The Hidden Crisis in Employment for Young Workers Exposed to Generative AI
Since the widespread availability of ChatGPT, workers aged 22 to 25 in occupations most exposed to generative AI have experienced a striking 16 percent reduction in employment, a fact that remains obscured in headline economic data. Older workers in the same roles have not seen this decline, highlighting an age-specific and occupation-specific disruption that is quietly reshaping the labor market. This phenomenon is not reflected in the aggregate unemployment rate, which remains deceptively stable, masking the significant damage to the entry-level tier of the workforce.
While broad economic indicators imply a healthy labor market, the reality facing young workers at the start of their careers is starkly different. The traditional “bottom rung” of the employment ladder, historically crucial for skill development and career progression, is collapsing. Standard reassurances that technological advancements create more jobs than they destroy, though technically accurate, miss the point entirely. The looming crisis is not a dystopian vision of robots replacing humans on a large scale, but rather a generation struggling to secure the foundational jobs that once shaped their professional judgment and expertise.
The Numbers Behind the Unemployment Rate
At first glance, headline figures suggest no cause for alarm. Yet, when focusing on recent college graduates, a different and concerning picture emerges. According to the Federal Reserve Bank of New York, the unemployment rate for recent graduates rose to 5.6 percent in the fourth quarter of 2025, while underemployment surged to 42.5 percent—the highest level since the pandemic. Underemployment, a more revealing metric, describes graduates working in roles unrelated to their qualifications, such as philosophy majors in retail or computer science graduates driving rideshare vehicles. These are individuals who followed the prescribed educational path, only to find the labor market no longer reciprocating their investment.
Researchers, controlling for firm-specific shocks, excluding tech companies, and even removing remote-friendly occupations from the analysis, find that this pattern persists. Early evidence suggests that in AI-exposed fields, workforce adjustments are occurring primarily through employment reductions rather than wage cuts, particularly in roles where AI substitutes rather than complements human labor.
Supporting this, a March 2026 labor market study by Anthropic, which analyzed usage patterns of its AI models, found similar trends. Entry-level roles in software development, customer service, and programming are thinning out rapidly. The jobs that traditionally served as gateways to industry competence are disappearing.
The Training-Ground Problem
What distinguishes this wave of automation from previous ones is the erosion of entry-level work, which has always been more than just a paycheck. Entry-level roles serve as the critical training ground where young professionals acquire tacit knowledge and practical skills that cannot be taught in classrooms. For example, a junior associate learns to spot problematic contract language after repetitive exposure; a junior developer understands bug patterns linked to architectural decisions; an analyst deciphers what senior partners expect in presentations. These lessons are absorbed through hands-on experience, often disguised as grunt work.
Generative AI excels precisely at these routine, lower-value tasks, undermining the very foundation upon which expertise is built. The optimistic narrative that AI frees humans for higher-value work assumes that newcomers can bypass the grunt work and still develop the necessary skills. Unfortunately, this assumption does not hold. Higher-value contributions depend on years of mastering foundational tasks. Removing the “bottom rung” of this ladder leaves the upper levels inaccessible to new entrants.
As a result, a 22-year-old denied a junior role today loses not only immediate income but also the essential sequence of experiences that would have shaped their competence by their late twenties. Firms that cut junior hiring now are unwittingly starving their future talent pipelines, leading to a shortage of mid-career professionals a decade hence. MIT Technology Review recently emphasized this point, framing entry-level hiring as an investment in future workforce quality, not merely a cost to be minimized.
Why Coding Bootcamps No Longer Guarantee Job Security
For over a decade, coding bootcamps were the standard policy response to technological displacement. Governments and universities invested heavily in these programs, promising that software skills would safeguard workers from automation’s worst effects. However, this promise has not aged well. The coding tasks taught—building CRUD applications, wiring APIs, writing test cases—are now well within the capabilities of current AI language models.
Graduates proficient only in these routine tasks find themselves competing with subscription-based AI tools capable of performing the same work more efficiently. This does not render coding useless, but junior coding jobs have become scarcer, a nuance often lost in career advice over the past three years.
The most resilient career strategy combines domain expertise with AI fluency. For instance, mechanical engineers who understand manufacturing and can direct AI models, or programmers with deep knowledge of financial regulations who can harness AI effectively, possess a rare skill set. AI alone cannot discern which industry-specific questions matter, nor can domain experts operate AI at the pace demanded by today’s markets. This hybrid expertise is increasingly scarce and valuable.
The Real Competition: Human Versus AI-Augmented Human
Media narratives often frame AI’s impact on jobs as a battle between humans and machines, but this oversimplification misrepresents the labor market dynamics. The actual competition is between colleagues who use AI tools effectively and those who do not. A graduate who leverages AI to complete a week’s work in a day displaces peers who lack such fluency.
This dynamic echoes the productivity revolution brought by spreadsheets in accounting during the 1980s, where the most successful professionals viewed the technology as a tool rather than a threat. However, the pace today is much faster. Data from the U.S. Bureau of Labor Statistics shows a slowdown in hiring for white-collar roles that traditionally expanded steadily across economic cycles. Companies aren’t conducting mass AI-driven layoffs but are instead quietly not replacing junior staff who leave and withholding the creation of new junior roles.
This subtle shift manifests as graduates sending hundreds of applications before receiving a single offer. Unlike a recession, which depresses overall employment and wages, this is a structural realignment of a specific career stage, unfolding faster than universities, employers, or policymakers can adapt.
Universities’ Missteps in Preparing Students for an AI-Driven Economy
Universities have largely responded by adding isolated AI-related courses or seminars, such as “AI for Business” or “Generative AI in the Humanities,” or by imposing coding requirements on non-technical majors. Treating AI as a discrete subject is insufficient.
AI literacy must become a foundational component woven into every discipline from the outset, because every profession now integrates these tools into its practice. For example, law schools that teach contract drafting without incorporating AI-assisted drafting techniques are setting students up to fall behind peers who master both. Similarly, journalism programs must teach students to use AI for research, transcription, and verification alongside traditional reporting skills.
Moreover, the bachelor’s degree no longer guarantees readiness for entry-level roles as it once did. Employers historically relied on first jobs to provide the final layer of training, but as AI absorbs many of those training tasks, this system is breaking down.
The Policy Gap Nobody Dares to Name
Governments worldwide have invested heavily in AI infrastructure, safety research, and national strategies. Yet, by comparison, funding to incentivize employers to hire and train early-career workers in this new environment remains negligible. This imbalance is concerning.
Computing resources and AI development will continue regardless because their economics are compelling. What requires deliberate policy intervention is sustaining an apprenticeship system when entry-level tasks vanish. Effective measures might include wage subsidies for genuine training positions, tax reforms recognizing junior hires as capital investments, and accreditation systems that distinguish meaningful workforce development from mere credential inflation.
Countries that address this challenge proactively will gain a generational advantage. Those that do not may face a dearth of mid-career professionals around 2032 to 2034, reflecting the absence of hires from 2025 and 2026. By then, retroactive fixes will be difficult and costly.
A Longer Historical Perspective
This disruption is not without precedent. Economist John Maynard Keynes famously predicted in 1930 that his grandchildren would work roughly 15 hours per week by the early 21st century, an idea that anticipated massive productivity growth. Technologically, Keynes was right; socially, he was mistaken. Instead of leisure, increased productivity fueled consumption, status competition, and longer working hours for those still employed.
Today’s situation echoes this pattern. Productivity gains from AI are real, but how those gains are distributed depends on institutional choices rather than technology alone. If entry-level hiring continues to be treated as a discretionary cost eliminated by AI, benefits will flow disproportionately to capital owners and credentialed elites. Conversely, if treated as a foundational investment in workforce development, productivity gains can be broadly shared.
Ultimately, these outcomes are policy decisions being made now, often by default rather than design.
Pragmatic Strategies for Young Workers
Given this structural challenge, simple advice like “learn AI” or “gain domain expertise” is necessary but insufficient. The critical combination is domain expertise plus AI fluency, but acquiring this combination is difficult when the entry-level roles that historically taught domain skills are disappearing.
Successful graduates employ three strategies. First, they develop narrow, concrete AI applications addressing real problems in specific fields—for example, automating a step in legal discovery or fine-tuning models for regulatory compliance. This portfolio substitutes for the apprenticeship firms no longer provide.
Second, they target smaller employers earlier in their careers. Large firms are cutting junior roles aggressively because they can integrate AI at scale. Smaller and regional firms still hire and train juniors, though they often cannot match large-company salaries. Graduates who gain four years of real experience at these firms will outcompete peers stuck in endless application cycles.
Third, they treat AI fluency as a baseline skill and honestly acknowledge that AI outpaces them in mechanical tasks. A junior’s true value now lies in judgment—knowing which questions to ask AI, assessing output reliability, and deciding which insights to escalate to human decision-makers. Though demanding senior-level skills from juniors is unfair, it reflects the evolving nature of entry-level work.
What Firms Should Stop Pretending
The allure of cutting costs by replacing junior analysts with AI models is understandable. It delivers immediate savings while maintaining output levels. However, this approach is short-sighted. Within a year or two, the firm’s pipeline of mid-level analysts dries up; within five years, senior analysts vanish because there were no juniors to promote.
The talent pipeline is not just a metaphor—it is a literal sequence of jobs. Removing any link breaks the chain and eventually undermines the entire structure. Firms that view junior hiring as a strategic, long-term investment will be stronger in 2030, despite short-term spreadsheet metrics suggesting otherwise. Those optimizing it away are hollowing themselves out quietly.
This lesson is difficult to communicate to CFOs, whose incentives reward immediate cost savings while the consequences emerge years later, often after they have moved on. The question firms must confront is profound: what happens when they discover they can operate quarter to quarter without hiring anyone under 26? And once this practice becomes widespread and junior roles vanish for years, can the ladder be rebuilt, or will the missing rungs harden into a permanent gap?
The headline unemployment rate will continue to look healthy. It always does. But the deeper, more consequential story is already unfolding. The pressing question is whether anyone with the authority to intervene will act before this quiet crisis becomes permanent.
Read more Here.
