CivicSabha 2.0 - Participatory AI Evaluation for Safe and Trusted Systems

Artificial Intelligence (AI) is now embedded across India's public systems - from health and agriculture, welfare, financial inclusion and governance platforms - shaping decisions at scale. While it brings major opportunities, it also poses risks like safety failures, contextual misalignment, and exclusion, endangering public trust and effective governance.

Building safe and trustworthy responsible AI systems requires robust, participatory evaluation that moves from broad principles to applied and localised context-sensitive systems rooted in Global South realities and aligned with emerging global best practices. CivicDataLab (CDL) and Karya convened a roundtable discussion "Participatory AI Evaluation for Safe and Trusted Systems" bringing together policymakers, international organisations, AI safety researchers, technologists, and civil society practitioners to examine the role of participatory evaluation of AI to enable a responsible and safe AI ecosystem to mitigate harms, risks, & exclusion. CDL convened this dialogue to foster cross-sector collaboration, share practical use cases, and identify pathways for responsible and inclusive deployment of AI in public systems.