Artificial Intelligence Jobs and AI Automation in the UK (2026): Which AI Roles Are Growing

There is a particular irony to the moment we are in. The same generative and agentic AI tools that are automating call centres, drafting legal memos and writing code are now starting to automate parts of artificial intelligence work itself. If you build AI for a living, the question is no longer abstract: which artificial intelligence jobs are growing in the UK, and which are quietly being absorbed by the very systems they help create?

This article looks specifically at how automation is reshaping AI careers, using current 2025 and 2026 UK data, named employers and salary benchmarks. It is not a general "robots are coming" piece. It is about the people who build the robots.

The Short Answer

Artificial intelligence jobs in the UK are growing overall, but the work inside them is being reshaped fast. According to ITJobsWatch data, the median AI engineer salary sat around £85,000 in the six months to May 2025, with the top decile above £101,000. AI-related demand has risen sharply since 2022, and UK job adverts mentioning AI reached roughly 5.6% in 2025, ahead of the US, Germany and Australia. At the same time, the IPPR has warned that up to around 8 million UK jobs could be exposed to automation in a worst-case scenario, with roughly 11% of tasks currently at risk. Within AI itself, routine work such as basic data labelling and simple prompt-tuning is being automated, while roles in MLOps, AI safety, agent orchestration and AI governance are expanding. Net effect, per most UK sources: more AI jobs, but a steep shift in required skills.

Will AI replace artificial intelligence jobs?

The honest answer, based on current UK evidence, is "not wholesale, but it is changing them deeply". The headline anxiety figure comes from the IPPR, whose 2024 analysis warned that up to roughly 8 million UK jobs could be at risk in a worst-case "second wave" of automation, with women, younger workers and lower-paid workers typically most exposed. Crucially, that same analysis put the central scenario far lower, at around 545,000 jobs displaced alongside a potential GDP uplift of about 3.1%. The gap between those two numbers is essentially a policy and adoption choice, not a foregone conclusion.

For AI roles specifically, the picture is more favourable than for the wider economy. The PwC AI Jobs Barometer has repeatedly found that occupations more exposed to AI have tended to see stronger demand and wage growth, not collapse, at least so far. AI practitioners typically sit on the building side of that equation rather than the displaced side. What automation does instead is hollow out the repetitive layers of AI work, the manual data cleaning, the boilerplate model wiring, the first-draft documentation, leaving humans to do judgement, integration and oversight. In our reading of the UK data, that means fewer junior "task" roles and more demand for people who can supervise AI systems end to end.

Which AI tasks is automation taking over?

Generative and agentic tools are now competent at a surprising amount of the day-to-day grind inside AI teams. The clearest examples are in data preparation and routine coding. Modern annotation platforms increasingly use computer-vision and language models to pre-label data, with human annotators reviewing and correcting rather than labelling from scratch, which compresses the volume of pure entry-level labelling work even as demand for expert "hard case" reviewers rises.

Here is how the split typically looks in 2026.

The pattern is consistent: AI is strongest at well-defined, repeatable sub-tasks, and weakest at ambiguous, accountable, cross-system judgement. That is why, according to multiple 2025 UK skills analyses, the skills attracting a premium are LLM deployment, evaluation frameworks, vector databases, RLHF and AI-safety methods, all of which sit above the automatable layer.

Which AI roles are growing in the UK?

Demand has clearly tilted towards deployment and oversight rather than pure experimentation. As organisations move from AI pilots to production, MLOps has become a genuine bottleneck, blending DevOps discipline with machine-learning knowledge, and it is one of the fastest-growing AI specialisms in UK hiring data going into 2026.

Roles that are expanding, in our reading of current UK demand, include:

• MLOps and ML platform engineers, who keep models running, monitored and retrained in production.

• AI engineers focused on LLM and agent integration, wiring foundation models into real products rather than training from scratch.

• AI safety, evaluation and red-teaming specialists, a category boosted directly by UK policy and by frontier labs.

• AI governance and assurance professionals, who manage risk, documentation and compliance.

• AI product managers, who decide what to build and where humans stay in the loop.

The salary signal supports this. According to ITJobsWatch data, the median AI engineer salary was around £85,000 in the period to May 2025, with entry-level benchmarks having risen from roughly £42,500 in 2023, and the top 10% earning above £101,000. Specialist 2026 UK salary guides typically place senior MLOps and applied-science roles comfortably into six figures, particularly in London. These are not the salary curves of a shrinking field.

Which AI roles are shrinking or being squeezed?

Not every AI job title is on the up. The most cited example is the standalone "prompt engineer". The role surged in 2023 and 2024, and LinkedIn data still showed a large year-on-year jump in prompt-related postings in 2025, but the consensus among 2025 and 2026 analyses is that pure prompt-writing is being commoditised. As models get better at interpreting messy instructions and at clarifying intent themselves, the value of crafting the single "perfect" prompt falls. Prompt skill is becoming a baseline expectation inside other roles rather than a job in its own right.

Below the senior tier, the squeeze is on routine and entry-level AI-adjacent work.

The risk here is structural rather than total. The concern flagged by the IPPR and others is that fewer junior rungs make it harder for newcomers to enter AI careers at all, which is a different problem from senior practitioners being replaced. Hedged honestly: these roles are not vanishing overnight, but their headcount per project is typically falling.

Where are the AI jobs, and who is hiring?

The UK's AI hiring map is concentrated but not solely in London. Google DeepMind continues to recruit across UK locations, with vacancies listed via Adzuna and LinkedIn spanning research, infrastructure and strategy. Amazon, through AWS and its applied-science teams, has been hiring across London, Cambridge, Manchester and Edinburgh, including applied-scientist roles tied to foundation-model work and supported by very large UK investment commitments. Microsoft has recruited for Copilot and Azure ML roles across Cambridge and London.

Beyond the hyperscalers, the UK has a distinctive layer of specialist employers and public institutions. Faculty, based in London, remains one of the country's best-known machine-learning consultancies, working across healthcare, government, retail and logistics. On the public side, the Alan Turing Institute partners with the Department for Science, Innovation and Technology (DSIT) on initiatives such as the Open Source AI Fellowship, which began in January 2026, runs for 12 months and reimburses salary costs of up to £85,000. Cambridge and Manchester, alongside Edinburgh, function as genuine secondary hubs, which matters for anyone unwilling or unable to relocate to the capital.

That spread of named employers, DeepMind, Amazon, Microsoft, Faculty and the Turing Institute among them, is one reason the UK AI labour market has stayed resilient even as automation reshapes individual tasks.

What is the UK government and regulator doing about AI and jobs?

The institutional backdrop matters because it is itself creating AI jobs. The body formerly known as the AI Safety Institute was renamed the AI Security Institute in February 2025 and operates as a directorate of DSIT. It runs frontier-model evaluation and a Challenge Fund covering safeguards, control, alignment and societal resilience, work that directly employs evaluators, red-teamers and research engineers.

Policy is pulling in the same direction. The government's AI Opportunities Action Plan, and DSIT's wider strategy, lean towards adoption and capability-building rather than restriction, which tends to expand demand for AI safety, assurance and governance skills. The IPPR, for its part, has argued for a "job-centric" industrial strategy so that productivity gains from automation are shared rather than concentrated. For an individual practitioner, the practical takeaway is that AI safety, evaluation and governance are now policy-backed growth areas, not niche concerns, and they are among the harder AI roles to automate because accountability cannot be delegated to a model.

How should you future-proof an AI career in 2026?

The defensible position, in our reading of the UK data, is to move up the value chain from tasks to systems and from execution to oversight. A few practical principles tend to hold.

First, specialise in deployment and reliability, not just modelling. MLOps, evaluation and platform skills are where production demand is concentrated. Second, learn to work with agents, orchestrating, supervising and correcting AI systems, rather than competing with them on raw output. Third, develop judgement-heavy capabilities, AI safety, governance, product sense and domain expertise, that depend on accountability and context. Fourth, treat prompting and basic coding assistance as table stakes rather than a career. The pattern across 2025 and 2026 UK sources is unambiguous: the parts of AI work that are easy to automate are being automated, and the parts that require human judgement are commanding the premiums.

Frequently Asked Questions: AI Jobs and Automation

Are artificial intelligence jobs still in demand in the UK in 2026?

Yes, typically. UK job adverts mentioning AI reached around 5.6% in 2025, ahead of several comparable economies, and demand for AI engineers has risen markedly since 2022. The composition is shifting towards deployment, safety and governance roles, but according to ITJobsWatch and multiple 2026 salary guides, overall AI hiring and pay have held up rather than declined.

Which AI job is most at risk from automation?

The clearest example is the standalone prompt engineer. As models get better at interpreting unclear instructions and clarifying intent, the value of crafting a single perfect prompt falls, and prompting is becoming a baseline skill embedded in other roles. Pure entry-level data labelling is also squeezed, because AI now pre-labels data and humans increasingly review rather than label from scratch.

How much do AI engineers earn in the UK?

According to ITJobsWatch data covering the period to May 2025, the median AI engineer salary in the UK was around £85,000, with the top 10% earning above £101,000. Entry-level benchmarks rose from roughly £42,500 in 2023. Senior MLOps and applied-science roles, particularly in London, typically reach well into six figures in 2026 salary guides.

Is MLOps a good career in 2026?

In our reading of UK hiring data, yes. As organisations move AI from pilots into production, MLOps has become a genuine bottleneck, combining DevOps and machine-learning skills. It is consistently listed among the fastest-growing AI specialisms going into 2026, and because it centres on reliability and oversight, it is harder to automate than first-pass coding or labelling tasks.

Will AI replace data scientists and machine learning engineers?

Not wholesale, on current evidence, but the work is changing. Automation absorbs routine analysis, boilerplate code and basic monitoring, while demand grows for engineers who integrate foundation models, design agent workflows and run evaluation. The PwC AI Jobs Barometer has found AI-exposed roles often seeing stronger demand, suggesting augmentation rather than straightforward replacement for skilled practitioners.

Which UK companies are hiring for AI roles?

Major employers include Google DeepMind, Amazon (across AWS and applied science), and Microsoft, recruiting across London, Cambridge, Manchester and Edinburgh. Specialist consultancy Faculty hires in London, while the Alan Turing Institute and DSIT, including the AI Security Institute, recruit for research, evaluation and fellowship roles. Vacancies are typically advertised via boards such as Adzuna and LinkedIn.

How many UK jobs could AI automate?

The IPPR has estimated that up to roughly 8 million UK jobs could be exposed in a worst-case scenario, with around 11% of tasks currently at risk. However, its central scenario put displacement far lower, at about 545,000 roles alongside a potential 3.1% GDP uplift. The outcome depends heavily on policy and how firms choose to adopt the technology.

Summary: AI Is Reshaping AI Careers, Not Ending Them

Artificial intelligence jobs in the UK are growing, but automation is rapidly redrawing what those jobs involve. Routine tasks, first-pass labelling, simple prompting, boilerplate code, are being absorbed, while MLOps, AI safety, evaluation, agent orchestration and governance are expanding, supported by named employers from DeepMind to Faculty and by DSIT and the Alan Turing Institute. Salary data from ITJobsWatch, with a median AI engineer wage around £85,000, points to a healthy field rather than a shrinking one. The strategic move for 2026 is to climb from tasks to systems and from execution to oversight, where human judgement still commands a premium.

Ready to find your next role in a fast-changing market? Browse current vacancies and career advice at artificialintelligencejobs.co.uk.