Is The UK Jobs Market Ready For AI's 'Transitionary Shocks'?
Illustration by Tracy Worrall
9 min read
AI is transforming the world of work. Will it lead to a spike in unemployment? What is being done to prepare for one? Noah Vickers reports
At the start of this year, Keir Starmer set out his plan to “turbocharge” the development of artificial intelligence across the economy. Unveiling an ‘AI Opportunities Action Plan’, he hailed it as a strategy to deliver “more jobs and investment in the UK, more money in people’s pockets, and transform public services”.
But buried within that plan led by entrepreneur Matt Clifford was a warning. “While some jobs will be replaced by AI, many will be augmented – and an unknown number will be created.”
It is this uncertainty that lies at the heart of concerns over the speed with which AI appears already to be displacing jobs and suppressing recruitment, particularly for graduates seeking entry-level roles.
A new study from King’s College London – which analysed millions of job postings and LinkedIn profiles from 2021 to 2025 – found that the release of ChatGPT in late 2022 marked a turning point for certain sectors.
Firms whose workforces were “highly exposed” to AI reduced total employment by 4.5 per cent on average. The effect was concentrated almost entirely in junior positions, which fell by 5.8 per cent.
The most pronounced impacts, however, have been seen in hiring intentions, as highly exposed companies became 16.3 percentage points less likely to post new vacancies over the period studied. The sharpest decline was seen in listings for technical roles like software engineers and data analysts, while customer-facing jobs, such as sales representatives, in fact saw a small increase.
Separate research by the IPPR think tank in March last year found that around 11 per cent of tasks in the UK jobs market were exposed to displacement by AI in its “here and now” form. Among the most exposed professions identified in that paper were personal assistants and secretaries, HR staff, marketing associates and writers.
But the proportion of exposed tasks across the overall labour market rises to 59 per cent once the technology has progressed into a second phase where “integrated” AI systems are given more access and ability to execute tasks.
Carsten Jung, the IPPR’s associate director for economic policy and AI, warns that this transition could be happening faster than anticipated.
“The one big development that was only in its infancy when we wrote [the paper] was agentic AI,” says Jung, referring to a version of the technology capable of autonomous decision-making with little or no human supervision.
“AI is now increasingly capable of doing multi-task workstreams. That will make it easier to move to phase two, where we integrate AI into systems that aren’t necessarily made for AI.
“All that means that the timeline is maybe a bit more near-term than I’d thought at the time, but the occupations [identified as] exposed are probably still the right ones.”
There will be an interim transitionary phase that could be quite brutal for some people
For experts attempting to forecast AI’s impact on the labour market, one of the key questions is how long the time lag might be between the displacement of jobs and the creation of new ones, which in turn could require major support to re-skill enough workers.
“It is quite likely that we will have, at least, transitionary shocks,” says Jung, who points out that “market signals” often come too late for employees deciding what sector to train themselves up in.
“If we don’t have government intervention to guide people into a different direction, then it might eventually be OK, but I think there will be an interim transitionary phase that could be quite brutal for some people.”
In a report last year exploring that transitionary phase, researchers at the Tony Blair Institute for Global Change (TBI) modelled four future scenarios.
In the most likely of those, “nearly 1.5 million workers lose their jobs as a result of AI, but crucially these redundancies do not occur all at once”. They instead occur gradually, peaking at close to 100,000 additional redundancies per year around 2040, when the rollout of AI is at its fastest.
Here, the time lag between job losses and “new task creation” – to fully unwind the effect of those losses – is estimated at five years.
But in the most damaging scenario, this lag would instead occur over 10 years. In that outcome, three million jobs would be displaced, albeit again spread over several years, with annual redundancies peaking at 274,000 in the mid-2030s, before “re-employment effects kick in” in the 2040s.
This scenario is not thought to be “particularly likely” though, as “it relies on AI being very good at replacing workers in their current jobs but not creating any new products or services”.
How much modelling of this kind is going on inside Whitehall? In answer to a recent written question on the topic, employment minister Diana Johnson said: “No current assessment has been made by the Department for Work and Pensions (DWP) on the potential impact of AI on employment.”
She added, however, that the government “is planning against a range of plausible future outcomes and closely monitoring the data that will help track if we are heading towards any of these outcomes”.
The convoluted construction of her answer hints that, in fact, the government has been trying its best to model how AI will impact the labour market but doesn’t want to alarm voters. A Whitehall source told The House that multiple pieces of research had been commissioned “none of it for external consumption”.
What we don’t want is AI to turn our society on its head
Labour MP Neil Duncan-Jordan points out that if companies make large-scale redundancies in favour of AI substitutes, they will be paying far less in national insurance (NI) contributions.
The MP for Poole says the government should consider requiring businesses to pay an equivalent level of tax for “each AI agent that performs tasks previously done by people”. This idea was recently given short shrift by Treasury minister Dan Tomlinson.
“The minister, rather foolishly I think, said that you can’t have NI on robots because they’re not employees. That wasn’t what I was getting at. That was a bit dismissive, to be honest,” Duncan-Jordan tells The House.
“It’s not crazy. It’s a bit like redundancy pay – which they would have paid anyway, of course – but then it’s about the ongoing loss of income to the state, effectively, through lack of national insurance, through lack of wages.
“What we don’t want is AI to turn our society on its head so that we have large-scale unemployment.”
Duncan-Jordan believes ministers should develop a comprehensive strategy to prepare for “the impact of AI on employment, our economy and society at large”.
His Labour colleague Allison Gardner adds that while there are strict legal requirements for employers who try to sack workers only to rehire them or someone else on less favourable terms, there is no such prohibition if they want to replace them with an algorithm.
“There are some questions there about how we could tackle that and put the onus on employers to justify it,” she says, “and make sure they support people that they are making redundant due to automation, that they are helping them skill up.”
Both the IPPR and TBI agree that a better ‘safety net’ will be needed for the most exposed workers. The UK has the third-lowest level of unemployment benefit among OECD countries – and that is a problem not just for those made redundant, but also for anyone who tries to jump before they are pushed.
“Currently, if you quit your job to retrain, you lose all your income,” says Jung. “If you have some savings, you’ll pay from your savings. If you don’t have any savings, you’ll be straight on – by international comparisons – very low welfare support.
“So, you really have a disincentive for people to be forward-looking and quit their job and retrain. That needs to change. We need to celebrate and support people in moving into new roles, in up-skilling, moving across the country to where new industries are.”
To facilitate that re-skilling, the government has created Skills England, an executive agency recently transferred from the Department for Education to the DWP.
The organisation is focused on 10 “priority sectors” for the UK economy, with digital technologies, adult social care, construction and engineering all expected to see the largest increases in job demand.
Alarmingly for Skills England, only 57 per cent of recent apprenticeship starts were aligned to one of those priority occupations.
Jung argues that getting enough workers into the AI-resilient occupations will require “bold signalling” from the government, who should be “really doubling down on those sectors where we know, very likely, there will still be demand”.
In September, Chancellor Rachel Reeves announced plans for a “youth guarantee”, under which eligible unemployed young people who have been on Universal Credit for 18 months without earning or learning “will be provided guaranteed paid work”.
While Jung welcomes the initiative, the devil be in the policy’s detail, as he asks: “Who will offer these jobs? Who will pay for them? Will they be in future-proofed industries?”
Lord Knight, who served as an employment minister in Gordon Brown’s government, meanwhile says he is concerned not only about young people struggling to get entry-level employment, but also middle-aged workers who after losing their jobs to technology, “lose hope that they can retrain and pivot into a new opportunity”.
He adds: “It’s right to give the DWP more strategic oversight about aligning employment with the sectors that we need to develop for the industrial strategy. But, as ever, the constraint is whether or not there is enough resource to go into the youth guarantee, whether there’s enough resource to go into adult skills.”
A government spokesman tells The House: “We are determined to make the most of the benefits AI can bring while managing the risks.
“As part of our drive to provide our future workforce with the AI skills they need, we are partnering with major tech firms to offer AI skills training to millions of workers, while also providing training in classrooms and communities for tech careers of the future.
