Are Entry-Level Jobs Disappearing Because of AI? (2026 Data)

A November 2025 Stanford Digital Economy Lab study found workers ages 22 to 25 in the most AI-exposed jobs saw a 16 percent relative decline in employment, while older colleagues held steady. Are entry-level jobs disappearing? The real story is more specific, and it starts at the task level.

The Short Answer

No, entry-level jobs are not vanishing across the board. But the data from 2025 and 2026 shows something real and worth taking seriously. Early-career workers in a specific set of AI-exposed roles are seeing hiring slow and employment shrink, while more experienced workers in the same fields have held steady. The effect is concentrated, not universal.

That distinction matters because it points to a plan rather than a panic. The jobs under the most pressure are the ones built mostly from tasks that current AI can already do well. If you know which of your daily tasks fall into that bucket, you can see your own exposure clearly and start shifting toward the tasks that still need a human. That is the whole game right now.

What Does the 2026 Data Actually Show?

The single most cited piece of evidence is the Stanford Digital Economy Lab paper Canaries in the Coal Mine by Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, released November 2025. Using payroll records from the largest payroll provider in the United States, the researchers found that workers ages 22 to 25 in the occupations most exposed to AI experienced a 16 percent relative decline in employment, even after controlling for firm-level shocks. Employment for older workers in those same occupations stayed stable or kept growing.

Hiring data tells a parallel story. SignalFire's State of Tech Talent Report, published May 2025, found that new graduates now make up just 7 percent of hires at big tech companies, with new- grad hiring down 25 percent from 2023 and over 50 percent from pre-pandemic 2019 levels. And the Federal Reserve Bank of New York's recent-graduate data for the first quarter of 2026 put the unemployment rate for recent grads at 5.7 percent and the underemployment rate at 41.5 percent.

SignalFigureSource (date)
Employment, workers 22 to 25 in most AI-exposed jobs16% relative declineStanford Digital Economy Lab (Nov 2025)
New-grad share of big tech hires7% of hires, down 25% from 2023SignalFire (May 2025)
Recent-grad unemployment rate5.7%NY Fed (2026 Q1)
Recent-grad underemployment rate41.5%NY Fed (2026 Q1)
US workforce with 10%+ of tasks exposed to LLMsAbout 80%Eloundou et al., Science (2024)

Are Entry-Level Jobs Disappearing, or Just Changing Shape?

This is the heart of the question, and the honest answer is changing shape, unevenly. Entry-level work has always been a bundle of routine, learnable tasks. That is exactly what made it entry-level. Drafting a first version of a memo, summarizing a document, cleaning a spreadsheet, writing boilerplate code, answering tier-one support tickets. Those tasks were the training wheels of a career. They are also the tasks current AI handles most confidently.

So the roles feeling the most pressure are not disappearing so much as losing their easiest tasks to software. The Eloundou et al. study in Science in 2024 estimated that around 80 percent of the US workforce could have at least 10 percent of their work tasks affected by large language models, and roughly 19 percent could see at least half of their tasks affected. Note the word tasks. A job is a container for many tasks, and AI is emptying some tasks out of the container faster than others.

Why Are Young Workers Hit First?

There are a few plausible reasons, and it is worth being clear that these are hypotheses supported by the data rather than settled facts. The leading hypothesis is composition. Junior roles are simply made of a higher share of automatable tasks, so when AI absorbs those tasks, junior roles feel it first and hardest. A senior analyst spends more time on judgment, client relationships, and ambiguous problems, which are harder to hand to a model.

A second hypothesis is substitution at the margin. When a manager can get a competent first draft from AI, the marginal value of hiring someone whose main job was producing first drafts drops. That does not eliminate the role, but it can shrink how many of those roles a team opens. The Stanford authors describe young workers as canaries in the coal mine precisely because their exposure shows up in the data earliest, not because the effect stops with them.

Is AI Causing This, or Just Correlated With It?

Here is where intellectual honesty matters. The 2025 and 2026 data shows a strong correlation between AI exposure and weaker early- career employment. It does not, by itself, prove AI is the sole cause. Real competing explanations exist. Tech companies over- hired during 2021 and 2022 and spent the following years correcting. Higher interest rates cooled hiring across the economy. Broader white-collar softening would hit recent grads regardless of AI.

What makes the Stanford finding notable is that the declines were concentrated specifically in occupations where AI is more likely to automate rather than augment human work, and the pattern held after controlling for firm-level shocks. That is stronger evidence than a raw correlation. Still, the right posture is calibrated concern, not certainty. AI is very likely one real force among several, and the task-level view helps you respond no matter how the causal debate resolves.

Bar chart comparing steady employment for experienced workers against a decline for entry-level workers in AI-exposed roles
Early-career workers in AI-exposed roles are under real pressure, while experienced workers in the same fields have held steady.

Which Entry-Level Tasks Is AI Actually Doing?

This is the practical question, and the answer is more encouraging than the headlines suggest. AI today is strongest at bounded, well-defined tasks with a clear input and output. Summarizing, drafting, formatting, translating, basic coding, data cleanup, and first-pass research. It is much weaker at tasks that require accountability, physical presence, real-time judgment under ambiguity, stakeholder trust, and owning an outcome end to end.

Anthropic's Economic Index, published February 2025, found roughly 57 percent of use leaned toward augmentation, where AI assists a human, versus about 43 percent automation. Even where AI is capable, a majority of real-world use still has a person in the loop. The winning move for a young worker is to become the person in that loop, the one who directs the tool, checks its output, and takes responsibility for the result. If you want to see this mapped to your own workday rather than to averages, that is exactly what the free task- level check on our homepage is for.

What Should Young Workers Do Right Now?

Start by separating your job from your tasks. List what you actually did last week, then sort each item into AI can mostly do this today, AI can assist but I own it, and AI struggles here. That single exercise tells you more about your position than any national statistic, because it is about you.

Then shift your weight. Spend less time being the fastest producer of first drafts and more time on the tasks in your second and third buckets, judgment, coordination, verification, and owning outcomes. Learn to direct AI tools well, because can use AI effectively is quickly becoming a baseline entry-level skill rather than a bonus. The workers who thrive will not be the ones who avoided AI. They will be the ones who moved up the value chain from doing the task to owning the result.

Get your free AI Job Risk Score. Tell us your job title and how you actually spend your time, and we will show you which of your tasks are exposed today. Free. 60 seconds. No sign-up required.

Get My Free Score

Frequently Asked Questions

Are entry-level jobs disappearing because of AI?

Not entirely, but early-career workers in AI-exposed roles are under real pressure. The Stanford Digital Economy Lab found a 16 percent relative decline in employment for workers ages 22 to 25 in the most AI-exposed occupations in its November 2025 study, while older workers in the same fields stayed stable. The effect is concentrated in specific roles, not spread evenly across all entry-level work.

How much has new-graduate hiring actually dropped?

At big tech companies, new graduates now make up just 7 percent of hires, with new-grad hiring down 25 percent from 2023, according to SignalFire's State of Tech Talent Report published May 2025. Startups showed a smaller decline. This reflects a broader pullback in early-career opportunities, though AI is one of several forces involved, alongside post-2021 over-hiring corrections.

Is AI definitely the cause, or could it be the economy?

Both likely play a role. The 2025 data shows a strong correlation between AI exposure and weaker early-career employment, but correlation is not proof. Higher interest rates and tech over- hiring corrections also cooled the market. What strengthens the AI case is that Stanford's declines concentrated specifically in automation-prone occupations and held after controlling for firm- level shocks. Treat it as one real force among several.

What is the unemployment rate for recent college graduates?

The Federal Reserve Bank of New York reported a 5.7 percent unemployment rate for recent college graduates ages 22 to 27 in the first quarter of 2026, with an underemployment rate of 41.5 percent. Underemployment means working in a job that does not typically require a degree. The Fed described conditions as continuing to be challenging for recent graduates entering the market.

Which entry-level tasks can AI already do?

AI is strongest at bounded tasks with clear inputs and outputs, such as summarizing, drafting, formatting, basic coding, translation, and data cleanup. The Eloundou et al. study in Science in 2024 estimated around 80 percent of US workers could have at least 10 percent of their tasks affected by large language models. It is much weaker at judgment, accountability, and owning outcomes end to end.

Does using AI make me safer or more replaceable?

Learning to direct AI well tends to make you safer, not more replaceable. Anthropic's Economic Index, published February 2025, found roughly 57 percent of AI use leaned toward augmentation, where a human stays in the loop, versus 43 percent automation. The durable position is being the person who directs the tool, checks its output, and takes responsibility for the result, rather than competing with it on raw output.

What single skill matters most for young workers now?

Ownership of outcomes. AI can produce a first draft, but someone still has to verify it, adapt it to context, coordinate with people, and stand behind the result. Building tasks like judgment, stakeholder trust, and end-to-end accountability into your role moves you up the value chain. Practically, start by auditing your own weekly tasks against current AI capability so you know exactly where to shift your effort.

Sources

  • Stanford Digital Economy Lab, Canaries in the Coal Mine, November 2025. Read the study.
  • SignalFire, State of Tech Talent Report 2025, May 2025. Read the report.
  • Federal Reserve Bank of New York, The Labor Market for Recent College Graduates, 2026 Q1. See the data.
  • Eloundou, Manning, Mishkin, and Rock, GPTs are GPTs, Science, 2024. View in Science.
  • Anthropic, Introducing the Anthropic Economic Index, February 2025. Read the index.

AI Job Risk Check uses task data from O*NET, provided by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA), used under the CC BY 4.0 license and modified by Phronesis Labs LLC. USDOL/ETA does not endorse this product.