Skip to content

Will AI Replace You? An Honest Look at 10 Different Industries

A claim-by-claim check of the AI job headlines against BLS projections for 2025 to 2035, Stanford and Challenger data, and the WEF survey, across ten industries, plus a 15-minute way to test your own role.

Every few weeks a number goes around: 80% of customer service jobs gone, 7.5 million clerks replaced. Many of those figures circulate without a primary source. If you are wondering whether your own job is next, you need three kinds of evidence, and they disagree in useful ways: what AI is technically able to do, what the government projects for each occupation, and what hiring and layoff data show so far. This page lines up the October 2026 versions of all three across ten industries, then gives you a 15-minute way to check your own role.

Three Measures, Three Different Answers

Exposure is not job loss. In November 2025, an MIT study found that current AI could already perform tasks accounting for about 12% of the US workforce (headlines often wrote “replace”). That measures technical capability, not what employers have done. The Bureau of Labor Statistics now rates 831 occupations from Low to Very high AI exposure and warns that “exposure does not imply job loss, productivity gains, automation probability, or wage effects.” Goldman Sachs’s famous 300 million figure carries the same limit: it counted full-time-job equivalents exposed to automation, with roughly two-thirds of US and European jobs exposed to some degree.

Projections are about jobs, not tasks. BLS published its 2025 to 2035 projections on August 27, 2026. Total employment is expected to grow 3.5%, from 170.3 million to 176.2 million, slower than the 10.9% of the past decade. Office and administrative support has the steepest fall of any major group, 4.0% or 752,100 jobs, while healthcare support is the fastest-growing group at 13.3% and healthcare practitioners and technical occupations grow 8.0%. BLS adds that “the uncertainty about potential impacts remains very high.”

The early signal is at the entry level. Stanford’s Digital Economy Lab updated its “canaries” study on August 12, 2026 using ADP payroll data. Employment of 22 to 25 year-olds in the two most AI-exposed quintiles of occupations fell about 11% between November 2022 and June 2026, while the three least-exposed quintiles grew about 10%. That puts them about 19% below where they would be, up from 15% in the July 2025 data. The same post says experienced workers show no comparable gap, that the authors “do not see widespread, economy-wide job displacement,” and that these are descriptive patterns, not causal estimates.

Layoff announcements add a fourth angle, with a warning label. Challenger, Gray & Christmas counted 573,195 announced US job cuts through September 2026, down 39% from the same period of 2025, and AI was cited in 120,136 of them, about 21%, the leading reason so far this year. Those are reasons cited in announcements, not audited causes. Oxford Economics, cited by Fortune, suspects some firms “dress up layoffs” as AI efficiency, and Yale’s Budget Lab, in the same February 2026 coverage, found “stability, not major disruption at an economy-wide level.”

Ten Industries Against the Government Numbers

BLS projects occupations, not industries, so each row below uses one representative job. All figures are the Occupational Outlook Handbook’s 2025 to 2035 projections, and the last column is what BLS itself gives as the reason.

IndustryJob checkedChange, 2025 to 2035Reason BLS gives
ManufacturingAssemblers and fabricators+1% (1,799,500 to 1,816,200)Automation and robotics reduce job growth
RetailCashiers-6% (3,106,300 to 2,905,700)Self-service checkout, online sales
Customer serviceCustomer service representatives-5% (2,666,000 to 2,524,100)Tasks continue to be automated
LegalParalegals0% (404,900 to 403,800)Demand limited by technology “including artificial intelligence”
FinanceTellers-13% (339,200 to 294,500)Online banking, kiosks, enhanced ATMs
AdministrativeData entry keyers-25.5% (from 131,800)Fourth on BLS’s fastest-declining list; office support overall is -4.0%
HealthcareMedical transcriptionists-4% (42,000 to 40,200)Physicians document encounters in real time
HealthcareRegistered nurses+6% (3,465,400 to 3,660,100)AI not mentioned in the outlook
EducationElementary school teachers0% (1,530,500 to 1,524,400)Falling public school enrollment; AI not mentioned
CreativeWriters and authors0% (140,300 to 139,900)AI use for writing “projected to dampen demand”
CreativeGraphic designers-2% (253,100 to 248,800)Automated design tools, such as AI, raise productivity
TransportationHeavy truck drivers+4% (2,221,200 to 2,305,000)Freight demand; driverless trucks not mentioned

Read the last column before the numbers. Only three of these outlooks name AI directly (paralegals, writers, graphic designers). The cashier, teller and customer service declines come from self-service, online banking and general automation, technologies that predate chatbots. And two of the largest groups, teachers and truck drivers, are moving for reasons BLS ties to enrollment and freight. Customer service is the cleanest test of the “80% gone” claim: BLS projects a 5% fall over ten years, about 141,800 jobs, not 80%.

The companies that moved fastest show why. In February 2024 Klarna said its AI assistant did the work of 700 customer service agents, handling about 75% of chats. In May 2025 its CEO told Bloomberg that chatbots had produced “lower quality” and that the company was hiring human agents again, because “there will always be a human if you want.” Healthcare shows the same split between task and job: transcription is shrinking, yet nurse practitioners (+41.0%) lead the BLS fastest-growing list, with data scientists (+34.6%) third. Robotaxis are real but still small next to the roughly 2.2 million truck-driving jobs: Waymo reported about 500,000 paid rides a week across 10 US markets earlier in 2026.

What Employers Say They Will Do

The World Economic Forum’s Future of Jobs Report 2025, a survey of employers, is the source of the “170 million created, 92 million displaced, net 78 million by 2030” figures. Those numbers cover all labour-market trends together (technology, climate, demographics and more), not AI alone, so they are not an AI job-creation forecast. The AI-specific findings are narrower. In the same survey, 77% of employers plan to reskill staff to work alongside AI, and 41% also expect to downsize where AI can replicate people’s work. The report expects clerical and secretarial workers to see the largest absolute decline, and lists postal clerks, bank tellers and data entry clerks among the fastest-declining roles.

Check Your Own Job in 15 Minutes

Try this yourself with a pen and last week’s calendar. First, list what you did, task by task, with hours. Second, mark each task as type A (turning structured inputs into outputs by rules, such as typing in data, transcribing, answering routine questions, producing first drafts), type B (judgment about people, exceptions or accountability), or type C (physical presence). Third, look up your occupation on BLS, read its job outlook and why it gives that outlook, and compare with your split.

A rough decision rule, which is a heuristic and not a forecast: if more than half your hours are type A and the buyer treats the output as interchangeable, your role looks like the declining rows above, and the move is to shift hours toward type B before your employer decides for you. If type A is under a third and you answer for results, expect tools to take the type A hours while you absorb more review and exception work. In between, the answer depends on whether your employer sells the output or the judgment.

This does not work when the cuts are not about AI. Layoffs announced as “AI efficiency” can be about over-hiring or costs, as the Oxford Economics comment above suggests, and no task audit predicts that. It also says less to people who are early in their careers, because the Stanford data shows the pressure on hiring into entry-level jobs rather than on people already in them. If that is you, the practical response is to get a type B task early, such as owning a client’s follow-ups or checking AI output against the source.

The one thing to do this week: write the task list and total the type A hours. If it is over half, pick one type B task and ask your manager to hand it to you.

What Each Measure Tells You
Exposure
What AI could touch. About 12% of the US workforce in the MIT study. Not a layoff forecast.
Projection
Where jobs are headed to 2035. Office support down 4.0%, healthcare up. BLS calls uncertainty very high.
Early signal
Ages 22 to 25 in exposed jobs: about 11% fewer since late 2022. Older workers show no gap.

Frequently Asked Questions

Which jobs are shrinking fastest according to the government?

In the BLS 2025 to 2035 ranking, word processors and typists (-34.4%), telephone operators (-27.6%), switchboard operators (-26.0%) and data entry keyers (-25.5%) fall fastest, followed by telemarketers (-21.4%), order clerks (-17.5%) and payroll and timekeeping clerks (-15.9%). Several of these are small occupations, so the percentages look larger than the job counts. BLS ranks the changes but does not blame AI for each one.

What skills should I build to stay employable?

The WEF survey ranks AI and big data, networks and cybersecurity, and technological literacy as the three fastest-growing skills, and 77% of employers plan reskilling. The Stanford data suggests experience matters: workers with experience show no gap, while 22 to 25 year-olds in exposed jobs do. That points to building judgment tasks early, such as handling exceptions, owning client relationships and verifying AI output, rather than only learning tools.

Is AI creating enough jobs to replace the ones it removes?

No one can say yet. BLS expects total US employment to grow 3.5% to 2035, with data scientists up 34.6% and nurse practitioners up 41.0%. The WEF’s net 78 million globally by 2030 reflects all trends, not AI alone. New jobs and lost jobs also differ in skills and location, so a net gain on paper can still mean a hard transition for the person whose role shrinks.

Are young graduates being hit harder?

The data points that way for those 22 to 25 in highly exposed occupations (about 19% below trend, per Stanford’s August 2026 update), but the authors stress it is descriptive, not proof of cause. They note the gaps shrink when education is accounted for and that some differences appear before generative AI was widely used.

Leave a comment