arXiv cs.CY· William G. Resh, Yi Ming, Xinyao Xia, Michael Overton, Gul Nisa Gurbuz, Brandon De Bruhl·· 6 天前AI 评分26
生成式 AI 对美国联邦劳动力是互补、增强还是替代？基于 RAG 的数百个职业序列评估
Complementarity, Augmentation, or Substitutivity? The Impact of Generative Artificial Intelligence on the U.S. Federal Workforce
AI 导读
一项研究用多阶段 RAG 系统评估生成式 AI 对数百个美国联邦职业序列的近期限影响，按互补、增强、替代三个维度打分，结果显示生成式 AI 主要互补和增强联邦岗位，替代潜力有限。即便替代得分最高的职业，也仅体现常规子任务的部分自动化，而非岗位消失。研究据此反对以岗位被取代为前提的粗放裁员，主张定向再培训、基于能力的人才规划与人机协作。
正文 · 原文
Abstract:This study applies a multi-stage Retrieval-Augmented Generation system, which grounds large language model assessments in curated evidence, to evaluate generative AI's near-term impact on knowledge, skills, and abilities across several hundred federal occupational series. Each occupation is scored along three dimensions of complementarity (AI enhances existing competencies), augmentation (AI requires workers to transform competencies), and substitutivity (AI replaces human tasks). Results reveal that generative AI primarily complements and augments federal roles, with limited substitution potential. Even occupations scoring highest on substitutivity reflect partial automation of routine subtasks rather than position elimination. These findings caution against blunt workforce reductions premised on displacement and instead support targeted reskilling, competency-based planning, and human-AI collaboration. The framework is adaptable to other organizations with structured competency models and job documentation.
| Comments: | 58 pages, 11 figures, 3 tables |
| Subjects: | Computers and Society (cs.CY); General Economics (econ.GN) |
| ACM classes: | I.2.7; I.2.11; I.2.1; I.2.3; I.7; J.4; J.1 |
| Cite as: | arXiv:2503.09637 [cs.CY] |
| (or arXiv:2503.09637v2 [cs.CY] for this version) | |
| https://doi.org/10.48550/arXiv.2503.09637 arXiv-issued DOI via DataCite |
Submission history
From: William Resh [view email]
[v1]
Wed, 12 Mar 2025 01:19:52 UTC (2,444 KB)
[v2]
Thu, 8 Oct 2026 20:11:57 UTC (2,256 KB)
来源：arXiv cs.CY · arxiv.org