UK employers are more cautious about hiring than at any point outside the pandemic, and AI is part of the reason. The CIPD’s quarterly Labour Market Outlook, drawn from a survey of more than 2,000 employers, found that one in six expect AI to shrink their workforce over the next year, against 6% who expect it to grow.
The squeeze falls hardest on early-career and lower-level professional roles. One in four large private-sector firms expect their headcount to fall. Across the private sector as a whole the figure is 17%, and in the public sector 20%, with finance and insurance, IT, and administrative support the hardest hit. The CIPD also cited the World Economic Forum’s Future of Jobs Report 2025, which says 40% of employers worldwide expect to reduce their workforce by 2030 in favour of AI automation.
Why firms can’t find the AI skills they need
A separate survey points to an AI skills gap. The survey of the UK’s AI labour market, commissioned by the Department for Science, Innovation and Technology, found that 97% of businesses reported at least one skills gap. Some 57% pointed to a shortfall in technical skills such as programming or modelling. The single biggest gap was understanding AI concepts and algorithms: 60% of respondents named it in 2025, up from 55% five years earlier.
More than a third of organisations, 35%, said they struggled to fill AI roles outright, most often because candidates lacked work experience or sufficient technical skill. Employers have widened their search rather than relying only on training: apprenticeships rose from 3% of AI hires in 2020 to 19% in 2025, and 38% of businesses now recruit from outside the UK. Even so, only 13% of graduate schemes include any AI training, and 88% of organisations lean on informal, on-the-job learning instead of a structured programme.
The global picture: fewer jobs vanish than change
The International Labour Organization’s 2025 update to its worldwide exposure index found that one worker in four holds a job with some exposure to generative AI, a figure assessed across nearly 30,000 tasks. Its average automation score across occupations fell slightly, from 0.30 in 2023 to 0.29 in 2025. The ILO said most exposed jobs would be transformed rather than eliminated, because they still need a person to do them.
One exception stands out. As image, voice, and video generation tools have improved, the exposure score has climbed specifically for media and web-related occupations. The ILO’s conclusion was that governments and employers need to manage the shift through social dialogue.
Anne-Marie Imafidon, a mathematician and STEM advocate who addressed the CIPD’s annual conference, told the conference that the future of work is already taking shape, and that nothing about it is inevitable.
