Answers are grounded in this paper and cite their sources.
Background Nigeria's mental health system serves less than 10% of those in need, with severe provider shortages and deep-seated stigma. Generative AI chatbots offer scalable support, yet how Nigerian adults experience attachment to non-human therapeutic agents remains unexplored. Objective To examine lived experiences of AI therapeutic relationships among Nigerian adults, focusing on attachment formation, cultural acceptability, and perceived authenticity within contexts of care scarcity. Methods Interpretative Phenomenological Analysis with 37 adults (18–45 years) who engaged AI mental health chatbots for ≥ 4 weeks. Semi-structured interviews explored relational experiences, trust development, privacy concerns, and comparisons with human/traditional healing. Results Four superordinate themes emerged: (1) Relational Ambivalence ; participants valued 24/7 availability and non-judgmental space yet struggled with perceived lack of genuine human reciprocity; (2) Cultural Navigation ; users actively adapted Western therapeutic frameworks to local idioms of distress, blending AI interactions with family counsel and spiritual practices; (3) Stigma Sanctuary ; chatbots provided confidential entry points circumventing community visibility and professional hierarchy; and (4) Fragile Dependency ; prolonged use generated attachment anxiety about potential app discontinuation or data breaches, compounded by absence of human escalation pathways. Most participants maintained "instrumental intimacy"; emotional reliance tempered by awareness of algorithmic artifice. Conclusion Nigerian adults develop pragmatic, contextually mediated attachments to AI therapeutic agents that reflect both global digital therapeutic trends and local specificities of care scarcity, stigma, and pluralistic healing. AI chatbots function not as human substitutes but as bridging objects temporary relational anchors in systems of profound unmet need. Findings suggest ethical deployment in LMICs requires culturally adaptive design, transparent algorithmic boundaries, and mandatory human handoff protocols to prevent dependency in unsupported contexts.
No takes yet. Share an insight, caveat, or question.
Isaiah et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: