What AI is really doing to jobs right now

Behind the layoff headlines: is AI really causing job losses?

AI is increasingly cited in job cuts. Challenger, Gray & Christmas, a career transition firm, recorded 87,714 US AI-related job cuts by May 2026, already above 54,836 for all of 2025.

But headline figures can overstate AI’s direct role. In our analysis of 54 companies and around 245,000 announced AI-related cuts in the year to May 2026, only a small share directly reflected AI replacing workers.

The stronger signal is that companies are rethinking how work is organised. Firms are adjusting teams, skills and investment priorities, sometimes redirecting labour budgets towards AI infrastructure. These changes are often framed as AI adoption, but reflect a broader restructuring of the business.

For now, the clearest signal is not fewer workers, but a shift in which roles, skills and investments matter most to organisations.

Redesign before replacement

AI’s near-term impact is work redesign: more automation, fewer routine roles and greater emphasis on judgement and client interaction.

In some organisations, this is already leading to leaner structures and less reliance on junior-heavy models. As a result, redundancy risk is increasingly concentrated in entry-level roles. A Stanford University study finds that employment for workers aged 22–25 in highly AI-exposed roles has fallen by around 13% since 2022.

Exposure does not always translate into decline

If exposure translated directly into employment decline, the most exposed roles would already be contracting sharply. They are not.

Across the US, Canada and Australia, high-exposure roles – including software, data and IT – continue to grow. Many are in high demand because they are central to the development and deployment of AI. France and Germany show broader weakness across most roles, reflecting broader economic conditions rather than a clear AI-driven effect.

Stagnating employment and softer pay growth in the UK suggest a broader labour market slowdown, rather than a shift driven by AI, though AI may help firms run leaner teams. Rising exposure is not yet translating into broad-based employment decline – and where hiring is slowing, the drivers appear cyclical rather than structural.

The real shift is more gradual than it looks

AI is becoming more visible in job postings, but the evidence is hard to interpret.

Indeed Hiring Lab data shows AI mentions rising in job ads since 2023, especially in Canada, the UK and Australia. But levels remain modest, and the term itself is imprecise: “AI” in job postings can mean anything from building models to using AI tools in recruitment.

This reflects early adoption – visible, but still uneven. In many organisations, AI is being layered onto existing workflows rather than changing how work is designed. A late-2025 McKinsey global survey found that nearly 90% of respondents said their organisations were regularly using AI in at least one business function. But 80% are yet to see significant business gains.

The constraint is organisational and investment-related rather than technological. Scaling AI requires better data, new skills, trusted outputs, governance and redesigned workflows – all of which take time. AI’s impact is therefore real, but limited: still early, uneven and constrained by both cost and the way organisations work.

How AI impacts office demand

AI will affect office demand indirectly and unevenly: tasks change first, then roles, then organisation structures, and only then real estate requirements.

While most firms have adopted AI, few have yet embedded it fully into the way work is organised. For now, hybrid working remains the dominant driver of office utilisation. For most organisations, AI may improve efficiency over time, potentially slowing headcount growth.

Meanwhile, AI firms are expanding rapidly. Headcount in AI companies roughly tripled since 2020 (Pitchbook), and London-based firms have leased more than 1 million sq ft since 2025 (Knight Frank Insight). AI may also create indirect upside for some non-AI firms where it helps reshape products, services or business models. For the wider market, this creates a mixed picture: selective growth alongside gradual efficiency.

What is becoming clearer is the role of the office. As routine tasks are automated, value shifts towards collaboration, training and decision-making.

AI is therefore likely to make office demand more concentrated: more focused by location, more demanding on quality and more purposeful in use.

A slower, more selective shift

AI is not yet reshaping office demand at scale because most firms are still experimenting with how it changes work.

The bigger shift will come later, if AI moves from a tool used by individuals to an enterprise-wide system that changes workflows, staffing models and decisions about space. When that happens, the impact will not be uniform. It will favour the organisations, locations and buildings best able to support new ways of working.

AI will not shrink the office overnight, but it will change what firms need from it.