Where Will Tomorrow's Experts Come From If We Stop Hiring Juniors?

Mon, 09 Nov 2026 00:00:00 +0000

Imagine a budget meeting at a software company. Its experienced engineers have started using AI assistants. They solve problems faster, write code with less effort, and need less support for routine tasks. Someone asks a perfectly reasonable question: if the current team can produce more, why open new junior positions?

The decision looks sensible. This year, the company saves money and delivers its projects on time.

But one item is missing from the cost spreadsheet: the beginners who now seem unnecessary are also part of the pool of experts the company may need tomorrow.

This matters because we are already seeing signs that entry-level opportunities are narrowing in some occupations.

Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen analyzed payroll records covering millions of US workers through June 2026. They found that employment among workers aged 22 to 25 in highly AI-exposed occupations was 19% below the path it would have followed had it evolved like employment among young workers in less-exposed occupations. There was no comparable gap among experienced workers. Most of the difference came from reduced hiring, rather than a wave of layoffs (Brynjolfsson et al., 2026).

Another study, by Samuel Westby, Alicia Sasser Modestino, and Peiran Cheng, examined US software developer job postings. After ChatGPT became available, junior vacancies fell by 14% to 15% relative to senior vacancies. Part of the change happened within the same job titles: employers started asking for more years of experience without necessarily changing the name of the position (Westby et al., 2026).

These findings do not prove that AI is the sole cause. The Stanford researchers acknowledge pre-existing trends and educational differences that complicate interpretation. Nor do they show an economy-wide collapse in employment, and most of this evidence comes from the United States. But the signal is strong enough to look beyond the next hiring cycle.

What disappears when a junior position disappears?

A newly hired developer might spend weeks fixing small bugs, reading code written by someone else, or documenting decisions made by a more experienced colleague. From the perspective of immediate productivity, some of those tasks look like ideal candidates for automation.

For a beginner, however, they are also opportunities to learn. Fixing a bug reveals why a seemingly correct solution failed. Reviewing someone else’s code develops an eye for problems they cannot yet anticipate. Feedback gradually teaches them which decisions require judgment and which can be handled by a rule.

Professional expertise does not arrive automatically with age. It develops through real problems, mistakes, corrections, and increasing responsibility. Many small tasks used to provide exactly those opportunities.

In a recent qualitative study, Sumin Yu and Taesup Moon explored this tension through interviews with 14 people connected to software engineering in South Korea. They describe how some entry-level tasks may shift toward workflows in which senior engineers work with AI, leaving juniors with fewer opportunities to learn by doing (Yu & Moon, 2026). Those interviews cannot predict a future shortage of experts, but they help identify the process we should be watching.

This creates an organizational paradox.

A company can become more efficient by training fewer beginners and hiring experienced professionals when it needs them. From its own perspective, that may be rational. But what happens if many companies make the same choice?

Experts do not materialize spontaneously in the labor market. They develop inside firms, teams, institutions, and communities of practice. Each organization that stops investing in that process can still benefit from expertise developed elsewhere. If too many pursue the same strategy, the labor market as a whole may invest too little in the next generation.

This is a coordination problem with delayed consequences. The savings appear in this quarter’s accounts. The cost, if it materializes, may arrive years later when experienced people retire or move into other roles.

There is, however, another possibility that we should take seriously: AI may also help those who do get hired become proficient faster.

In a study published in The Quarterly Journal of Economics, Brynjolfsson, Danielle Li, and Lindsey Raymond examined the use of an AI assistant among 5,172 customer support agents. Productivity increased by an average of 15%, with especially large gains among less experienced workers. They also found evidence of learning associated with the tool (Brynjolfsson et al., 2025).

That changes the diagnosis. AI can remove tasks that once gave beginners practice, but it can also teach, provide feedback, and expose newcomers to techniques that used to be available mainly through experienced colleagues. The outcome will depend on which tasks are automated, which decisions remain the worker’s responsibility, and how learning is organized.

Conceptual illustration. Both mechanisms may coexist, and we do not yet know which will dominate.

We also do not know whether companies will need the same number of senior workers in the future. Some roles will change, and certain capabilities may be acquired faster. Predicting an inevitable shortage of experts would go beyond the evidence.

The more useful question is whether we are automating entry-level work faster than we are creating new ways for people to gain experience.

Answering it requires companies to measure something that rarely appears in their AI adoption dashboards. Alongside completed tasks, time saved, and lower costs, they should track how many newcomers enter the organization, which responsibilities they learn to take on, and how long it takes them to solve difficult problems independently.

We do not need to preserve repetitive work simply to keep beginners busy. We can design better learning paths: gradual exposure to real projects, decisions reviewed by experienced professionals, and AI tools that help people understand mistakes rather than conceal them. That takes time and resources. Developing expertise always has.

For now, the studies point to a labor market entry point under pressure and two forces that may pull in opposite directions: fewer chances to practice and greater capacity to learn with assistance. We do not yet know which will dominate.

But organizations are already making decisions to reduce junior hiring, while their long-term consequences remain absent from the spreadsheets.

When the time comes to replace an expert, publishing a job opening that asks for ten years of experience will not be enough. Someone must have had the chance to acquire that experience over those ten years.

References

Brynjolfsson, E., Chandar, B., & Chen, R. (2026). Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence. Stanford Digital Economy Lab, revised August 2026. Read study.

Westby, S., Modestino, A. S., & Cheng, P. (2026). Generative AI and the Redefinition of Entry-Level Software Work. IZA Discussion Paper No. 18723. Read study.

Tucker, L. (2026). You’re (not) Hired: Artificial Intelligence and Early Career Hiring in the Quarterly Workforce Indicators. U.S. Census Bureau, CES-26-27. Read study.

Yu, S., & Moon, T. (2026). Who Will Become the Next Senior? How Generative AI Erodes the Development Pathway in Software Engineering. Preprint, arXiv:2607.17067. Read study.

Brynjolfsson, E., Li, D., & Raymond, L. (2025). Generative AI at Work. The Quarterly Journal of Economics, 140(2), 889–942. Read article.

Labor Markets | CRiSS-LAB

Where Will Tomorrow's Experts Come From If We Stop Hiring Juniors?

References