Generative AI as Seniority-Biased Technological Change (papers.ssrn.com)

A new empirical study finds that generative AI adoption is producing a seniority‑biased shift inside U.S. firms: using résumé and job‑posting records for nearly 62 million workers across 285,000 firms (2015–2025), researchers identify adopters by flagging dedicated “AI integrator” job postings and estimate effects with difference‑in‑differences and triple‑difference designs. They document a sharp divergence beginning in 2023Q1—junior employment in adopter firms falls relative to non‑adopters while senior employment continues to rise. Crucially, the junior decline is driven mainly by reduced hiring rather than increased separations, signaling a change in entry‑level demand rather than mass layoffs. The paper also maps heterogeneity that matters for AI/ML practitioners and policymakers: wholesale and retail trade show the largest junior‑level declines, and educational effects follow a U‑shaped pattern—mid‑tier graduates suffer the biggest losses while elite and low‑tier graduates are less affected. These findings imply that generative AI is reshaping career ladders by compressing or slowing entry pipelines, altering skill demand at different seniority layers, and reshaping internal labor allocation. For the AI/ML community, this points to the importance of designing adoption strategies, reskilling pipelines, and evaluation metrics that account for changes in hiring patterns and the uneven distributional impacts across sectors and education strata.