Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of AI (arxiviq.substack.com)

A landmark empirical study from Stanford researchers provides one of the first large-scale, near real-time analyses of AI’s impact on the U.S. labor market using payroll data from ADP, America’s largest payroll provider. Tracking individual-level employment from 2021 through mid-2025, the study reveals a significant 13% relative decline in employment among early-career workers (ages 22-25) in occupations highly exposed to generative AI since late 2022. Crucially, the negative effects are concentrated in roles where AI automates tasks—substituting human labor—rather than augments it, highlighting that the nature of AI integration fundamentally shapes workforce outcomes. This paper’s findings are significant for the AI/ML community because they move beyond theoretical speculation, offering robust, data-driven evidence of AI’s uneven disruption. Young, entry-level workers—reliant on codified knowledge easily replicated by AI—emerge as vulnerable "canaries in the coal mine," whereas experienced workers who leverage tacit knowledge remain less affected. Using a Poisson event study regression that controls for firm-level economic shocks, the authors isolate AI exposure as a key driver of these employment shifts, signaling urgent implications for education, business strategy, and policy. The research underscores the need to pivot workforce training toward AI-complementary skills like creativity and complex problem-solving, and advises businesses to favor augmentative AI deployment to preserve early-career opportunities. Ultimately, the study sets a critical benchmark for understanding AI’s tangible, immediate labor market effects amid an evolving technological landscape.