A new study reveals a hidden cost of artificial intelligence: the disappearance of entry-level jobs. Workers aged 22 to 25 in AI-exposed occupations experienced a 16% relative decline in employment after generative AI spread, according to a November 2025 working paper from the Stanford Digital Economy Lab. More experienced workers in the same roles did not see the same drop. Jobs with low AI exposure also held steady.

The finding points to a quiet shift in how companies use AI. Firms appear to be replacing the junior tasks that once gave people their first foothold. Software developers, customer service representatives, computer programmers and information systems managers face the biggest changes.

The Data Behind the Trend

The Stanford study controlled for other factors that influence hiring. A separate report from Anthropic in March 2026 supports the conclusion. The Federal Reserve Bank of New York adds more context: the unemployment rate for recent college graduates rose to 5.6% in late 2025. The underemployment rate hit 42.5%, the highest since the pandemic.

Young workers now often submit hundreds of applications before getting an offer. Surveys show elevated anxiety, financial precarity and burnout among those in extended job searches. If AI closes the door on early jobs, the consequences will include delayed independence and postponed family formation.

Why This Matters

Entry-level jobs are the economy's training system. Junior analysts learn which numbers to trust. Young developers learn how production systems fail. New marketers observe customer behavior beyond dashboards. Early-career legal and financial staff see how rules, judgment, deadlines and relationships interact. When AI absorbs drafting, triage, coding and administrative preparation, firms gain short-term efficiency but society loses long-term capability.

The personal toll is real. Graduates face delayed careers and financial strain. The broader economy risks a less capable workforce in the future.

The 'Learn to Code' Myth Unravels

For over a decade, the advice to young workers was clear: learn to code. Federal initiatives and university expansions built on that premise. The premise no longer holds. AI handles the layer of work those programs targeted, translating specifications into routine code, reproducing patterns and debugging predictable errors. Supervising AI systems and verifying their output are now more relevant skills.

A Call for Change

Universities, community colleges and professional programs must embed AI literacy, data literacy, prompt-based workflow skills and domain judgment into ordinary degrees. Every graduate should understand how to use AI tools, check their output and combine them with human expertise. Governments need to incentivize businesses to hire and train early-career workers. Companies must recognize that building a long-term workforce experienced in AI starts with entry-level hires.

Students also bear responsibility. They must become fluent in AI and learn to apply that knowledge across fields. The traditional path through entry-level work is shifting. Adapting now can prevent a deeper crisis.