How AI Is Removing the First Rung of Career Ladders

PwC calls the labor-market pattern “seniorization”: work remains, but employers redesign it for people who already have experience. Fortune’s account of the firm’s findings describes entry-level work morphing into something many young applicants cannot get.

That diagnosis captures a structural risk hiding inside the productivity pitch for artificial intelligence. Generative AI can draft summaries, organize research, answer routine customer questions and produce first-pass administrative work. Those tasks also happen to be how junior employees learn what their organizations sell, how decisions get made and what a plausible answer looks like.

A company can remove those tasks from its payroll model. It cannot assume that judgment, relationships and institutional knowledge will appear later by magic.

The Job Survives, but the Entry Point Moves

An entry-level job is usually a bundle of two things: useful output and supervised learning. Employers pay junior workers to complete routine assignments, while managers correct mistakes and gradually hand over more complicated work. The arrangement is inefficient in the narrow sense that training consumes time. It is productive over a longer horizon because beginners become experienced employees.

If software produces a market scan, drafts sales copy or sorts support requests, a manager may need fewer people doing first passes. The remaining work often involves checking the system, resolving exceptions, applying context and accepting responsibility for the result.

Those are more senior duties. An AI-generated answer can arrive in seconds, but someone still needs enough domain knowledge to notice when it is polished nonsense. Employers consequently seek workers who can supervise automation from day one. New graduates encounter the familiar recruiting joke in upgraded form: the entry-level opening requires experience because the tasks that once provided experience have been automated.

CNBC reported on September 23 that companies are rethinking graduate and junior hiring as AI eliminates more “grunt work.” Digiday has documented similar concerns in marketing, where employers are acknowledging the pressure on roles that once brought new talent into the industry. Modern Retail has traced the same debate through retail businesses and their suppliers.

The pattern can spread well beyond occupations that disappear outright. A company might retain the same function while cutting its junior headcount, increasing the experience required for each opening and asking a smaller team to produce more with AI. Employment totals alone may miss that change. The door still exists; the handle has moved upward.

AI is One Cause in a Crowded Room

Current hiring weakness does not provide a clean measurement of jobs lost to AI. Employers also freeze recruitment when demand softens, financing stays expensive, budgets tighten or executives expect uncertain trading conditions. Companies have reduced junior hiring during previous downturns without a chatbot anywhere near the org chart.

The evidence supplied by recent industry reporting establishes that employers are redesigning some roles around AI. It does not isolate how much of the overall decline in entry-level recruitment comes from automation. Answering that question would require comparisons over time between occupations with different levels of AI exposure, alongside controls for industry demand, company size and broader hiring conditions. The public record cited here remains too thin for a confident job-loss total.

Corporate incentives add another wrinkle. Executives may attribute a hiring freeze to AI because “productivity transformation” sounds more strategic than “we cut the budget.” They may also understate automation when job losses attract political or employee scrutiny. Neither narrative should substitute for headcount, vacancy and task-level evidence.

Even so, AI affects the business case before it produces a visible layoff. A manager deciding whether to replace a departing analyst can compare the salary and supervision costs with a software subscription and extra review by existing staff. Leaving the vacancy empty may look attractive on a quarterly spreadsheet.

That spreadsheet rarely carries a line for the analyst the company will need three years later.

The Talent-Pipeline Bill Arrives Later

Companies can respond to a shortage of experienced workers by recruiting from competitors. One employer pays to train; another offers a higher salary once the worker becomes productive. Businesses have played that game for decades, but widespread reductions in junior hiring make it harder for the system to produce enough experienced people for everyone.

This is a collective-action problem wearing an efficiency badge. Each company can defend its own decision to hire fewer beginners. An industry cannot indefinitely demand five years of experience while reducing the jobs that generate years one through four.

The cost will not fall evenly. Applicants with internships, professional contacts or family support have more ways to acquire credible experience outside a standard job. People who need paid work to build a record have fewer. Career changers face a related obstacle because they may possess workplace judgment without the industry-specific background demanded by an AI-supervision role.

Workers who do get hired may also learn differently. Automation can accelerate development when a junior employee uses it under close supervision, compares outputs and receives feedback. It can weaken development when the employee merely forwards machine-generated work or monitors a process they never learned to perform.

That tension sits inside the broader deskilling shock: roles can shift from execution toward managing automated output, while the knowledge required to manage that output still comes from practice. Removing practice and retaining oversight is an unstable training model.

“AI Skills” Cannot Carry the Whole Training Plan

Job advertisements increasingly ask for AI fluency, but the phrase can cover everything from writing prompts to evaluating model errors, protecting confidential information and redesigning a workflow. A short course may teach a tool’s interface. It cannot confer the commercial judgment that comes from seeing customers complain, projects fail and experienced colleagues explain why.

Employers therefore face a choice about where learning happens. Apprenticeships can combine paid work with structured instruction. Rotational programs can expose junior staff to several functions. Supervised project work can preserve learning even when AI completes much of the first draft. Internal mobility can give support or operations workers a route into higher-skilled roles instead of restricting recruitment to outside candidates with ready-made credentials.

Each option costs money and managerial attention. That is also why vague calls for reskilling deserve scrutiny. Training programs are easy to announce, course-completion numbers are easy to publish, and neither guarantees a route into paid work.

Eric Hazan, writing in Project Syndicate, argues that policymakers should focus on rebuilding the bottom rung of the career ladder. Governments considering subsidies or education reforms will need measures tied to outcomes: paid placements, completion rates, subsequent employment and wage progression. Otherwise, public money can flow to credential providers while employers continue asking applicants to arrive fully trained.

Colleges and universities face a parallel test. Adding “AI” to a course title may help marketing, but students need to know whether employers helped design the curriculum, whether the work resembles actual junior assignments and whether graduates obtained jobs. A credential with no occupational doorway is an expensive waiting room.

What Employers Need to Disclose

Better evidence would make this debate less dependent on executive anecdotes. Companies could report how graduate recruitment has changed, which tasks AI now performs, how many apprentices or trainees they support, and whether internal promotions are filling roles once supplied by entry-level hiring.

Investors also have reason to ask. Cutting junior payroll can raise near-term productivity per employee, one of those metrics that looks immaculate until the organization lacks people ready to become managers, specialists or client leads. A company relying on experienced external hires should explain how it expects that labor pool to remain available if its peers adopt the same strategy.

The strongest case for automation remains straightforward: businesses should not preserve repetitive work solely because it has historically served as training. Software can remove drudgery, improve output and let workers tackle more demanding assignments sooner. Keeping obsolete tasks alive would turn career development into corporate make-work.

The harder obligation is to replace the learning embedded in those tasks. That requires designed pathways, supervision and paid opportunities to exercise judgment before being held responsible for it. Employers that capture automation’s savings are in the best position to fund those pathways, while educators and governments can test whether the routes lead somewhere.

When the first rung requires prior experience, the career ladder has become a poaching system.