↓ Skip to main content
Expert LinkedIn
Digital Economy Professional Development Entrepreneurship Future of Work

AI Didn't Kill the Junior Job. Founders Did.

9 min read
Raj Sharma
Raj Sharma Tech Entrepreneur & Digital Marketing Maverick

Every founder loves an efficiency story until the spreadsheet starts eating the apprenticeship layer.

If you replace junior work with AI and call it “operating leverage,” you are not merely saving salary. You are liquidating the practice field where tomorrow’s operators, product leads, managers, and domain experts are supposed to be grown.

A steel ladder rises upward, but its lower rungs are being ground into a glowing black server rack, leaving a gap no one can climb.
Short-term AI efficiency looks smart until it starts consuming the ladder your next leaders were supposed to climb.

I say this as someone who built a business on jugaad and thin margins. When cash is tight, the temptation is obvious: keep one expensive senior, give them a model, cancel the analyst requisition, and tell yourself you have discovered modern management. In the short term, the math looks beautiful. In the long term, it is the sort of beauty that ruins companies.

We are spending expertise like inherited wealth
#

Brookings put the sharpest language on the problem this month: today’s AI productivity gains are being captured mainly by people who built deep judgment before AI arrived, which means the boom may be less self-sustaining than it looks. In “Borrowed expertise”, Niam Yaraghi argues that firms are pulling value from an inherited stock of expertise while quietly starving the pipeline that replenishes it.

The productivity evidence itself is real. In the NBER paper “Generative AI at Work”, Erik Brynjolfsson, Danielle Li, and Lindsey Raymond found that access to a generative AI assistant lifted productivity for 5,179 customer-support agents by 14% on average, including a 34% improvement for novice and low-skilled workers. That is exactly the kind of result that makes a founder reach for the hiring freeze.

But read it carefully and the story is not “juniors are unnecessary.” The story is that juniors are borrowing the accumulated patterns of the best seniors faster than before. That is a very different proposition. Borrowing expertise is not the same thing as building it.

The distinction matters most when work stops being routine. In Harvard Business School’s “Navigating the Jagged Technological Frontier”, consultants using AI completed in-frontier tasks 12.2% more often, 25.1% faster, and with higher quality. But when the task moved outside AI’s frontier, AI-assisted participants were 19% less likely to reach the correct answer. Same workers. Same tool. Different terrain. That is what founders forget: the easy work trains the judgment required for the hard work.

The junior job was never just about output
#

The junior analyst does not exist because a company enjoys paying someone to format slides at 11:40 p.m. The junior role exists because real organizations need an apprenticeship layer. Someone learns the clients. Someone learns where the data is dirty. Someone learns which “small exception” has sunk three launches before. Someone learns the difference between a plausible answer and a safe one.

When companies erase that layer, they are not merely automating clerical labor. They are removing repeated exposure to consequence.

That is why the World Economic Forum’s June report on entry-level work and AI matters more than the usual reskilling sermon. Its framework focuses on four things companies keep pretending are separate: job access, job design, talent pipelines, and education-system alignment. They are not separate. They are one machine. Break the entry point and the rest of the machine eventually starts coughing.

The Forum’s follow-up article on the greatest risk of replacing early-career roles with technology quotes NYU dean Angie Kamath saying that firms which eliminate entry-level jobs too aggressively risk “destroy[ing] the pipeline that produces future managers, leaders, and institutional memory.” That is not HR poetry. That is operating reality.

Hannah Calhoon, VP of AI at Indeed, put numbers around the warning: junior-level job postings fell 7% year over year in 2025 while senior-level postings rose 4%. This is what a pipeline failure looks like in its polite phase. Nobody panics because the dashboard still looks calm. You only notice the damage later, when you need a manager who understands the business all the way down to the wiring and discover you never trained one.

The market is already telling us something broke
#

If you want a less philosophical signal, look at graduate behavior.

In April, Indeed Hiring Lab’s “For New Grads Looking for Work, the Struggle Is Real - But Not for All” showed recent-graduate unemployment at 5.7% in the fourth quarter of 2025, the highest in three years. The indexed share of bachelor’s graduates creating or modifying Indeed profiles jumped 67% between 2023 and 2025. Master’s graduates saw a 61% jump over the same period. That is not just platform growth. It is a labor-market distress flare.

The article makes one especially important point: fields with durable pipelines such as nursing, education, and apprenticeships held up better because the institutional bridge between learning and work remained intact. Open-market white-collar fields suffered more. In other words, the sectors doing the most talking about AI-enabled efficiency are also the sectors most at risk of sawing off their own on-ramp.

That same pattern surfaces in the World Economic Forum’s July argument that AI presents a livelihood problem, not just a jobs problem. More than one in three young workers are already in occupations with medium to high exposure to AI-driven task change. Reskilling alone will not save a system that keeps collapsing the first rung of the ladder. You cannot retrain people into experience if no one is willing to let them accumulate any.

The oversight fantasy is where this gets dangerous
#

The favorite founder rebuttal goes like this: “Fine, juniors will do less grunt work and more oversight.”

Sounds elegant. Mostly nonsense.

In “The oversight paradox”, two World Economic Forum contributors make the uncomfortable point regulators and executives prefer not to say aloud: the competence needed to oversee AI is built through practice, and AI is taking over the very practice that keeps that competence alive. If a junior lawyer spends two years reviewing AI-drafted contracts instead of writing them, at what point do they lose the feel for omission, structure, and risk? If a junior PM learns to approve model-generated specs before learning to reason through tradeoffs from scratch, what exactly have they become? Not an operator. A quality-assurance layer for a machine they are increasingly less qualified to challenge.

This is where the HBS jagged-frontier result stops being academic. If AI works brilliantly on in-frontier tasks and fails outside them, then the person supervising it must know which is which. That knowledge does not come from clicking thumbs-up on model output. It comes from doing real work long enough to smell when an answer is slick but hollow.

The company that says “humans stay in the loop” while ensuring humans never develop loop-worthy judgment is not building safety. It is staging compliance theatre.

Smart founders will redesign the ladder, not remove it
#

The answer is not nostalgia. Nobody needs to preserve the sacred tradition of twenty-tab spreadsheet slavery just because previous generations suffered through it. The answer is to redesign early-career work so AI strips away drudgery without stripping away formation.

Some institutions have already started acting like this is serious.

California’s AI-Ready California initiative launched on July 14 with an initial cohort of 500 participants across community-college learners and unemployment claimants. It treats AI literacy as foundational workforce infrastructure, not a premium perk for technical elites. Randstad, in the Forum’s June piece on redesigning entry-level work, describes mandatory AI-skills acceleration, a “human remains in the lead” principle, and a move toward job architectures centered on judgment, oversight, and human connection.

That is the correct direction. For founders, translated into plain English, it means four things.

First, automate tasks, not development. Let AI draft the first spreadsheet or summarize the call, but make the junior explain the assumptions, defend the conclusion, and fix the mistakes.

Second, redesign junior roles around context gathering, verification, escalation judgment, and customer exposure. If your junior people never touch the messy edges of the business, they will never become the seniors who can run it.

Third, keep structured non-AI reps. Pilots still train without autopilot. Your analysts, recruiters, marketers, and operators need the equivalent.

Fourth, measure bench strength like a real asset. If you track CAC, burn multiple, and gross margin with religious devotion but cannot tell me how many people in your company are on a credible path to expert judgment, you are not running lean. You are running blind.

The dirtiest secret in this whole conversation is that many founders do know this. They are simply hoping they can rent senior judgment from the market later. Maybe for a while they can. But if everyone stops growing juniors at the same time, the market for ready-made adults eventually empties out.

AI did not kill the junior job. It merely exposed how many leaders saw training as waste the moment a cheaper-looking substitute appeared. The founders who win the next decade will not be the ones who automated away their apprentices. They will be the ones who used AI to make those apprentices dangerous faster.

Seeing this pattern inside your team, agency, or startup? Send me the messy version, not the polished one.

Email me at raj.sharma@tlnw.uk

References
#

AI Content Notice

This article was created using artificial intelligence technology. Whenever possible, we include references and sources to support the information presented. Readers are encouraged to consult these sources for further information. While we strive for accuracy and provide valuable insights, readers should independently verify information and use their own judgment when making business decisions. The content may not reflect real-time market conditions or personal circumstances.

Related Articles

Tags

#entry-level jobs #AI adoption #startup hiring #apprenticeship #talent pipeline