The entry-level job is being dismantled from both ends at once

Meta has begun telling its job candidates that the AI assistant sitting inside the coding test is there because it is what the job will look like.

This is more representative of the developer environment that our future employees will work in, and also makes LLM-based cheating less effective. — Meta Platforms Incorporated

The interview is being remade to resemble the work the company expects. When Meta began piloting this in July 2025, Amazon was still instructing recruiters to disqualify candidates caught using AI tools in an interview, and Anthropic had banned the practice, then reversed itself ^[1]. By February 2026 the requirement had swung the other way and hardened: Accenture now requires AI tool use for senior promotions and monitors weekly log-ins, KPMG folds it into annual reviews, Meta tracks lines of AI-generated code, Google factors it into engineer reviews, and Conductor demands fluency of every new hire ^[2]. The reason the room changed is that the junior job's tasks were the curriculum, and the tools have absorbed the tasks. What entry-level employees did all day — scheduling, compliance, data entry — is exactly what the new agents target. OpenAI's Computer Use tools are aimed at routine white-collar work, and the company's president says the technology is now meant for anyone who does computer work, not just coders ^[3]. The IMF's managing director, Kristalina Georgieva, put the mechanism plainly: the tasks AI removes are the ones entry-level jobs are made of ^[4]. The juniors who are still hired are, increasingly, hired to build the agents that absorb the other junior work — when Meta was cutting thousands of roles in April, Salesforce's Marc Benioff was hiring a thousand graduates and interns to build its agent platform ^[5]. At the same time, the floor beneath the gate was being thinned. Global banks have cut junior analyst classes by as much as two-thirds as they integrate AI, and JPMorgan's Jamie Dimon says plainly that the technology will eliminate jobs — and that some firms may use it to cover for people they never should have hired ^[6]. Tech shed more than 165,000 jobs in early 2026, led by Oracle, Amazon and Microsoft, and Block cut 40 percent of its staff on Jack Dorsey's stated logic that a smaller team using the tools it is building can do more, and do it better ^[7]. National Australia Bank announced on October 19 that it will cut up to 2,500 roles — about 6 percent of its staff — and reduce its external consultants, explicitly to redirect capital toward technology, at a A$300 million restructuring charge ^[8]. Google DeepMind's Demis Hassabis has said 2026 would show the beginnings of an impact at the junior level ^[9]. The edge is no longer anecdote. The Bank of Korea found that 94 percent of South Korea's youth job losses between 2022 and 2026 sat in AI-exposed sectors — 285,000 fewer jobs for people under 30, led by a 31 percent fall in IT services ^[10]. US entry-level postings fell 35 percent between January 2023 and June 2025, steepest in software development, data engineering and financial analysis ^[11]. To be fair to the doubters, none of this was settled even recently. As late as mid-2025, Goldman Sachs, Brookings and the World Economic Forum found AI adoption had not triggered mass layoffs, and Goldman saw no sign the technology had moved most labor-market outcomes ^[12]. OpenAI's Sam Altman has called the layoff wave "AI washing": ordinary cost-cutting wearing an AI costume ^[7]. White House AI advisor David Sacks points to a 5.6 percent rise in new-graduate hiring in April 2026 as proof the collapse story is wrong ^[5]. And Bank of America is hiring 4,000 interns ^[6]. These read as exceptions inside a pattern the numbers now measure, not as an even split. The telling detail is how the restructure gets narrated. Announcing 3,500 job cuts in September 2025, ANZ chief executive Nuno Matos was unequivocal ^[13].

This decision has absolutely nothing to do with AI or any technology. — Nuno Matos

NAB's Andrew Irvine, by contrast, ran both lines in a single announcement: he called AI an epochal shift on the scale of the steam engine, electricity and the computer, yet attributed the 2,500 cuts to a slowing economy, higher rates and fintech competition — not to the technology the savings are funding ^[8]. The people who study how novices become experts describe what this does. At UC Santa Barbara, Matt Beane says the junior analyst, banker or educator no longer gets a shot at participating in the work, because the novice has become optional to it; Brookings' Molly Kinder warns that if every firm skips early-career training, the entire pipeline of talent starts to collapse ^[11]. The record shows the apprenticeship being priced out — its tasks absorbed, its floor thinned, its gate rebuilt to test supervision of the tools that absorbed the tasks. What it does not show is where the next generation of seniors is supposed to come from. Nothing in it says.