Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesNo—Goldman Sachs did not predict that 300 million people will lose their jobs. Its 2023 estimate described the equivalent of 300 million full-time jobs whose tasks could be exposed to automation as generative AI changes work. Exposure is not a count of layoffs, nor proof that entire jobs will disappear.
What Goldman Sachs meant by “300 million jobs”
Goldman Sachs Research’s April 5, 2023 estimate concerned the equivalent of full-time jobs exposed to potential automation through changes to work processes. It was a modeled estimate based on occupational tasks—not a forecast that 300 million people would be unemployed. Goldman Sachs’ explanation of the estimate said roughly two-thirds of U.S. occupations were exposed to some degree of AI automation, and that roughly a quarter to half of the workload in exposed occupations could potentially be replaced.
Those figures describe potential task automation. They do not say that every exposed task will be automated, that all affected work will be removed rather than reorganized, or that a particular worker has a corresponding chance of losing a job. Goldman economists Joseph Briggs and Devesh Kodnani wrote that most jobs and industries were only partially exposed and therefore more likely to be “complemented rather than substituted” by AI.
How the ILO’s 2025 estimate differs
A separate assessment by the International Labour Organization and Poland’s NASK, published in 2025, found that one in four workers worldwide is in an occupation with some degree of generative-AI exposure. The share rises to 34% in high-income countries. The index assessed nearly 30,000 occupational tasks, with expert validation and harmonized ILO data; its unit is workers in occupations with exposure, not equivalent full-time jobs potentially automated. Read the ILO–NASK global analysis.
#1 Best Overall
The ILO–NASK assessment identifies clerical work as the most exposed broad category. It also finds rising exposure in some highly digitized media-, software-, and finance-related occupations. Yet many tasks still require human involvement, so the report concludes that transformation is more likely than full replacement. Its one-in-four measure should not be treated as a newer version of Goldman’s 300 million estimate: the studies use different methods, dates, and units. The ILO’s 2025 mean automation score was 0.29, compared with 0.30 in its 2023 analysis, following methodological refinements that also reduced variation in scores.
Exposure is not the same as job loss
An occupation can contain tasks AI might assist with or automate without the whole role becoming unnecessary. Whether exposure leads to fewer jobs depends on what employers implement, how work is redesigned, whether demand grows, and whether new tasks or roles emerge. A task-level exposure score is therefore not an individual worker’s probability of dismissal.
Rank #2
It also helps to distinguish several kinds of evidence. A model estimates what could be affected; vacancy trends show changes in hiring activity; headcount measures employment levels; layoffs record job separations. An observed association between AI exposure and slower hiring is not by itself proof that AI caused each change. None of these measures, on its own, is a global tally of jobs lost to AI.
What employment evidence says—and what it cannot prove
The OECD’s 2023 review
The OECD’s 2023 Employment Outlook reviewed empirical studies available at the time and found little to no aggregate employment effect attributable to AI in that evidence base. It discussed limited adoption, slow implementation, employers relying on attrition, and the creation of new tasks as reasons aggregate effects might be small or delayed. The review also noted evidence of reduced vacancy posting in some AI-exposed firms and displacement findings in particular sectors. Because it predates much of the later adoption of generative AI, it is a dated baseline, not evidence that AI cannot reduce employment. See the OECD Employment Outlook 2023.
Recommended Free Tools
Goldman Sachs’ 2026 labor-market analysis
Goldman Sachs Research’s September 3, 2026 analysis describes slower job-opening growth since late 2022 in more AI-exposed industries, with differences by country and sector. It reports that a 10% occupational exposure is associated with a 0.1 percentage-point drag on annual headcount growth in France, Canada, and the United States. That is a reported association, not proof of a causal effect or a global job-loss rate.
The same analysis says U.S. call-center employment was 39% below trend, Canada’s was 33% below trend, and Germany’s was 27% below trend. Goldman attributes these patterns to AI-related headwinds; “below trend” is not a count of jobs definitively lost to AI. The report also suggests hiring headwinds may be stronger for junior workers, while describing economy-wide hiring headwinds as limited. It does not provide an audited global count of AI-caused replacements. Read Goldman Sachs Research’s 2026 analysis.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Which jobs are most exposed to generative AI?
The ILO–NASK index puts clerical occupations highest in exposure. Some media, software, and finance work also has exposure because many tasks in these fields are digitized. That does not mean everyone in those occupations faces the same risk: roles combine different tasks, and exposure does not establish that an employer will automate them or eliminate a position.
The ILO–NASK report also cautions that effects can diverge between occupations accustomed to rapid digital change, such as software development, and those where limited digital skills may make adaptation harder. Exposure is best read as a signal about tasks that may change—not a ranking that predicts which individual workers will be replaced.
Free tools Windows power users keep installed
One-click scans. No signup required.
Best Value
So, will AI replace 300 million jobs?
The evidence supports a narrower conclusion: Goldman Sachs estimated that work equivalent to 300 million full-time jobs could be exposed to potential automation, while the ILO’s later global assessment finds widespread occupational exposure but says transformation is more likely than full replacement. More recent analysis points to hiring headwinds in some exposed sectors and possible effects on junior hiring, but the cited evidence does not establish that 300 million jobs have been, or will be, lost to AI.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
