Elon Musk’s Doge-style dynamics are catching on across the Atlantic.
The British government is embracing lay-offs and artificial intelligence as a means of mending public finances, and improving services.
What’s happening? The UK’s Chancellor of the Exchequer, Rachel Reeves, is confident that civil service job numbers in the UK can be reduced by 10,000.
To do so, the government will need to fire people, and plans to invest in the use of technologies like artificial intelligence throughout the public sector.
So what? Whatever one makes of the legal, security, and operational implications, Elon Musk is using the Department of Government Efficiency (DOGE) in the US to write a playbook for governance in the modern age. It appears that some of the Silicon Valley billionaire’s new rules are being applied across the Atlantic.
How? Keir Starmer, the UK Prime Minister, is embracing his own version of the DOGE doctrine — replacing civil servants with artificial intelligence, while attempting to purge deadweight costs from public administration so more money can be spent on frontline services, including a push to significantly increase defense spending.
Why? Starmer has said the civil service has grown “overcautious and flabby” since the Covid-19 pandemic, and needs to be cut back. The 10,000 job cuts would represent approximately 2 percent of the total staff number. He wants to save £45 billion ($58bn) annually by applying digital technologies more widely across the government.
He added that simplification and automation of delivery, form-filling, communications and legal or policy documentation are all areas of bureaucratic sludge ripe for an overhaul. At the same time, Technology Secretary Peter Kyle has said he will “help give AI innovators in Government the freedom they need to chase an exciting idea and build prototypes almost immediately.”
Sound Doge-y? Somewhat. There are many differences that make the approach in Britain distinct. No marauding teams of hired-gun computer scientists have invaded ministries in Westminster. Nor have diversity and inclusion initiatives been specifically targeted.
Instead, the drive for efficiency is being centered on returning to pre-Brexit, and pre-pandemic levels of provision, during which more than 100,000 civil servants were hired to cope with the burden on public services.
The recent revelations that top US officials used Signal to discuss sensitive information about military operations in Yemen, and added a journalist to the chat by accident, will reinforce concerns that government agency shake-ups could expose them to cyber-risk and reduce government capacity across the pond. It also shows that a significant downsizing of the government workforce leaves major policy decisions to a smaller and smaller group of people.
What’s the UK’s plan? It’s big on vision, but light on detail and logistics. The UK government’s recent AI Opportunities Action Plan promised to “shape the AI revolution on principles of shared economic prosperity” and “improved public services.” It notes that “government purchasing power can be a huge lever for improving public services.”
Maker or taker? The question remains as to which companies the UK government will spend with. At present, there are no purely domestic firms building foundational model capabilities at a scale comparable to the American market.
Nor are there any companies of sufficient scale providing underlying compute capabilities that are crucial to developing cutting-edge models that can challenge market leaders in the US. Although that could change over time as UK-based technology companies like Arm and Graphcore continue to invest in new products that could provide a basis for more domestic competitiveness in the UK.
For the most part, these things are still American made. On a recent visit to the US, the UK tech secretary touted Britain as a prime destination of US investment. The pitch is that the government is “rewiring Britain’s economy to run on AI”, and that the UK’s $92 billion AI sector could position it as the “second leading AI nation in the democratic world.”
Downstream applications, involving the fine-tuning of US-developed foundational models, the creation of application-specific instances, and ingestion of UK data with both generative and predictive artificial intelligence are likely to be most of the menu.
Tim Gordon, a partner at Best Practice AI, told Inferences that “advantages for the government come more from being in a good foundational position to implement artificial intelligence technologies, rather than just implementing it anyway.”
The challenge of overhauling the government's legacy data systems, data rights, funding mechanisms, and other factors mean that the shining light on the hill may still be a long way off.
“If you have the data plumbing, organisational processes, incentives and governance in place then you will already have captured many of the potential benefits. Of course that is all hard work, and perhaps slightly less glamorous.” Gordon added.
Judging by the areas in which Starmer has said he expects savings to be made, the UK government is listening to more grounded advisors who claim that task automation, error-detection and streamlining using more conventional software is the goal.
Though it is difficult to rule out the influence of more “pro-innovation” accelerationists who believe that generative models may replace entire functions, at a superhuman level of performance, relatively soon.
“You're not going to get the types of savings being touted by using AI to rework the entire state, within three years, before the next election. Even if some important gains can be made at the margin,” Gordon told Inferences.
Case in point, the UK government just announced £1m in funding for technology companies building education-specific applications using AI that could be rolled-out in public schools to supplement teaching capacity. This is workaday AI innovation, for paltry sum, with technology available off the shelf – no frontier models required. Nor does it imply any fundamental re-working of the state apparatus.
What could go wrong? A spat is already brewing between the UK government and the National Education Union (NEU), the largest teaching union in the UK. In response to government plans to “reduce workload” and “slash bureaucracy” using AI, the teacher’s union said it would resist any move that “seeks to de-professionalise, deskill or replace teaching assistants.”
These concerns could sweep the civil service, and other sectors that face automation and cuts with a view to increasing the use of AI software, and reducing the number of people doing actual work. Concerns will also abound that the destruction of human jobs will contribute to underemployment, and a long-term decline in the relevance and competence of skilled workers in the UK economy.
As part of its plans to deploy AI in public sector reform, the UK government has explicitly said that “creativity, critical thinking, and emotional intelligence are still absolutely vital.” But critics of the government plan, who see an imitation of Musk’s DOGE as a genuine threat, warn that the UK faces a considerable challenge with inequality already. They argue that the Treasury should back “digital access, collective rights to information and new e-learning roles” before moving ahead with sweeping cuts and a whole-hearted embrace of AI as an optimizer.
The upshot? Trump’s campaign to significantly reshape civil service roles in international development, public health, environmental protection, education, publicly-funded media, and much more, has seen a broad swathe of federal government sheared off, disrupting long-standing, government-funded initiatives. Serious questions remain as to whether government’s efficiency is more important than government’s effectiveness.
What we’re reading:
Initial thoughts on the Signal chat security breach.
Remarks by the Vice President at the American Dynamism Summit.
Reporting from MIT Tech Review on rest-of-world perspectives on US models under Trump.
What we’re looking ahead to:
14 - 16 May: B7 Summit, Ottawa.
4 June: AI+ Expo and Ash Carter Exchange in Washington, DC.
June (expected): The UN Internet Governance Forum.
15 - 17 June: G7 Leaders’ Summit in Kananaskis, Alberta.
24 - 25 June: North Atlantic Treaty Organization (NATO) Summit.
6 - 7 July 2025: Annual BRICS Summit, Rio de Janeiro, Brazil.
9 - 11 July 2025: AI for Good Global Summit.
9 - 23 Sep 2025: UN General Assembly (UNGA 80), New York.