AI, Business and the Future of the Workforce

Introduction

1. PauseAI UK is part of a global movement dedicated to mitigating the risks posed by frontier artificial intelligence. We advocate for an international pause on the development of advanced AI systems until they are proven safe and subject to democratic oversight. PauseAI UK emphasises that uncontrolled frontier AI development poses unprecedented risk to the labour market and to economic equality amongst other more severe risks. A proactive framework is essential to ensure that technological advancement does not outpace our ability to protect working people during a time of unparallelled economic disruption.
2. PauseAI UK urges the Committee to recognise that the current trajectory of AI development, with the automation of human cognitive labour, is a paradigm shift that does not compare to previous historical developments like the industrial revolution¹. It is a mistake to think that our civilisation has been through similar transitions before. The stated goal of the major AI developers is to build systems that outperform humans at most/all economically valuable tasks². This is fundamentally incompatible with a labour-based economy. Without an immediate pause on development, and a total rethink of the socio-political value of employment³, the UK risks extreme political and economic destabilisation.

1. The future of AI

How is AI likely to evolve over the next decade? What are the scenarios we should plan for?

3. Given the speed of development, it is very difficult to accurately predict the evolution of Artificial Intelligence (“AI”) over the next decade. As recently as 2022, it would have been nearly impossible to predict ‘the ChatGPT moment’ and to foresee all the ways in which AI has now rapidly spread throughout society and become deeply integrated in the economy.
4. However, we can extrapolate from the trends and draw conclusions based on the incentives and stated goals of the companies developing frontier AI models. In our view, AI development will continue to drive towards Artificial General Intelligence (“AGI”) which aims to perform as well as or better than humans at nearly all economically valuable tasks. This is the explicit objective of most major AI developers (including OpenAI, Google DeepMind, and Anthropic).
5. It is critically important to recognise what this means: if the technology advances in the way that the developers aim for and which they believe is likely to happen⁴ — then very soon there will be little to no cognitive work left for which humans have a competitive advantage. Simply put, if a human is currently paid to do a job which mostly involves thinking and working on a computer, then AI will soon be able to do that job better, faster, and cheaper.
6. We would urge the Committee not to fall into the trap of doubting whether AI can actually reach general human-level competence. Many experts believe it is both possible and imminent. For example, in February 2026, British entrepreneur and CEO of Microsoft AI, Mustafa Suleyman, stated his belief that we will see “human-level performance on most, if not all, professional tasks … [and] most of those tasks will be fully automated by an AI within the next 12 to 18 months”⁵.
7. There is already strong evidence for the erosion of white-collar work. In February 2026, Twitter co-founder Jack Dorsey announced that his multi-billion dollar payments company, Block, is laying off nearly half its workforce because AI “fundamentally changes what it means to build and run a company.”⁶ Meta is reportedly planning to cut around 20% of its workforce in 2026 due to AI⁷.
8. The advent of agentic AI (particularly tools like Anthropic’s Claude Code and OpenAI’s Codex) have resulted in software engineers reporting that they no longer write any computer code. Boris Cherney, inventor of Claude Code, says that the technology is “going to be very disruptive, and it’s going to be painful for a lot of people”⁸.
9. This trend will not stay localised to computer engineering⁹. In boardrooms across the country, the question asked when making a decision on a new hire is now “can an AI do this job instead?” Entry-level hiring freezes are a clear early warning sign. As AI capabilities continue to rapidly increase, we expect to see more and more skilled work performed exclusively by agentic AI.
10. Given the above, the Committee mustexpand its scope beyond scenarios where AI merely augments the human worker (e.g. by enhancing efficiency, speed, and value when performing existing job roles) and plan for circumstances where autonomous AI agents are capable of entirely replacing human roles — performing equivalent or better work, in a fraction of the time, and at a fraction of the cost. This will be a fundamentally different labour market.
11. PauseAI UK strongly urges the Committee to investigate the implication of increasing AI capability leading to mass unemployment across large sectors of the economy. The Committee must plan for scenarios where millions of professionals no longer have an income and cannot provide for themselves or their dependants. The Committee must also plan for the second-order consequences of job displacement, including (for example) the power concentration which will occur as a small handful of AI companies control large parts of the economy, and the political disempowerment which comes with increasing wealth inequality and power imbalance between the public and corporations.
12. PauseAI UK recommends that the Committee strongly consider that any intervention short of a global pause in the development of advanced AI systems will be insufficient to prevent a massive destabilisation of the economy and social order.

2. AI adoption

Where is AI currently being deployed across the UK economy (by sector, region, firm size, and public/private) and how is it being used? How does adoption vary across different parts of the economy and workforce and why; what are the main barriers to AI adoption?

13. PauseAI UK does not have a view to share regarding sectoral AI deployment or practical barriers to adoption.

3. Infrastructure for the AI Transition

What public and private infrastructure is required to support safe, responsible and effective AI adoption by 2035? How much of this infrastructure currently exists? Which elements should be publicly funded, and which through private investment? Where are the remaining market gaps?

14. We note that the Inquiry outline states “the Government’s AI Opportunities Action Plan includes a twentyfold expansion of public AI hardware by 2030 and seeks to leverage private investment through initiatives such as the US–UK Tech Prosperity Deal (with £30 billion committed by major technology firms)”.
15. This is an extraordinarily consequential investment in hardware resources which will be used to accelerate the development of AI technologies, and therefore will accelerate the risks and challenges which are set out in this response. PauseAI UK strongly recommends that the Committee increase its focus on the objective of ensuring that infrastructure development is both “safe” and “responsible”, as the question indicates.
16. Notwithstanding that PauseAI UK advocates for an immediate pause in development of advanced frontier AI generally, with respect to the underlying technical infrastructure specifically, PauseAI UK recommends that the Committee consider the following:

4. Impacts on work & workers

What impact can AI have on productivity? Which tasks and occupations are most exposed to automation? What are the effects of AI on employment, job security, employment rights, pay, job quality, health and safety, and wellbeing? Which sectors, regions and groups are most likely to be affected, and what are the implications for inequality?

20. PauseAI UK urges the Committee to look beyond productivity and efficiency forecasts and confront the reality of job displacement. While AI adoption is expected to enhance output, in a competitive market these gains accrue almost exclusively to capital owners, while human labour faces a total loss of competitive advantage. The drive for “AI efficiency” will soon be a threat to the professional classes, as the marginal cost of cognitive labour approaches zero.
21. While all industries are exposed to automation, computer-heavy, white-collar, cognitive roles are the first to be impacted. This represents a large majority of economic work in the UK. Professional copywriters, graphic artists, and translators have already been substantially displaced by models that can generate equivalent output at near-zero marginal cost. As mentioned above, the deployment of agentic tools has fundamentally altered the software engineering sector, with reports of engineers no longer writing original code but merely “orchestrating” automated outputs. This has turned a high-value career into a precarious role. Computer-heavy roles including middle management, data entry, and legal research, are all roles which primarily comprise text-based cognitive work — exactly what current AI is adapted for.
22. We are particularly concerned by the erosion of entry-level roles. As businesses use AI to perform the work previously assigned to junior staff, the training ground for the next generation of UK professionals is vanishing. This creates a widening skills gap: without junior roles, there is no pathway to senior expertise, leading to a long-term attrition of human talent.
23. The consequences for worker wellbeing and societal stability are potentially catastrophic. Beyond the risk of mass unemployment and job insecurity, the relentless drive for efficiency is already causing profound anxiety and stress as human workers are forced to compete with machines that do not sleep¹⁰. Workers lose agency, bargaining power, and purpose as AI increasingly encroaches on their professional and economic territory. Without a strategic global pause to rethink our national economic structure, we risk a future where millions are economically irrelevant, leading to unprecedented political disempowerment and the upending of our social contract.

5. Skills, Education & Transitions

What skills will workers and managers need to work effectively with AI? Are current education and adult learning systems adequate? What reskilling and upskilling models are most effective? How can Government and employers support augmentation and upskilling when introducing AI?

24. PauseAI UK strongly believes that the common focus on “lifelong learning” and adult retraining is a fundamentally inadequate response to advancing AI technology. This approach mistakenly treats AI as a simple tool, like a laptop or a calculator, rather than a technology that learns and improves on its own. The core problem is the massive difference in the speed of human and machine learning. While it takes years for a person to gain expertise through education, AI capabilities are advancing so quickly that new jobs will be created and become fully automated in less time than it takes a student to complete a degree.
25. This “reskilling” narrative also misses the fact that the goal of frontier AI development is to master almost all economically valuable work. Some suggest that when confronted with redundancy, workers should move “up the ladder” into higher-value roles that involve managing or “orchestrating” AI systems. However, this fails to account for the moving goalposts of automation. If a worker begins retraining today to become an “AI orchestrator,” the rapid advance in AI autonomy means that by the time they are qualified, the AI will likely be able to manage itself.
26. If the Government nevertheless intends to pursue a strategy of retraining in the near term, education must shift its focus away from technical skills that AI can easily copy and toward human-centric resilience. We should prioritise roles that require physical presence and direct human accountability. However, skilled physical labour provides only a temporary buffer while robots are still primitive and expensive to build. This human advantage will disappear as hardware becomes cheaper and more capable¹¹. There are very few job categories that are likely to remain “human-only” in an age of AGI. Accordingly, we must be clear: retraining is not a solution to the impending crisis, and it is dangerous to act upon this misapprehension.
27. Ultimately, the challenge facing the UK is not about workers developing new skills, but is instead about a looming crisis of human competitiveness. Our education system cannot solve this problem because the problem is not about the quality of training, it’s about the speed of the technology. Government strategy must therefore move beyond retraining and start planning to create a society where a person’s wellbeing is no longer tied to their market value.

6. Government Strategy, Regulation & Rights

How prepared is the Government and are regulators for the potential impacts of AI? Is the existing UK regulatory framework sufficient to address current and emerging employment-related AI issues? What support should the Government be offering to businesses, employers and employees?

28. The UK Government and its regulators are currently unprepared for the systemic impacts of advanced AI. Our existing regulatory framework is a patchwork of rules that were not designed for a technological paradigm shift of this magnitude. While we have the Information Commissioner’s Office (ICO) regulating data privacy in the context of AI, the Competition and Markets Authority (CMA) for market fairness, and sector-specific oversight bodies like the Financial Conduct Authority (FCA), there is no single regulator with the specific mandate or technical expertise to oversee the development and deployment of frontier AI models. This piecemeal approach is insufficient for a technology that moves across all sectors at once.
29. We urgently need dedicated AI regulation capable of enforcing safety standards before potentially dangerous models are developed or released to the public. Current policy levers are focused on accelerating AI adoption without first ensuring the ultimate stability of the UK economy. If the current efforts racing toward wide deployment of AI across the economy results in fracturing our labour market and social fabric, the benefits of increased productivity may be irrelevant to the welfare of most people and Britain’s prosperity and security, especially since most owners of AI companies and infrastructure are outside of the UK.
30. The Government should recognise that there is a strong public mandate for government intervention of AI: polling consistently shows that voters want strict regulation and oversight of AI¹². Rather than solely funding adoption, the Government should prioritise policy levers that ensure safety-first development. This includes the regulation of the hardware — the data centers and compute power that makes these technologies possible (see response to Question 3 above). Without controlling the physical infrastructure of AI, the UK cannot hope to manage its outcomes.
31. The Committee must also confront the reality of what success looks like for major frontier labs. Their stated goal is to build artificial general intelligence (AGI) that can outperform humans on all economically valuable tasks. If they succeed, there is no remaining role for human workers.
32. We urge the Government to take a leadership role in guarding against AI risks. The UK must lead in advocating for a global treaty that mandates a strategic pause on AI development. Until we have a proven way to align these systems with human interests and a clear plan for a post-labour economy, the risks of continuing this arms race far outweigh the rewards.
33. Finally, the Government must prepare for the political and economic turbulence that rapid automation will bring. Because reskilling is not a long-term solution when AI learns faster than humans, we must rethink the role of human labour in the market. The Government must act to ensure that workers displaced by AI are not left without a means to support themselves and their families.
34. We note that this question invites comment generally on preparedness of the UK Government for the impacts of AI. At this time, no Government is prepared.
35. Many of the world’s foremost experts are urgently sounding the alarm about the significant existential risks posed by AI. Day by day we are seeing increasing harms from AI; from political destabilisation caused by deepfakes and AI-fuelled misinformation¹³ to dangerous and sophisticated cyberattacks on our critical infrastructure¹⁴. These are just early warning signs. Powerful AI will put incredibly dangerous technologies in the hands of malicious actors, enabling them to cause devastation with minimal costs. These dangerous technologies include engineered pandemics and bioweapons which have the potential to kill billions¹⁵. These technologies already exist — they are now being made dangerously accessible.
36. Beyond the risks mentioned above, many experts expect that superhuman AI (that is, AI which is much more intelligent than any human) will be uncontrollable¹⁶and may lead to the extinction of the human species. This is not a science-fiction story or a fringe point of view — it is one shared by the world’s leading AI experts¹⁷ and leaders of technology companies¹⁸ alike. When the experts closest to this unprecedented technology tell us there is a significant probability that AI will cause human extinction, the UK Government must take that warning extremely seriously.