As big tech circles, UK government struggles to reap promised AI benefits

The UK government’s grand plan for AI in the public sector is struggling in the face of growing technological challenges, a report by the Parliamentary Public Accounts Committee (PAC), a bipartisan group of elected members of parliament, has found.

Many of these problems will be familiar to anyone who has tried to make AI work inside an organization: the dead hand of obsolete systems, poor quality data, and a chronic lack of skilled people to implement the technology.

But beyond these issues lies another problem that could prove just as difficult: the monopolistic power of tech vendors that control the AI technology the government so badly desires.

Coming only weeks after the Government Digital Service (GDS) was created to drive AI, the committee’s initial assessment in the AI in Government report is a sobering reality check.

For the birds

The committee’s report identifies several areas of concern, starting with poor-quality data “locked away in out-of-date legacy IT systems.” Of the 72 systems previously identified as being legacy barriers, 21 hadn’t even yet received remediation funding to overcome these problems, it found.

It also noted a lack of transparency in government data use in AI, which risked creating public mistrust and a future withdrawal by citizens of their consent for its use. Other problems included the perennial shortage of AI and digital skills, an issue mentioned by 70% of government bodies responding to a 2024 National Audit Office (NAO) survey.

Additionally, government departments were running AI test pilots in a siloed way, making it difficult to learn wider lessons, said the committee.

“The government has said it wants to mainline AI into the veins of the nation, but our report raises questions over whether the public sector is ready for such a procedure,” said committee chair, Sir Geoffrey Clifton-Brown.

“Unfortunately, those familiar with our committee’s past scrutiny of the government’s frankly sclerotic digital architecture will know that any promises of sudden transformation are for the birds,” he added.

AI oligopoly

There’s a lot at stake here. AI is often talked up by the ministers as the key to overhauling the state, getting it to work more efficiently and cheaply. It’s a story that has become hugely important in many countries. If progress slows, that promise will be questioned.

In its report, the committee drew attention to the market power of a small band of AI companies. The tech industry has a tendency towards monopolies over time, it said, but with AI it was starting from this position, which might lead to technological lock-in and higher costs, hindering development in the long term.

According to the Open Cloud Coalition (OCC), a recently formed lobby group of smaller cloud providers backed by Google, the UK government’s struggles with AI mirror what happened with cloud deployment from the 2010s onwards, which included the lack of competition.

“This report shows that the dominance of a few large technology suppliers in the public procurement of AI risks stifling competition and innovation, while also hampering growth, exactly the same problems we’ve seen with cloud contracts,” commented Nicky Stewart, senior advisor to the OCC.

Cloud and AI are symbiotic, she noted, and the domination of one or both by a small group of mostly US tech companies risks building monopolies it might be difficult to escape from.

“Without reform, the government will remain over-reliant on a handful of major providers, limiting flexibility and access to innovative, leading edge technology, whilst locking taxpayers into expensive, restrictive agreements,” she said.

Sylvester Kaczmarek, CTO at OrbiSky Systems, a UK company specializing in integrating AI into aerospace applications, agreed that supplier dominance could stifle innovation, but remained just as skeptical of AI’s projected cost savings. Implementation was always where technologies proved themselves, he pointed out.

“Are savings over-sold? Most likely, in the short run,” said Kaczmarek. “There is a lot of groundwork to be laid before large-scale, reliable AI deployment can safely deliver meaningful savings. [governments need to] prioritize realistic roadmaps and more comprehensive value.”