Model Drift and AI in Humanitarian Contexts: Strengthening Humanitarian Action in the Age of AI

Building on the previous HOISA panel on AI and Emerging Tech for Humanitarian Action, this incoming panel discussion gathers expert panelists to discuss how rapidly AI is being deployed in high-stakes humanitarian contexts from early warning systems to evidence collection. This movement, however, also points to a gap in its implementation as AI systems carry subtle shifts in model behavior, known as model drift, that can potentially distort outputs, undermine trust, and jeopardize decision-making.

Critical for the South Asian subcontinent, but also beyond, HOISA invites you to this discussion that will combine diverse perspectives — practice, ethics, and research — to frame the risks and identify pathways toward scalable, evidence-based approaches. 

Expert Panelists:

• Anisa Abeytia (op-ed author, AI governance expert)

• Asma Derja, Founder of the Ethical AI Alliance

• Alphoncina Lyamuya, PhD Candidate, USC Annenberg