Skip to main content

POID Working Paper

When GenAI increases inequality: Evidence from a university debating competition


This paper evaluates the impact of Generative Artificial Intelligence (GenAI) on productivity and work inequality. While existing literature has focused on the impact of GenAI on writing tasks, I exploit a unique setting to analyse productivity improvements on a task involving verbal interactions and requiring high cognitive and social skills. I run a Randomized Controlled Trial with undergraduate students in a debating competition. Participants were randomly assigned to either GenAI support for debate preparation or to conventional internet resources. The small overall impact of GenAI I find hides significant heterogeneity between students. In particular, and contrary to previous findings, high-performing students and those with a stronger academic background benefit significantly more from GenAI than their lower-performing counterparts. Moreover, high-skilled individuals with access to GenAI experience large improvements in their perception of time needed to prepare debates, while low-ability students do not. I suggest a possible explanation to reconcile these results with previous literature: when tasks require higher-order skills, and answers cannot be directly extracted from the AI and copy pasted, high-skilled workers are likely to benefit more of the advantages of GenAI. These findings could have important implications for the broader understanding of potential economic and social impacts of these technologies.


Antonio Roldan

18 June 2024     Paper Number POIDWP096

Download PDF - When GenAI increases inequality: Evidence from a university debating competition