IA · 25 August 2026 · 4 min read
Stanford Study: AI Disproportionately Hits Entry-Level Jobs with 19% Slump
In brief: The impact of artificial intelligence on the labor market is not showing up as a tide of across-the-board layoffs, but as a steep barrier to entry for early-career professionals. Updated research from Stanford University economists analyzing real-world payroll data shows employment for workers aged 22 to 25 in AI-exposed roles is 19% lower than in protected fields. While senior staff leverage AI to amplify output, the foundational tasks traditionally assigned to junior workers are increasingly handled by automated agents.
by Team Mocchi's
For years, debates over the economic impact of artificial intelligence swung between apocalyptic forecasts of mass joblessness and optimistic promises of new industry creation. Real-world empirical data, however, reveals a more nuanced and asymmetric reality: automation is steadily pulling the bottom rungs off the corporate career ladder.
An updated study by Stanford University economists led by Erik Brynjolfsson — titled "Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence" — examined aggregated payroll data from HR services provider ADP, cross-referencing it with occupational AI exposure benchmarks and the Anthropic Economic Index. The findings illustrate a stark generational divide.
A 19% employment gap for workers aged 22 to 25
As reported by Ars Technica, overall employment across the entire workforce in heavily AI-exposed fields has not seen catastrophic drops. The dynamic shifts dramatically, however, when focusing on workers aged 22 to 25: within the top 40% of occupations most affected by AI, junior headcount has dropped approximately 11% since 2022. In contrast, employment for the same age group in the least affected 60% of occupations expanded by 10%.
This translates to a 19% relative employment deficit today, up from the 13% gap identified in the previous edition of the study a year ago. The contraction is most pronounced in software engineering, copywriting, customer support, and administrative data processing — fields where frontier large language models and autonomous agents have achieved operational maturity.
Why seniors thrive while juniors get left behind
Stanford’s analysis explains how enterprises are restructuring their teams around AI capabilities. Experienced professionals possess organizational context, domain intuition, and architectural judgment required to direct and audit generative models; for them, AI functions as a productivity multiplier that cements their value.
Conversely, the tasks historically assigned to entry-level hires — drafting boilerplate code, initial research, document classification, and first-line triage — are now handled instantly and cost-effectively by automated agents. Organizations are choosing to bypass hiring for these junior positions, minimizing onboarding costs while delegating routine chores to software.
The long-term risk of a broken career ladder
This trend creates a systemic issue for the tech industry and the broader knowledge economy. If junior workers are not brought into active workflows to handle foundational work, the natural progression through which raw talent develops into senior expertise breaks down.
The researchers describe young workers as "canaries in the coal mine": representing the most fluid and least contractually protected segment of the labor market, they serve as an early indicator of how enterprise automation will reshape organizational structures. Without intentional training frameworks, businesses risk facing a severe deficit of senior talent within the next decade.
Mocchi's take
For technology agencies and enterprise software teams, this study serves as a critical call to rethink talent pipelines. Cutting junior positions simply because generative models handle basic coding or documentation is a short-sighted strategy that will hollow out the senior engineering base required five to ten years down the road. Our industry must revamp early-career onboarding: instead of training juniors on manual repetitive tasks, we need to mentor them from day one on architectural design, critical output validation, and the oversight of autonomous AI agents.