Understanding how AI is reshaping work requires an accurate account of the tasks workers perform. Workers are often the first to see how their jobs are changing, as new responsibilities and tools often appear in everyday work before these changes reach formal descriptions. Interviews with workers can help capture these emerging changes in job tasks. Yet existing methods support standardized ratings across workers, while open-ended conversations with workers remain costly to collect and synthesize at scale.
We introduce the Stanford Task Network (TaskNet), an occupational taxonomy grounded directly in workers’ accounts of their everyday work. To construct TaskNet, we develop a mixed-initiative system that induces task inventories directly from workers. The system conducts interviews, standardizes reports into individual task lists, and aggregates these lists into occupational task inventories.
Applied to six occupations spanning a range of education requirements and AI exposure, our system interviews 132 workers and produces 202 task statements from 2726 reports, all of which were accepted by O*NET Center analysts, compared to roughly 65 % of workers’ write-in task suggestions collected through the current O*NET data collection process, a major alternative source for emerging tasks. Every statement is validated by 318 incumbents in 7923 ratings of relevance and importance. Relative to the task data from the O*NET 31.0 database for the sampled occupations, TaskNet covers 75% of the 118 published O*NET tasks and adds 79 tasks with no O*NET counterpart and 58 that refine existing statements. It covers 89 % of tasks reported by held-out workers, compared with 79 % for O *NET, and a larger share of the activities described in employer job postings (72.5% vs. 70.2%). Downstream, TaskNet changes AI exposure estimates in both directions, raising them for Software Developers and Lawyers, two occupations where AI’s effects are widely debated. Because such estimates guide workforce policy and training, more complete and up-to-date task lists can help direct these efforts toward the workers most likely to be affected. TaskNet provides a foundation for keeping occupational information current and grounded in workers’ experiences as AI reshapes work.