Edward Egan

Headlines warn of a looming ‘jobpocalypse’, however the actuality is extra advanced. Reasonably than merely inflicting a wave of job losses, the financial literature suggests generative AI may affect the labour market via a number of – doubtlessly offsetting – channels: productiveness positive factors, job displacement, new job creation, and compositional shifts. The stability between these results, quite than displacement alone, will form AI’s combination influence on employment. The most recent analysis means that general results stay restricted up to now, however there are some early indicators of AI’s influence. I discover that, since mid-2022, new on-line vacancies in essentially the most AI-exposed roles have decreased by greater than twice as a lot because the least uncovered group. This highlights the necessity for ongoing monitoring as AI adoption accelerates.
How will AI have an effect on employment?
To assist us assume via this advanced query, we are able to use a ‘task-based’ framework (Acemoglu and Restrepo (2019)). This strategy stems from the concept that jobs are made up of an outlined set of duties. Reasonably than taking a look at broad occupations or industries, it’s extra helpful to know how explicit duties could be automated, augmented or created by new applied sciences like AI. The influence on any given job will then rely on the combination of various duties inside that function.
For instance, in finance, AI may assist automate knowledge assortment and reporting, which is a big a part of a junior analysts’ function, whereas senior portfolio managers would possibly use AI to scan market sentiment or simulate threat eventualities – therefore utilizing AI to streamline decision-making. This can assist clarify why some roles could also be displaced by AI whereas others could turn into extra productive, regardless of being in the identical trade.
We are able to broadly simplify this framework into 4 key channels via which AI could have an effect on the labour market:
- Productiveness (Augmentation): AI could make employees extra productive by automating repetitive duties, liberating employees up for different higher-value actions. If companies use positive factors to develop manufacturing, this could improve demand for labour in non-automated duties.
- Displacement (Automation): AI may automate a big share of (if not all) duties in some roles, lowering demand for labour in sure jobs.
- Reinstatement (New Duties): Traditionally, technological improvements create new duties that we couldn’t have imagined earlier than. For instance, in an AI context, this might imply the emergence of latest roles which assist customise and combine AI instruments into companies’ workflows. For the reason that begin of 2023, there was a big improve in demand for these employees (generally known as Ahead-deployed Engineers).
- Compositional (Reallocation): Even when combination employment doesn’t change considerably, AI is prone to reallocate jobs between sectors. Some industries would possibly shrink, others develop, and a few employees might want to retrain to adapt their abilities accordingly.
Many of the public debate focusses on the proof across the ‘displacement’ channel. However maybe an important message to remove from this publish is that the long term web influence of AI on employment will rely on the stability of those results, in addition to the velocity of AI improvement and adoption. Since these forces may unfold over totally different time horizons, understanding how they in the end stability out stays extremely unsure at this stage.
What does the proof say up to now?
Regardless of widespread hypothesis about AI-driven job losses, the mixture proof for the UK stays restricted. A current Choice Maker Panel Survey discovered that AI has had little impact on employment up to now, with solely a minor discount anticipated in coming years. Equally, the Enterprise Insights and Situations Survey stories simply 4% of AI-using companies (23% of all companies) decreased their workforce on account of AI, whereas solely 7% of future adopters anticipate reductions. In the meantime, knowledge from Certainly reveals that demand for AI-related abilities has elevated within the UK not too long ago (Chart 1), suggesting some early proof for the ‘reinstatement’ impact, as new duties that require AI-related abilities have gotten extra widespread.
Chart 1: Share of Certainly job postings referencing AI abilities (per cent)

Supply: Certainly. Information to October 2025.
Proof from the US additionally suggests the story is extra nuanced. Researchers on the Yale funds lab discover no important combination labour market disruption up to now, noting that shifts in job composition started earlier than AI’s widespread adoption. Whereas some have attributed the rise in youth unemployment to be on account of AI, evaluation from the Financial Innovation Group and the Monetary Instances finds that broader macroeconomic elements are nonetheless prone to be extra necessary. Encouragingly, survey knowledge from the Federal Reserve Financial institution of New York reveals most AI-using companies are at the moment retraining employees quite than chopping them. This underscores that displacement is just one channel of AI’s labour market influence, with upskilling and new job creation additionally taking part in an necessary function in future dynamics.
Digging deeper: slowing in AI-exposed occupations and for junior employees
Whereas general employment results appear muted, there could also be some early indicators of influence in additional AI-exposed occupations. My evaluation of UK knowledge finds a damaging relationship between posting of latest on-line job vacancies and AI occupational publicity. In different phrases, the extra uncovered a job is to AI, the much less doubtless a agency is to publish a brand new emptiness in that place. This relationship is much more pronounced if we group jobs into AI publicity quintiles (Chart 2). Right here, I discover that new on-line job postings in essentially the most AI-exposed roles have dropped by virtually 40% relative to mid-2022, greater than double the autumn within the least uncovered group. Whereas these findings corroborate comparable work by McKinsey, it could possibly be the case that these occupations are merely extra uncovered to a cyclical slowing within the financial system, so this proof suggests correlation quite than proving any causation.
Chart 2: Proportion change in new on-line job postings since mid-2022 by AI occupational publicity quintile

Notes: ONS on-line emptiness knowledge by SOC is experimental so needs to be handled with warning and is probably going topic to future revisions. Six-month averages are used to easy volatility and lacking knowledge. Division for Schooling (DfE) use Felten et al (2021) measure of AI occupational publicity and map this to UK labour market knowledge.
Sources: DfE (2023) and Experimental ONS on-line emptiness knowledge.
Latest tutorial analysis additionally finds sooner falls in vacancies and employment in AI-exposed occupations, significantly concentrated in junior positions. Henseke et al (2025) discover that, by mid-2025, UK job postings have been 5.5% decrease in AI-exposed occupations than they might have been if pre-ChatGPT developments had continued. Equally, Teeselink (2025) finds that extremely uncovered UK companies decreased employment by 4.5% (concentrated virtually solely in junior roles) and have been 16 proportion factors much less prone to publish new vacancies. Within the US, analysis finds early-career employees in essentially the most AI-exposed occupations have skilled a 13% relative decline in employment, whereas much less uncovered and extra skilled employees in the identical roles have been largely unaffected (Brynjolfsson et al (2025)). Analysis from Hosseini Maasoum and Lichtinger (2025) largely corroborates this, discovering that the adjustment has largely taken place through decreased hiring quite than elevated layoffs.
However regardless of rising proof, AI doubtless stays an amplifier quite than the only real driver of the slowing in youth employment. Most research acknowledge that there’s a lack of high-quality knowledge and important challenges with disentangling specific causality, particularly given the tightness (and subsequent loosening) of the labour market since ChatGPT’s launch in November 2022. So, whereas AI could also be amplifying results for hiring of latest entrants in AI-exposed sectors, the broader slowdown seems to additionally replicate typical labour market downturns, the place youthful and fewer skilled employees are disproportionately affected.
What about longer-term forecasts?
Forecasts range considerably, however most counsel the outlook is much less extreme than headlines suggest. Eventualities of UK job displacement on account of AI vary from zero to round eight million over the long term (IPPR (2024), Tony Blair Institute for World Change (2024), PwC (2018)), however most evaluation expects this to be largely offset by the creation of latest roles and better productiveness, consistent with historic proof from earlier technological advances (Hötte et al (2023)).
The important thing threat is that if productiveness positive factors are extra restricted than anticipated and if new jobs and duties should not created shortly sufficient to offset these misplaced to automation. This might result in a brief rise in unemployment, although the magnitude would rely closely on the velocity of AI adoption and measurement of the displacement impact (Goldman Sachs (2025)).
One other threat to the long-term outlook stems from the event of extra superior types of AI (akin to ‘Synthetic Basic Intelligence’). This publish doesn’t discover what this might imply for the labour market, however some counsel the impacts could possibly be extra extreme (Restrepo (2025)).
Conclusion
Present proof suggests AI has had little impact on general labour market dynamics up to now. Nevertheless, my evaluation and different analysis finds indicators of AI amplifying the slowdown in hiring in AI-exposed occupations. Wanting forward, the impacts could possibly be broader if AI’s productiveness positive factors disappoint or if new roles don’t emerge shortly sufficient. This might pose a threat of upper unemployment which may take a while to unwind because the labour market adjusts. Due to this fact, it’s important to watch not solely displacement results, but in addition how AI is impacting productiveness, job creation charges and compositional shifts. Creating extra subtle metrics for monitoring these elements might be key to understanding the transition to an AI-augmented financial system. Finally, the long term web influence of AI on employment will rely on the stability of the consequences outlined on this weblog and the velocity of AI improvement and adoption.
Edward Egan works within the Financial institution’s Worldwide Surveillance Division.
If you wish to get in contact, please electronic mail us at bankunderground@bankofengland.co.uk or depart a remark under.
Feedback will solely seem as soon as accepted by a moderator, and are solely printed the place a full identify is equipped. Financial institution Underground is a weblog for Financial institution of England employees to share views that problem – or help – prevailing coverage orthodoxies. The views expressed listed here are these of the authors, and should not essentially these of the Financial institution of England, or its coverage committees.
Share the publish “Generative AI: degenerative for jobs?”
