Indexing the Impact of AI within the O*NET System: A Review of Methods and Development of Recommendations
Citation. Putka, D. J., Voss, N. M., & Lewis, P. (2026, June 18). Indexing the impact of AI within the ONET system: A review of methods and development of recommendations* (Report No. 051). National Center for O*NET Development.
Summary. This National Center for ONET Development report reviews scalable methods for estimating AI’s impact on work and recommends a 16-index suite for the ONET system. The proposal grounds indices in job analysis and a contemporary job-performance model, separating AI capability from observed use, automation from augmentation, and task, KSAO, vacancy, contextual-performance, and occupation-level units. It presents the suite as a measurement-development proposal requiring reliability and validity evaluation, not as evidence that the proposed scores already represent realized workforce effects.
Key claims
- Of the 19 core AI-impact studies reviewed, 16 used O*NET data in some way; the report classifies methods by AI operationalization, initial target (tasks, KSAOs, or vacancies), and occupation-level aggregation strategy.
- Task-centric AI-impact estimates can overstate AI’s relationship to whole-job performance because contemporary performance models include contextual performance as well as task performance and because jobs contain interdependencies that simple element-level aggregation can miss.
- The proposed 16-index suite spans declarative knowledge, skills, software skills, task performance, contextual performance, overall job performance, and vacancies, with separate automation and augmentation indices where applicable.
- The report recommends rating composites across multiple LLMs and repeated runs to estimate model- and run-specific error, then evaluating the indices through a preregistered-style nomological network that tests expected convergence and discrimination among indices and against external measures.
- Vacancy data can offer timely, within-occupation variation but should not be treated as a comprehensive or quality-assured job description without explicit inclusion and quality rules.
Evidence & limitations
- This is a June 2026 National Center for ONET Development report prepared for the ONET program, not a peer-reviewed empirical study; its 16 indices are recommendations rather than validated deployed measures.
- The report’s method review covers 19 core studies and its proposed index architecture is conceptual; it does not estimate the indices or establish realized effects on productivity, wages, staffing, learning, or worker experience.
- The report is public-research plausible as an official ONET-program report, but no stable public document-level landing page or anonymously downloadable copy was located during ingest; the frontmatter URL therefore identifies the official ONET Center rather than a verified report page, and the ingested Drive binary is retained by hash for provenance.