识别人工智能开发人员

Executive Summary Assessing the artificial intelligence workforce is critical for developing effective trainingpipelines and policy. Assessment requires reliable measures of AI talent demand andsupply. Most existing methodologies adopt an occupation-or skills-based approach toidentify the AI workforce, both of which can overestimate the size of the AI workforceas they blur together three different labor markets: (1) people building AI systems, (2)people adopting AI tools in other roles, and (3) workers whose tasks are exposed toAI-enabled change. In this report,we apply a more precise definition of theAIdevelopment workforceto assess the U.S. demand and supply of people who design,train, fine-tune, scale,and deploy AI systems. We defineAI development jobs as roles that require specialized knowledge, skills, andabilities and directly contribute to the technical development of AI systems.1Using thisdefinition, we developeda machine learning(ML)classifier that identifies AIdevelopment roles in job postings data. Weranour classifier over a dataset of U.S. jobpostings from January 2010 to February 2026 to estimate the demand for AIdevelopment roles. We then estimatedthe size of the AI development workforce usinga dataset of U.S. employee profiles. Wefound: ●Approximately1.6 million AI development job postingsin theUnited Statessince 2010, including 331,445 postings in 2025.●Approximately519,000 AI development workersin theUnited Statesas ofMarch 2026.●AI development roles are a small portion of the total U.S. workforce, accountingfor less than 1% of both total labor demand and employment.●AI development is concentrated in highly technical occupations, althoughamong these occupations the proportion of roles that directly support AIdevelopment varies widely. These estimates are substantially smaller than previous counts of the AI workforcebecause we disaggregated AI development roles from the broader workforce adoptingAI tools in other roles and workers whose tasks are exposed to AI-enabled change.The technical workforce that develops AI systems is smaller and more specialized thanpreviously understood. Our future publications and talent tracking tool,PATHWISE,will incorporate our AI development jobs definition and provide an empirical basis onwhich to develop workforce and education policy recommendations to bolster the AIworkforce. Introduction The AI workforce is foundationalto crafting anycomprehensive national AI strategy.Governments design policy to grow it, companies compete for it, and researchers trackit to gauge national capacity. Given its importance to national competitiveness andeffective workforce development, it is essential to clearly define and measurethispopulation. Existing efforts to measure the AI workforce adopt an occupation-or skills-basedapproach to identify the AI workforce. Although well-established and replicable, bothapproaches can overestimate the AI workforce, including an overly broad set of AI-related roles.2These approaches miss cutting-edge technical roles that do not fit intoexisting occupation or skills buckets. We developeda methodology tospecificallyidentify theworkforce driving the design, training, and deployment of AI systems.Ourfocus on workers involved in technical AI development provides a moreprecise AIworkforceanalysis and enables more targeted policy analyses and recommendations. We define AI development jobs as roles that require specialized knowledge, skills, andabilities (KSAs) andthatdirectly contribute to the technical development of AIsystems––for example, systems built around ML models, including natural languageprocessing (NLP), computer vision,and large language models (LLMs). Using thisdefinition, we quantitatively identified the demand for,and current employment of,thetechnical AI workforce in theUnited States.In this report, we describe ourmethodology for identifying AI development jobs and the current employment of AIworkers, and present initial findings. Future publications on this data will includeanalysesas well astargeted workforce and education policy recommendations for theAI workforce. U.S. Job Postings and Profiles Data We usedU.S. job posting and worker profile data provided by Lightcast.*Each jobpostingcontainsthe full job description, including job duties and responsibilities. Eachworker profile includes a worker’s employment and education history and acquiredskills. Lightcast enhances both datasets by mapping them to federal workforce taxonomies such as theU.S.Bureau of Labor Statistics’ Standard Occupational Codes(SOC) and Lightcast’s in-house occupation and skills taxonomies. Measuring Demand for AI Talent An important aspect in understanding the AI workforce is defining what constitutes anAI job. As outlined inrecent CSET analysis, we define AI development jobs as rolesthat require specialized KSAs andthatdirectly contribute to the technical developmentof AI systems.*Technical development work includesthedevelopment of models,