Genai is deleting the entry-level rung and the résumés pile up
The first rung of the career ladder has turned into a trapdoor. Instead of stepping up, new graduates are dropping straight into a statistical void: hiring of 22- to 25-year-old software developers in the U.S. has fallen almost 20 % since 2022, according to aggregated LinkedIn and Indeed data. The culprit is not a recession; it’s the generative models that learned to do the junior chores faster and cheaper than any human intern.
BBVA’s research unit drilled into the payrolls of 2,300 companies and found the smoking gun: productivity per worker jumped 4 % after GenAI adoption, yet headcount growth for under-30s flatlined. The bank’s economists stress the nuance—this is not mass firing, it is hiring freeze by algorithmic substitution. The tasks that once taught rookies how to read legacy code, draft pitch decks or tag legal precedents are now one prompt away from a cloud API.
The canary cohort is already coughing
Stanford’s 2025 “AI & Labor” report baptised Gen Z workers the “canaries in the coal mine” because their daily routines sit in the bull’s-eye of large-language-model competence: summarisation, templating, translation, basic scripting. The study isolates a 13 % relative drop in employment for young professionals in highly exposed occupations even after controlling for sector and macro shocks. Wages for those who keep their seats have not budged downward; the adjustment is purely quantitative—fewer bodies at open-plan desks.
Mark Zuckerberg told investors in February that Meta’s internal coding assistant now handles “the bottom half of our complexity curve,” the exact turf where junior engineers cut their teeth. Anthropic CEO Dario Amodei goes further: he predicts half of all entry-level white-collar roles evaporate within five years, pushing office unemployment into double digits. The prediction feels less outrageous each quarter. U.K. job board Adzuna logged a 30 % plunge in listings tagged “graduate,” “entry,” or “junior” since ChatGPT’s public release; Canada’s youth-employment rate is back to 1998 levels.
Spain never fixed its ladder; now someone sawed off the bottom
Spain, already the EU’s youth-unemployment champion at 25 %, faces an extra twist. Eurostat shows only 76.5 % of Spanish graduates find any job within three years of tossing the mortarboard—below the bloc’s 80 % average and sinking. The country relies on temp agencies and training contracts to absorb first-time workers; both are precisely the channels GenAI hollows out first. A Madrid-based HR director at a global consultancy admitted on background that his firm now “hires seniors or nothing,” because client contracts demand immediate productivity and AI handles the gap.
The long-term scar is measurable: the Center for American Progress calculates six months of joblessness at 22 translates into $22,000 lost cumulative earnings over the next decade. Compound that with Spain’s 27 % youth rate and the equation turns brutal. Meanwhile, vacancies for senior roles on Indeed Spain rose 4 % year-over-year; entry-level offers shrank 7 %.
Exceptions exist, but they sound almost medieval. Nutritionists face 0.4 % unemployment—algorithms still can’t plate hospital meals. Construction site supervisors hover at 0.7 %. Computer science graduates, once the safe bet, wrestle 7.5 % unemployment, the highest of any technical degree. The irony stings: the people who trained the machines can’t get past HR’s firewall.
Companies are not hiding the shift. A leaked Amazon internal memo last month instructed managers to “prefer AI output over L4 coding assignments,” referencing the level that normally swallows fresh B.Sc.s. Google’s staffing plan for 2026 shows 40 % fewer intern slots in software, offset by a 15 % rise in AI-research Ph.D. recruiting. The message is unambiguous: come back when you can train the model, not when you need it to train you.
Policy answers are late and timid. The EU’s fresh AI Act says nothing about displaced juniors; Spain’s new digital talent visa targets senior engineers from Latin America, not its own graduates. Universities keep selling four-year degrees whose first employer no longer exists.
The upshot? A generation that did everything right—degrees, languages, GitHub portfolios—walks into offices where the receptionist is a chatbot and the only human vacancy demands five years of experience. The ladder hasn’t just moved; it’s been uploaded to the cloud, accessible only to those who never needed the first rung.