NextFin News - The Bank of England is drawing a straight line between artificial intelligence and a two-track UK labour market: firms adopting the technology are seeing higher productivity, but the gains are arriving alongside weaker hiring and pressure on lower-value jobs. That tension matters because the central bank is no longer talking about AI as a distant productivity story. It is now treating it as an active force in a labour market already showing slower hiring, softer vacancies and rising anxiety about entry-level work.
AI Is Lifting Output, But The Labour Market Is Paying The First Bill
The Bank’s recent remarks come against a backdrop of slower British growth and a labour market that is still functioning, but not with the easy hiring conditions of the post-pandemic rebound. The Bank Rate was left at 3.75% on 29 July 2026 by a 6-3 vote, underscoring how policymakers are still balancing inflation risks against weak trend growth. In May, Governor Andrew Bailey said meaningful productivity gains usually come from new products and activities, not just automating existing tasks, and he argued that AI could help only if the economy also invests in education and skills. In June, the Bank’s conference on transformative artificial intelligence framed the topic around structural change and monetary policy, not just software adoption.
That framing is important because the story is not merely that AI improves efficiency inside one firm. The channel is broader: when a company can produce more output with fewer workers, or with a smaller share of junior staff, the savings show up first in margins and cash flow, then in slower hiring, and only later in aggregate productivity statistics. The early evidence from the UK points in that direction. A survey of UK job postings found that AI or related tools and programmes appeared in 9.4% of postings at the end of June, a record high, while job postings overall fell 11% between the start of 2026 and July 17 and were 32% below pre-pandemic levels. In the same period, graduate postings were at their lowest since 2020. The labour market is not collapsing, but it is becoming more selective.
The Bank is essentially arguing that this is not a simple cyclical squeeze. A cyclical slowdown would normally hit broad hiring and then reverse when demand improves. Here, the pressure appears to be falling hardest on tasks that AI can substitute or compress, while jobs requiring judgment, specialised technical skill or direct AI oversight are still being created. That is why the AI productivity story and the jobs story are not opposites. They are the same mechanism seen from two sides: firms that adopt AI can do more with less, and workers whose tasks sit near automation thresholds face weaker demand.
That mechanism is already visible in the structure of job ads. AI-related skills are increasingly clustered in senior and specialist roles, while retail, manufacturing and routine white-collar functions have faced double-digit declines in vacancies. If that pattern holds, the real economic effect is not just fewer jobs. It is a reshuffling of who gets hired, at what level, and at what wage. That matters for productivity because the first wave of gains often comes from removing friction and reassigning tasks, not from an immediate surge in national output per hour.
The obvious question is whether this is a temporary adjustment or the beginning of a more durable regime shift. The answer, for now, leans structural. Unlike a one-off demand shock, AI adoption changes production functions, task allocation and the skill premium inside firms. Those changes do not revert simply because the business cycle turns. But the labour-market impact can still be cyclical on top of that structural layer: if the economy weakens, companies will use AI as a justification to delay hiring even more aggressively, and junior workers will absorb the first hit. That is why the Bank’s messaging sounds more cautious than celebratory.
Why The Market Is Not Yet Pricing The Full Labour Risk
The market’s easiest read on AI is still the productivity bull case: higher output, lower costs, better margins and eventually faster growth. That is the part investors can model. The harder part is the second-order consequence: if firms use AI to cut labour demand faster than new revenue appears, national income may rise more slowly than company profits, at least initially. That gap matters for wages, consumer spending and the shape of the recovery. A productivity gain that arrives through headcount reduction can be disinflationary for firms and painful for workers at the same time.
That second-order effect is especially relevant in the UK because the macro backdrop is already fragile. The Bank has said the economy’s potential growth has slowed markedly since the financial crisis, and Bailey has linked that to weaker productivity, slower labour supply and a less favourable growth environment. In that setting, AI is not just another tech upgrade. It is one of the few mechanisms that could lift trend productivity. But if adoption happens by replacing the least expensive labour first, the near-term social cost can be larger than the output gain. In practical terms, the payback may be delayed for the economy while arriving quickly for firms and shareholders.
“Really meaningful gains to productivity tend to come from wholly new products and activities, not so much from automating our existing tasks,” Andrew Bailey said in a speech at Sheffield’s Cutler’s Feast in May.
That sentence is the key to the Bank’s thinking. It implies that automation alone is not enough to solve the UK’s growth problem. If companies simply trim labour costs without creating new products or expanding demand, the economy gets a cleaner income statement but not necessarily a bigger economy. The central bank’s AI conference in June reinforced that point by putting labour markets, innovation and structural change in the same conversation. The question is not whether AI saves money. It is whether that saving becomes broad-based investment, higher capital formation and new output, or just a thinner payroll.
The comparison with prior technology waves helps. Past general-purpose technologies tended to create a lag: firms adopted them first for cost control, then reorganised production, and only later did the productivity gains become visible in the macro data. That lag is one reason the current episode is likely to look cyclical in the data before it looks structural in the economy. Hiring weakens now; measured productivity improves later. The timing mismatch creates the false impression that one has little to do with the other. In reality, the labour-market pain is often the leading indicator.
There is also a distributional angle. AI tends to reward workers who can direct, supervise or complement the technology, while reducing demand for workers whose tasks can be standardized. That creates a two-speed market inside the same economy: higher pay and more openings in specialist roles, thinner opportunities in entry-level work. Such bifurcation does not show up cleanly in headline unemployment right away, especially if firms respond by freezing vacancies rather than firing staff. But it does appear in the mix of postings, the composition of wage growth and the share of jobs explicitly asking for AI skills.
The strongest counter-thesis is that this is mostly a normal late-cycle labour slowdown dressed up in AI language. Hiring has weakened because employers are cautious, wage pressures remain elevated and the economy has been sluggish; AI may simply be the latest excuse, not the cause. That argument has force because broad labour-market weakness can swamp any technology effect in the short run. If vacancies keep falling across all sectors, if AI-related postings stop rising, and if junior hiring recovers when demand improves, the present narrative would look overstated. But the burden of proof is shifting against that counter-thesis because the composition of hiring is changing even before the cycle turns.
The falsifying signal for the structural-automation view is specific: if AI-related job postings fall back below 7% of all UK postings while overall vacancy levels recover to their 2025 average and graduate hiring returns to pre-2024 trends, then the current episode would look like a cyclical compression rather than a regime change. Until then, the balance of evidence says AI is not just adding productivity. It is also deciding who gets left out of the labour market first.
What It Means For Growth, Wages And Policy
The near-term beneficiaries are the firms that can convert AI into lower operating costs, faster workflows and higher output per employee. The exposed groups are entry-level workers, routine white-collar roles and sectors where task standardisation is high. Over the medium term, the question becomes whether those cost gains are recycled into capex, new products and hiring or whether they remain trapped as margin expansion. The long-term answer will determine whether AI becomes a genuine UK growth engine or simply a more efficient way to run a weak labour market.
For policy, the implication is uncomfortable. If AI raises measured productivity but suppresses job creation in the near term, the Bank may face a more complicated combination of slower wage growth, weaker demand and pockets of labour-market damage. That can matter for monetary policy because it changes how quickly inflation pressure fades and how much slack is hidden beneath the headline figures. The Bank’s current posture suggests it sees this as a structural shift that will need skill formation, adaptation and time, not just a policy response to a single weak print.
In the short term, the base case is a continuation of the two-speed labour market: AI-exposed firms keep adopting, hiring stays selective, and productivity headlines improve before labour-market pain eases. In the medium term, the upside case is that AI investment starts creating new products and roles fast enough to offset substitution, which would support growth without a further rise in unemployment. The downside case is that firms use AI primarily to avoid hiring, especially at the bottom of the ladder, while demand stays soft and the UK labour market loses another layer of dynamism.
What would prove this view wrong? A broad re-acceleration in vacancies, a rebound in graduate hiring, and evidence that AI postings are rising alongside total hiring rather than replacing it. Until then, the Bank’s message is clear: AI is already making firms more productive, but it is doing so in a way that leaves the labour market worse off before it looks better.
NextFin News - AI is not just lifting output in Britain; for now, it is also deciding which jobs never come back.
Explore more exclusive insights at nextfin.ai.
