"Generate" the Future of Work through AI: Empirical Evidence from Online Labor Markets

Large language model-based generative AI is reshaping online labor markets, producing a pronounced displacement effect and an overall contraction in submarkets where required skills align with core LLM functionalities, according to a study leveraging data from a leading online labor platform [1]. The study, authored by Xingchen Xu and submitted to arXiv on 9 Aug 2023, documents that both demand and supply decline in affected submarkets, but the reduction in supply is comparatively smaller, intensifying competition among freelancers [1]. This heightened competition is especially pronounced in programming-intensive submarkets, a pattern the researchers attribute to skill-transition effects: by lowering the human-capital barrier to programming, ChatGPT enables incumbent freelancers to enter programming tasks [1]. The analysis further shows these transitions are not homogeneous, with high-skilled freelancers contributing disproportionately to the shift [1]. The findings arrive amid a long-running debate over technological unemployment, a term popularized by John Maynard Keynes in the 1930s, who described it as "only a temporary phase of maladjustment" [7]. Historical examples include artisan weavers displaced by mechanized looms, and contemporary cases range from retail cashiers replaced by self-service tills to broader automation of intellectual tasks [7]. The World Bank's 2019 World Development Report argued that while automation displaces workers, technological innovation creates more new industries and jobs on balance [7]. The online labor platform data analyzed in the study reflects dynamics within global value chains, where firms outsource design, production, and distribution activities across borders [9]. As states become increasingly vulnerable to economic interdependence, fewer are relying on global value chains and instead are pushing for protectionist policies, creating uncertainty over the future structure of these markets [9]. The study's emphasis on strategic responses by market participants aligns with insights from behavioral economics, a field that examines how psychological factors cause decisions to deviate from traditional economic models [8]. Behavioral economics, which began as a distinct field in the 1970s and 1980s, is increasingly used to understand real-world decision-making in areas such as market design and public policy [8]. The paper has undergone multiple revisions, with the latest version posted on 25 Jun 2026 [1]. The research was conducted within arXivLabs, a framework allowing collaborators to develop and share new arXiv features while adhering to values of openness, community, and user data privacy [1].