Exploring User Heterogeneity in Human Delegation Behavior towards AI

As artificial intelligence (AI) can increasingly be used to support decision-making in various areas, enhancing the understanding of human-AI collaboration is more important than ever. We study delegation between humans and AI as one form of collaboration. Specifically, we investigate whether there exist distinct patterns of human delegation behavior towards AI. In a laboratory experiment, subjects performed an image classification task with 100 images to be classified. For the last 50 images, the treatment group had the option to delegate images to an AI. By performing a cluster analysis on this treatment, we find four types of delegation behavior towards AI that differ in their overall performance, delegation rate, and their accuracy of self-assessment. Our results motivate further research on delegation of humans to AI and act as a starting point to research on human collaboration with AI on an individual level