AI and Jobs at Risk in the UK: Sector-by-Sector Analysis and Retraining Pathways

Last updated: 31 March 2026

The question of which jobs are at risk from AI is one of the most important — and most misunderstood — workforce planning questions of our era. Misunderstood because the public debate oscillates between two unhelpful extremes: the catastrophists who predict mass unemployment within the decade, and the optimists who argue that every previous wave of automation created more jobs than it destroyed, so this one will too.

The truth is more nuanced and more actionable. AI automation does not eliminate occupations uniformly — it eliminates specific tasks within occupations, transforming the nature of roles rather than simply removing them. The practical implication for employers and L&D leaders is that the question to ask is not “which jobs will disappear?” but “which tasks within our roles will AI handle — and what does that mean for the skills our people need?”

This guide sets out the evidence base, provides a sector-by-sector analysis, and maps the funded retraining routes available to UK employers and workers.

The Evidence Base: What the Research Actually Says

Three major research sources inform the UK picture on AI automation risk. They agree on more than they disagree, but the nuances matter.

ONS: Automation and the Future of Work

The Office for National Statistics’ analysis of automation risk across UK occupations is the most UK-specific and methodologically rigorous source. Its key findings:

• Approximately 1.5 million UK workers are in roles where over 70% of tasks could be automated by current AI and robotics technology — these are the highest-risk roles
• A further 7 million workers are in roles where 20–70% of tasks could be automated — these are the transformation roles where hybrid human-AI working is the most likely near-term outcome
• Women are disproportionately represented in high-automation-risk occupations — primarily because of the concentration of women in administrative and customer service roles
• Younger workers (16–24) and workers without Level 3 qualifications are more exposed, as their role profiles are more task-routine than those of more experienced or more qualified colleagues

Critically, the ONS analysis distinguishes between technical feasibility (can this task be automated?) and economic deployment (will organisations actually automate it, and when?). The gap between these two is often significant — many tasks that are technically automatable will remain human-performed for years because of cost, organisational inertia, regulatory constraints, or customer preference.

WEF Future of Jobs Report 2025

The World Economic Forum’s Future of Jobs 2025 report provides the most cited global projections. Its headline figures — 85 million jobs displaced and 97 million new jobs created by 2025, later revised to project a net positive of around 12 million new roles by 2030 — are frequently quoted but often misunderstood.

The WEF identifies the fastest-growing role categories as: AI and machine learning specialists; data analysts and scientists; information security specialists; green economy specialists; care workers; and education and training professionals. The fastest-declining categories are: data entry clerks; administrative and executive secretaries; bank tellers and related clerks; cashiers; and postal service clerks.

McKinsey: The Economic Potential of Generative AI

McKinsey’s 2023 analysis — updated with UK-specific modelling — estimates that generative AI could automate the equivalent of 60–70% of employees’ time in certain knowledge work occupations, primarily through the automation of text generation, data analysis, customer interaction, and code writing. The McKinsey estimate is higher than the ONS figure because it focuses specifically on generative AI capabilities, whereas the ONS analysis covers the broader automation landscape.

McKinsey’s most important finding for UK employers is that the roles most exposed to generative AI are disproportionately in higher-wage, knowledge-work occupations — a different pattern from previous automation waves, which primarily affected manual and routine cognitive work. This means that AI displacement is not just a concern for lower-skilled workers; it affects solicitors, accountants, financial analysts, and marketing professionals whose work involves significant information processing and text generation.

High-Risk UK Roles: The Detailed Picture

Across the three research sources and the UK-specific ONS data, the following roles consistently appear in the highest-risk categories. “High risk” here means that more than 50% of the role’s tasks are technically automatable by current AI technology, and that economic deployment of those capabilities is plausible within a five-year horizon.

Administrative Assistants and Personal Assistants

Diary management, email drafting, document preparation, data entry, meeting minutes, travel booking — the majority of tasks in this role family are already being handled by AI tools in early-adopting organisations. The role is not disappearing entirely, but it is transforming from execution to oversight and judgment. The most resilient version of this role focuses on stakeholder relationship management, complex decision support, and the management of the AI tools themselves.

Timeline: Significant task displacement already underway; role transformation rather than elimination likely within 3–5 years for most organisations.

Data Entry Clerks and Records Clerks

These roles are among the most technically automatable and the least likely to benefit from the “human judgment adds value” protection that preserves many other roles. Optical character recognition, intelligent document processing, and AI-powered data extraction have reached the point where structured data entry is economically viable to automate at scale in most industries.

Timeline: High automation already deployed in banking, insurance, and logistics; widespread across all sectors within 2–3 years.

Customer Service Agents (Routine)

The distinction here is important. AI customer service tools have advanced significantly — modern conversational AI handles a high proportion of routine enquiries (balance queries, order tracking, standard complaints, policy information) with customer satisfaction scores comparable to human agents. However, complex complaints, vulnerable customer handling, and high-value retention conversations remain human territory — and growing in importance as AI handles the routine volume.

Timeline: Routine query handling substantially automated in most sectors within 2–3 years. Complex customer interaction growing as a human-specialist function.

Bank Tellers and Counter Service Roles

The combination of online banking, ATM functionality, and AI-powered query handling has been eroding this role for over a decade. The remaining bank teller population is disproportionately involved in complex customer conversations — but the total headcount continues to decline as branch networks contract.

Timeline: Continued structural decline; not primarily an AI story but AI accelerates an existing trend.

Legal Clerks and Paralegals (Routine)

Contract review, due diligence, legal research, and document drafting — the routine cognitive work that forms a large proportion of junior legal roles — is already being handled by AI legal tools in early-adopting law firms. The displacement pattern in professional services follows McKinsey’s finding: it is the junior, high-volume, relatively routine work that disappears first, while senior judgment and client relationship work grows in relative importance.

Timeline: Significant task displacement in large law firms and corporate legal teams already; wider professional services sector within 3–5 years.

Some Accounting and Bookkeeping Roles

Transactional bookkeeping, invoice processing, reconciliation, and standard reporting are highly automatable. AI accounting tools (Xero, QuickBooks, Sage with AI layers) already handle much of this in SME accounting. The accounting profession is not disappearing — but the junior bookkeeper role is transforming rapidly, and the skills premium is shifting toward advisory, tax planning, and complex financial judgment.

HGV and Delivery Drivers (Longer Timeline)

Autonomous vehicle technology is developing but deployment at scale faces significant regulatory, infrastructure, and technical barriers in the UK context. The timeline for material displacement in road transport is longer — most plausibly in the 2030–2040 window for long-haul HGV, with last-mile delivery automation possibly sooner in urban areas. This is included here not because displacement is imminent but because it is a large occupational group and planning for transition needs to start well in advance.

Medium-Risk Roles: The Transformation Picture

Medium-risk roles are those where AI will handle a significant proportion of current tasks but where the remaining work — requiring judgment, relationship management, or contextual complexity — is substantial enough that the role persists in transformed form. These roles require proactive upskilling rather than retraining.

Paralegal and Legal Assistant Roles (Complex)

Beyond the routine tasks described above, the paralegal role involves client-facing work, case management coordination, and contextual judgment that AI currently cannot replicate. Organisations that upskill their paralegal population to work effectively alongside AI — using AI for research and document review while focusing human effort on judgment and client work — will find that the transformed role is both more productive and more valuable.

Junior Finance and Analyst Roles

Financial modelling, variance analysis, and management reporting are AI-augmented rather than AI-replaced. AI tools can build models, run scenarios, and generate initial narratives — but the judgment about which scenarios matter, what the analysis means for decisions, and how to communicate findings to leadership remains human. Junior finance professionals who develop strong AI tool competency alongside analytical judgment are well-positioned; those who rely on being faster at spreadsheet work than AI tools are not.

Marketing Coordination and Content Roles

AI content generation tools have transformed the economics of content production. Copy that took a copywriter a day can be drafted in minutes. But the strategic judgment — what to say, to whom, in what context, with what tone — remains human. Marketing roles are evolving from content production toward content strategy, audience insight, and campaign orchestration. The retraining need is not replacement but augmentation.

Middle Management — Data and Reporting Focus

Managers whose primary value is aggregating information and producing reports are at risk. AI dashboards and business intelligence tools can produce the reporting that used to require a middle layer of management. Managers who add value through team development, stakeholder navigation, and complex problem-solving are more resilient — but many in this category have developed their careers around information aggregation rather than judgment.

Lower-Risk and Growing Roles

For completeness — and because workforce planning needs to include where to redirect retraining investment — it is worth identifying the role categories where demand is growing and AI risk is low.

Care Workers and Health Professionals

Demographic change is driving structural growth in care demand that no amount of AI can offset. Care roles require physical presence, empathetic relationship management, and real-time judgment in unpredictable environments — none of which AI can replicate at scale. The care sector faces a significant workforce shortage, not a surplus. The retraining opportunity here is from high-risk roles into care — a transition that is supported by specific funded routes.

Teachers, Trainers, and L&D Professionals

Demand for education and training professionals is growing, not declining. AI is augmenting the tools available to educators and trainers, but the relationship between teacher and learner — particularly in the complex, affective dimensions of learning — remains irreducibly human. The fastest-growing sub-category within this group is AI literacy trainer — professionals who can help organisations build the capabilities to work effectively with AI tools.

Trades: Electricians, Plumbers, HVAC Engineers

Physical trades operating in unstructured, variable environments face very low AI automation risk. The green economy transition is actively growing demand for electricians (EV charging infrastructure, solar installation, heat pump installation) and other trades. The combination of low automation risk and growing demand makes this a high-priority retraining destination for workers displaced from routine cognitive roles.

AI and Data Specialists

The most obvious growth category — but one where the retraining pipeline is significantly behind current demand. AI engineer, machine learning engineer, data scientist, and AI product manager roles are growing rapidly and command significant salary premiums. The challenge is that the retraining pathway into these roles from routine cognitive positions is long — typically 18–24 months of serious study — and not suited to every worker. Investment in these retraining pathways is worthwhile for the right populations, but should be targeted rather than universal.

Sector-by-Sector Analysis

The following sector analysis draws on ONS occupation data, Skills England’s inaugural Skills Assessment (2025), and sector-specific research to provide an overview of AI impact by industry.

Financial Services

High admin and processing risk. UK financial services employs around 1.1 million people. The sector has the highest concentration of routine cognitive work of any major industry — transaction processing, claims handling, compliance checking, document review, and customer query management. AI deployment is already significantly advanced in large banks and insurers, primarily in back-office automation and AI-assisted customer service.

The retraining priority in financial services is twofold: upskilling existing staff to work alongside AI tools (AI literacy and AI-augmented workflow training), and retraining those displaced from routine processing roles into higher-judgment functions such as relationship banking, complex claims handling, and financial advice.

Retail

Cashier and inventory risk; buying complexity retained. UK retail employs around 3 million people, with cashier and checkout roles forming a significant proportion. Self-service technology has been displacing checkout roles for 15 years; AI-powered checkout (computer vision systems that automatically process purchases) is accelerating this in large stores. Online retail’s growth further reduces the need for in-store customer-facing staff.

The roles growing in retail are those that AI cannot replicate: complex customer consultation (specialist retail, luxury), personal shopping, and the “experiential retail” functions that differentiate physical stores. Retail workers displaced from checkout and routine stock roles need retraining into adjacent service roles or out of the sector entirely.

Manufacturing

Operational risk moderate; digital skills demand growing. UK manufacturing is already highly automated in production, and AI adds relatively little incremental displacement risk to production-line roles in the near term — the major automation shift in manufacturing happened in the 1990s–2010s through robotics. The AI risk in manufacturing is more concentrated in planning, quality assurance, and logistics coordination roles.

The growing skills need in manufacturing is digital — operating and maintaining AI-powered production systems, interpreting data from connected machinery, and contributing to continuous improvement in automated environments. Advanced Manufacturing apprenticeship standards and digital skills programmes are the primary training response.

Public Sector

Admin risk high; care growing. The UK public sector employs around 5.5 million people. Administrative roles across central government, local government, HMRC, and DWP are highly exposed to AI automation — the AI Opportunities Action Plan explicitly targets public sector AI deployment as a productivity driver. At the same time, the NHS and social care system face growing demand for clinical and care staff that no amount of administrative efficiency can replace.

The public sector retraining challenge is particularly acute because it involves large numbers of workers in relatively stable employment who may not proactively seek retraining. Employer-led programmes, supported by the Growth and Skills Levy and Skills Bootcamp provision, are the most effective vehicle for this population.

Professional Services (Legal, Accounting, Consulting)

Junior roles at high risk; senior judgment growing. As noted in the role analysis above, professional services faces a distinctive pattern: AI displaces the junior, high-volume work while simultaneously increasing the productivity and thus the earning power of senior professionals. This compresses the traditional career pipeline — fewer junior roles are needed to support the same number of senior professionals — which has implications for graduate entry-level hiring and for how professional skills are developed.

Organisations in professional services should be planning for smaller junior cohorts with higher AI capability, and should be investing in AI literacy as a core professional development requirement rather than an optional add-on.

The Retraining Imperative: Why Early Movers Win

The evidence consistently shows that organisations that invest early in workforce transition — rather than waiting until roles become redundant — gain significant competitive advantage. The advantages of early action include:

• Talent retention: Workers who receive retraining investment are more loyal and more likely to stay. The cost of retraining an existing employee is typically 20–30% of the cost of recruiting a replacement with the target skills.
• Productivity during transition: Workers who understand AI tools before they are mandated to use them adopt more quickly, make fewer errors, and reach productive use faster than those who encounter tools as “imposed change.”
• Employer brand: Organisations visible as investing in workforce transition attract better quality candidates in a tight labour market for AI-adjacent skills.
• Cost of funded provision: The funded routes available for AI retraining — Skills Bootcamps, levy-funded apprenticeships, AI Skills Boost — are available now. There is no guarantee they will be funded at the same level in 2027 and beyond as the government’s fiscal position evolves.

Funded Retraining Routes: The UK Landscape in 2026

One of the most important practical points in this guide is that significant funded provision exists right now for AI-related retraining. Many employers are unaware of the full range of options, or believe the funded routes are only available for new hires rather than existing staff. Neither is true.

Growth & Skills Levy: AI and Digital Apprenticeship Standards

The Growth and Skills Levy (replacing the Apprenticeship Levy from August 2025) funds apprenticeship standards and — under the reformed model — approved shorter qualifications. For AI and digital retraining, the key standards are:

• Data Analyst Level 4 — 15 months, covers data analysis, visualisation, and insight; accessible to workers with a range of prior experience
• AI Data Specialist Level 4 — 15 months, focused on AI model development and data science applications; more technical
• Digital Marketer Level 3 — 15 months, covers digital marketing including AI-powered marketing tools; good retraining route from traditional marketing and admin roles
• Digital and Technology Solutions Professional Level 6 — 30 months, deeper technical qualification for more significant career transitions
• Business Administrator Level 3 — 15 months, increasingly incorporating AI tool use in its curriculum; relevant for workers in transforming admin roles

Levy-paying employers can fund these programmes from their levy balance with no additional cost. Non-levy employers co-invest 5% with the government covering 95%.

Skills Bootcamps: AI, Data, and Digital

Skills Bootcamps are 12–16 week intensive programmes with a guaranteed job offer, interview, or progression opportunity on completion. They are designed precisely for the lateral retraining market — workers who need to develop new skills quickly and want a structured, time-bound pathway.

Available AI and digital Bootcamp themes include: data analysis and data science; AI fundamentals and AI tool use; digital marketing and content creation; software development; cybersecurity; and cloud computing. Employer co-investment is 10% for SMEs (fewer than 250 employees) and 30% for large employers. Individuals not in employment can access Bootcamps with no co-investment requirement.

AI Skills Boost Programme

The government-backed AI Skills Boost Programme provides access to AI literacy training at three levels:

• Awareness level — free, self-paced; provided through tech company partnerships (Microsoft, Google, AWS); suitable for broad workforce AI literacy
• Applied level — subsidised; structured programmes with assessment; suitable for workers needing to use AI tools in their roles
• Advanced level — partially funded; qualification-bearing programmes for workers moving into AI-adjacent roles

Free Essential Digital Skills / Digital Entitlement

Adults without a Level 1 qualification in essential digital skills are entitled to free provision funded by DfE. This is the baseline provision for workers in high-risk roles who currently have limited digital capability and need a foundation before progressing to AI-specific training.

Sector-Based Work Academy Programmes (SWAPs)

For workers who have already been displaced, DWP’s Sector-Based Work Academy Programmes combine pre-employment training (typically 6 weeks), work experience, and a guaranteed job interview with a participating employer. SWAPs are available in growth sectors including healthcare, green economy, digital, and logistics. They are particularly relevant for workers from retail, administration, and financial services who are being displaced and need a structured pathway into a new sector.

Building a Retraining Programme: A Framework for L&D Leaders

The following framework gives L&D leaders a structured approach to building a retraining programme that responds to AI displacement risk in their organisation.

Step 1: Assess Role Exposure

Map your workforce by role family and assess each role family’s exposure to AI automation using the ONS risk categories as a starting framework. Supplement with your own analysis of which tasks within each role are currently being or could be automated by the AI tools your organisation uses or is considering deploying.

Step 2: Segment by Urgency

Not all exposed roles have the same urgency. Segment by: how soon displacement is likely (technical feasibility plus your organisation’s AI deployment timeline); the scale of the population affected; and the availability of retraining pathways. Prioritise high-urgency, large-population segments.

Step 3: Map Retraining Options

For each priority segment, identify the retraining options — both what you want workers to be able to do and which funded routes could support their development. Be realistic about retraining timelines: a data entry clerk retraining as a data analyst in 15 months via an apprenticeship standard is achievable; the same worker becoming a machine learning engineer in 12 months is not.

Step 4: Design the Programme

Combine funded provision (Skills Bootcamps, levy standards, AI Skills Boost) with internally designed contextual training (how to use AI in your specific tools and workflows). Build in a support structure — managers need to support learners through the transition, not just send them on a course.

Step 5: Measure What Matters

Measure not just training completion but role-level outcomes: can workers in the retraining cohort perform the new tasks their role requires? Have the roles they are transitioning into been filled, and at what quality level? What is the retention rate of workers who have completed retraining programmes?

What L&D Leaders Should Do in 2026: Checklist

• Mapped all role families against ONS automation risk categories
• Identified which roles in your organisation are in the high-risk, medium-risk, and transformation categories
• Assessed your AI deployment timeline — when will AI tools be introduced into which roles?
• Calculated the retraining population: how many workers are in high-urgency, high-exposure roles?
• Reviewed Growth and Skills Levy balance and identified which AI/digital apprenticeship standards match your retraining needs
• Identified Skills Bootcamp providers in your region offering AI, data, and digital programmes
• Enrolled qualifying workers in the AI Skills Boost Programme for foundational AI literacy
• Audited digital skills baseline across the workforce — identified workers who need Digital Entitlement provision before progressing to AI-specific training
• Communicated your retraining programme to affected workers — transparency reduces resistance and increases uptake
• Built manager capability to support learners through the retraining transition
• Established a measurement framework for retraining outcomes, not just completion rates
• Set a 12-month target for the proportion of high-risk role workers who have started a retraining pathway

Frequently Asked Questions

Which UK jobs are most at risk from AI?

ONS analysis identifies roles with the highest automation risk as those involving routine cognitive tasks — data entry clerks, administrative assistants, customer service agents handling routine queries, bank tellers, filing and records clerks, and some paralegal and bookkeeping roles. The pattern is consistent: roles involving the processing of structured data, following rule-based processes, and producing standardised written outputs are most exposed. Roles requiring physical dexterity in unpredictable environments (care workers, trades), professional judgment with high stakes (clinicians, senior lawyers), and interpersonal relationship management (teachers, counsellors) are substantially less exposed and in many cases are growing in demand.

What funded training is available for AI-displaced workers in the UK?

Several funded routes are available. The Growth and Skills Levy funds AI and digital apprenticeship standards including Data Analyst Level 4 and AI Data Specialist Level 4. Skills Bootcamps for AI, data, and digital skills require 10–30% employer co-investment with the remainder publicly funded. The AI Skills Boost Programme provides government-backed AI literacy provision including free and subsidised courses. The Digital Entitlement provides free essential digital skills qualifications for adults without a Level 1 digital qualification. For workers without employment, DWP’s Sector-Based Work Academy Programmes offer pre-employment training combined with work experience placements in growing sectors.

How long does workforce retraining for AI take?

Retraining timelines vary by the depth of transition required. AI literacy upskilling — helping workers in AI-exposed roles use AI tools alongside existing skills — typically takes 4–12 weeks of structured learning followed by 3–6 months of supported adoption. Lateral retraining into adjacent roles (for example, an administrative assistant becoming a data analyst) typically takes 6–12 months via a Skills Bootcamp or modular qualification pathway. Full career transition into a substantially different occupation typically requires 12–24 months. The funded routes available in the UK are broadly matched to these timescales: AI Skills Boost for quick upskilling, Skills Bootcamps for lateral retraining, and apprenticeship standards for deeper career transition.