Nonhuman humanitarianism: when 'AI for good' can be harmful
A decolonial critique of 'Karim', a psychotherapy chatbot (built by X2AI) deployed to Syrian refugees in Lebanon via a small-scale pilot, examining the power asymmetries and ethical risks of positioning an AI mental-health tool as humanitarian aid for a highly vulnerable displaced population.
Publisher
Information, Communication & Society (Taylor & Francis)
Published
8 Apr 2021
Added
2 weeks ago
Key Findings
- Analyzes the deployment of 'Karim', an Arabic-language psychotherapy chatbot, to Syrian refugees in Lebanon through a roughly 60-person pilot
- Argues that framing an unproven AI mental-health tool as 'AI for good' humanitarian aid obscures power asymmetries between technology providers and displaced populations who have little practical ability to consent, refuse, or seek alternatives
- One of the earliest rigorous academic case studies of AI mental-health tools deployed to a vulnerable, non-Western population outside a controlled clinical-trial context
Methodology Notes
Critical/decolonial media-studies analysis of a single case (the Karim chatbot pilot), drawing on humanitarian-technology and power-asymmetry theory rather than a clinical-outcomes methodology.
Sources
Taylor & Francis Online (primary)
Archived snapshot (Wayback Machine) — preserved against link rot
Authors
Mirca Madianou
Tags
Cite This
APA
Mirca Madianou (2021). Nonhuman humanitarianism: when 'AI for good' can be harmful. Information, Communication & Society (Taylor & Francis). https://www.tandfonline.com/doi/full/10.1080/1369118X.2021.1909100