Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
An artificial intelligence “hit squad” unit will be set up at the heart of Whitehall with a remit to shrink the size of the UK civil service and bolster public sector productivity.
Deputy Prime Minister Oliver Dowden plans to form a task force of 30 “high-end, technically capable” experts in AI and data engineering with an annual budget of about £5mn, to begin the process of transforming public services.
Speaking on Monday on the margins of an AI training course for civil servants in London, he said the technology was the “closest thing you have to a silver bullet in terms of driving efficiency for the taxpayer”.
The unit will expedite how information is compiled from different departments, sieve out irrelevant data and allow officials to make decisions more quickly, he added.
“This is about trying to get a hit squad, a sort of crack squad, that is going to go out there and bring a high level of expertise to try and identify innovative solutions to projects,” he said.
The unit’s priorities will be to tackle welfare fraud, assist the processing of asylum claims, and overhaul the interface between the public and the NHS, Dowden added.
Allowing AI a role in decisions about welfare abuse and asylum claims would be controversial given the way algorithms have been shown to amplify racial bias.
A childcare benefit scandal in the Netherlands, which forced the government to resign in 2021, involved racial profiling that was “baked into” the design of the self-learning algorithm that flagged potentially fraudulent claims, according to Amnesty International.
Dowden insisted the use of the technology in the UK public sector would be used as a “co-pilot . . . alongside human decision-making processes”.
Laura Gilbert, chief analyst and director of data science at Downing Street, said the government would be “very careful with the data”, adding that “the human in the loop is incredibly important” when applying AI. She added the government “wouldn’t be looking to break that paradigm”.
Dowden said the AI unit would also look to apply the technology to correspondence and call handling, arguing it would be able to “genuinely drive efficiencies”. This in turn would allow the government to further reduce the size of the civil service, which is already set to shrink by 66,000 jobs.
“I think you can get more for less,” he said, describing AI’s potential to act as “a significant downward driver” in the number of personnel employed across Whitehall, which has had a “very big expansion” as a result of Brexit and Covid-19.
He said that while most government decisions involve a trade-off between resources and output, AI meant “you can both enhance outputs, so get better services for people, and with lower inputs, so lower cost”.
Economists have voiced scepticism about AI’s ability to turbocharge public sector productivity, however, arguing that upgrading basic IT systems, increasing capital investment and bolstering management capacity may be more effective solutions.
Gilbert argued AI would create a “happier workforce” in Whitehall, freed from administrative tasks. Some officials “waste a whole day a week” on admin, Dowden added.
He said he would ask the AI unit to “crack on” with launching about five big projects and 10 small projects, instead of devising an ambitious, overarching cross-department AI strategy akin to trying to “boil the ocean”.
Gilbert said the government was already using roughly 20 open source models, and would continue to engage with AI systems devised by leading private companies such as OpenAI and Anthropic “where appropriate”.
She said the government was not relying solely on large language models, but was investigating other types of AI too.
