Is AI Really Killing Entry-Level Jobs?

If you're a grad this year, chances are you've heard this story on repeat: The robots got the bottom rung before you could grab it. It writes itself, neatly. ChatGPT can draft the memo, summarize the deck, clean up the spreadsheet that a 22-year-old used to do for her first gig; why hire the 22-year-old?

It's an elegant tale, and the fear underpinning it is valid. In Monster's 2026 survey of the graduating class, 89% said they were afraid that AI or automation would eliminate their jobs at the entry level, up from 64% last year. But the facts underpinning the panic are less clear-cut and, in some ways, more illuminating than the headline itself. The job market truly is less hospitable than it was three years ago. The reason for its recalcitrance, however, isn't what most people are stating.

How rough is the job market for new graduates?

It's not as apocalyptic as it sounds, though it is rough. The New York Fed tracks workers aged 22 to 27 who hold a bachelor's degree, and put their unemployment rate at about 5.6% in the first quarter of '26. While that number itself isn't shocking, it's unprecedented in that it's above the national unemployment rate for the whole workforce, which almost never happens. Having a degree used to confer a lower risk of unemployment; right now, it doesn't.

The starker fact is the rate of underemployment: 41.5% of recent graduates working at jobs that don't even require a degree. A barista, a warehouse worker, retail floor staff: the degree hangs on the wall, and the work is paid as if it were invisible. It's been described as the worst job market in 37 years. While that claim is debatable, the experience is real: hundreds of applications, deafening silence, and jobs that disappear the week you're offered them.

Is AI really the culprit?

This is where the headline diverges from the data. Economists haven't found many AI fingerprints in overall employment figures so far. According to Anthropic, there was no statistically significant increase in unemployment among those in professions most exposed to AI since late 2022. The Stanford AI Index shows similar trends; mass layoffs are nowhere to be seen in the big-picture data. An IMF report from Denmark found no significant change in wages or hours.

If AI were silently emptying offices, the aggregate national data would reflect it. It doesn't. Instead, what the data reveals is a decline that extends far beyond what a chatbot can do. Sectors with no exposure to AI, such as HR and civil engineering, have also seen hiring plummet. That points to the mundane reasons for the slowdown: companies overhired during the post-pandemic boom and are now sitting on their hands, plus high interest rates and trade uncertainty that make every hiring manager cautious about adding someone they might have to lay off later.

The freshest clue points somewhere else entirely. A June 2026 New York Fed analysis pinned roughly 64% of the rise in young-graduate unemployment on the retreat from remote work, not AI. Employers are reluctant to drop inexperienced grads into remote roles, where the on-the-job mentoring that turns a hire into a professional is hard to deliver. It's the same logic that makes the entry rung so fragile in the first place.

That being said, "AI is not the whole story" does not mean "AI is not a factor." The one place its influence is detectable is exactly where these grads are. Anthropic's own data found that among workers aged 22 to 25, the rate of landing a job in a heavily AI-exposed occupation fell by roughly 14% compared to 2022, while those over 25 saw no such drop in the same fields. The door is closing, but only on those who just arrived.

Why are the most vulnerable the young?

The tasks a robot can now excel at are almost identical to the tasks in an entry-level job. In essence, a first job is a paid apprenticeship: performing the tedious research, drafting the initial outlines, cleaning up data, and learning the business along the way. Automating those tasks saves not only a salary but also removes the rung from which people climb.

The firms don't often realize the cost of that exchange until later. Nobel Prize winner Kenneth Arrow spoke of "learning by doing": skill acquisition happens on the job, not before. In 2026, MIT researchers have warned that companies automating away their entry-level jobs are depleting the pool of future senior talent. They save money now and will eventually run out of experienced staff.

Which entry-level jobs are at risk?

Not uniformly. The fastest-declining jobs are those whose tasks an AI can perform without much supervision: customer service, receptionist and administrative roles, basic accounting and bookkeeping, legal assistants and editorial support, and rudimentary programming. These are traditional "first jobs," and they all involve routine, repeatable tasks that are now susceptible to automation.

This impact is most evident in software development. "Learn to code" was considered sound advice for the past decade, but the entrance to that field has narrowed significantly. The number of new software-engineering postings dropped by roughly 15% in early 2026, and junior listings were the first to go; employment for developers aged 22 to 25 had fallen nearly 20% by mid-2025 compared to late 2022, while older engineers held steady. Tools like GitHub Copilot and Cursor now handle much of the boilerplate and first-pass debugging that used to be part of a junior's training. One senior developer with an AI assistant can do what previously took a senior plus a junior. Coding has become a metered service, and part of the bill is being paid by the juniors who weren't hired. Not surprisingly, AI skills now show up in about 42% of software job postings, up from just 8% in 2022. The job still exists, but the entry exam has changed.

What are people doing instead?

Largely, they're not waiting. As the main door has closed, graduates are finding alternative routes, and the sheer scale of it is the real story. About 57% of Gen Z now work a side job, or a main one that's unconventional, from content creation and gig apps to selling templates and digital goods. Nearly two-thirds say multiple income streams are essential to their financial security.

Among new grads specifically, survey data shows about 38% are considering launching their own business, a third are eyeing gig work, and a growing number are looking at skilled trades, the one sector an AI model can't touch. Some of these ventures are genuine entrepreneurial spirit. Many are simply improvisation dressed up as a brand, the same survival math that has put this generation on track to be poorer than their parents. A side gig isn't a retirement plan, but when a full-time salaried position won't even answer your email, you build a raft with the materials you have.

So, what should you do?

Based on the data, rather than the feeling of panic:

• Become an expert in what you fear. The job market is rapidly adjusting to AI fluency. Graduates who succeed are not avoiding these tools but using them more effectively than their managers can. Trying to compete with software on price is a losing game; being the person who commands it is not.
• Seek out fields less affected by AI. The most resilient jobs are the ones AI is not yet good at: anything involving physical work, anything that runs on trust and face-to-face contact, and the skilled trades.
• Treat your side hustle as a bridge, not a destination. Stacking income makes sense for now. Just don't let "I sell templates" quietly become the entire plan, with no benefits, no stability, and no way out.
• Don't take a structural problem personally. A 41.5% underemployment rate doesn't mean 41.5% of graduates are slacking off. The market is broken in a specific, measurable way. Knowing that won't land you a job, but it will stop you from blaming yourself for an economy you didn't build.

The honest version of this story is less dramatic than "AI took the jobs," and more uncomfortable, because there's no single villain to outlast. The entry door really is harder to open: AI is pressing on it at the edges, and a cautious, overstaffed, tariff-rattled economy is pressing on the rest. To a 23-year-old firing a hundredth application into the void, the cause hardly matters. What does matter is that the ladder is still there. For now, you just have to build the first few rungs yourself.