"She's Like a Person but Better": Characterizing Companion-Assistant Dynamics in Human-AI Relationships

TL;DR

This study investigates digital companionship as an emergent spectrum between social companionship and task-oriented assistance in human–AI relationships. Using a mixed-methods design with $N=202$ survey respondents and $N=30$ in-depth interviews of high-engagement ChatGPT and Replika users, it reveals fluid use where users blend companion and assistant roles, underpinned by shared user characteristics across platforms. The findings show stronger anthropomorphism and affective responses to Replika, yet similar cognitive trust and potential spillovers to general AI attitudes across both systems. The work discusses bounded personhood, stigma, and societal implications, offering design and policy guidelines to support emotionally engaging yet properly contextualized AI companions. Overall, digital companionship emerges as a hybrid social infrastructure with broad implications for user well-being, human relationships, and AI governance.

Abstract

Large language models are increasingly used for both task-based assistance and social companionship, yet research has typically focused on one or the other. Drawing on a survey (N = 202) and 30 interviews with high-engagement ChatGPT and Replika users, we characterize digital companionship as an emerging form of human-AI relationship. With both systems, users were drawn to humanlike qualities, such as emotional resonance and personalized responses, and non-humanlike qualities, such as constant availability and inexhaustible tolerance. This led to fluid chatbot uses, such as Replika as a writing assistant and ChatGPT as an emotional confidant, despite their distinct branding. However, we observed challenging tensions in digital companionship dynamics: participants grappled with bounded personhood, forming deep attachments while denying chatbots "real" human qualities, and struggled to reconcile chatbot relationships with social norms. These dynamics raise questions for the design of digital companions and the rise of hybrid, general-purpose AI systems.

Figures (7)

• Figure 1: A conceptual model of digital companionship. Digital companionship is shaped by the interplay between user and chatbot characteristics during interaction. It emerges from shared interaction affordances that cut across chatbots’ intended marketing and include both humanlike and non-humanlike allures and challenges, common across systems. These common affordances generate fluid assistant–companion dynamics that often exceed a chatbot’s intended use, with implications for users and for society more broadly. (Images adapted from flaticon.com).
• Figure 2: Example ChatGPT (A) and Replika (B) interfaces (November 2025).
• Figure 3: Mean ranks of ChatGPT and Replika use cases (from most to least important). Error bars depict the standard error of the mean.
• Figure 4: Means (A, B, C) and distribution (D, E, F) of characteristics of ChatGPT and Replika users. Error bars depict the standard error of the mean. (* = $p$ < .05; ** = $p$ < .01; *** = $p$ < .001; FDR-adjusted.)
• Figure 5: Mean ratings of chatbot perceived attributes (A), user attitudes and emotions toward chatbots (B), and general attitudes toward AI (C). Error bars depict the standard error of the mean. (* = $p$ < .05; ** = $p$ < .01; *** = $p$ < .001; FDR-adjusted.)