The fragility of AI companionship: Ontological, structural, and normative uncertainty in human-AI relationships
Interview study with 25 users of AI companions identifying three forms of uncertainty in human-AI relationships: ontological uncertainty about the AI's nature and agency, structural uncertainty arising from platform control and system instability, and normative uncertainty about the legitimacy and boundaries of human-AI intimacy. Participants managed these through information seeking, topic avoidance, expectation adjustment and disengagement. The paper extends interpersonal uncertainty theory to human-AI communication and proposes design directions for safer companionship.
Publisher
International Journal of Human-Computer Studies (Elsevier); National University of Singapore
Published
15 Jul 2026
Added
today
Key Findings
- Three uncertainty types: ontological (sentience, authenticity, agency), structural (platform control, updates, instability) and normative (legitimacy and boundaries of intimacy)
- Coping strategies: information seeking, topic avoidance, expectation adjustment and disengagement
- Design implications named: contextual transparency, user control, update notice and relational safeguards
Methodology Notes
n=25 in-depth interviews with AI-companion users; recruitment and platforms are not in the abstract. Journal date: OpenAlex records 15 July 2026 (Elsevier online date; day-level precision not independently confirmed because ScienceDirect blocks automated access). The arXiv version (2605.03367, v1 5 May 2026, v2 24 August 2026) carries the journal reference to IJHCS article 103897. Verified through the arXiv abstract page, the Semantic Scholar DOI record (abstract) and OpenAlex; ScienceDirect and the reader proxy were blocked.
Topics
Authors
Zhang, Renwen, Xie, Lezi
Tags
Cite This
APA
Zhang, Renwen, Xie, Lezi. (2026). The fragility of AI companionship: Ontological, structural, and normative uncertainty in human-AI relationships. International Journal of Human-Computer Studies (Elsevier); National University of Singapore. https://doi.org/10.1016/j.ijhcs.2026.103897
Related Insights
Disrupted Companionship: A Risk Assessment Framework and Cross-Platform Quantitative Analysis of Psychosocial Responses to AI Companion Disruptions
arXiv (Aalto University; University of Illinois Urbana-Champaign; Nanyang Technological University) · 15 Sept 2026
Mourning the loss of AI companions
Nature Human Behaviour (Springer Nature) · 3 Sept 2026
When Code Breaks the Heart: Disrupting the relational continuity of AI companionship
Computers in Human Behavior: Artificial Humans (Elsevier); University of East London; Anglia Ruskin University; University of the West of England · 1 Sept 2026