The promise of AI is vast, but its full economic impact is still emerging. Nevertheless, labour market effects are already visible: job adverts have dropped most for the occupations most exposed to AI. While this has raised concerns – especially for young graduates – the real story is more complex, with AI acting alongside broader economic forces. For their part, to avoid long-term talent gaps, businesses need to rethink both how they integrate AI and how they continue to invest in future skills.
Deep transformation takes time: a lot of AI’s promise is still in the future
AI and large language models (LLMs) are being rolled out at speed across sectors. In McKinsey’s latest global survey, nearly 80% of the world’s largest companies – typically with tens of thousands of employees – reported using AI in at least one business function. In the UK, while smaller firms are lagging behind, over a third of mid-sized businesses (with more than 250 employees) say they are now using these technologies.
Yet, for all the momentum, broad-based productivity improvements remain elusive to date. Although 92% of global businesses plan to ramp up investment in generative AI in the next three years, just one percent believe their efforts have reached maturity. Only around 20% report a tangible impact on enterprise-level earnings. The tools exist, but integrating them into workflows and changing how people work is proving challenging.
This paradox – widespread usage but unrealized gains – reflects a gap between surface-level adoption and deep transformation. Many companies have deployed tools like chatbots and employee copilots – the “horizontal” use cases. These can be easier to scale but, at present, deliver modest, diffuse benefits. By contrast, 90% of the more powerful “vertical” use cases – where AI fully automates specific business processes – remain stuck in pilot mode.
Hiring hesitation: the slowdown in professional jobs due to AI-exposure
If significant productivity gains from AI are still in the future, you might expect companies to maintain their usual approach to recruitment – all other things equal. But other things are not equal, either at the macroeconomic level or when it comes to the anticipated impact of AI.
At the macro level, while the data for June and July 2025 suggests a slight uptick in hiring intentions, the broader picture is of a marked cooling in the labour market. Between the three months ending in May 2022 and May 2025, the number of vacancies fell by 43%, from 1.3 million to 0.7 million. Analysts expect unemployment to rise from 4.6% to 4.8% by the end of the year.
Much of the job market slowdown reflects broad economic forces. Unprecedented uncertainty from geopolitical tensions, higher interest rates to curb inflation, cautious households saving more and consuming less, tighter public finances, and rising employee costs are all contributing to a slower pace of recruitment. The net balance of businesses expecting to increase rather than decrease employment has fallen – from +2.4% in July 2022 to just +0.2% in July 2025 – reflecting fewer firms expanding and more planning to shrink their workforce.
However, there are also signs that the advent of AI and LLMs is dampening hiring intentions. Since enterprise-wide productivity gains have yet to materialise, this cannot be in response to large-scale output improvements. But a mix of observed task-level time savings, and the anticipation of significant – albeit uncertain – future productivity gains, especially as the technology and its applications mature, is prompting companies to review their workforce strategies and pause aspects of their recruitment.
This effect is particularly pronounced in roles where AI has the most potential to reshape human work, either by automating or augmenting it. Since the three months ending in May 2022, the overall volume of online job ads has declined by 31%. However, while the reduction for occupations with low AI exposure was 21%, job ads dropped by 38% for roles with high exposure to AI and LLMs (Exhibit 1). Some of the biggest declines were in jobs that have been predicted to have the highest impacts from generative AI. These include software developers and other IT workers, as well as professionals in data, design, media, research, legal, HR, finance, and business.
Young graduates: navigating a tougher job market
These changes have been noticed by employers, job seekers, and recruitment professionals alike, giving rise to concerns about young people’s job prospects in particular. Has the slowdown hit entry-level roles – especially those for new graduates – harder? Our analysis suggests that the story is nuanced, but that young graduates are potentially facing a triple-whammy: a general labour market slowdown, a sharper decline in graduate-level job openings, and reduced demand for lower-skilled roles which, despite not requiring degree level qualifications, are often taken up by new graduates.
At the macroeconomic level, rises in unemployment and declines in vacancies may favour those who already have a job, rather than those seeking their first posting. Historical data shows that youth unemployment typically rises faster and recovers more slowly than overall unemployment; young people are more likely to be in temporary roles, concentrated in cyclical sectors, or simply not yet established in the labour market. Indeed, while the UK’s overall unemployment has increased from 3.8% in the three months ending in April 2022 to 4.6% in 2025, the rate for those aged 16 to 24 has gone from 10.9% to 14.3%.
As far as graduate roles are concerned, however, it is not yet clear that vacancies have declined any more in jobs likely to be taken up by young people: job ads for occupations typical for both young and older graduates (with a bachelor’s degree or higher) have declined by 33% since the three months ending in May 2022 (Exhibit 2). This likely reflects the similarity of roles and overlap in the tasks performed by younger and older professionals. However, when it comes to recruitment practices, organisations are often more likely to hire a more experienced candidate.
Finally, young graduates are facing challenges for a third, less obvious, and less AI-related, reason: many start out in jobs that don’t require a degree. In the England and Wales Census 2021, nearly 30% of graduates aged 24 or less worked in low-skilled jobs—as sales assistants or retail cashiers, in bars, restaurants, coffee shops, leisure parks, and storage facilities, or as care workers (Exhibit 3). Apart from call centre operators, these are typically not roles that are highly exposed to AI, because they require significant physical dexterity and social interactions. However, they have faced a more significant drop in vacancies than roles more typically held by older people. In contrast to professional roles, this slowdown is more likely due to weak consumer demand and rising employment costs, not AI.
A smarter approach: reshape work, don’t stall talent
It is logical for organisations to reflect on what their future workforce might look like and take time to jointly plan their approach to AI adoption and recruitment.
However, if entry-level hiring continues to slow – whether for AI-exposed professional roles or for the kinds of lower-skilled jobs that often give graduates their start – organisations risk leaving gaps in their future workforce. Pausing recruitment may seem prudent but risks becoming a missed opportunity to attract the next generation of talent, build capabilities, and strengthen resilience. And once that pipeline breaks, it’s hard to rebuild.
To avoid this, employers need to take a twin-track approach. First, focus on deliberate integration of AI into workflows – not just deploying tools, but redesigning roles and processes to maximise their value. This might mean investing in change management, rethinking job design, reskilling and training, and building the internal capability to continuously adapt. Crucially, it also means identifying which tasks are best automated – and which need human creativity, judgment, and relationships. Done well, this can help ease workforce shortages and elevate human roles to focus on more complex, higher-value work, making AI a productivity multiplier, not just a cost-cutting tool.
Second, keep investing in people – especially early-career talent. Today’s entry-level hires will shape tomorrow’s AI strategies, culture, and competitiveness. Doing so requires a shift in mindset: treat graduate and junior roles not as short-term costs, but as long-term assets. AI may be causing growing pains now, but its long-term success depends on how people and technology evolve together. Businesses that maintain their talent pipelines – even during economic and technological transitions – will be better placed to scale innovation, stay resilient, and lead the next wave of transformation.