--- title: AI and jobs url: https://ftrlg.com/en/wiki/ai-and-jobs/ lang: en published: 2026-10-16 version: 1 summary: How AI is changing people's work: which tasks it takes over, which jobs it helps and which it threatens. Educated office workers are hit for the first time, and beginners first. cite_as: "future.log, «AI and jobs», https://ftrlg.com/en/wiki/ai-and-jobs/" license: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ --- # AI and jobs **AI and jobs** is about how artificial intelligence is changing people's work: which tasks it takes over, which occupations it helps and which it threatens with job cuts. The site makes no forecasts of its own: whatever is expected by 2030 is stated here with a reference to whoever expects it. For decades automation hit routine manual and clerical work, and the higher the education and pay, the safer a job was considered.[^1] [Large language models](https://ftrlg.com/en/wiki/large-language-model/) turned this picture upside down: they affect educated office workers the most.[^3] The first measurable trace is not mass layoffs but that fewer young people are being hired in such occupations; across the economy as a whole there is no noticeable shift yet.[^11][^9] ## History - **17.09.2013** Carl Frey and Michael Osborne of Oxford estimate that about 47% of US employment is in occupations at high risk of computerisation, and the higher the wages and education, the lower the risk.[^1] - **05.2016** OECD economists respond that if one looks not at whole occupations but at individual tasks, on average 9% of jobs across 21 OECD countries can be automated.[^2] - **17.03.2023** Researchers from [OpenAI](https://ftrlg.com/en/models/openai/) and the University of Pennsylvania, in “GPTs are GPTs”: for 80% of US workers, language models could affect at least 10% of their tasks, and for 19%, at least half; higher-income occupations are more exposed.[^3] - **14.01.2024** The IMF: AI affects almost 40% of employment worldwide and about 60% in advanced economies.[^4] - **07.01.2025** The World Economic Forum publishes a survey of more than 1,000 employers: by 2030 they expect 170 million new jobs and 92 million displaced ones, a net gain of 78 million; on average 39% of a worker's skills will change or become outdated.[^5] - **10.02.2025** [Anthropic](https://ftrlg.com/en/models/anthropic/) releases its first Economic Index: [Claude](https://ftrlg.com/en/models/claude/) is used most often in software development and for writing; 36% of occupations use AI for at least a quarter of their tasks, and in 57% of cases it augments a person, while in 43% it does the work for them.[^6] - **17.06.2025** Amazon CEO Andy Jassy writes to employees that the company expects AI to reduce its corporate workforce in the next few years.[^7] - **26.08.2025** The Stanford Digital Economy Lab, in “Canaries in the Coal Mine?”: employment of 22–25-year-olds in the occupations most exposed to AI fell by 13% relative to the rest; experienced workers show no decline.[^8] - **01.10.2025** The Budget Lab at Yale: “the broader labor market has not experienced a discernible disruption since ChatGPT's release 33 months ago”.[^9] ## How it works AI changes not whole occupations but the tasks within them. That is why different approaches give different answers: 47% “at risk” if one counts whole occupations (US), and 9% if one counts tasks (21 OECD countries).[^1][^2] In every occupation, AI speeds up some tasks, can do some itself, and for now cannot do others at all. It is important to distinguish whether AI **augments** a person or **replaces** them. According to Anthropic, in chat Claude augments slightly more often, while through the API, where companies build the model into their own processes, it does the work itself in about three quarters of cases.[^12] Microsoft researchers, who compiled the most detailed list of occupations by AI applicability, warn: “It is tempting to conclude that occupations that have high AI action applicability score will be automated … This would be a mistake”: the consequences of new technologies are hard to predict.[^10] The impact comes not through layoffs but through hiring. Companies are less likely to take on newcomers for tasks that AI already performs, while keeping experienced employees: the model cannot yet replace their knowledge.[^11] ## Data ### Which occupations AI is closest to, and which it is furthest from | source | most affected or declining | least affected or growing | |---|---|---| | Microsoft Research, 22.12.2025[^10]: AI applicability to an occupation's tasks, based on 200,000 conversations with Bing Copilot; not a forecast of replacement | interpreters and translators, historians, writers and authors, sales representatives of services, CNC tool programmers, broadcast announcers and radio DJs, customer service representatives, telemarketers, political scientists, mathematicians | zero score: dredge operators, bridge and lock tenders, water treatment plant operators, foundry mold and coremakers, rail-track laying equipment operators, pile driver operators, floor sanders and finishers; close to them: orderlies, maids and housekeeping cleaners, roofers | | OpenAI and the University of Pennsylvania, 2023[^3]: share of tasks that a language model makes twice as fast | mathematicians, tax preparers, financial quantitative analysts, writers and authors, web and digital interface designers | no exposed tasks at all: athletes and sports competitors, short-order cooks, bus and truck mechanics and diesel engine specialists, motorcycle mechanics, dishwashers, stonemasons, electrical power-line installers | | World Economic Forum, 01.2025[^5]: employers' expectations through 2030; not only because of AI | fastest declining: postal service clerks, bank tellers, data entry clerks, cashiers, secretaries, bookkeeping clerks, graphic designers | fastest growing: big data specialists, fintech engineers, AI specialists, software developers; the largest gains in numbers: farmworkers, delivery drivers, nurses, personal care aides, teachers | ### Studies | study | what was measured | result | |---|---|---| | Stanford, “Canaries”, 12.08.2026[^11] | employment of young people in AI-exposed occupations, from ADP payroll data; a description, not proof of cause | for 22–25-year-olds, **19%** lower than if it had grown like that of their peers in other occupations, and the gap is widening; **no** mass displacement across the economy | | Anthropic, 15.01.2026[^12] | whether AI augments or does the work itself; company data | in Claude chat, augmentation **52%**, automation 45%; through the API, about three quarters automation | | OpenAI, GDPval, 05.10.2025[^13] | 1,320 tasks from 44 occupations: experts blindly compared the work of models and humans; a test by a model developer | the best model, Claude Opus 4.1, produced **47.6%** of deliverables as good as or better than the experts' | | Challenger, Gray & Christmas, 01.10.2026[^14] | job cuts announced by US employers, with the reason named by the companies themselves | since the start of 2026, AI has been named as the reason for **120,136** cuts, about 21% of all and more often than any other reason; for the whole of 2025, 54,836 | | Federal Reserve Bank of New York, Q2 2026[^15] | the labor market for recent US college graduates; causes are not assessed | unemployment about **5.6%**, share working in jobs that do not require a degree **42%** | > **[EDITORS' CONCLUSION]** The data do not yet show what people fear: there is no noticeable displacement across the economy as a whole, and models perform at expert level in roughly half of the test tasks. But pressure is already visible in two places: in the hiring of young people in exposed occupations, and in companies increasingly naming AI themselves as the reason for job cuts. First in line are tasks involving text, data and dealing with customers, while manual work and caring for people are the least affected. Caveats: the occupation lists measure AI applicability, not replacement; the 2030 forecasts are employers' expectations; announced job cuts are companies' words, not statistics. ## Criticism In 2024 MIT economist Daron Acemoglu calculated that AI will bring “no more than a 0.66% increase in total factor productivity (TFP) over 10 years”, and less than 0.53% after adjusting for hard tasks; by his estimate, about 4.6% of all tasks are profitable to automate.[^16] MIT economist David Autor, with the caveat “My thesis is not a forecast but an argument about what is possible”, writes that AI, used well, could give middle-skill workers access to decisions that today only experts make, and thereby rebuild the middle class.[^17] In May 2025 Anthropic CEO Dario Amodei, as paraphrased by Axios, warned that AI could wipe out half of all entry-level white-collar jobs and push unemployment up to 10–20% within one to five years. He himself described a possible world this way: “Cancer is cured, the economy grows at 10% a year, the budget is balanced — and 20% of people don't have jobs.”[^18] OpenAI CEO Sam Altman in his essay “The Gentle Singularity” (06.2025): “There will be very hard parts like whole classes of jobs going away, but on the other hand the world will be getting so much richer so quickly that we'll be able to seriously entertain new policy ideas we never could before.”[^19] On 16.06.2025 Geoffrey Hinton said that “for mundane intellectual labor, AI is just going to replace everybody”, while AI will be slow to master physical skills, so until humanoid robots show up, “a good bet would be to be a plumber”.[^20] ## Evolution On 07.05.2026 the Budget Lab at Yale repeated the check with a different method and again said that “we find no strong evidence of impacts as of yet” of AI on employment and wages, adding that “AI seems quite likely to eventually leave its mark on the labor market, even if it has not already”.[^21] According to the World Economic Forum survey, “if the world's workforce was made up of 100 people, 59 would need training by 2030”, and 11 of them are unlikely to receive it.[^5] On 25.06.2026 the US Department of Labor allocated about $50 million in grants for rapid reskilling of workers; the stated reason was explicitly “the broad and accelerating economic transformation driven by AI”.[^22] ## See also [Large language model](https://ftrlg.com/en/wiki/large-language-model/) · [Autonomous AI agents](https://ftrlg.com/en/wiki/autonomous-ai-agents/) · [Agentic engineering](https://ftrlg.com/en/wiki/agentic-engineering/) · [Vibe coding](https://ftrlg.com/en/wiki/vibe-coding/) · [AI safety](https://ftrlg.com/en/wiki/ai-safety/) ## References [^1]: C. B. Frey, M. A. Osborne. The Future of Employment: How Susceptible Are Jobs to Computerisation? — Oxford Martin School, 17.09.2013. [oxfordmartin.ox.ac.uk](https://www.oxfordmartin.ox.ac.uk/publications/the-future-of-employment) [^2]: M. Arntz, T. Gregory, U. Zierahn. The Risk of Automation for Jobs in OECD Countries: A Comparative Analysis — OECD Social, Employment and Migration Working Papers No. 189, 05.2016. [doi.org](https://doi.org/10.1787/5jlz9h56dvq7-en) [^3]: T. Eloundou, S. Manning, P. Mishkin, D. Rock. GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models — arXiv, 17.03.2023. [arxiv.org](https://arxiv.org/abs/2303.10130) [^4]: Kristalina Georgieva. AI Will Transform the Global Economy. Let's Make Sure It Benefits Humanity — IMF Blog, 14.01.2024. [imf.org](https://www.imf.org/en/Blogs/Articles/2024/01/14/ai-will-transform-the-global-economy-lets-make-sure-it-benefits-humanity) [^5]: Future of Jobs Report 2025 — World Economic Forum, 01.2025. [weforum.org](https://reports.weforum.org/docs/WEF_Future_of_Jobs_Report_2025.pdf) [^6]: The Anthropic Economic Index — Anthropic, 10.02.2025. [anthropic.com](https://www.anthropic.com/news/the-anthropic-economic-index) [^7]: Message from CEO Andy Jassy: Some thoughts on Generative AI — Amazon, 17.06.2025. [aboutamazon.com](https://www.aboutamazon.com/news/company-news/amazon-ceo-andy-jassy-on-generative-ai) [^8]: E. Brynjolfsson, B. Chandar, R. Chen. Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence — Stanford Digital Economy Lab, 26.08.2025. [digitaleconomy.stanford.edu](https://digitaleconomy.stanford.edu/publications/canaries-in-the-coal-mine/) [^9]: M. Gimbel, M. Kinder, J. Kendall, M. Lee. Evaluating the Impact of AI on the Labor Market: Current State of Affairs — The Budget Lab at Yale, 01.10.2025. [budgetlab.yale.edu](https://budgetlab.yale.edu/research/evaluating-impact-ai-labor-market-current-state-affairs) [^10]: K. Tomlinson, S. Jaffe, W. Wang, S. Counts, S. Suri. Working with AI: Measuring the Applicability of Generative AI to Occupations — arXiv, v6, 22.12.2025. [arxiv.org](https://arxiv.org/abs/2507.07935) [^11]: E. Brynjolfsson, B. Chandar, R. Chen. Canaries in the Coal Mine? (update, data through June 2026) — Stanford Digital Economy Lab, 12.08.2026. [digitaleconomy.stanford.edu](https://digitaleconomy.stanford.edu/app/uploads/2026/08/Canaries_August2026.pdf) [^12]: Anthropic Economic Index report: Economic primitives — Anthropic, 15.01.2026. [anthropic.com](https://www.anthropic.com/research/anthropic-economic-index-january-2026-report) [^13]: T. Patwardhan et al. GDPval: Evaluating AI Model Performance on Real-World Economically Valuable Tasks — arXiv, 05.10.2025. [arxiv.org](https://arxiv.org/abs/2510.04374) [^14]: Challenger Report: September 2026 — Challenger, Gray & Christmas, 01.10.2026. [challengergray.com](https://www.challengergray.com/wp-content/uploads/2026/10/Challenger-Report-September-2026.pdf) [^15]: The Labor Market for Recent College Graduates — Federal Reserve Bank of New York, data for Q2 2026. [newyorkfed.org](https://www.newyorkfed.org/research/college-labor-market) [^16]: Daron Acemoglu. The Simple Macroeconomics of AI — NBER Working Paper 32487, 05.2024. [nber.org](https://www.nber.org/papers/w32487) [^17]: David Autor. Applying AI to Rebuild Middle Class Jobs — NBER Working Paper 32140, 02.2024. [nber.org](https://www.nber.org/papers/w32140) [^18]: Jim VandeHei, Mike Allen. Behind the Curtain: A white-collar bloodbath — Axios, 28.05.2025. [axios.com](https://www.axios.com/2025/05/28/ai-jobs-white-collar-unemployment-anthropic) [^19]: Sam Altman. The Gentle Singularity — 06.2025. [blog.samaltman.com](https://blog.samaltman.com/the-gentle-singularity) [^20]: Godfather of AI: They Keep Silencing Me But I'm Trying to Warn Them! — The Diary of a CEO, 16.06.2025. [youtube.com](https://www.youtube.com/watch?v=giT0ytynSqg) [^21]: Ryan Nunn. AI Is Probably Not (Yet) the Reason for Labor Market Weakening — The Budget Lab at Yale, 07.05.2026. [budgetlab.yale.edu](https://budgetlab.yale.edu/research/ai-probably-not-yet-reason-labor-market-weakening) [^22]: Rapid Reskill Employment Recovery National Dislocated Worker Grants (TEGL 15-25) — U.S. Department of Labor, 25.06.2026. [dol.gov](https://www.dol.gov/agencies/eta/advisories/tegl-15-25) — future.log, «AI and jobs», https://ftrlg.com/en/wiki/ai-and-jobs/ · CC BY 4.0