Young workers are losing entry-level jobs to AI, Stanford research finds

Key points

• Stanford University economists found that workers aged 22 to 25 in the most AI-exposed jobs held employment levels 19 percent below peers in less-exposed fields, as of August 2026.
• That gap was 13 percent when the same research team published an earlier version of the study in 2025.
• Older workers appear largely unaffected by the same trend so far.
• The updated paper is titled "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence."
• The trend identified last year is persisting and widening, not fading.

Stanford economists have a word for it: canaries. As in the birds miners once carried underground to detect poison gas before it reached humans. Their updated research paper uses that image deliberately, and the numbers behind it are hard to brush off.

The study, first flagged by Ars Technica AI, tracks what has been happening to young people entering the workforce in jobs that AI can plausibly do. Think roles like junior data analyst, entry-level coder, or basic paralegal work. The researchers call these "AI-exposed" occupations, meaning jobs where a large language model (the technology behind chatbots like ChatGPT and Claude) can already handle a meaningful chunk of the daily tasks.

What does the data actually show?

Workers aged 22 to 25 in those AI-exposed fields are now employed at levels 19 percent below their counterparts in jobs AI has not touched much. A year ago, the same gap sat at 13 percent.

That six-point jump in twelve months is the part that matters. It suggests the trend is not a blip. It is moving in one direction, and it is picking up speed.

Older workers? Largely fine, for now. The researchers found no comparable employment drop among more experienced employees in the same fields. The theory is straightforward: a company considering whether to hire an AI tool or a 23-year-old to do basic research will increasingly choose the tool. A 45-year-old with a decade of client relationships and judgment calls under their belt is a different story.

Should young job-seekers be worried?

Worried is the wrong word. Informed is better. The data does not say AI is eliminating careers wholesale. It says the traditional "first rung" of certain career ladders is getting shorter or disappearing.

If you are a recent graduate aiming for a role in law, finance, coding, or any field heavy on structured information tasks, the practical takeaway is simple: the skills that used to get you the interview are now table stakes. What sets a candidate apart today is knowing how to work alongside AI tools, not compete against them.

Pricing on AI tools ranges from free (ChatGPT's basic tier, for instance) to around £20 a month for professional plans. Learning to use them costs far less than ignoring them.

One privacy note worth knowing: when you practise with these tools using real workplace documents, the default settings on most consumer AI apps send that text to the company's servers. Keep sensitive client or employer information out of free-tier tools unless the provider explicitly promises not to train on your data.

Common questions

Does this mean AI is replacing all jobs?

No. The research is specific to entry-level roles in fields where AI can already handle basic, structured tasks. Jobs requiring physical presence, human judgment, or long-term relationships show no comparable drop yet.

Which jobs count as "AI-exposed" in the study?

The researchers focus on occupations where a significant share of daily tasks involve processing information, writing, coding, or legal research, work that current AI tools can handle with reasonable accuracy.

Where can I read the actual study?

The paper "Canaries in the Coal Mine?" is produced by Stanford University economists; search the title directly for the latest August 2026 edition.