The Minds Behind the Prompts: How Mental Representations Shape User Interaction with Generative AI

This study investigates how users’ mental representations of generative AI influence their interaction patterns with AI systems. Through an online survey with 211 participants, we analyzed users’ perceptions of AI using adjective ratings and role classifications, followed by analyses of their preferred prompting styles across task scenarios. Using k-means clustering, we identified four groups of AI users (optimistic pragmatists, neutral observers, skeptical realists, and enthusiastic advocates) and three mental representations of AI (all-rounder, helper, and tool). Chi-squared analyses revealed significant relationships between mental representations and prompting styles in the essay-writing task, where users who had more faith in AI tended to adopt conversational and co-creative prompting styles, while those viewing AI as merely a tool favored directive approaches. By integrating perspectives from cognitive psychology and speech communication, this study provides empirical evidence linking users’ mental representations of AI to their communication styles, contributing to the theoretical development of human-machine communication.