The easy starter work is moving, and the learning has to move with it.

The first job was never only a paycheck.

It was also a classroom with a manager, a deadline, a messy inbox, a customer who did not explain things clearly, and a mistake you could still recover from.

A new worker learned by doing small things: draft the memo, check the spreadsheet, summarize the meeting, clean up the slide, research the question, answer the easy ticket, sit in on the call, write the first version nobody expected to be perfect.

Some of that work was boring. Some of it was inefficient. Some of it deserved better tools years ago.

But it taught people how the real work felt.

That is why the entry-level AI story is bigger than this year’s hiring numbers. Pay attention to the small work: the draft, the check, the cleanup, the second try. That is where people used to become useful.

The practice layer is moving

Strada Institute surveyed nearly 1,500 executives and senior talent leaders in 2026 and found a mixed picture, not a simple collapse. For 2026, Strada reports that about half of employers expect AI use to increase entry-level hiring, while a much smaller share expect it to decrease hiring. But the shape of the job is changing. Forty-two percent of employers said AI has increased analytical or judgment-based responsibilities for entry-level workers. Forty-one percent said it has reduced foundational or skill-building tasks.

That is the first-rung problem.

A beginner may be asked to check, guide, revise, or supervise machine-made work before they have practiced enough of the underlying task to know what good looks like.

The old ladder was not fair to everyone. Plenty of people were shut out of it. Plenty of junior work was repetitive, poorly paid, or hidden behind privilege and networks. But a ladder with missing rungs is not fixed by telling people to jump higher.

Handshake’s Class of 2026 report shows the pressure from the student side. Job postings for that class were 12 percent below pre-pandemic levels on the platform, and graduating seniors were much more pessimistic about starting their careers than they were two years earlier. That is platform evidence, not the entire labor market. Still, it helps explain the mood. AI is part of the anxiety, but the larger worry is immediate: fewer entry-level openings.

The same report also shows students adapting fast. Eighty-five percent of seniors said they use AI tools. Many use them for research, brainstorming, resumes, and recruiter messages. AI mentions on Class of 2026 resumes appeared more than nine times as often as they did for the Class of 2022.

So this is not a story about young people refusing to learn AI.

The problem is that tool access is not the same as apprenticeship.

A person can use AI to produce a cleaner first draft and still not know why the draft is wrong. They can ask for a summary and miss the thing the summary left out. They can generate a spreadsheet formula and not notice the bad assumption. They can polish the answer before they understand the question.

That does not make them lazy. It means the learning environment changed.

ZipRecruiter’s 2026 graduate survey points to the same missing-rung issue from another direction. Work experience is one of the clearest advantages for new graduates. In that survey, 81.6 percent of recent graduates with work experience were currently employed, compared with 40.7 percent of those without it. Do not turn that exact gap into a universal rule; it is survey evidence, not payroll data. The useful point is simpler: experience still opens doors.

Cengage’s public article, “AI Is Changing Entry-Level Jobs. How Will Workers Build Experience?,” uses a plain phrase for the risk: an “experience gap.” If students cannot reliably gain practice after graduation, more of that practice has to be built before graduation and inside early jobs on purpose.

Build the rung on purpose

If AI removes some beginner tasks, schools and employers have to design new beginner tasks deliberately.

Not busywork. Not fake assignments. Not “use AI and hope judgment appears.” Real practice surfaces.

A good beginner task does more than produce an answer. It shows where the worker is learning. Before cutting one, ask:

  • What did a beginner used to learn by doing this?
  • Where did mistakes get caught?
  • What does the worker have to understand before AI can safely help?
  • What proof shows they understand the work behind the output?
  • Who gives feedback before the mistake becomes expensive?

A junior analyst still needs to learn what a bad number feels like. A new teacher still needs to see how a lesson plan changes in a real classroom. A young marketer still needs to hear why a sentence sounds wrong for the customer. A new paralegal still needs to understand why a citation matters before a machine helps assemble the text.

AI can help with all of that. It can also hide the practice if nobody builds the practice back in.

The manager question is simple: where will judgment be learned now?

For students and new workers, the answer is not to avoid AI. Build proof around it: projects, notes, before-and-after drafts, decisions, corrections, and examples of what you checked.

For schools, make AI use visible enough that learning can still be seen.

For employers, stop treating entry-level work as disposable friction. Some of it is the training system.

If the first rung disappears, someone has to build a new one.

Otherwise the workplace keeps the output and loses the apprenticeship.

What to watch

  • Do entry-level job descriptions expect review and judgment before offering practice?
  • Are schools teaching professional AI use, or only warning students about misuse?
  • Are internships, apprenticeships, and project-based work expanding enough to replace lost practice?
  • Can a new worker show how they checked, revised, and learned from AI-assisted work?
  • Are employers measuring what junior work teaches before they automate it?

Source note

Sources include Strada Institute’s 2026 employer survey on entry-level hiring in the AI era, Handshake’s Class of 2026 report on AI and the workforce ahead, ZipRecruiter’s 2026 graduate survey, and Cengage’s public analysis of AI and entry-level experience. These sources do not prove one simple collapse story. They point to a narrower problem: entry-level work is being reshaped, and the training function has to be rebuilt deliberately.