Files
Abstract
Novel artificial intelligence (AI)-based decision support tools (DSTs) promise to make pesticide application more efficient. However, the adoption of existing, non-AI, DST by farmers is low, and farmers seem to prefer recommendations from human advisors. Additionally, for medical applications, there is evidence of users’ reluctance against (potentially superior) AI-based recommendations - a phenomenon known as Algorithm Aversion. This study is the first to investigate Algorithm Aversion in the farming context specifically with respect to farmers' intention to use an AI-DST for wheat fungicide application. We conducted a preregistered online survey with a representative sample of German farmers in autumn 2024. The analysis is based on a novel Bayesian probabilistic programming workflow for experimental studies. The approach allows jointly analysing an extended version of the Unified Theory of Acceptance and Use of Technology (UTAUT) with a willingness-to-pay-experiment. We find that Algorithm Aversion plays an important role in farmers’ decision-making. Our results emphasize the importance of user-friendly tech design, inform extension services on resource allocation, and stress the need for policy to support AI-DST adoption. This is the first study quantifying Algorithm Aversion in farmers’ decision-making. It forms the foundation for future research on the underlying causes of Algorithm Aversion. Additionally, we show how probabilistic programming can improve experimental research.