Will AI give the UK a new commercial edge? Keir Starmer thinks so. But are his plans for world domination realistic? Or is he shilling technologies that don’t yet, and may never, exist?
On 13 January 2025, the UK’s prime minister, Keir Starmer, gave a speech in which he set out his government’s plans and ambitions concerning artificial intelligence (AI). And while I think it began as a clunky and awkward attempt to connect with the common folk – AI can learn to play a game even faster than Luke Littler but I’ve never seen it throw darts! They’ve got a big screen with the brain pulsating . . . it’s really complicated! – Keir’s speech developed into something rather interesting.
I think.
As long as the government has thrashed out some detail behind the scenes. The funding, for example, and the massive volumes of power and water all those data centres are going to need. Also, the mechanisms behind data quality and assurance (rubbish in = rubbish out is a fundamental truth of AI, after all).
And providing that we have solid definitions of terms such as ‘safety’; ‘safe access to the unique resource of our NHS for resource’; and ‘making AI work for the people’.
And above all, as long as everybody understands that maximising the UK’s role in and benefits from the great AI roll-out will take a long time. Possibly a long, long time. Certainly longer than the 4 or 5 years Keir has until the next election.
So, my immediate response to the speech was to think: yeah, as long as Keir has that lot sorted there is quite a lot to be cheerful about. That impression grew when I read the accompanying action plan. There’s a lot of interesting stuff here. But there’s also a touch of the ‘We love AI, it’s going to change the world! Rah rah rah!’ cheerleading that does my head in online. And that concerns me, because . . .
AI is not a silver bullet
Starmer is wrestling with an economy hamstrung by low productivity, particularly in the public sector. He assured us that AI will slash the routine tasks he claims hold services back, and will allow staff to reconnect with their original reasons for entering the public sector (atruism, presumably). This combination will, he seems to imply, inevitably boost public sector productivity.
The thing is, having worked in the UK public sector and, like most Brits, interacted with it extensively, I’m not at all sure this is correct. I know AI will greatly improve processes for healthy systems with high-quality data which they manage well. I also believe that many public sector organisations are not optimally functional or brilliant with data. So the government has to solve some chunky problems before AI can be let loose.
For example, how is AI going to tackle the stubborn problems of persistent and severe absence in schools and pupils currently missing from education? I agree with the prime minister when he states that AI will help teachers to reduce admin and tailor their teaching to individuals more effectively. But it’s hard to see how you do that for pupils who are not in school to be taught. If you hang around for AI to make perfectly differentiated lesson plans that entice all disengaged students back to their classrooms, you’ll have lost a generation.
As a side note, page 5 of the action plan (PDF version) states that ‘The public sector should rapidly pilot and scale AI products and services and encourage the private sector to do the same.’ Starmer’s speech also referred to giving ‘safe’ access to NHS data. I really, really hope they have robust risk assessment and monitoring around all of this. Failure could be catastrophic for some of the nation’s most vulnerable people, and ruin credibility and public trust.
AI may be hard to greenwash
The government’s AI action plan refers extensively to ‘AI growth zones’ and ‘sovereign compute’. There are indications that planning applications for data centres will be expedited. This might play to our national strengths because we have plenty of coastline – and data centres use loads of water for cooling purposes. Desalination technology could help to meet that need at coastal locations, but first we have to power the things (the data centres, not the coastal towns!)
Let us recall that the government’s first actions on coming to power included restricting the winter fuel subsidies available to some older people. Even though household energy bills are skyrocketing. So the same government’s apparent intention to guarantee power supply for new data centres is interesting. But I digress. My real point is this: how will the government reconcile this (genuninely valuable) ‘investment in our future’ with climate change?
On the day of Starmer’s speech, a couple of the Just Stop Oil people decided to spray paint Charles Darwin’s grave in a baffling piece of climate change activism. Who will break it to them that we will soon have multiple power-hungry, state-of-the-art data centres to harness as-yet unscoped, often unproven, technology?
‘But what about renewables?’ I hear you cry. And you are correct. While data centres are not great candidates for solar or wind power, they can use clean-energy technologies. British company Rolls Royce is making small modular (nuclear) reactors with potential in this area. However, while support for nuclear energy is rising in the UK, it remains controversial. So . . . er . . . good luck to whoever gets that PR job.
Is AI overhyped?
I see much to applaud in the government’s report and action plan. AI matters; it will evenutally change many of the roles, rules and processes we currently take for granted. It will obliterate some. It is a really exciting technology with loads of scope and commercial value. I am fascinated to see how it unwinds. But I will be patient, because the maturation of AI will take decades.
Because AI cannot mature until humans shape it. It is a tool, not a magic spell. I’d be very much richer if I had £1 for every time I’ve read or heard something to the effect of ‘AI will make everyone unemployed so they’d better give us all a universal basic income now or we’ll be eating out of bins by next year’ or (more often) ‘At Completely Made Up Company Ltd. we pride ourselves on being AI driven. In everything. Always. We don’t even answer emails unless ChatGPT tells us how to; we love AI so much we’ve actually lost the ability to complete basic human tasks, that’s how committed we are’. The extremity of the doom-mongers often suggests fear more than understanding.
AI has brakes and limits, just like other technologies. Limits on computing power, data quality and the ability of humans to harness it for genuinely useful purposes. As a species we are often better at imagining possibilities than we are at making them happen. And hysteria at large-scale technological change is nothing new. Remember the Luddites?
That said, we have Daisy the AI granny, so maybe AI has hit its peak already. We – and Keir Starmer – will have to wait and see. Hopefully the government is busily solving the other big challenges while we do so.
Comments are closed