Why 43% of recent graduates are underemployed and why the traditional degree-to-job pipeline no longer functions the way it was designed to.
America's entry-level job market is not keeping pace with the fields graduates are entering. Bachelor's degrees awarded annually grew from approximately 1.24 million in 2000 to 2.0 million in 2022, but the mismatch is not about raw volume — it is about concentration. Graduates are flooding into knowledge-work fields where AI automation is eliminating the first rung of the career ladder fastest, while entry-level job postings across those sectors have fallen 35% since early 2023. That is a structural condition, not a recession effect, and it is being made permanent by two compounding forces: algorithmic hiring screening out candidates before a human reviews them, and a degree system still steering students toward roles that were abundant when the curriculum was designed.
The headline number sits at 43%. Nearly 43% of U.S. college graduates aged 22 to 27 were underemployed as of December 2025, the highest rate since the pandemic, according to the New York Federal Reserve Bank. Unemployment for that cohort hit 5.7% in Q1 2026, well above the 4.2% overall rate. Industries that traditionally absorbed new graduates shed 9,000 jobs per month since 2023, and only 19% of graduates now say it is a good time to find a quality job, down from over 70% in 2022.
Most economists, including those cited in Bloomberg's April 2026 reporting on the underemployment surge, point not to AI but to a frozen hiring market as the primary driver. Both explanations contain substantial truth. A frozen market produced the crisis. AI is now structurally preventing recovery by removing the entry-level work that even a healthy hiring environment would need to restore first.
More than half of organizations currently constraining entry-level hiring cite AI-driven automation as the primary reason, according to a D2L survey published in University Business in May 2026. 30% of HR leaders are shifting hiring away from entry-level positions toward mid-level talent entirely. Big tech hiring of recent graduates fell 25% in 2024 versus 2023 and is down 50% versus pre-pandemic levels, according to SignalFire. PwC UK cut approximately 200 entry-level roles explicitly citing generative AI. LinkedIn's Chief Economic Opportunity Officer Aneesh Raman published a New York Times op-ed that captured the structural problem in one line: he saw the bottom rung of the career ladder breaking.
Degree supply grows while entry-level postings fall sharply. Created via Gemini.
The mechanism is not mass replacement but task elimination. AI now handles drafts, summaries, basic code, first-pass analysis, and document review that used to be assigned to junior staff. Those tasks were not just work. They were the training through which junior workers developed into senior ones. Removing them does not just cut headcount today. It severs the pipeline through which institutional knowledge transfers to the next generation of workers inside any organization.
The major mismatch problem compounds the crisis. Computer science graduates carry a 6.1% unemployment rate in 2026, the seventh highest of any major, despite commanding the highest starting salary at $80,000. CS degrees more than doubled from 51,696 in the 2013 to 2014 academic year to 112,720 in 2022 to 2023, flooding the market precisely as tech companies conducted more than 150,000 layoffs in 2024 and over 100,000 more in 2025. Experienced displaced workers now compete directly with new graduates for generalist software engineering roles that themselves dropped roughly 30% year-over-year according to Handshake data.
Healthcare tells the opposite story. Nursing graduates carry a 1.4% unemployment rate in 2026, the lowest of any major by a substantial margin, driven by an aging population and clinical staffing shortages that automation cannot resolve. Universities are not producing enough nursing graduates to meet healthcare demand, while producing computer science graduates at twice the rate the market can currently absorb.
CS unemployment at 6.1% versus nursing's 1.4% gap. Created via Gemini.
The result for working-class graduates is a credential that costs the same regardless of what it promises. A borrower who financed a computer science degree carries the same debt as one who financed a nursing degree, with structurally different outcomes. Workers who chose their major based on hiring boom conditions in 2022 and 2023 are navigating a market that shifted beneath them before they graduated.
Fixing the supply-demand gap requires changes at two levels simultaneously. Universities need real-time labor market data feeding into enrollment guidance, not four-year-old hiring projections from when current seniors were freshmen. Employers need to rebuild entry-level pipelines rather than eliminating them, because the senior talent they need in 2030 is currently the junior talent they are declining to hire in 2026.
A generation that cannot get a first job cannot develop into the experienced workforce an AI-intensive economy will still need to supervise, direct, and correct the systems doing the work.