Cultural Bias and Representational Gaps in Conversational Generative AI for Mental Health: A Systematic Review
Proceedings of the ACM on Human-Computer Interaction · 30.04.2027 · 11 (2) · 1-43
DOI: 10.1145/3861024 ↗Kurz gefasst
Systematische Übersicht in der Zeitschrift Proceedings of the ACM on Human-Computer Interaction, online erschienen am 18.11.2026. Verfasst von Oritsetimeyin Arueyingho und 4 weiteren Autor:innen. Die Arbeit nennt 243 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Engineering and Physical Sciences Research Council.
Abstract (Originalsprache)
Limited access to mental health support due to infrastructural limitations and workforce shortages has contributed to growing interest in the use of Generative Artificial Intelligence (GenAI) for scalable mental health support. Despite their promise, these technologies pose risks, such as hallucinations and misalignment with local experiences of mental health challenges in the Global South. While there is growing attention to safety and harm mitigation, limited research has examined how cultural representation, linguistic coverage, and dataset provenance shape these technologies, particularly for Global South populations, and how cultural sensitivity can be improved to support mental health care. This systematic review synthesises 29 empirical studies published between 2020 and April 2026, predominantly concerning conversational GenAI for mental health. Using a keyword-based search strategy, critical narrative synthesis, and reflexive deductive thematic analysis, we examined how GenAI systems encode cultural assumptions. Our findings reveal representational gaps, limited culturally grounded narratives, and predominantly additive adaptation, whereby cultural elements are introduced while underlying Western diagnostic and therapeutic frameworks generally remain unchanged. Also, we consolidate cross-study evidence into strategies that have implications for clinical and non-clinical collaborative practices for mental health support. Lastly, we outline implications for Human-Centred AI and Computer-Supported Cooperative Work and directions for culturally grounded conversational GenAI that supports equitable mental health care.
Förderung
- Engineering and Physical Sciences Research Council
Zitieren
APA
Arueyingho, O., Wonah, M., Lawrence, M. O., Nkwo, M., & Gooch, D. (2027). Cultural Bias and Representational Gaps in Conversational Generative AI for Mental Health: A Systematic Review. Proceedings of the ACM on Human-Computer Interaction, 11(2), 1-43. https://doi.org/10.1145/3861024
Vancouver
Arueyingho O, Wonah M, Lawrence MO, Nkwo M, Gooch D. Cultural Bias and Representational Gaps in Conversational Generative AI for Mental Health: A Systematic Review. Proceedings of the ACM on Human-Computer Interaction. 2027;11(2):1-43. doi:10.1145/3861024
BibTeX
@article{Arueyingho2027,
author = {Arueyingho, Oritsetimeyin and Wonah, Margaret and Lawrence, Morolake Oladayo and Nkwo, Makuochi and Gooch, Daniel},
title = {Cultural Bias and Representational Gaps in Conversational Generative AI for Mental Health: A Systematic Review},
journal = {Proceedings of the ACM on Human-Computer Interaction},
year = {2027},
volume = {11},
number = {2},
pages = {1--43},
doi = {10.1145/3861024}
}
EUVIMED