A typology of consumer relationships with generative AI: understanding the Good, the Bad, and the Ugly

This commentary proposes an interpretive framework for understanding consumer-GenAI relationships based on Relational Models Theory (RMT). It advances a typology of four elementary consumer-GenAI relationship models – Companion, Advisor, Collaborator, and Broker. Each relationship model captures a distinct set of relational expectations and norms that shape how consumers interact with, evaluate, and emotionally respond to GenAI systems. The typology is illustrated empirically using a large corpus of one million ChatGPT conversations, from which a random sample of 5000 human-ChatGPT dialogues was selected for analysis. Guided by principles of linguistic pragmatics and regex matching, the analysis demonstrates how relational models enact in everyday prompting language. The four models are then interpreted through the metaphor of the Good, the Bad, and the Ugly, offering insight into the social dynamics that underlie consumer-GenAI relationships. The paper derives implications for consumer experience designers, GenAI tool developers, and policymakers by highlighting the importance of relational sensitivity in GenAI system design, marketing positioning and regulatory approaches.