If you are early in an office career, the honest answer is that it has already started, beginning with whoever was hired last. This is no longer a projection from task models. It is visible in payroll data. How far it goes is set by cost curves and adoption speed, not by whether the technology can do your job in a demonstration, and while it happens, the returns to education are splitting into two tracks. Which track your children ride is now a purchasing decision.

What the payroll data already shows

Stanford's Digital Economy Lab, working with payroll records covering millions of workers, found a 13 percent relative decline in employment for workers aged 22 to 25 in the most AI-exposed occupations since generative AI tools spread, while older workers in the same occupations grew 6 to 9 percent.[1] By April 2026, employment for that youngest cohort in highly exposed roles was shrinking at 3.8 percent per year.[1] The affected occupations are the ones parents consider safe: accountants and auditors, operations managers, software developers, customer service.[1] The machine did not take the jobs. It took the entry to the jobs, which is the same thing on a fifteen-year delay.

How big it gets

Goldman Sachs' task-level analysis put the equivalent of 300 million full-time jobs worldwide in exposure, with two-thirds of US and European occupations exposed to some degree.[2] McKinsey's 2025 update estimates 57 percent of current US work hours are technically automatable with technology that already exists, and puts 2.9 trillion dollars of US economic value on the table by 2030, contingent on how fast organizations restructure around it.[3] Read those numbers the way capital does: not as a forecast of unemployment, but as a budget for it.

The education split

The college wage premium looks healthy at a distance, roughly 80,000 dollars for the typical graduate against 47,000 for high school alone.[4] But the premium has been flat since 2000 while the price of the degree rose about 40 percent, and the mobility statistics underneath are unambiguous: the share of children who out-earn their parents fell from 90 percent for those born in 1940 to 50 percent for those born in the 1980s.[5] The average degree buys less than it did. The distribution around that average is doing the real work, and it is widening.

The tutoring line

Education research has known its most important number for forty years: Benjamin Bloom found that one-to-one tutoring moves student achievement about two standard deviations, the difference between an average student and an exceptional one.[6] The constraint was always cost. It no longer is. A World Bank randomized trial in Nigeria found six weeks of GPT-4 tutoring, supervised by teachers, produced gains equivalent to roughly 1.5 to 2 years of ordinary schooling, among the most cost-effective interventions ever measured.[7]

Two-sigma instruction is now purchasable, and what is purchasable is purchased unevenly. Families that buy comprehensive AI tutoring are buying their children into the cohort the labor market will still bid for. Families that do not are enrolling, by default, in the other cohort. The Institute's education program describes this arrangement without euphemism: the literate class is now an enrollment decision, and the enrollment window is open.

Sources

  1. Brynjolfsson, E., Chandar, B., Chen, R., "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence," Stanford Digital Economy Lab, 2025. https://digitaleconomy.stanford.edu/publications/canaries-in-the-coal-mine/
  2. Goldman Sachs Global Investment Research, "The Potentially Large Effects of Artificial Intelligence on Economic Growth," 2023; see https://www.goldmansachs.com/insights/articles/how-will-ai-affect-the-us-labor-market
  3. McKinsey Global Institute, "Agents, robots, and us," November 2025; figures as reported in Fortune, November 25, 2025. https://fortune.com/2025/11/25/why-ai-wont-take-your-job-partnership-agents-robots-mckinsey/
  4. Federal Reserve Bank of New York, "Is College Still Worth It?", Liberty Street Economics, April 2025. https://libertystreeteconomics.newyorkfed.org/2025/04/is-college-still-worth-it/
  5. Chetty, R. et al., "The Fading American Dream: Trends in Absolute Income Mobility Since 1940," Science 356, 2017. https://opportunityinsights.org/paper/the-fading-american-dream/
  6. Bloom, B., "The 2 Sigma Problem," Educational Researcher 13(6), 1984. https://journals.sagepub.com/doi/10.3102/0013189X013006004
  7. World Bank, "From Chalkboards to Chatbots: Evaluating the Impact of Generative AI on Learning Outcomes in Nigeria," 2025. https://documents.worldbank.org/en/publication/documents-reports/documentdetail/099548105192529324