentrated in London and the South East.
The study argues that this creates a risk that AI will make already productive businesses and regions even more productive, while areas and sectors which have struggled with low productivity are left further behind.UK lacks infrastructure for AI boom
The UK currently has only around 3% of the world’s computational power - much of which is not optimised for AI - leaving the country dependent on cloud computing provided by large technology companies.
The researchers warn that expanding computing capacity will place enormous demands on the UK’s energy infrastructure. They estimate that a twentyfold increase in UK computing capacity could increase the energy consumed by AI infrastructure from around 3.6 terawatt-hours (TWh) in 2020 to 72 TWh – more than a quarter of the UK’s total electricity consumption.
This creates a tension between the Government’s AI ambitions and its commitment to reaching net zero, particularly given constraints in the electricity grid and limited energy storage capacity. The researchers also highlight the water required to cool data centres as a potential environmental constraint.Poor data and outdated technology
The study identifies the quality and accessibility of government data as another major obstacle. A National Audit Office survey found that 62% of responding government bodies identified a lack of access to good-quality data as a major barrier to implementing AI. Government data is often held in fragmented systems and legacy IT infrastructure, while data held by local authorities can be siloed across incompatible systems.
The researchers argue that these problems will make it difficult to deliver the Government’s plans for AI-enabled public services without significant investment in underlying digital infrastructure. They also warn that the public sector faces persistent digital skills shortages, with problems particularly acute in local government.Skills shortages could limit AI adoption
Despite a relatively strong range of AI-related training in the UK, demand for AI skills continues to outstrip supply.
Smaller businesses are particularly disadvantaged because larger companies can offer higher salaries and attract workers with scarce AI expertise. The researchers also highlight a lack of in-work training, vocational education and further education provision compared with the emphasis placed on higher education.
These problems could limit the Government’s ambitions to use AI to improve productivity across the economy and make public services more efficient.Decade of AI policy leaves UK with major gaps
The researchers argue that Labour’s approach is part of almost a decade of increasingly active government policy on AI, stretching back through the Conservative governments of Theresa May, Boris Johnson and Rishi Sunak. They describe this pattern as “hyperactive incrementalism” - characterised by short-termism, fragmentation, limited evaluation and repeated policy initiatives.
While the UK has developed an internationally significant AI sector, the researchers say this should not be confused with widespread adoption of AI across the economy. Successive governments have focused heavily on developing the UK’s AI sector, they argue, but have been less successful at encouraging businesses and public services across the country to adopt AI. The result is a significant gap between the Government’s ambitions for AI and the infrastructure, skills and organisational capacity needed to deliver them.AI is not a ‘magic bullet’
The researchers argue that there is currently little evidence that AI can, by itself, deliver the broad improvements in productivity, government efficiency and public services envisaged by policymakers.
They also warn that rapid AI expansion could create new problems, including threats to data protection and civil liberties, surveillance, deskilling and unemployment, and increased environmental pressures.What the researchers say
AI has enormous potential, but it cannot simply be treated as a technological fix for deep-rooted economic and public-service problems. If the Government wants AI to deliver widespread benefits, it first needs to address the longstanding weaknesses in infrastructure, data, skills and public-sector capacity.
“AI adoption is currently concentrated in the businesses, sectors and regions that are already among the most productive,” said Darcy Luke from The University of Edinburgh’s BRAID (Bridging Responsible AI Divides) Research Programme. “Without policies that encourage much wider adoption, there is a real risk that AI will reinforce existing inequalities rather than reduce them.”
The researchers conclude that while Labour’s plans represent the most comprehensive attempt by a UK Government to harness AI, the scale and depth of the barriers involved mean that AI may ultimately become “another problem for an already problem-laden government, rather than the solution it is hoped to be.”
Publication details
The paper was published in the journal The Political Quarterly.