Guide · Your team and AI
Will AI take my job? An honest answer, with the evidence
Nobody can honestly promise what will happen to any one job, and so far the evidence shows no widespread job losses from AI in the UK or the US. It does show fewer openings in some early-career roles where the work is easy to write down and hand over, so the honest answer is a mixed one.
By Andrew Ward, Founder of ScorchsoftChecked 12 minute read
In short
- So far, around half of UK businesses said AI had no impact on their headcount, and just under 7% of medium-sized businesses reported a fall (ONS, June 2026).
- The early warning is for people starting out: employment of 22 to 25-year-olds in the most AI-exposed US jobs is about 19% below where it would otherwise be (Stanford, ADP payroll data to June 2026).
- AI touches tasks; people carry outcomes. Deciding what good looks like, checking the work and making sure it achieved its purpose are likely to grow in value as AI improves.
- If you manage people, decide and say what the time AI saves is for. That choice belongs to the organisation, not to the people doing the work.
Written for: Anyone in a UK office or care job wondering what AI means for their work, and the managers they ask.
Is AI going to take my job?
Probably not in the way the headlines suggest, but nobody can promise it for any particular job. AI is changing which parts of work people do. So far it hasn’t caused widespread job losses, though it is making some entry-level work harder to find.
It’s a fair question, and a common one. In Ipsos’s 2026 AI Monitor, 22% of people in Britain said they fear AI will replace them in their job within five years, against a 32-country average of 35%. In Henley Business School’s survey of 2,900 UK full-time workers (June 2026), 36% feared AI would replace their role, and 44% of Generation Z. The two surveys asked different questions, so the figures differ, but both say the same thing: a large minority of UK workers are worried, and they deserve a straight answer.
What this guide covers
UK and US evidence about jobs and AI, as of 29 October 2026. Most of the strongest research is from the US, and we say so where it is. It doesn’t cover employment law, redundancy or your rights at work; for those, see Acas (opens in a new tab) or a qualified adviser.
What does the evidence say so far?
Large studies in the US and official UK surveys find no economy-wide job loss from AI yet. The clearest warning sign is fewer openings for people at the start of their careers in AI-exposed jobs, and even that isn’t proven to be caused by AI.
| Source | Who and when | What it found |
|---|---|---|
| Office for National Statistics (ONS), Artificial intelligence in UK businesses | UK businesses with 10 or more employees, 5 to 28 June 2026 (published 20 July 2026) | About 35% of businesses use AI, up from about 12% in late 2023. Around half of businesses said AI had no impact on headcount; just under 7% of medium-sized businesses reported a decrease. The ONS calls these “early indications”. |
| Department for Science, Innovation and Technology (DSIT) with LinkedIn, A snapshot of entry-level hiring in the UK | LinkedIn hiring data for the UK, to April 2026 (published 8 June 2026) | Entry-level hiring is weak but falling broadly in line with the wider market. Some of the biggest falls are in roles where AI capabilities have grown, “but this is not yet causal evidence of AI’s impact on hiring”. |
| Stanford Digital Economy Lab, Canaries in the Coal Mine? (Brynjolfsson, Chandar and Chen) | US payroll data from ADP covering millions of workers, to June 2026 (revised 12 August 2026) | No evidence of widespread, economy-wide job displacement. Employment of 22 to 25-year-olds in the most AI-exposed jobs is about 19% below where it would be had it kept pace with less-exposed peers, mainly through less hiring rather than layoffs. Experienced workers show no comparable gap. |
| The Budget Lab at Yale, Tracking the Impact of AI on the Labor Market | US labour market, monthly survey data to September 2026 (updated 14 October 2026) | No “clear evidence of labor market disruption associated with AI”. Unemployment has grown somewhat faster among AI-exposed workers than comparable unexposed workers since late 2022, but the Budget Lab calls these estimates volatile and says to read them cautiously. |
| Department for Education (DfE), The impact of AI on UK jobs and training | UK occupations, sectors and regions (published 28 November 2023) | Professional jobs, especially in finance, law and business management, are the most exposed to AI. The report measures exposure, not “whether a job will be augmented (aided) or replaced (substituted)”. |
Two patterns stand out in the Stanford data. Where AI does the task instead of the person, jobs for newcomers are fewer. Where it helps the person do the task, employment is flat or rising, especially for experienced workers. Our reading: that difference often comes down to how an organisation chooses to use the tool, not the tool itself.
The Stanford authors are careful about their own findings. They call them “early, descriptive indicators”, not proof of cause.
The pattern weakens when education is taken into account, and some of it started before generative AI. The gap is also larger in their payroll sample than in national surveys. That is why the honest word is “warning”, which is what a canary in a coal mine is.
What doesn’t the evidence tell us?
It doesn’t say your job is safe, and it doesn’t say mass unemployment is coming. Exposure scores aren’t forecasts of job loss, the entry-level signal isn’t proven to be caused by AI, and most of the strongest research is from the US.
- It doesn’t say any particular job is safe. Nobody can promise that, and you should be wary of anyone who does.
- It doesn’t say mass unemployment is coming. The data shows nothing like it so far, and forecasts disagree widely.
- Exposure isn’t replacement. A job being “exposed” means AI can touch some of its tasks. The DfE report says plainly that it can’t tell whether a job will be aided or replaced.
- The entry-level squeeze isn’t proven to be caused by AI. UK hiring overall has been falling since late 2022, and DSIT’s key messages say the roles hit hardest are seeing less hiring at every level of seniority, not just for juniors.
- Most of the strongest evidence is American. UK sectors, employment law and protections differ.
- Learning AI isn’t an insurance policy. It is a sensible bet, not a promise. Organisations make decisions about headcount that individuals don’t control.
You may have heard that “AI won’t take your job, but someone using AI will.” It is half true: using AI well can make you quicker at some tasks and better at judging the results. But the line assumes the number of jobs stays fixed, and it puts the whole burden on the individual. When fewer people are hired, that isn’t the fault of those who weren’t.
What makes someone more valuable as AI improves?
Carrying the outcome: deciding what good looks like, checking what AI produces, noticing when something is going wrong and making sure the work achieved what it was for. AI can touch more and more tasks, but someone still has to own the result.
The AI Stairway™ puts it in five words: AI touches tasks, people carry outcomes. A tool can draft the email, summarise the meeting or fill in the form. It doesn’t notice three weeks later that nobody acted on the summary, or ring the client to check the letter made sense. Carrying the outcome means owning that: knowing what the work is for, and checking that it got there.
Research on skills points the same way. A Microsoft Research survey of 319 knowledge workers (Lee and colleagues, April 2025; research by an AI vendor) found that with AI, critical thinking shifts toward checking information, fitting answers together and looking after the task as a whole. The Stanford study found experienced workers, whose know-how is learned through practice rather than written down, holding their ground. So the parts of your job that are hardest to write down are also the parts that gain most from your judgement.
Darren is a payroll and bookkeeping senior at a 40-person accountancy practice. He has watched AI draft client emails in seconds and asks his manager, Gareth, “Is this coming for my job?” Gareth doesn’t promise anything. He says what the evidence shows, that the practice hasn’t cut anyone because of AI and isn’t planning to, and that Darren’s value is in what the tool can’t do: knowing when a payroll figure looks wrong and making sure every client is paid correctly. Then he asks Darren to help decide where AI should and shouldn’t touch payroll. Darren becomes the person whose expertise makes the AI’s output trustworthy.
A simple habit makes this concrete. Before you use AI on anything that matters, ask the question the course keeps coming back to: “Who’s carrying this, and how will we know it landed?” If the answer is you, you’re doing the part of the work that gains value as the tools improve.
What should managers do with the time AI saves?
Decide it on purpose, and tell your team before they have to guess. Time saved can go into better quality, more customers, development for junior staff, or fewer hours, and each is a choice the organisation makes, not something AI decides.
When people hear that AI saves time, many quietly do the sum: “If I need less time, will they need less of me?” The fastest way to make that worse is to say nothing. The organisation chooses what to do with the time, so say what you’ve chosen, even if the answer is “we haven’t decided yet, and we’ll tell you before we do”.
| Where the time goes | What the team hears |
|---|---|
| Better checking and quality on the work that matters most | “Your judgement is the point.” |
| Work you never had time for: the backlog, the improvement, the client you couldn’t serve | “We’re growing with this, not shrinking.” |
| Coaching and development, especially for junior staff | “We still need people to learn the job properly.” |
| Breathing space or fewer hours | “We’re not just raising the bar again.” |
| Fewer people, without saying so | “Every saving you find costs a colleague.” People stop sharing what works. |
Two points need particular care. First, junior roles: if AI does the routine tasks that used to teach newcomers the job, plan how they will learn it now, or you won’t have experienced staff in five years. Second, if you ever do plan to reduce headcount, that is a change to people’s jobs with legal duties attached. Take advice before you say or decide anything.
How do you answer a colleague who asks about their job?
Name the worry, give the mixed picture with its date, say what nobody knows, and offer one practical thing to do this week. Don’t promise their job is safe, and don’t tell them to upskill or else.
| When | Do | Don’t |
|---|---|---|
| Opening | Name it early: “Lots of people are wondering what AI means for their job.” | Skip it, or leave it until after the tool tips. |
| The evidence | Give the mixed picture, with sources and dates. | Pick only the reassuring numbers, or only the frightening ones. |
| The future | Say what nobody knows. | Promise that jobs are safe, or predict that they aren’t. |
| Closing | Offer something to do this week. | End on “adapt or else”. |
| Saved time | Say that the organisation chooses how to use saved time, and say what it has chosen. | Imply that any job loss would be the individual’s fault for not keeping up. |
A script you can adapt (owner: the line manager)
- “It’s a fair question, and I won’t pretend anyone can promise what happens next.”
- “What the evidence shows so far is no widespread job losses from AI, but fewer openings for some people starting out. That’s as of now, and it may change.”
- “Here, we’ve decided to use the time AI saves for [your choice]. If that changes, you’ll hear it from me first.”
- “The part of your job I value most is [the outcome they carry]. That’s the part AI can’t do for you.”
- “Would you help us decide where AI should and shouldn’t touch your work?”
Some people will be more than a little worried. If the worry is affecting sleep, health or life outside work, our guide to AI anxiety at work has a calmer look at it and where to get support. If the worry is less about jobs and more about the pace, see AI fatigue.
What can you do this week?
Three small moves, none of which needs a budget: write down the outcome your work serves, try AI on one task you’d check anyway, and, if you manage people, tell your team what the time AI saves is for.
Step 1: Write your outcome line (owner: you)
Pick one piece of work that matters and finish this sentence: “The outcome this serves is … and I’ll know it landed when …” Keep it where you’ll see it. It is the part of the job to hold on to.
Step 2: Try AI on one task you’d check anyway (owner: you)
Use an approved tool on something low-stakes, then check it properly. You learn where it helps and where it goes wrong, which is the judgement that makes you worth asking. If your organisation hasn’t approved a tool yet, see our guide to shadow AI on why that matters.
Step 3: Say what saved time is for (owner: the manager or owner)
At your next team meeting, say in one sentence what the organisation will do with time AI saves. If you have peer AI champions, ask them to collect where time is really being saved. Our guide to AI champions covers how to support them.
Module 7 of The AI Stairway™, Forming an AI Culture, practises this conversation as a scene with the course’s cast, and covers how good habits spread through a team. If you’d rather start with your own habits, the free introductory module, Know Your AI Blind Spots, takes about 22 minutes.
Questions people ask
Which jobs are least exposed to AI?
In the Department for Education’s 2023 analysis, the least exposed were manual roles in unpredictable settings, such as roofers, plasterers, hospital porters and cleaners. Our reading is that in hands-on care, AI is more likely to change the paperwork than the care itself. Being less exposed promises nothing, and more exposed doesn’t mean a job will go.
Is it worse for young people and graduates?
There are early signs that it is harder to get a first job in some AI-exposed roles. US payroll data shows employment of 22 to 25-year-olds in the most exposed jobs about 19% below trend, mainly through less hiring. In the UK, DSIT found entry-level hiring weak but falling broadly in line with hiring overall. Neither proves AI is the cause.
Should I learn AI to protect my job?
Learning to use AI well is a sensible bet: it makes you quicker at some tasks and better at judging the output. It can’t promise anything, and anyone who sells it as a safeguard is overclaiming. The most durable part of your work is carrying the outcome, which AI makes more valuable, not less.
Is my employer allowed to replace me with AI?
That is a question about your contract and UK employment law, which this guide doesn’t cover. Acas says employers must consult workers about redundancy and follow a fair process, and it has free, impartial guidance on both. You can also ask a union representative or an employment adviser about your own situation.
How often is this page checked?
We recheck the evidence at least every six months, and sooner when a major source updates. The date it was last checked is at the top of the page. The research changes quickly, so read every figure with its date.
Find your own AI blind spots
Know Your AI Blind Spots is the free introductory module of The AI Stairway™: about 22 minutes, on your phone or desktop, with three fixes you try in your own AI tool.
Start freeSources
Every figure on this page comes from one of these. Links open in a new tab.
- Artificial intelligence in UK businesses: 2023 to 2026 (opens in a new tab). Office for National Statistics, 20 July 2026.
- A snapshot of entry-level hiring in the UK (opens in a new tab). Department for Science, Innovation and Technology, 8 June 2026.
- Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence (opens in a new tab). Stanford Digital Economy Lab (Brynjolfsson, Chandar and Chen), revised 12 August 2026.
- Tracking the Impact of AI on the Labor Market (AI Labor Market Tracker, current update) (opens in a new tab). The Budget Lab at Yale, updated 14 October 2026.
- The impact of AI on UK jobs and training (opens in a new tab). Department for Education, Unit for Future Skills, 28 November 2023.
- The British reserve: navigating the AI transition with caution (Ipsos AI Monitor 2026) (opens in a new tab). Ipsos UK, 25 June 2026.
- Hype versus reality: a 12-month AI pulse check (opens in a new tab). Henley Business School, 4 June 2026.
- The Impact of Generative AI on Critical Thinking (Lee and others, CHI 2025) (opens in a new tab). Microsoft Research, April 2025.
- Redundancy (opens in a new tab). Acas, read 29 October 2026.