Government AI rollout on the rocks over inadequate pay

Civil Service struggling to attract skilled AI specialists

New research suggests that the rollout of AI across the Civil Service is stalling over staffing, with Civil Service pay scales proving insufficient to attract the skilled specialists required to lead the initiative.

The recruitment bottlenecks mean that from a budget of £12 million in the first year for the “Incubator for Artificial Intelligence” (i.AI), only £5m was spent and, of that, just £3.7m was spent on staffing costs. As a result, the unit launched fewer projects, delaying government plans to increase the use of AI in the Civil Service.

The aim of i.AI is to promote and accelerate the adoption of AI across Whitehall in a bid to automate processes, speed-up decision making and improve productivity.

The current government has strongly pushed AI as a way of cutting the cost of the Civil Service by 15 per cent over the long term, although the initiative was established in November 2023 under the previous government by then Deputy Prime Minister Oliver Dowden. He claimed that it would help reduce the size of the Civil Service, while the current government has emphasised cost cutting.

The Incubator was merged with a number of other government offices in January 2025 as part of a “blueprint for digital government”, but could boast more than 40 members of staff before the merger.

However, research commissioned by Westminster think tank Reform – not to be confused with the political party of the same name – indicates that the AI push is unlikely to succeed without a radical change to the usual approach by government to such high-tech rollouts. Radical change in this instance means considerably higher pay and partnerships with tech companies and academia.

“Transformation will not happen unless government fundamentally rewires itself to adopt the new tools much more quickly. Compared to private industry… public service adoption of AI is extremely limited [and] the public sector is still too risk-averse to even test AI in many cases, and lacks the focus and investment required to see projects through to deployment at scale,” concluded the report. [PDF]

Pay, it pointed out, is a particularly acute issue: “Many top machine learning engineers make salaries 10 times what they would be paid in the civil service – last year OpenAI researchers were reportedly hired on over $800,000 a year. Even in the UK, one technologist interviewed estimated that the equivalent compensation for some AI engineers in government would be £650,000. Plenty are paid even more than that.”

It recommended new frameworks for pay for critical AI roles in the Cabinet Office, although the government would struggle to justify US levels of pay for British civil servants, however specialised and highly skilled.

The report also pointed out that it is an easier and lower risk option for leaders in the Civil Service to simply demand more money in their next budget to improve services than implementing a technology roll-out that may or may not work.

The issue of low pay for skilled specialists in the Civil Service is becoming increasingly acute. With, for example, IT security roles that are little or no better paid than ordinary physical security roles.

Indeed, in Parliamentary testimony before the House of Commons Science, Innovation and Technology Committee earlier this year, Dr Laura Gilbert, a former director of the unit, also pointed out that the Civil Service’s outdated hiring processes also acted as a barrier. Many senior digital roles, she continued, are filled by non-technical staff that, she admitted, she wouldn’t hire for AI roles.