The Effect of Exposure to Generative AI on Crowdworkers’ Adoption: A Field Experiment on Amazon Mechanical Turk

The disruption of generative AI (GenAI) enables machines to produce content resembling human output, providing workers on online labor platforms (OLPs) with assistance in crowdwork. However, the promise of integrating GenAI into crowdwork is not simply taken for granted due to the inexplicable nature of GenAI and workers’ skepticism about GenAI tools being viable for the crowdwork they have been performing. To investigate the potential diffusion of GenAI for crowdwork on OLPs, this study examines whether and why workers are willing to adopt GenAI for crowdwork. Drawing upon the Diffusion of Innovation theory, we design a field experiment that manipulates crowdworkers’ exposure to ChatGPT, a representative GenAI tool leveraging large language models. The experiment investigates how crowdworkers perceive the trialability, observability, compatibility, and relative advantage of GenAI, and how these perceptions affect their adoption behavior on OLPs. Furthermore, by distinguishing between early and late adopters of innovation, this study uncovers the heterogeneous effect of GenAI exposure. Our research expects to contribute to the research on OLPs by illuminating the mechanisms driving crowdworkers’ adoption of GenAI and offer practical implications to OLP stakeholders seeking to integrate GenAI into crowdwork effectively.