Charting the UK's £45 Billion AI Claim
How much money can 'Humphrey’ really save the government?
‘A smaller and more efficient state’. Sir Keir Starmer was sounding a lot like Elon Musk this week as he unveiled his plans to make the UK public sector more productive.
The vibe shift came as his Labour administration announced it would scrap NHS England (link). The bureaucracy was duplicating other parts of the health service and causing inefficiencies at taxpayers' expense. It had to go.
The move, which includes the axing of thousands of jobs, sounded extremely reasonable, but you can only imagine the uproar if a Conservative government tried this.
‘They’re privatising the NHS, the b*****ds,’ would’ve been the cry from the left.
Alas, Labour once oversaw the creation of the NHS, the party often gets a pass on all things healthcare and the decision has actually gone down reasonably well.
This is partly because voters increasingly understand the UK’s delicate financial state.
The country’s debts have soared (link) and economic growth is sluggish (link). Labour’s big play to boost productivity across the UK economy is to focus on the public sector, where 45% of national income will be spent in 2024/25 (link).
Productivity rates in the UK have been worryingly low for years, so by making these gains in the public sector, where the government apparently has most control, it should have a substantial impact.
Here’s what Chancellor Rachel Reeves said in her Budget speech last August (link):
“We are setting a 2% productivity, efficiency and savings target for all departments to meet next year, by using technology more effectively and joining up services across government.”
It should also be noted that the government is employing hundreds more tax collectors to boost Treasury revenues (link).
So, going into the start of 2025 we’ve got a fairly clear picture of what Reeves and Starmer want to do: Boost economic growth, achieve this by increasing productivity in the public sector and raise more taxes by hiking levies whilst increasing the number of tax collectors.
They are also going to ‘use technology more efficiently’. But what does that even mean?
Well, we got a further sense of it in January, when the UK government promised to deploy AI tools to improve the NHS (link):
“A new package of AI tools – nicknamed ‘Humphrey’ – will be available to civil servants in an effort to modernise tech and deliver better public services to set the country on course for a decade of national renewal.”
The announcement followed a review of the government’s technology stack (link), finding that public sector workers were being held back by “archaic technology”:
“With nearly half of public services unable to be accessed online, people are spending too much time applying for support in person, including time on hold or travelling to council offices.
Public sector workers are also wasting time sifting through physical letters. This means response times are unnecessarily long with British citizens paying the price and wasting valuable time on government admin.”
It’s here, in the State of Digital Government report, that we’re introduced to a bold claim by the government: Whitehall is missing out on £45 billion per year of unrealised savings and productivity benefits, as 4-7% of public sector spend could be achieved through full potential digitisation of public sector services.
Crucially, there are no footnotes attached to the “£45 billion per year” claim, but that doesn’t stop later government announcements citing the State of Digital Government report (link).
In fact, it looks like the figure has been repeatedly used by the UK government, including in Starmer’s speech on 13 March (link):
“We don’t want a bigger state, or an intrusive state, an ever-expanding state.
“A state that demands more and more from people as it fails to deliver on core purposes. So we’ve got to change things.
“Now the good news is technology can massively help. If we push forward with digital reform of government – and we are going to do that, we can make massive savings, £45 billion savings in efficiency. AI is a golden opportunity.”
Technology Secretary Peter Kyle also referenced the £45 billion “jackpot” on the same day (link), as it was announced that new AI teams would be deployed into Whitehall departments.
“One in 10 civil servants will work in tech and digital roles within the next five years, with 2,000 tech apprenticeships turbo charging the transformation,” the government stated.
Most of the British media accepted the £45 billion claim as stone-cold fact and simply repeated the figure after Number 10 trailed Starmer’s speech. Other journalists, including the FT’s Lucy Fisher, shared their scepticism (link).
The government was subsequently forced to share some of its workings. Here’s what they disclosed: £36 billion of the estimated efficiencies would derive from “simplifying and automating delivery across the public sector”, a further £4 billion would come from “migrating service processing to cheaper online channels” and the final £6 billion? This could be saved by reducing “fraud and error with digital compliance solutions”, according to officials.
This answer is unsatisfactory to me. What time frame are we working with here, when does it reach the full £45 billion per year and what financial probabilities surround the government’s thinking?
Without some fully fleshed-out workings, it looks like this “jackpot” will remain elusive, though I bet we will see the figure again soon.
Around the same time tech journalist Chris Stokel-Walker was able to find out how the government is actually using generative AI.
He FoI’d Peter Kyle and discovered that the Technology Secretary had asked ChatGPT for advice on why the adoption of AI was so slow in the UK business community – and which podcasts he should appear on (link).
“ChatGPT is fantastically good, and where there are things that you really struggle to understand in depth, ChatGPT can be a very good tutor for it,” Kyle had previously told journalists.
That £45 billion in savings just might remain elusive, folks.
The FT’s New CEO
Jon Slade, currently Chief Commercial Officer, will take up the role of CEO on 1 July. Slade has 30 years of media industry experience, more than 20 of which have been spent at the FT.
He held various roles in the outlet’s advertising team, before becoming CCO and joining the FT’s board in 2014. He is currently responsible for three quarters of the FT Group’s more than £500mn annual revenues, the FT said.
D is For ‘Disinformation’
As the hostilities between Russia and Ukraine continue, remember this: the Russian state has been practicing organised disinformation campaigns since the start of the 20th Century. The Bolsheviks adopted the methods of the Tsarist secret police.
Those tactics included forgeries, rumour mills and other deceptions. One of the most infamous early campaigns was Operation Trust. Initiated by the Cheka, fake anti-Soviet groups were established in the 1920s to lore monarchists and other anti-Bolsheviks into the hands of the secret police.
This not only helped the Soviets capture their adversaries, but also discredit émigré groups in Western Europe. Disinformation was later codified and cemented into the practices of the KGB when it established a Department D within its First Chief Directorate in the 1950s.
The First Chief Directorate was responsible for running foreign agents and one Vladimir Putin would reportedly serve in its ranks as part of his KGB career. Deception is baked into the playbook.
Remember the ‘little green men’ who appeared in Crimea in 2014? Putin described them as a “self-defence group” at the time (link), that was well before he launched his ‘special military operation’.
The Decline of Intelligence
The FT’s John Burn-Murdoch has posted a rather worrying thread over on X (link) showing that literacy rates have been declining. “Recent results from major international tests show that the average person’s capacity to process information, use reasoning and solve novel problems has been falling since around the mid 2010s,” he wrote.
Quentin Skinner
Since Quentin Skinner has a new book out (link), I thought I would catch-up on the intellectual historian’s Liberty Before Liberalism.
He explains in the essay that a so-called ‘neo-Roman’ theory of liberty was popular before the classical liberalism of the 19th Century became the go-to theory for freedom.
The neo-Romans, including Machiavelli, the Commonwealthmen of the mid-17th Century, and the early Whigs (John Locke) of the late 17th Century and early 18th Century, perceived liberty through the prism of slavery. Namely, you were only free so long as you weren’t dependent on anyone else.
Since the neo-Romans also viewed the state as an artificial person (outlined in Hobbes’ Leviathan), any nation or dominion which depended on another country couldn’t — by definition — by free.
Such thinking, as well as the highly influential Cato Letters of John Trenchard and Thomas Gordon, inspired the Patriot cause in the American colonies. It has since fallen out of fashion.
On a related note, my essay on the First Amendment can be found here (link) and my piece on The Craftsman can be found here (link).
The Political Press Box
Find my latest long-form audio interviews of political communicators here. Please like, subscribe and listen.
📧 Contact
For high-praise, tips or gripes, please contact the editor at iansilvera@gmail.com or via BlueSky (link). Follow on LinkedIn here.
210 can be found here
219 can be found here
218 can be found here
217 can be found here
216 can be found here