In 2022, most economists expected AI automation to primarily affect routine manual work — the kind of jobs that follow predictable physical steps. That prediction turned out to be wrong. The large language models that emerged from 2023 onwards proved far more capable at cognitive, text-based tasks than at fine motor skills. The UK labour market has been adjusting to that surprise ever since.
The Tasks Most at Risk
Automation economists distinguish between tasks and jobs. Very few jobs consist of a single task that a machine can replicate entirely. Most roles involve a bundle of activities — some of which AI can handle, some of which it cannot.
The tasks most vulnerable to current AI capabilities share a common profile: they are text-based, follow recognisable patterns, and do not require physical presence or real-time relational judgement. That profile describes large portions of:
- Legal work: Contract review, due diligence, first drafts of standard agreements
- Financial services: Transaction categorisation, report generation, basic client communications
- Content and media: Routine article writing, SEO content, simple translations, image editing
- Administrative functions: Data entry, scheduling, email triage, standard document processing
- Call centres: Tier-one customer service queries that follow scripts
This does not mean these sectors are haemorrhaging jobs en masse. Firms are primarily using AI to handle higher volumes with existing headcounts rather than making immediate redundancies. The disruption is manifesting as a slow squeeze on hiring rather than dramatic layoffs.
Where New Roles Are Emerging
The same AI wave creating pressure in some areas is generating demand in others. The shortfall, at least for now, is in workers who can work alongside AI systems rather than be displaced by them.
AI trainers and evaluators — people who assess AI outputs for accuracy, bias and safety — are in significant demand. The role requires domain expertise (a lawyer reviewing legal AI outputs, a doctor assessing clinical AI) rather than technical skills.
Prompt engineers and AI integration specialists are being embedded across industries. This is a function that barely existed in 2022 and is now appearing on job boards from FTSE 100 companies to local government.
AI safety and ethics roles are growing, particularly in regulated industries and in government. The passage of the EU AI Act and ongoing UK AI regulatory discussions have created demand for professionals who understand both the technology and its governance implications.
Human-AI collaboration roles — where the competitive advantage is the ability to direct, refine and verify AI outputs rather than produce them from scratch — are appearing across professional services, marketing and education.
The Government's Position
The UK government's AI Opportunities Action Plan, published in January 2025, committed to making the UK a global AI hub. The target of 200,000 new AI-related jobs by 2030 is ambitious but not unreasonable given current hiring trends.
The government has also established AI Growth Zones — designated areas where infrastructure and planning restrictions are relaxed for AI data centre development. These create construction and infrastructure jobs alongside technology roles.
Training is a key component. The government has partnered with several providers to offer reskilling programmes for workers in at-risk occupations. Take-up has been modest so far, and the quality of provision varies significantly.
What Workers Are Actually Doing
Survey data from the CIPD and others suggests a more complex picture than headlines imply. Fewer than one in ten UK workers reported any change to their role specifically attributable to AI in 2025. Among those who had experienced change, the most common description was "AI helps me do my job faster" rather than "AI is replacing my tasks."
This is consistent with an adoption curve that is still in its early stages. The productivity gains from AI in professional services are significant but unevenly distributed — concentrated in firms that have invested in integration, training and change management. Many organisations have AI tools available to staff but lack the internal capability to use them effectively.
The Productivity Dividend
The core economic argument for optimism is that productivity gains from AI should, over time, generate growth that creates new employment. This happened with previous general-purpose technologies — electricity, computers, the internet. In each case, aggregate employment eventually rose even as specific occupations declined.
The honest caveat is that the transition is neither smooth nor geographically even. The productivity dividend accrues first to firms and sectors that can adopt quickly. Workers in sectors with slow adoption cycles face a prolonged period of uncertainty. Geographies where at-risk occupations are concentrated will feel displacement before they feel new job creation.
What This Means Practically
If you're a knowledge worker in the UK in 2026, the evidence suggests a few practical conclusions:
- Identify your AI-replaceable tasks and start experimenting with the tools before your employer does it without you. Workers who understand how AI can augment their role are harder to displace than those who don't.
- Invest in the skills AI cannot replicate: nuanced client relationships, ethical judgement, creative problem-solving, leadership and communication.
- Do not ignore retraining resources. Government-funded provision is patchy, but employer-funded programmes and sector-specific qualifications are becoming more widely available.
- Watch the hiring data, not the hype. Aggregate UK employment has remained robust. Sector-level disruption is real but concentrated. Staying informed about your specific field is more useful than generalised anxiety about automation.
The AI transition is real and it will reshape the UK labour market — but it is not a cliff edge. It is a restructuring that rewards adaptability, investment in skills, and the willingness to work alongside tools rather than against them.