Will AI take graduate jobs? What the UK evidence says

On the evidence so far, AI is not taking graduate jobs wholesale. In the UK, it is changing what graduate jobs involve faster than it is removing them: 87% of employers in the Institute of Student Employers’ 2026 survey expect AI to reshape graduate and apprentice roles, but 82% expect it to replace no more than one in ten entry-level roles over the next three years. Graduate job adverts are at their lowest for the time of year since 2020, which job site Indeed links to cost pressures and uncertainty, although research does find early signs that junior jobs most exposed to AI are being hit hardest. Graduates who use AI well and honestly, and can show real judgement and evidence of their work, are likely to be best placed.

Is AI reducing graduate jobs in the UK?#

Graduate hiring is down, but AI is not the main reason given so far. Indeed Hiring Lab found that UK job adverts with “graduate” in the title were about 7% lower on 10 July 2026 than a year earlier, the lowest for the time of year since 2020. But the fall is broad. Overall UK job adverts were 32% below their pre-pandemic level, most job categories had fallen, and Indeed says employers are pulling back on junior hiring “in response to the cost pressures and uncertainty that have characterised 2026”.

Across the economy, AI has not yet changed headcount much. The Office for National Statistics found that around 35% of UK businesses with 10 or more employees used AI by June 2026, up from about 12% in late 2023. Around half of businesses using AI said it had made no difference to their headcount, many were unsure or said it did not apply, and among medium-sized businesses just under 7% reported a fall.

Is AI hitting entry-level jobs hardest?#

There are early signs that it is, in the jobs most exposed to AI:

• UK, 2021 to 2025. A King’s College London study of millions of job adverts and LinkedIn profiles found that firms whose workforces were highly exposed to AI cut total employment by 4.5% on average, with the effect “concentrated almost entirely in junior positions”, which fell by 5.8%. Customer-facing roles held up.
• US, to 2026. Stanford researchers using US payroll data found that employment of 22 to 25-year-olds in highly AI-exposed occupations was about 19% below where it would have been had it kept pace with less-exposed jobs. They stress that these are descriptive patterns rather than proof that AI caused them, and that they do not see widespread job losses across the economy.

A Bank of England staff blog, reviewing UK and US research in January 2026, concluded that AI “likely remains an amplifier rather than the sole driver” of weaker youth hiring, with the wider economic slowdown also playing a big part. The views are the author’s, not the Bank’s.

What is known, and what is only predicted?#

It helps to separate what has happened from what employers expect.

In the ISE survey, 58% of employers expected minor changes to entry-level tasks and 29% significant changes, and 43% said entry-level roles had already changed because of AI without being formally redesigned. In the CIPD survey, almost two-thirds (62%) of the employers expecting cuts thought clerical, junior managerial, professional or administrative roles were most likely to go. These are expectations, not outcomes, and plans can change.

How do employers use AI in recruitment?#

Less than many applicants think, so far. In the Prospects Early Careers Survey 2026, 65% of respondents believed employers use AI to shortlist applicants always or most of the time. The ISE’s 2025 Student Recruitment Survey found a more mixed picture:

• Automation is used mainly for tests with clear scoring, such as numerical reasoning (57% fully automated), situational judgement (53%) and verbal reasoning (49%).
• In-person interviews, group tasks, role plays and case studies are 94% to 100% fully human-led.
• AI-only systems are rare, most notably in game-based assessments (15%).
• More change is expected: 62% of employers expect to use AI in recruitment, and 70% expect to automate more of their selection over the next five years.

In practice, expect automated tests and screening early on, and people at interview and assessment centre stages. Our guide to video interviews covers one stage where recorded answers are common.

Can you use AI to write job applications?#

Often yes, for some parts, but the rules vary by employer, and misuse is being noticed.

Most applicants now use it. Prospects found that 73% of respondents who had applied for a job used AI for at least one part of the process, up from 55% the year before, most often to edit a CV or cover letter (57%), prepare for interviews (47%) or write a CV or cover letter (46%).

Employers are split. In the ISE’s 2025 survey, 46% had no problem with candidates using AI at some stages, 29% advised against it but took no action, and 45% had given candidates no guidance at all. Among employers that allowed AI, 61% allowed it for CVs and cover letters and 41% for online application questions. But concern is growing. The share of employers who frequently came across suspected cheating more than doubled, from 7% to 15%, and the most common problem was candidates using AI in interviews without permission. By early 2026, two-thirds of employers believed graduates and apprentices were using AI to misrepresent their skills, up from about half in 2025.

The Civil Service guidance is a clear example of where many employers draw the line. You may use AI to research the organisation, refine your ideas, check spelling and grammar, and practise interview questions. You must not use it to invent or inflate your experience, write generic answers, complete tests or answer interview questions, and the Civil Service says it may reject applications where AI is used inappropriately.

A sensible rule: check each employer’s guidance. Where there is none, use AI to prepare and polish, never to invent, and never during a live test or interview.

Which skills are growing in value?#

• Judgement and people skills. Employers in the ISE survey expect critical thinking, AI literacy, communication and adaptability to grow in importance, and routine admin, basic data and writing tasks to matter less.
• Using AI at work. Indeed found that a record 9.4% of UK job adverts mentioned AI at the end of June 2026. The share is spreading beyond tech into marketing, HR and finance, where adverts mentioning AI are growing even as overall adverts fall.
• Experience learned on the job. The Stanford team found employment falling for young workers in jobs built on codified knowledge that can be learned from textbooks and procedures, but rising for experienced workers in jobs that rely on know-how gained through practice and mentoring.

The ISE also found employers almost twice as likely to build digital literacy on the job as to recruit for it, which suggests many expect to teach these skills after you join.

What should graduates do now?#

• Learn the AI tools used in your field, and practise checking their output. Keep one or two examples of work you improved with AI and then checked yourself.
• Use AI to prepare, not to pretend. Draft in your own words and use AI for feedback. Make sure every claim is true and you can talk about it at interview.
• Build evidence of real work. Projects, internships, part-time jobs and society roles all count, especially with results you can explain.
• Practise the human stages. Interviews and assessment centres remain almost entirely human-led, so rehearse out loud.
• Get careers advice about AI. In Prospects’ survey, 27% of students who had received careers advice about AI felt their opportunities would increase, compared with 12% of those who had not.

What does this mean for choosing a career?#

Look at the tasks in a job, not just its title. The evidence so far suggests routine admin, basic data and writing tasks are the most exposed, while work that depends on judgement, relationships and experience built up on the job is more resilient. Indeed found the lowest share of AI mentions, below 1%, in in-person work such as personal care, cleaning and driving. But no graduate career is simply “safe” or “doomed”, and forecasts are uncertain.

Data roles show what this can look like. Nearly half of UK data and analytics adverts on Indeed now mention AI, the highest of any category, which suggests employers want analysts who can work with AI tools and check what they produce. That does not guarantee demand for any one role, and routine reporting is the kind of basic data task employers expect to matter less. Our research on data analyst jobs sets out the data-specific evidence.

For parents, the most useful help is practical: encourage your son or daughter to build evidence of their skills, practise interviews and use their careers service. Our guide for parents on the graduate job market explains the wider picture.