Almost 40% of global employment is exposed to AI according to the International Monetary Fund, yet only roughly 13% of US job-cut plans in early 2026 actually cited artificial intelligence as the reason, per Challenger, Gray & Christmas. That gap shapes the answer to how AI replaces jobs. The sections below walk through the mechanism and map it across the industries where substitution is moving fastest, drawing on data from the World Economic Forum, Goldman Sachs Research, the IMF, the U.S. Bureau of Labor Statistics, Eurostat and the Reuters Institute.
Key Takeaways
- The IMF estimates that almost 40% of global employment is exposed to AI, with the share rising to about 60% in advanced economies.
- Goldman Sachs Research finds that AI could expose the equivalent of 300 million full-time jobs to automation worldwide and automate tasks accounting for 25% of all US work hours.
- The World Economic Forum’s Future of Jobs Report 2025 projects 92 million jobs displaced and 170 million created over the 2025-2030 period, a net gain of 78 million roles.
- In March 2026, Challenger, Gray & Christmas counted 15,341 AI-cited US layoffs, 25% of total cuts that month and the leading single reason.
- US office and administrative support has the highest task-automation share at 46%, followed by legal at 44% and architecture and engineering at 37%.
- Clerical and secretarial workers, including cashiers, ticket clerks, administrative assistants, postal clerks, bank tellers and data entry clerks, are projected to see the largest absolute decline over the 2025-2030 period.
How Does AI Replace Jobs at the Task Level?
Goldman Sachs Research estimates that generative AI could expose the equivalent of 300 million full-time jobs to automation and automate tasks accounting for 25% of all work hours in the United States. Goldman’s task-share breakdown shows US office and administrative support jobs at 46% automatable, legal work at 44%, architecture and engineering at 37%, life, physical and social sciences at 36%, and business and financial operations at 35%, while construction sits at 6% and installation, maintenance and repair at 4%.
Most jobs and industries are only partially exposed to automation and are more likely to be complemented rather than substituted by AI, according to the same Goldman analysis.
By the numbers: According to Goldman Sachs Research, AI could automate tasks accounting for 25% of US work hours, with office and administrative support exposed at 46% and legal work at 44%. Most jobs are only partially exposed, meaning AI tends to compress task bundles rather than erase whole roles outright.
Computer vision and robotics show how this plays out on the factory floor.
Manufacturing: Where Computer Vision and Robotics Hit First
Manufacturing-adjacent occupations, according to Goldman Sachs Research, sit at the lower end of the exposure scale, with installation roles at 4% task automation and construction at 6%. Physical tasks remain expensive to automate, so AI shows up first in the information layer around production lines.
The substitution pattern in manufacturing typically looks like this:
- Computer-vision quality inspection replaces visual inspection roles on standardized parts.
- Predictive-maintenance models reduce reactive-repair headcount but require new technician profiles to operate them.
- AI-driven demand forecasting compresses planner and scheduler teams.
- Generative-design tools shrink the engineering hours per product without eliminating engineering roles outright.
The World Economic Forum’s Future of Jobs Report 2025, according to that body’s data, projects that information-processing technologies, including AI, will transform 86% of businesses over the 2025-2030 period. Adoption data per Eurostat: 13.5% of EU enterprises with at least 10 employees used AI technologies in 2024, up from 8.0% in 2023, with 41.2% of enterprises with 250 or more employees using AI compared with 11.2% of small enterprises. Larger context on enterprise adoption sits in our Microsoft 365 statistics data, where Copilot rollouts are an early example of the same pattern.
Customer Service and the Chatbot Deflection Wave
World Economic Forum analysis of the Future of Jobs Report 2025 finds that telephone operators, insurance claims clerks, and bill collectors face the highest risk of substitution by AI, because their work involves structured, repetitive data tasks that current language and vision models execute well. Conversational agents handle L1 triage that historically supported large call centres, leaving human agents on harder cases.
The Future of Jobs Report 2025 also found that 41% of employers worldwide intend to reduce their workforce as AI automates certain tasks. In Goldman Sachs’s task framework, business and financial operations, which include many customer-facing administrative functions, sit at 35% task automation.
Finance and Banking: Back-Office and Entry-Level Analyst Roles
The World Economic Forum’s Future of Jobs Report 2025 names bank tellers and data entry clerks among the fastest-declining roles over the 2025-2030 period, alongside cashiers, ticket clerks, postal service clerks, and administrative assistants. Challenger, Gray & Christmas reports that financial institutions cut 9,397 jobs in the first quarter of 2026, down 41% from the 15,982 cuts during the same period in 2025.
The mechanism inside banks looks like this:
- Entry-level analyst tasks (pulling data, drafting initial memos, reconciling reports) are increasingly handled by copilot tools.
- Compliance review and KYC onboarding rely on document-processing models that reduce the analyst hours per file.
- Algorithmic trading and risk modelling never employed many people to begin with, so AI’s effect there is incremental.
- Branch-level teller demand keeps falling as customers shift to mobile, with AI playing a supporting rather than leading role.
The same pattern is now reaching banking back offices in Europe.
Retail: Self-Checkout, Inventory Bots, and Demand Forecasting
Cashiers and ticket clerks are among the clerical roles WEF projects to see the largest absolute decline over the 2025-2030 period. Retail combines self-checkout in-store with supply-chain forecasting behind the scenes, which quietly reduces back-office headcount.
Retail support functions overlap heavily with the office and administrative support category that Goldman Sachs Research classifies as 46% task-automatable, the highest single bucket in its analysis.
Legal: Contract Review and E-Discovery
Goldman Sachs Research places legal work at 44% task automation, second only to office and administrative support, because contract review, document discovery and routine drafting overlap heavily with what large language models execute well. At the same time, the U.S. Bureau of Labor Statistics projects that employment of paralegals and legal assistants will grow about 1% from 2024 to 2034, slower than the average for all occupations, with about 39,300 annual openings projected each year and 367,220 paralegals and legal assistants currently employed at a median annual wage of $61,010 as of May 2024. The pair captures the task-versus-job distinction: 44% of legal tasks are automatable, but paralegal headcount is flat to slightly growing because legal demand keeps climbing.
Key finding: The IMF estimates that about 60% of jobs in advanced economies are exposed to AI, with roughly half likely to benefit through productivity gains and the other half facing reduced labour demand. Legal work, at 44% task automation per Goldman Sachs, sits squarely in the second bracket for routine contract and discovery functions.
Healthcare: Administrative Roles vs. Clinical Roles
The IMF analysis published in January 2024 found that almost 40% of global employment is exposed to AI, rising to about 60% in advanced economies, with roughly half of exposed jobs benefiting from productivity gains and the other half facing the prospect of lower labour demand and reduced hiring. Challenger, Gray & Christmas reports that healthcare companies and health-product manufacturers announced 23,520 job cuts in the first quarter of 2026, the highest Q1 total on record for the sector, surpassing the previous high of 22,950 set in 2023. Healthcare’s Q1 record reflects reimbursement pressure more than direct AI substitution; AI’s footprint sits in coding and claims, while clinical roles stay insulated.
Media and Journalism: AI Summarization and the Search-Referral Collapse
Press Gazette tracking shows more than 500 journalism jobs were eliminated in the first three months of 2026, building on more than 3,400 journalism job cuts across the U.S. and U.K. during 2025, with the Washington Post alone cutting more than 300 of its roughly 800 journalists. In a January 2026 Reuters Institute survey of 326 senior media leaders across 51 countries, two-thirds (67%) of newsrooms said they have not saved any jobs so far as a result of AI efficiencies. AI is replacing the traffic that funds journalism more than the journalists themselves; AI Overviews compress referral traffic, which compresses ad revenue.
That gap deserves a section of its own.
The Exposure-Versus-Displacement Gap
The IMF’s January 2024 analysis frames AI exposure as a measurement of how many jobs could be touched by the technology, not a forecast of how many will disappear, with about 60% of advanced-economy jobs exposed and roughly half of those potentially benefiting from productivity gains. Against that exposure rate, Challenger data shows AI was cited in 99,470 US job-cut announcements between 2023 and March 2026, equal to 3.5% of all layoff plans over that window. The gap between 40-60% potential exposure and 3.5% cumulative AI-cited cuts is the central number in the displacement debate; exposure is a leading indicator, not a current count.
Goldman Sachs estimates that over the longer term, AI automation will displace roughly 6 to 7% of the U.S. workforce, equivalent to approximately 11 million workers, while still creating jobs in the buildout of power and data-center infrastructure required to sustain the AI boom. Displacement will rise through years of task substitution and selective layoffs; career decisions should weigh the task-level breakdown more than headline exposure.
Which Workers Face the Highest Substitution Risk
The Future of Jobs Report 2025 names telephone operators, insurance claims clerks, bill collectors, bookkeepers, payroll clerks and tellers as the professions facing the highest risk of substitution by AI, on the grounds that their work involves structured, repetitive data tasks that current language and vision models execute well. The same WEF report identifies clerical and secretarial workers (cashiers, ticket clerks, administrative assistants, executive secretaries, postal service clerks, bank tellers and data entry clerks) as the roles expected to see the largest absolute decline in numbers over the 2025-2030 period.
For readers tracking AI job loss data at the role level on workplace deployment, three patterns emerge:
- High-exposure clerical roles tend to compress before they collapse, with attrition replacing layoffs in many cases.
- Mid-skill professional roles (paralegal, junior analyst, copy editor) see task substitution rather than role elimination, with surviving headcount handling more output per person.
- Physical and care-based roles remain mostly insulated, though their administrative surrounds are exposed.
Challenger commentary on the March 2026 report noted that companies are shifting budgets toward AI investments at the expense of jobs, with the most direct replacement visible in technology companies where AI can replace coding functions, while other industries are still testing the limits of the technology.
Frequently Asked Questions (FAQs)
Office and administrative support has the highest task-automation share at 46%, per Goldman Sachs Research, with legal work at 44% and architecture and engineering at 37%, while construction sits at 6% and installation, maintenance and repair at 4%.
The World Economic Forum projects 92 million jobs displaced and 170 million created over the 2025-2030 period, for a net gain of 78 million. Goldman Sachs estimates AI will eventually displace roughly 6 to 7 percent of the US workforce, approximately 11 million workers, over the longer term.
Exposure measures the share of tasks AI could perform, while layoff data measures cuts companies actually attribute to AI. The IMF puts exposure at about 60 percent in advanced economies. Challenger data shows AI was cited in roughly 13% of 2026 US job-cut plans year-to-date.
The U.S. Bureau of Labor Statistics projects paralegal employment will grow about 1% from 2024 to 2034, slower than the average for all occupations, with about 39,300 annual openings projected and 367,220 paralegals and legal assistants employed at a median annual wage of $61,010 as of May 2024.
A Reuters Institute survey of 326 senior media leaders across 51 countries published in January 2026 found that two-thirds (67%) of newsrooms have not saved any jobs so far as a result of AI efficiencies.
Conclusion
The honest answer to how AI replaces jobs is that it is replacing tasks first and roles second, with the gap most visible in the difference between almost 40% global exposure per the IMF and roughly 13% of 2026 US layoff plans actually citing AI per Challenger, Gray & Christmas. Office and administrative support, legal, customer service, retail back-office, and journalism see the fastest substitution this year.
Manufacturing line work, healthcare clinical roles and construction stay insulated. Clerical and structured-data roles face the highest near-term risk. For ongoing tracking, the AI job creation data and the broader workforce trends in our Claude vs ChatGPT data coverage sit alongside the trends documented here.
