The Productivity Puzzle: Artificial Intelligence (AI), Workers and the Missing Gains 

More than half of UK firms are now using artificial intelligence (AI) and almost none of them report it changing their workforce. Research published by the British Chambers of Commerce (BCC) in March 2026, carried out with Atos and the University of Essex, found 54% of firms actively using AI, up from 35% in 2025, 25% in 2024, and 23% in 2023. Over the same period, 95% of small and medium enterprises (SMEs) using AI reported no change in workforce size and 86% said job roles had remained unchanged. 

However, while adoption is rising steeply, the effects on employment and productivity are not showing up as prominently as predicted.  

Current landscape of research 

PwC’s 2026 Hopes and Fears survey of 2,023 UK workers showcased that one in five UK employees now use AI every day. Importantly though, use is heavily concentrated by seniority with 85% of senior executives reporting using AI at work, compared with 73% of managers and 35% of non-managers.  

There are different use cases and effects experienced by survey respondents that are worth noting. Among those who use it, 70% said it had improved the quality of their work and 63% said it allowed them to use more of their skills. However, at the same time, 45% respondents said their role had become more complex and 44% reported a heavier workload. In the same survey, around one in eight respondents said their employer limits access to tools, and the same proportion said they were confused about company policy, with a further 12% saying they needed more training to use AI effectively. 

According to a survey of 2900 UK workers, conducted by Henley Business School at the University of Reading, found that optimism with regards to AI has edged up. But 61% respondents still feel overwhelmed by the pace of change. A word that workers most often chose to describe their feelings about AI was ‘cautious’.  

However, there are two distinct phenomena taking place. The first is that use is spreading unevenly, reaching senior staff faster than everyone else. It is also shallow, in the sense that AI is being applied to discrete tasks without changing how work is organised around it. The second indicates a significant share of those using AI are doing so without clear policy, adequate training, or any reduction in workload. This is expected when a tool is added to existing processes rather than used to redesign them. 

Differing explanations 

The Business and Trade Committee noted this year that around 70% of workers are in jobs where AI could reshape their tasks or replace them altogether. However, a question that was left unanswered was about why the current wave of adoption is producing so little visible change. 

The TUC’s Building a Pro-Worker AI Innovation Strategy argues that adoption is widespread but shallow because workers are excluded from decisions about how the technology is introduced, and that productivity gains will not follow until that changes. The BCC’s recommendations on adoption focusses on capability and monitoring rather than governance, including setting up an AI Labour Market Observatory, a network for AI champions for SMEs, Growth and Skills Levy funding for AI literacy, and tax relief for investment in AI and training.  

Both identify shallow adoption and absent productivity gains as they differ on the cause, and therefore on the remedy. However, there is a finding that complicates both, the one in ten SMEs adopting deeper, bespoke AI are more likely than generic users to expect headcount reductions, and 14% of SMEs investing in AI training anticipate reductions over the next 12 months. The BCC’s own analysis suggests that advanced AI adoption may drive broader restructuring, which suggests employment effects follow the depth of adoption rather than its spread. 

It is also worth noting that concern is not confined to workers, with polling by the King’s College London Policy Institute in April 2026 finding employers almost as worried as workers about AI’s impact on jobs, at 63% and 69% respectively. 

Who is responsible? 

The Department for Science, Innovation and Technology has been abolished and had its remit moved to the new Department for Business, Innovation, Science and Trade under Jonathan Reynolds. Online safety and digital duties has moved to the Department for Digital, Culture, Media and Sport. The Prime Minister has established an AI Taskforce, chaired by Lord Vallance, which sits in the Office for the Prime Minister and the Cabinet. 

While the Taskforce holds the Prime Minister’s authority it does not hold the skills funding, the regional delivery machinery or the regulatory levers that workforce transition would require, and these now sit with other departments. The AI challenge is also broader than the problem the Vaccines Taskforce, on which it is modelled, was built to solve. A cross-government Future of Work Unit has been set up to monitor the impact of AI on jobs, alongside a commitment to help 10 million people build practical AI skills, but its relationship to the Taskforce and to the departments holding the skills budgets has not been set out publicly. 

The Business and Trade Committee heard evidence through 2026 but is yet to report. The Joint Committee on Human Rights published Human Rights and the Regulation of AI on 14 September, recommending a dedicated AI Bill, a single statutory regulator, and stronger safeguards on automated decision-making. It concluded that a human in the loop should not count as meaningful oversight where that person lacks the information or independence to scrutinise the outcome, which bears directly on how AI is used to manage people at work. 

Questions to be answered 

With the above landscape in place, there are still some questions to be answered. These include:  

1. Who in government is now accountable for AI and the workforce? The Taskforce has the Prime Minister’s authority but not the budgets, and the Future of Work Unit’s relationship to it has not been explained. 
2. Is the problem that workers are shut out of decisions about deployment, or that firms lack the capability to adopt well?  
3. Rules on consultation and oversight assume a person makes the decision. What happens when the system acts on its own? 
4. Should workplace AI be dealt with through legislation, through bargaining between employers and unions, or through conditions attached to public money? 
5. What does the skills system need to do differently, and who delivers that locally? 

Join the discussion 

The All-Party Parliamentary Group for Data and Emerging Technologies are convening a roundtable on 9 November 2026, chaired by Daniel Zeichner MP, with the All-Party Parliamentary Group for Skills, Careers and Employment, to put these questions to unions, employers, researchers and government. If you work on AI adoption, skills or workforce policy and would like to contribute, please get in touch with Lavanya Rangarajan at lavanya.rangarajan@policyconnect.org.uk.