Is AI Coming for Entry-Level Jobs? What the Data Shows

Not in the way most headlines suggest, and the difference matters for what you should do. There is no wave of AI layoffs in the numbers. What three separate research groups have found is narrower and, if you are young, more relevant: the bottom rung of certain office careers is getting harder to climb onto, because hiring slows rather than because anyone gets fired. It is early enough that researchers still disagree about the cause. It is not too early to plan around it.

Start with the wage number

Your paycheck is losing ground this year, and AI is not why. Prices are outrunning pay across the whole economy, for exposed and unexposed workers alike. That is a this-month problem. AI is a five-year problem about which ladder you climb. Most of what is written about this has those the wrong way round, which is how people end up panicking about the wrong thing while the actual damage shows up in the grocery bill.

Question 1: is AI causing mass job losses?

Question 2: is the entry rung closing?

Question 3: is AI lowering wages?

Occupations with the lowest AI exposure

Exposure is far narrower than the conversation implies. Somewhere between 30% and 40% of workers have no measured exposure at all, and fewer than 10% are in highly exposed occupations.

Government projections point the same way, and they are worth reading closely because the Bureau of Labor Statistics now names AI directly as a reason demand will soften in sales, design and administrative support.

Where AI labor market research disagrees

Any article that tells you this is settled is selling something, so here is the other side laid out properly.

• The Economic Innovation Group looked at the same period with survey data and five different exposure measures, and found unemployment rising for young workers whether or not their jobs were exposed.
• The Yale Budget Lab tracks whether the mix of occupations in the economy is shifting faster than normal, and reports it is not.
• The New York Fed, examining job postings in May 2026, found junior and senior roles in exposed occupations declining at similar rates, which directly contradicts the entry-rung finding.
• The Economic Policy Institute calculates that nearly 98% of the rise in young-graduate unemployment came from more people entering the labour force rather than losing jobs.

There is also a timing problem. Some analyses find job-posting declines beginning before ChatGPT was released. And the exposure measures themselves disagree with each other, which is part of why researchers reach opposite conclusions from similar data.

So what do you actually do

Short version: keep the job you have if it is in an exposed field, because every study shows the damage falls on people trying to get in rather than people already inside. If you are starting out, weight work that requires you physically present and pays for judgment rather than procedure. Get paid while you train instead of paying to train. And build one source of income that does not require anyone to hire you.

Common questions

Is AI taking jobs right now?

AI-attributed job loss is defined as employment decline a firm reports as caused by adopting AI. By that measure it is not happening at scale: unemployment is 4.2% and the best firm-level estimate attributes under 0.4% of 2026 employment change to AI. But only about 20% of firms use AI in production, so current figures measure a technology that is barely deployed.

Is AI making it harder to get a first job?

The entry rung is defined as the first job in a career that requires little experience and teaches the rest. The evidence points to that rung narrowing: three independent datasets found early-career employment declining in AI-exposed occupations, through slower hiring rather than layoffs. Researchers still disagree about how much of it is AI versus a hiring correction and interest rates.

Which jobs are least affected by AI?

AI exposure is defined as the share of an occupation’s day-to-day tasks that current AI systems can perform. Hands-on, in-person work scores lowest on every exposure framework built so far: construction trades, home health and nursing assistants, therapy assistants, mechanics, grounds maintenance. Government projections have healthcare support growing 12.4% and construction 5.2% through 2034 against 3.1% overall.

Should I get a degree or learn a trade?

Underemployment is defined as working in a job that does not require your qualification, and it is the number that makes this decision hard. The research cannot decide it for you. What it shows: recent graduates face 5.6% unemployment and 41.5% underemployment, and the unemployment advantage a degree carried in 1979 has inverted. A paid apprenticeship costs nothing up front. A degree financed with debt is a heavier bet than it was a generation ago.

Sources: US Bureau of Labor Statistics, The Employment Situation June 2026, Real Earnings May 2026, Occupational Outlook Handbook, and Employment Projections 2024-34; Brynjolfsson, Chandar and Chen, Canaries in the Coal Mine, Stanford Digital Economy Lab; Hosseini Maasoum and Lichtinger, Harvard; Anthropic, Labor market impacts of AI; Humlum and Vestergaard, NBER Working Paper 33777; Tomlinson et al., Microsoft Research; Federal Reserve Bank of Dallas; Federal Reserve Bank of New York; Economic Innovation Group; The Budget Lab at Yale; Economic Policy Institute; Stanford Institute for Economic Policy Research; Federal Reserve Bank of Atlanta Working Paper 2026-4. Figures current as of July 2026.