AI Education
Will AI Take Your Job? What the 2026 Data Shows
What the latest research says about AI and jobs, which roles are most and least exposed, and how to check your own risk in about 3 minutes.
Probably not, at least not in 2026 and not all of it. The data we have shows AI changing jobs faster than it is ending them: US unemployment was 4.2% in September 2026, and studies from Yale and Anthropic find no broad rise in unemployment among workers in AI-exposed jobs. Some groups are under real pressure, though. Young people starting out in fields like software and customer service are being hired less, and companies named AI as the reason for about a fifth of announced US layoffs this year.
When I first wrote this article in April, most of the headlines were forecasts. Six months later there's a lot more real data, so I rebuilt the article around it. You'll find what the numbers say, which jobs are most and least exposed, why your daily tasks matter more than your title, and a few things worth doing now.
Will AI take most jobs?
No major forecast expects that. The World Economic Forum's Future of Jobs Report 2025 expects 170 million jobs to be created and 92 million displaced worldwide between 2025 and 2030, a net gain of 78 million (WEF, January 2025). The churn is big, though. New and lost jobs together add up to 22% of today's jobs, and the WEF expects 39% of workers' existing skills to change or become outdated by 2030.
Other estimates point the same way. In 2023, Goldman Sachs economists calculated that generative AI could expose the equivalent of 300 million full-time jobs to automation (Goldman Sachs, April 2023). Exposed doesn't mean lost. Their 2025 update is more concrete: if AI is widely adopted, it could displace 6 to 7% of the US workforce, and they expect the effect to be temporary, typically fading after about two years (Goldman Sachs, August 2025).
The IMF estimated in 2024 that almost 40% of jobs worldwide are exposed to AI, about 60% in advanced economies, and that roughly half of the exposed jobs may benefit from it (IMF, January 2024). McKinsey went further on the technical side and found that today's technology could technically automate 57% of US work hours. The same report says this doesn't have to mean mass job losses, because more than 70% of today's most sought-after skills are used in both automatable and non-automatable work (McKinsey Global Institute, November 2025).
Companies are also slower than their own predictions. In McKinsey's 2025 State of AI survey, 32% of respondents expected AI to shrink their workforce within a year. In the 2026 survey, 14% said it actually had (McKinsey, August 2026).
Not everyone is this calm. In May 2025, Anthropic CEO Dario Amodei told Axios that AI could wipe out about half of all entry-level white-collar jobs within one to five years and push unemployment to 10 to 20% (Axios, May 2025). That's a forecast. The early data on young workers, further down, is the reason to take it seriously anyway.
What the 2026 data shows so far
The overall job market is weak, but so far the research can't pin that on AI. Layoff announcements that name AI are a different story, and so is hiring of young people.
| Source | What it found | Date |
|---|---|---|
| US Bureau of Labor Statistics | Unemployment 4.2%, payrolls up 29,000 in September 2026 | Oct 2026 |
| Yale Budget Lab | No clear effect of AI on employment or wages in AI-exposed jobs yet | May 2026 |
| Anthropic | No systematic rise in unemployment for highly exposed workers since late 2022 | Mar 2026 |
| Stanford Digital Economy Lab | Employment of 22- to 25-year-olds in the most exposed jobs about 19% below trend | Aug 2026 |
| Challenger, Gray & Christmas | AI cited in 120,136 announced US job cuts, January to September 2026 | Oct 2026 |
| McKinsey | 14% of respondents say AI reduced their workforce in the past year | Aug 2026 |
| Monster | 89% of recent and upcoming graduates worry AI could replace entry-level roles, up from 64% | Mar 2026 |
Start with the big picture. US payrolls grew by just 29,000 in September 2026 and unemployment was 4.2% (BLS, October 2026). Yale's Budget Lab compared occupations with high and low AI exposure and concluded that "the results do not yet show clear indication of labor market effects of AI." It puts much of the weak job growth down to slower immigration (Yale Budget Lab, May 2026). Anthropic's economists came to a similar result with a different method: no systematic increase in unemployment for highly exposed workers since late 2022 (Anthropic, March 2026).
Layoff announcements look louder. Employers cited AI as the reason for 120,136 announced US job cuts in the first nine months of 2026, about 21% of all announced cuts and the most-cited reason this year (Challenger, Gray & Christmas, October 2026). I'd read that with some care. Challenger records the reason a company announces, and I suspect "AI" sometimes sounds better to investors than "demand is down". Total announced cuts were also 39% lower than in the same period of 2025.
Why entry-level workers feel it first
The clearest early signal shows up in hiring. In August 2026, Stanford's Digital Economy Lab updated its "Canaries in the Coal Mine" study with ADP payroll data through June 2026. Employment of 22- to 25-year-olds in highly AI-exposed jobs, such as software developers and customer service reps, now sits about 19% below where it would be if it had kept pace with young workers in less exposed jobs. A year earlier the gap was 15% (Stanford Digital Economy Lab, August 2026).
A few details in that study matter if you're worried about your own job:
- Experienced workers in the same jobs show no comparable gap.
- The drop comes mainly from companies hiring fewer young people, not from firing more of them.
- The decline is concentrated in jobs where people use AI to automate tasks. Where AI mostly helps people do their work, employment is flat or rising.
The authors are careful to say these are descriptive patterns, not proof that AI caused them. Anthropic's data shows a smaller version of the same thing. The rate at which young workers find jobs in the most exposed occupations fell by an estimated 14% compared with 2022, a result that is only just statistically significant, and nothing similar showed up for workers over 25.
Graduates have noticed. In Monster's 2026 State of the Graduate Report, 89% of recent and upcoming graduates said they worry AI could replace entry-level roles, up from 64% a year earlier (Monster, March 2026). PwC's 2026 AI Jobs Barometer adds a useful detail for anyone starting out: entry-level roles that are most exposed to AI are seven times more likely to ask for senior-level human skills such as leadership, creativity or face-to-face work (PwC, June 2026).
Which jobs are most at risk from AI?
Jobs built on text, data and screens. Across the big studies, the same roles keep coming up: programmers, customer service reps, data entry workers, translators and writers.
| Role | What the research found | Source |
|---|---|---|
| Computer programmers | Claude usage already covers about 75% of their tasks, the highest of any occupation | Anthropic, 2026 |
| Customer service representatives | Second-highest coverage in Anthropic's data, and high in Microsoft's study too | Anthropic, 2026; Microsoft, 2025 |
| Data entry keyers and clerks | 67% task coverage; one of the WEF's fastest-declining jobs | Anthropic, 2026; WEF, 2025 |
| Interpreters and translators | Highest AI applicability score of all occupations | Microsoft, 2025 |
| Writers, authors and journalists | Among the highest AI applicability scores | Microsoft, 2025 |
| Bank tellers and postal service clerks | Among the fastest-declining jobs to 2030 | WEF, 2025 |
| Cashiers, administrative assistants and executive secretaries | Largest expected job losses in absolute numbers | WEF, 2025 |
Microsoft's researchers built their ranking from 200,000 anonymized Bing Copilot conversations (Tomlinson et al., July 2025). One of the authors, Kiran Tomlinson, pointed out afterwards that the list shows which jobs might find AI most useful, not which ones will disappear. That applies to every list in this section. Exposure means AI can do some of the tasks in a job, which can make the person doing it faster just as easily as it can make them unnecessary.
One finding surprised me. In Anthropic's data, the most exposed group of workers earns 47% more on average than workers with no exposure, and is far more likely to hold a graduate degree (Anthropic, March 2026). This time the most exposed jobs are mostly well-paid office jobs.
Which jobs are safest from AI?
Jobs where most of the work is physical or hands-on with people. Anthropic found that about 30% of workers are in jobs where AI covers none of their tasks yet, for example cooks, motorcycle mechanics, lifeguards, bartenders, dishwashers and dressing room attendants. Microsoft's lowest-scoring jobs look similar: phlebotomists, nursing assistants, hazardous materials removal workers, painters' helpers, embalmers, and tire repairers and changers.
Safe is relative. These lists measure what today's chatbots can do, and robotics is a separate question. But if your work depends on being physically present, handling unpredictable situations or earning a person's trust face to face, AI is a tool for the paperwork around your job rather than a replacement for it.
The real question: which of your tasks will AI handle?
Most jobs are a bundle of different tasks. AI takes over some of them, speeds up others and leaves the rest alone, so the useful question is which parts of your week fall into which group.
The research backs this up. Stanford's data shows that jobs where people use AI to help with their work held up, while jobs where AI takes over whole tasks lost young workers. Anthropic found that actual AI use still covers only a fraction of what the technology could do in theory. Even for computer and math jobs, Claude usage covers 33% of tasks.
That's why two people with the same title can face very different risk. Take two financial analysts. One spends most of the week pulling data together and formatting reports. The other spends most of it advising clients and deciding what the numbers mean. AI can already do a lot of the first person's week and much less of the second's.
Five things you can do now
- Use AI on your real work this week. Employers pay for it: PwC found an average wage premium of 62% for AI skills in 2026, and jobs that require specific AI skills grew 69%, against 9% for all jobs. McKinsey found that demand for AI fluency in US job postings grew sevenfold in two years. Start with ChatGPT or Claude on something you already do, and read why learning AI matters in 2026 if you're not sure where to begin.
- Write down your tasks. List everything you do in a normal week and mark each one: could AI do this today, in two years, or not for a long time? The last group is where your time and training should go.
- Spend more time on the parts of your job that need you. AI is good at drafting and pattern-matching. Leading people, negotiating and working face to face are still yours, and PwC found that AI-exposed entry-level roles increasingly ask for exactly those skills.
- Get good at checking AI output. Someone has to decide when the AI is wrong, and that person is hard to replace. Learning how to write clear prompts is the fastest way to see where the tools fall short.
- Go deep in one area. Stanford found that employment rose among experienced workers in jobs that rely on tacit knowledge, the kind you only pick up by doing the work for years. In my view, deep domain knowledge combined with AI skills is the safest position you can be in.
Find out your own risk score
Statistics about millions of jobs only go so far. What you probably want to know is where you stand.
The AI Job Displacement Calculator asks 14 questions about your industry, role, daily tasks and AI skills, and covers 48 job profiles. It takes about 3 minutes, needs no signup, and gives you a risk score from 0 to 100, your two biggest risk factors and a first recommendation. The optional $27 Career Defense Report adds a task-by-task breakdown, a skills roadmap, a 30-day action plan and alternative career paths.
It's a self-assessment built on published research such as the WEF and Anthropic studies above. It can't predict what your employer will do, but it shows you where to focus.
Frequently asked questions
Will AI take my job in 2026?
For most people, no. US unemployment was 4.2% in September 2026, and studies from Yale and Anthropic have found no broad rise in unemployment among workers in AI-exposed jobs. The risk is higher if you're early in your career in an exposed field such as software or customer service, where companies are hiring fewer young people.
How many jobs has AI replaced so far?
Nobody has a precise count. US employers cited AI as the reason for 120,136 announced job cuts from January to September 2026, according to Challenger, Gray & Christmas. Those are announcements, with reasons given by the companies themselves. Researchers at Yale and Anthropic haven't found a matching rise in unemployment among AI-exposed workers yet.
Which jobs will AI replace first?
Jobs made up of routine screen-based tasks are the most exposed. Anthropic's data puts computer programmers, customer service representatives and data entry keyers at the top, Microsoft's study ranks interpreters and translators highest, and the WEF expects the fastest declines for postal service clerks, bank tellers and data entry clerks.
Which jobs are safe from AI?
Jobs that are mostly physical or hands-on with people, such as cooks, mechanics, nursing assistants and phlebotomists. About 30% of US workers are in jobs where Anthropic found no AI use covering their tasks yet.
Is AI making it harder for graduates to get hired?
In AI-exposed fields, it looks that way. Stanford researchers found that employment of 22- to 25-year-olds in the most exposed jobs is about 19% below trend, mainly because companies hire fewer of them. The authors say the data shows a pattern rather than proof that AI is the cause.
How many jobs will AI create?
The World Economic Forum expects 170 million new jobs worldwide between 2025 and 2030, against 92 million displaced, for a net gain of 78 million. Not all of that is due to AI. LinkedIn counted at least 1.3 million new AI-related job opportunities over two years in its January 2026 report.
How can I make my career AI-proof?
No job is fully AI-proof, but you can lower your risk. Learn to use AI tools on your real work, shift time toward tasks that need judgment and personal contact, and build deep knowledge in one area. The free AI Job Displacement Calculator gives you a risk score and your two biggest risk factors.
Where to go from here
Take the free risk assessment, then pick one task from your list and try it with AI this week. Worrying about the headlines won't change your score. Knowing which parts of your work are exposed gives you something to act on.
If you'd like one short, practical AI note a week, The Risk Memo goes out every Tuesday.
Researched with AI tools, written and fact-checked by Dominic.
Sources
- World Economic Forum, The Future of Jobs Report 2025 (Jan 7, 2025)
- Goldman Sachs, Generative AI could raise global GDP by 7% (Apr 5, 2023)
- Goldman Sachs, How will AI affect the global workforce? (Aug 13, 2025)
- IMF Blog, Kristalina Georgieva, AI Will Transform the Global Economy. Let's Make Sure It Benefits Humanity (Jan 14, 2024)
- McKinsey Global Institute, Agents, robots, and us: Skill partnerships in the age of AI (Nov 2025)
- McKinsey, The state of AI in 2026 (Aug 25, 2026)
- Axios, Ready or not, AI is starting to replace people (May 30, 2025)
- US Bureau of Labor Statistics, The Employment Situation, September 2026 (Oct 2, 2026)
- The Budget Lab at Yale, AI Is Probably Not (Yet) the Reason for Labor Market Weakening (May 7, 2026)
- Anthropic, Labor market impacts of AI: A new measure and early evidence (Mar 5, 2026)
- Challenger, Gray & Christmas, Job Cut Announcement Report, September 2026 (Oct 1, 2026)
- Stanford Digital Economy Lab, No Widespread Displacement, but the AI Employment Gap for Young Workers Has Widened to 19% (Aug 12, 2026)
- Monster, 2026 State of the Graduate Report (Mar 27, 2026)
- PwC, 2026 Global AI Jobs Barometer (Jun 15, 2026)
- Tomlinson et al., Working with AI: Measuring the Applicability of Generative AI to Occupations (Jul 10, 2025)
- LinkedIn Economic Graph, Labor Market Report (Jan 2026)
Last updated: