Who Trains the Trainers?

Two organizations looked at the same evidence this year and made opposite decisions.

Most of the industry cut entry-level hiring. IBM tripled it in the US for 2026, including software development roles that AI is widely assumed to have made unnecessary.

Both groups were reading accurate data. The difference is what they concluded from it.

What the data says

Stanford's 2026 AI Index found that employment among software developers aged 22 to 25 fell roughly 20% from its 2022 peak. Over the same stretch, every age group above 26 continued to grow.

The share of juniors and recent graduates in IT employment dropped from about 15% to about 7% in three years. Entry-level hiring at the top 15 technology companies fell 25% between 2023 and 2024, and the decline continued into 2026. Ravio's hiring data puts entry-level hiring down 73% year over year, with engineering among the hardest-hit functions.

Stanford researchers gave the pattern a name: seniority-biased technological change. AI substitutes for junior labor and leaves senior roles intact.

The mechanism is not mysterious. AI took the exact tasks juniors were historically hired to perform, boilerplate, test generation, documentation, CRUD scaffolding, routine bug fixes, basic data processing. Those were the training wheels. They are gone.

One more number, and it is the one that should worry you: only 12% of the tech workforce now has under three years of experience.

The error is not the automation. It is the conclusion.

Here is where the two groups diverge.

The common conclusion: AI can do what a junior does, so we no longer need juniors.

IBM's conclusion, from its HR chief Nickle LaMoreaux: the junior job as it existed two or three years ago can mostly be done by AI now, so rewrite the role rather than delete it. IBM is rewriting entry-level job descriptions around customer interaction and AI oversight rather than routine coding, and is tripling its US entry-level hiring for 2026. This is not a company that doubts what automation can do. It is a company that separated the tasks from the position.

IKEA's parent company reached the same place from a different direction. Its assistant Billie now resolves the large majority of customer inquiries, up from about 47% in 2023 to roughly three quarters today. Instead of cutting the call-centre workers whose routine workload had vanished, Ingka retrained a portion of them as remote interior design consultants. That channel now generates around €1.3 billion a year. A cost centre became one of the company's fastest-growing revenue lines.

Neither company denied the automation. Both refused to treat a task list as a job description.

The bill arrives later, and it is large

Cutting junior roles is nearly free this quarter. That is precisely what makes it dangerous.

Juniors are not only cheap labor. They are the mechanism by which an engineering organization reproduces itself. Mentoring, code review culture, the slow and unglamorous transfer of context that turns a new hire into someone who can be trusted with a production system, all of it runs through the bottom rung.

Remove it and nothing breaks immediately. It breaks in five to ten years, when the seniors being fought over today retire and no one behind them was ever given the chance to learn.

The reversals have already started. Klarna's CEO said publicly that the company went too far, citing declining quality, and began rehiring. Ford brought experienced engineers back after automated systems produced quality problems. Roughly one in three hiring managers who eliminated a role because of AI later rehired for the same or a similar position. Gartner forecasts that by 2027, half of the companies that attributed headcount cuts to AI will rehire for those functions under different titles.

There is also a less flattering explanation for some of this. The market rewards the story. Block capped its headcount at roughly 12,000 and cut about a tenth of its workforce, and investors rewarded the cost-cutting narrative with repeated stock pops, even as the company later walked back parts of the cut. Investors respond better to "we cut because of AI" than to "we over-hired." The people affected experience the same outcome either way, but only one version is good for the share price.

What the rewritten role looks like

If the old junior tasks are automated, what is left is the part that was always more valuable and harder to teach:

• Accountability for something real. One feature, shipped and owned, including when it breaks at 2am.
• Verification. Someone has to determine whether the generated output is actually correct. That skill is now scarce and rising in value.
• Context. Why the system is shaped this way. What broke in 2023. Which customer this constraint exists for.
• Judgment under ambiguity. Not execution against a spec, deciding what the spec should have said.

That is a harder job than the one it replaces, and a more interesting one.

The uncomfortable read

Companies eliminating junior roles now will be competing for scarce mid-level talent later in the decade. The ones that keep a training pipeline will be the ones selling access to it.

The technology did not remove the need for people who grow into senior engineers. It removed the tasks we used to use as an excuse to hire them. Those are not the same thing, and the companies that understand the difference are quietly building an advantage that will take a decade for anyone else to copy.

Research notes

• Stanford Digital Economy Lab, Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of AI (Aug 2026): software-developer employment among 22–25-year-olds fell roughly 20% from its 2022 peak, driven by reduced hiring; the paper describes the pattern as “seniority-biased technological change.”
• IBM announced plans to triple U.S. entry-level hiring for 2026, with roles refocused on customer interaction and AI oversight (Bloomberg, TechCrunch, Fortune).
• Ravio’s Tech Jobs Report 2025 found entry-level hiring down 73% year over year, with engineering among the hardest-hit functions.
• Ingka’s assistant Billie resolved about 47% of inquiries in 2023 and roughly 74% by 2026; its remote-sales channel generated about €1.3B in FY22 and €1.25B in the most recent fiscal year.
• Klarna began rehiring and redeploying human agents after service quality declined (Forbes, Invezz).
• Ford brought back roughly 350 veteran engineers after automated quality systems underperformed (TechCrunch).
• Gartner forecast that half of companies attributing headcount cuts to AI will rehire for those functions under different titles by 2027.
• A Robert Half survey found that 32% of U.S. hiring managers who eliminated a role because of AI later rehired for the same or a similar position.
• Block capped headcount at roughly 12,000 and cut about 10–14% of its workforce; its stock moves across 2023–24 varied with each event.