Toto-Rex's story
Toto-Rex is helping a 23-year-old graduate find a first job.
The graduate applies for junior coding and customer service roles.
Fifty applications. Two interviews. No offers yet. Everyone keeps saying: 'AI took those jobs.'
Toto-Rex wants to know what the data actually show. Toto-Rex's story is made up. The data in this video are real.
The short answer
In US payroll data, young workers in AI-exposed jobs fell behind. The data do not prove AI is the cause.
The details
A Stanford team used US payroll data through June 2026. They compared employment changes since November 2022.
They focused on workers aged 22 to 25 in jobs with high AI exposure (jobs where AI could affect many tasks).
By June 2026, this group's employment was about 19% below where it would have been.
That benchmark assumes it grew at the same rate as same-age peers in less-exposed jobs.
The researchers found no evidence of widespread, economy-wide job displacement linked to AI.
The gap comes mainly from fewer hires. More people leaving their jobs (job separations) did not drive it.
Experienced workers show no comparable gap. How AI is used matters too.
The declines are concentrated in jobs where AI tends to automate tasks.
Where AI is used more to support workers, employment is flat or rising, especially for experienced workers.
The authors add two cautions. Some of these trends began before generative AI spread widely.
The gap between job groups shrinks once education is taken into account.
They call their findings descriptive. They do not prove the cause.
Yale's Budget Lab also tracks AI and the US labour market.
Its October 2026 update finds no clear AI-related labour-market footprint in the measures it tracks.
The studies use different data and methods.
A gap in one group can coexist with no clear change in broader measures.
Neither finding, on its own, establishes how much AI caused.
Useful to know: when you hear 'AI is killing jobs', ask: fewer hires, or more people leaving? And in which group?
Real-life examples
Real life: the UK graduate market.
In the Institute of Student Employers' 2026 report, graduate hiring among the large employers it surveyed rose 4%, after an 8% fall the year before.
The growth was uneven: 45% of employers increased their graduate intake, 15% kept it the same, and 40% reduced it.
62% of employers saw more graduates applying for roles aimed at school and college leavers, up from 40% in 2025.
Employers forecast 7% graduate growth for 2026/27, again driven by only a few employers.
Apprentice hiring rose 4%, and employers forecast 2% for 2026/27.
These UK figures measure something different from the US payroll data: hiring by large UK employers, not employment across whole occupations.
Real life: a second US study.
Harvard economists Seyed Hosseini and Guy Lichtinger studied US CV and job-posting data from more than 280,000 firms, from 2015 to 2025.
They identified AI adoption through job adverts seeking people to integrate generative AI into a firm's work.
From early 2023, junior employment at adopting firms fell sharply compared with other firms, while senior employment trends stayed largely unchanged.
The decline came mainly from slower hiring, not from more people leaving.
It is a working paper, and, like the Stanford study, it uses observational data.
Real life: US graduate unemployment.
The Federal Reserve Bank of New York tracks recent US college graduates aged 22 to 27.
In the second quarter of 2026, their unemployment rate stayed elevated at about 5.6%.
Among employed recent graduates, 42% were in jobs that typically do not require a bachelor's degree.
These figures show what is happening, not why.
Real life: what business leaders say.
In May 2025, Anthropic's chief executive, Dario Amodei, told Axios that AI could wipe out half of all entry-level white-collar jobs within one to five years.
In 2026, IBM said it planned to triple its entry-level hiring in the US, after rewriting junior job descriptions around AI.
IBM's HR chief, Nickle LaMoreaux, said that without entry-level hiring, there would be no talent pipeline in three to five years.
These are predictions and plans, not data.
Back to Toto-Rex
So, did AI take the graduate's jobs? The data cannot answer that for one person.
In the US payroll data, the gap for young workers in highly AI-exposed jobs comes mainly from fewer hires.
Two US studies find associations between AI exposure or adoption and weaker junior hiring. Yale's broader tracker finds no clear AI-related footprint.
Employment held up better in jobs where AI mostly supports workers than in jobs where it mostly automates tasks.
In the UK, hiring by large graduate employers rose in 2025/26, but unevenly.
So the honest answer for Toto-Rex's graduate is: maybe, partly, or not at all. The data cannot say which.
Summary
In short: in US payroll data, young workers in AI-exposed jobs fell behind, mainly because fewer were hired.
That describes a pattern, not a cause.
And the broader measures Yale tracks show no clear AI footprint.
Three key terms
AI exposure
It measures how much AI could affect a job's tasks. It does not show whether an employer actually uses AI.
Job separations
A worker's employment with an employer ends. For example: through resignation or dismissal.
In the study, the gap (for young workers in AI-exposed jobs) came mainly from fewer hires, not more separations.
Descriptive, not causal
The data show a pattern. They do not prove what caused it. So the study does not prove that AI caused the gap.
Further reading
Start with the Stanford paper: its method, data, and the authors' own cautions.
Then the Yale Budget Lab tracker, which the Lab plans to update regularly.
For the UK graduate market, see the Institute of Student Employers'
2026 report.
Sources
Brynjolfsson, Chandar, and Chen, 'Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence', Stanford Digital Economy Lab, revised 12 August 2026.
The Budget Lab at Yale, 'Tracking the Impact of AI on the Labor Market', updated 14 October 2026.
Institute of Student Employers, Student Recruitment Report 2026, 14 October 2026.
Hosseini and Lichtinger, 'Generative AI as Seniority-Biased Technological Change', working paper, May 2026 version. Federal Reserve Bank of New York, The Labor Market for Recent College Graduates, 2026 Q2.
Axios, 'Behind the Curtain: A white-collar bloodbath', 28 May 2025. IBM Think, 'The bottom rung returns as AI reshapes entry-level jobs', 2 March 2026.