Skip to content
Big Brain BattleTwo articles. One vote.
Live Battle · AI & Automation

Is AI already taking jobs from young workers?

Since ChatGPT arrived in late 2022, graduates have been told the bottom rungs of the career ladder are disappearing. Some data supports that. Some says it is too early to tell. Derek Thompson reads new Stanford payroll research as the strongest evidence yet that AI is already squeezing 22 to 25 year olds in exposed jobs such as software development and customer service. Economists at the Yale Budget Lab look at the whole US labour market and find no discernible disruption so far. Both pieces use US data and both take the evidence seriously. The disagreement is about which data answers the question, and how much weight early signals deserve. Read both cases, then vote for the argument you find more convincing, even if you dislike its conclusion.

Voting closes · 11 October 2026

ProvocationFirst article

The Evidence That AI Is Destroying Jobs For Young People Just Got Stronger

Derek Thompson, Derek Thompson (Substack) · AI & Automation

New payroll data covering millions of US workers shows employment falling for 22 to 25 year olds in the jobs most exposed to AI, while older workers and less exposed jobs hold steady. Thompson, who had come to doubt the AI jobs story, now says this is the most compelling evidence yet that AI is already affecting young people's work, not just in some future scenario.

Read the original →
  • Stanford economists using ADP payroll data found a 13 per cent decline in employment for 22 to 25 year olds in highly AI-exposed jobs, such as software developers and customer service agents, since ChatGPT arrived.
  • The decline shows up even within the same companies, which the authors say rules out firm-wide shocks such as interest rate rises as the main explanation.
  • Losses cluster in jobs where AI automates tasks rather than assists people. The authors suggest, tentatively, that AI overlaps most with the textbook knowledge new graduates bring.
Evidence from the source
But I’m updating in the direction of trusting my initial gut instinct. I think we’re looking at the single most compelling evidence that AI is already affecting the labor force for young people.

The strongest objection

The study is observational, not a causal test, as the authors say themselves. Junior hiring in software and customer service could be falling for other reasons, such as post-pandemic over-hiring or a slowing economy, and the result covers some exposed occupations rather than young workers in general. Thompson has also changed his view on this question several times in five months.

“I think we’re looking at the single most compelling evidence that AI is already affecting the labor force for young people.”

DefenceSecond article

Evaluating the Impact of AI on the Labor Market: Current State of Affairs

Martha Gimbel, Molly Kinder, Joshua Kendall and Maddie Lee, The Budget Lab at Yale · AI & Automation

Looking across the whole US labour market, the Budget Lab finds no discernible disruption in the 33 months after ChatGPT's release. The mix of jobs is changing at roughly the pace of earlier technology shifts, the change began before generative AI, and measures of AI exposure and use show no sign of being related to changes in employment or unemployment. On this reading, fears of AI-driven job loss are running ahead of the data.

Read the original →
  • The mix of jobs in the US economy is shifting only about one percentage point faster than when the internet arrived, and that shift was already under way in 2021, before ChatGPT.
  • The share of workers in the most AI-exposed occupations has stayed stable, and there is no clear rise in AI exposure among unemployed workers.
  • For recent graduates the gap with older workers has widened slightly, but the authors say this could reflect a slowing job market and warn that the sample is small.
Evidence from the source
Overall, our metrics indicate that the broader labor market has not experienced a discernible disruption since ChatGPT’s release 33 months ago, undercutting fears that AI automation is currently eroding the demand for cognitive labor across the economy.

The strongest objection

It answers a broader question than the one most graduates are asking. An economy-wide average can hide a sharp hit to a small group, and the authors concede that their survey sample for recent graduates is small and that their exposure measure is a theoretical proxy that may underestimate disruption.

“While anxiety over the effects of AI on today’s labor market is widespread, our data suggests it remains largely speculative.”