Factories, Flowers, and Trees: Visual Metaphors of People’s AI Mental Models

Abstract

Despite efforts to build AI literacy programs for laypeople, solutions often overlook their existing perceptions and mental models of AI, which can play an important role in shaping how literacy programs are designed and taught. People's mental models of AI influence how they react to and engage with these systems, but they are difficult to capture. While studies in human-centered computing have relied on proxy or explicit techniques to capture mental models in human-AI interaction contexts, we focus on implicitly capturing and analyzing mental models through an open-ended design activity in which participants physicalize and visualize their understanding of AI using tangible materials such as Legos and Play-Doh. We conducted a workshop in which participants worked individually and in groups to build physical and/or visual representations of AI through design activities and educational interventions. We also administered pre- and post-surveys to gauge changes in participant perceptions. We found that participants' mental models of AI drew on analogies and metaphors, and extended beyond the commonly assumed ``black box'' framing. In this poster, we present a qualitative coding of participants' visual representations of AI, cataloging the themes and elements that appeared before and after our interventions.