Jobs of the future: what the 2030 data actually shows
The WEF growth and decline lists with their grades, the method behind every job forecast and why the same question yields 47% or 9%, the ILO exposure numbers kept apart from displacement, our three scenarios with radar scores and bands, and the skills that hold under all of them.
How job forecasts are made, and why the method matters
Every list of "jobs of the future" rests on one of three methods, and the method decides what the numbers can and cannot tell you.
| Method | Example | What it measures | Typical error |
|---|---|---|---|
| Employer survey, extrapolated | WEF Future of Jobs 2025: executives from more than 1,000 large employers asked what they expect by 2030[1] [2] | Expectations of large firms, weighted to the global workforce | Employers are optimistic about hiring and vague about cuts; small firms and the informal economy are not in the sample |
| Occupation-level automation probability | Frey and Osborne: experts labelled 70 occupations as automatable or not, a classifier extended the labels to 702, and 47% of US employment landed in the "high risk" band[3] | Technical feasibility of automating a whole occupation | Treats an occupation as one task; the 47% was widely reported as a job-loss forecast, which the paper did not claim |
| Task-level exposure | Arntz, Gregory and Zierahn repeated the exercise at task level across 21 OECD countries and found 9% of jobs at high risk[4]; the ILO's 2025 index scores occupations by the share of tasks generative AI can perform[5] | Which tasks in a job could be done by the technology | Exposure is not displacement; it says nothing about demand, prices, regulation or what the freed time is used for |
The jump from 47% to 9% on the same question is the most useful fact on this page. Neither team was wrong; "occupation" and "task" are different units, and the headline depends on the unit. Keep that in mind for every list below, including ours, which sits at the end in the forecast section, separate from the measured data.
Interest is cyclical. "jobs of the future" draws about 3,600 US searches a month, down 33% year on year but with a three-year compound growth rate of 17%; "future of work ai" (390 a month) fell 47% in a year while "will ai replace my job" (210) rose 91% (DataForSEO, retrieved 2026-11-09). The vocabulary is shifting from the abstract to the personal, and the personal question has its own page: Will AI replace my job?
Growing roles: what the WEF survey actually says
The WEF report projects 170 million jobs created and 92 million displaced by 2030, a net gain of 78 million, with the churn touching 22% of today's jobs.[1] A Those are survey extrapolations, not observed counts, and they describe the roles the surveyed employers expect to need.
Two lists come out of the report and they are routinely confused. The fastest-growing roles by percentage are almost all technical. The largest-growing roles by headcount are almost all not.
| Fastest-growing by percentage, 2025–2030 | Largest net growth in headcount |
|---|---|
| Big data specialists (+113%) | Farmworkers, labourers and other agricultural workers (+34.9 million net: 49 million created, 14.1 million lost) |
| FinTech engineers (+93%) | Light truck and delivery drivers (+9.8 million) |
| AI and machine-learning specialists (+82%) | Software and applications developers (+6.7 million) |
| Software and applications developers (+57%) | Shop salespersons (+4.3 million net) |
| Security management specialists (+53%) | Food processing and related trades workers (+4.3 million) |
| Data warehousing specialists (+49%) | Car, van and motorcycle drivers (+4.1 million) |
| Autonomous and electric vehicle specialists (+48%) | Nursing professionals (+3.1 million) |
| UI and UX designers (+48%) | Food and beverage serving workers (+2.8 million) |
| Light truck or delivery services drivers (+46%) | University and higher-education teachers (+1.9 million) |
| Internet of Things specialists (+42%) | Personal care workers (+1.6 million) |
| Data analysts and scientists (+41%) | Secondary education teachers (+1.6 million) |
Source: WEF Future of Jobs Report 2025 as summarised by Euronews, 1 February 2025.[2] B Farming alone accounts for 35 million of the 78 million net gain, about 45%. The report's growth is driven by demographics, the green transition and the cost of living at least as much as by technology, and most of the jobs it expects to add are physical, local and poorly paid relative to the percentage list. A list that shows only the left column describes the labour market of a few hundred thousand specialists.
Declining roles
The declining list is where the survey is most consistent with the task-exposure research: clerical work built on structured data entry and retrieval.
| Fastest-declining by percentage | Largest net loss in headcount |
|---|---|
| Postal service clerks (−34%) | Cashiers and ticket clerks (−13.7 million net; 16.3 million displaced) |
| Bank tellers and related clerks (−31%) | Administrative assistants and executive secretaries (−6.1 million) |
| Data entry clerks (−26%) | Material-recording and stock-keeping clerks (−2.6 million) |
| Cashiers and ticket clerks (−20%) | Accounting, bookkeeping and payroll clerks (−1.7 million) |
| Administrative assistants and executive secretaries (−20%) | Data entry clerks (−0.5 million) |
| Printing and related trades workers (−20%) | Graphic designers (−0.4 million) |
| Accounting, bookkeeping and payroll clerks (−18%) | |
| Material-recording and stock-keeping clerks (−16%) | |
| Door-to-door sales workers, news and street vendors (−14%) |
Source: as above.[2] B Cashiers and clerks were declining before generative AI, through self-checkout, online banking and enterprise software. Graphic designers are the one creative occupation with a projected net loss, and it is small. No analyst, lawyer, manager, teacher or developer role appears among the fastest decliners, even though those occupations have the highest generative-AI exposure in the ILO index. That gap between exposure and expected decline is the subject of the next section.
Exposure is not displacement: the ILO numbers
The ILO's refined index scores occupations by how many of their tasks generative AI can plausibly perform, then grades them. Globally, one in four workers is in an occupation with some exposure; 3.3% of global employment falls in the highest exposure category.[5] [6] A Exposure rises with national income: 34% of total employment in high-income countries against 11% in low-income countries.[7] A Women's employment is roughly twice as concentrated in the top category as men's, 4.7% against 2.4%, because clerical and administrative work is both highly exposed and disproportionately female.[6] A
The ILO is careful about what the top band means: tasks with high automation potential, not jobs that will be eliminated. The controlled studies explain why the distinction holds. In a quasi-experiment across 5,172 customer-support agents, a generative-AI assistant raised issues resolved per hour by about 14–15% on average, over 30% for the least experienced, close to zero for the most experienced.[8] A In a preregistered experiment with 453 professionals, ChatGPT cut time on writing tasks and compressed the quality gap between weaker and stronger writers.[9] A Among 758 consultants, GPT-4 improved speed and quality on tasks inside the model's competence and made people 19 percentage points less likely to be correct on a task just outside it.[10] B
So "exposed" knowledge work is, so far, augmented work: the tool does part of the task, the novice gains most, and the expert's job shifts toward checking. PwC's analysis of close to a billion job postings found that occupations more exposed to AI still grew in number between 2019 and 2024, by 38% against 65% for less-exposed occupations, and that advertised wages for roles asking for AI skills carried a 56% premium in 2024.[11] A Slower growth, not decline, is the observed pattern for exposed occupations to date. Whether it stays that way as agents take over whole workflows rather than single tasks is exactly what our scenarios differ on.
Transformed roles: most knowledge work
The WEF figure that applies to the most readers is not about jobs at all: on average, 39% of workers' current skill sets are expected to be transformed or become obsolete by 2030.[2] B Of every 100 workers, 59 are expected to need reskilling or upskilling by then, and 11 of them are unlikely to get it, which the report translates into more than 120 million workers at medium-term redundancy risk.[1] A
Read together with the ILO exposure gradient, the picture for an analyst, a marketer, a lawyer or a developer is not a vanishing job but a job whose task mix changes: fewer first drafts, more briefing, checking, judgement and coordination. Our three scenarios for knowledge work to 2035 describe what that looks like in each case, with the signposts we track each quarter.
The demographic backdrop pulls the other way from the automation story. The OECD projects the working-age population to fall 8% across its members by 2060, by more than 30% in a quarter of them, with the ratio of older people to the working-age population rising from 31% in 2023 to 52% in 2060.[12] B Labour shortage, not surplus, is the base case for most rich countries, which is one reason employers keep forecasting net job growth while exposure indices rise.
Our three scenarios, with the radar numbers
Our Future Demand Radar scores twelve productivity markets to 2035 under three scenarios. The relevant category here is "future-of-work skills and AI literacy", which bundles the demand behind this page and its neighbours. The figures are model outputs with stated uncertainty, not facts.
| Scenario | Assumption about work | Radar score (p10–p90) | Class | US demand multiple, 2030 (p10–p90) |
|---|---|---|---|---|
| A, conservative | AI adoption gradual; copilots stay features inside existing software; job change tracks the WEF survey | 42.5 (38.0–46.9) | Monitor | 1.4× (1.07–1.81) |
| B, baseline | Copilots and delegated agents routine in knowledge work by 2028–2030; new demand around briefing, verifying and orchestrating; the AI Act literacy duty sustains training demand | 45.8 (41.4–50.2) | Monitor | 2.2× (1.69–2.83) |
| C, accelerated | Capable agents absorb much routine knowledge work; the individual's value is judgement, orchestration and recovery | 48.2 (43.8–52.5) | Validate in 12 months | 3.0× (2.31–3.86) |
Our baseline scenario expects the roles that grow for knowledge workers to be the ones that sit between a goal and a set of agents: specifying work, checking it, owning the outcome and coordinating the people the outcome affects. Under scenario C the same roles grow faster and the first-draft layer of most professions thins out within the decade; under scenario A the WEF table above is roughly what 2030 looks like. The category's 12-month trend is slightly negative while its three-year growth is strong, which is why none of the scenarios reaches the "build now" threshold of 55 and all of them say: check again in a year.
Durable skills: what survives all three scenarios
If the job titles are uncertain, the skills that hold value across the scenarios are less so, because they are the ones the controlled studies keep pointing at. The durability rating is our editorial judgement informed by the evidence cited.
| Skill | Why it survives | Evidence | Where to build it |
|---|---|---|---|
| Judgement about the frontier: knowing which tasks a model handles and which it does not | The losses in the studies come from tasks delegated on the wrong side of the line[10] | B | Human skills in an AI economy |
| Verification and acceptance testing | Novices gain most and check least; the premium moves to whoever catches the error[8] [10] | A | Our ranked list of AI skills for 2030 |
| Specification and briefing | The management skill that replaces prompting as agents take whole workflows | Editorial, from the agent literature | The personal AI operating system builder, which turns your task list into briefs |
| Domain depth | Tools raise the floor; the ceiling is set by knowing what good looks like[9] | A | Your field; apply AI where you know least, check with what you know best |
| Care, coordination and trust | The headcount growth list is nursing, teaching, personal care and sales, which are relational before they are technical[2] | B | Human skills in an AI economy |
| Learning method itself | 39% of skills transformed by 2030 means the half-life of any one skill is short[2] | B | Spaced practice on real work rather than courses; see the learning plan on the AI skills page |
How to use this for your own plan
- Find the unit. Ask whether a claim about your job concerns the occupation (Frey and Osborne), the tasks (ILO, Arntz) or employers' hiring intentions (WEF). A 47% and a 9% can both be true.
- Map your tasks, not your title. List your twenty most common tasks and mark which a current assistant does acceptably, which it does with checking, and which it cannot. The middle column is your job in 2030 under our baseline scenario.
- Weight the headcount list. More people will be nurses and delivery drivers than fintech engineers, and those roles have wage and condition problems that technology does not solve.
- Watch three signposts. Agent adoption in your sector, the ILO's next exposure update, and whether exposed occupations keep growing in job postings. If they start shrinking in absolute terms, move your planning from scenario B to C.
- Invest in the durable table. Every row holds under all three scenarios; most of the percentage-growth list holds under one.
What to do next
Open the Future Demand Radar to see the twelve categories side by side with their bands and signposts, then decide which scenario you are planning against. The personal AI operating system builder turns the task map from step 2 into a working setup: which tasks to delegate, which to verify, which to keep. We revise this page when the WEF, ILO and PwC publish their next editions; subscribe below for the update.
Frequently asked questions
What are the fastest-growing jobs by 2030?
By percentage, the WEF Future of Jobs Report 2025 lists big data specialists (+113%), fintech engineers (+93%), AI and machine-learning specialists (+82%), software developers (+57%) and security management specialists (+53%). By headcount the picture is different: farmworkers (+34.9 million net), delivery drivers (+9.8 million), software developers (+6.7 million), shop salespersons and food processing workers (+4.3 million each) and nurses (+3.1 million). Both lists are employer-survey extrapolations, not observed counts.
Which jobs are declining fastest?
In the same report: postal service clerks (−34%), bank tellers (−31%), data entry clerks (−26%), cashiers and ticket clerks (−20%), administrative assistants (−20%), printing trades (−20%), accounting and payroll clerks (−18%) and stock-keeping clerks (−16%). By headcount, cashiers and ticket clerks lose most (−13.7 million net). Most of these declines predate generative AI.
Will AI eliminate 47% of jobs?
No study says that. Frey and Osborne (2017) estimated that 47% of US employment was in occupations whose tasks were technically automatable at high probability; Arntz, Gregory and Zierahn (2016) redid the analysis at task level and found 9% of jobs at high risk across 21 OECD countries. Exposure to automation is not a forecast of job loss. The ILO (2025) finds one in four workers in an occupation with some generative-AI exposure and 3.3% in the highest exposure band.
Are knowledge jobs safe from AI?
Knowledge jobs are the most exposed in the ILO index and, so far, the least displaced. Controlled studies find AI assistance raises output most for less experienced workers and shifts experienced workers toward checking; PwC's job-posting data shows exposed occupations still growing between 2019 and 2024, though more slowly than less-exposed ones. Our baseline scenario expects transformed jobs rather than vanished ones; under our accelerated scenario the first-draft layer of many professions thins out within the decade.
What skills will matter most for future jobs?
Across all three of our scenarios: judgement about which tasks a model handles, verification of AI output, specification and briefing, domain depth, and relational skills such as care, coordination and trust. The WEF expects 39% of current skill sets to be transformed by 2030, which makes the ability to learn a new skill quickly a skill in itself. Our AI skills for 2030 page ranks the technical ones by durability.
Sources and evidence
- Future of Jobs Report 2025 – press release (French edition, 8 January 2025) — World Economic Forum (2025), report B reports.weforum.org/docs/WEF_Future_of_Jobs_2025_Press_Release_FR.pdf (retrieved 2026-11-09)
- Jobs market at a crossroads: which are the fastest growing and declining jobs? (summarising WEF Future of Jobs Report 2025) — Yanatma, Servet (Euronews) (2025), report B www.euronews.com/business/2025/02/01/jobs-market-at-a-crossroads-which (retrieved 2026-11-09)
- The future of employment: How susceptible are jobs to computerisation? — Frey, Carl Benedikt; Osborne, Michael A. (2017), observational B doi.org/10.1016/j.techfore.2016.08.019 (retrieved 2026-11-09)
- The Risk of Automation for Jobs in OECD Countries: A Comparative Analysis (OECD Social, Employment and Migration Working Papers No. 189) — Arntz, Melanie; Gregory, Terry; Zierahn, Ulrich (2016), report A doi.org/10.1787/5jlz9h56dvq7-en (retrieved 2026-11-09)
- Generative AI and Jobs: A Refined Global Index of Occupational Exposure (ILO Working Paper 140) — International Labour Organization (2025), report A www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-o (retrieved 2026-11-09)
- Generative AI and Jobs: A Refined Global Index of Occupational Exposure – highest exposure category and gender gap — International Labour Organization (2025), report A www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-o (retrieved 2026-11-09)
- Generative AI and Jobs: A Refined Global Index of Occupational Exposure – exposure by country income group — International Labour Organization (2025), report A www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-o (retrieved 2026-11-09)
- Generative AI at Work — Brynjolfsson, Erik; Li, Danielle; Raymond, Lindsey (2025), observational A doi.org/10.1093/qje/qjae044 (retrieved 2026-11-09)
- Experimental evidence on the productivity effects of generative artificial intelligence — Noy, Shakked; Zhang, Whitney (2023), RCT A doi.org/10.1126/science.adh2586 (retrieved 2026-11-09)
- Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality — Dell'Acqua, Fabrizio; McFowland III, Edward; Mollick, Ethan R.; Lifshitz-Assaf, Hila; Kellogg, Katherine; Rajendran, Saran; Krayer, Lisa; Candelon, François (2023), RCT B doi.org/10.2139/ssrn.4573321 (retrieved 2026-11-09)
- 2025 Global AI Jobs Barometer (press release: AI linked to a fourfold increase in productivity growth) — PwC (2025), report B www.pwc.com/gx/en/news-room/press-releases/2025/ai-linked-to-a-fourfol (retrieved 2026-11-09)
- OECD Employment Outlook 2025 – press release: OECD job markets remain resilient but population ageing will cause significant labour shortages and fiscal pressures (9 July 2025, reprint) — OECD (2025), report B www.hellenicshippingnews.com/?p=1096804 (retrieved 2026-11-09)