A new working paper on arXiv builds a labor market in which AI systems enter as workers. In the model, AI agents apply for jobs, pass screenings, get hired, and can be dismissed, just like human workers. Firms post vacancies, and both human and AI workers search with friction, meaning matches are not instantaneous and depend on market conditions.
The framework is a search-and-matching model, a standard tool in labor economics, but with AI as an explicit participant. The paper's title emphasizes that displacement is endogenous: rather than assuming AI replaces workers as an external shock, the model lets displacement arise from the interactions of firms, human workers, and AI applicants. The abstract also mentions product demand slopes, suggesting the model accounts for demand-side effects, though details are limited.
Because the source is only an abstract, no results or quantitative findings are available. The contribution so far is conceptual: a way to think about AI not as a tool or a shock, but as a worker inside the same labor market institutions that govern hiring and firing.