The public sector as a battleground for AI governance
Latin American governments are deploying AI at a growing pace: predictive crime models, triage systems in public health services, algorithms for allocating social benefits, facial recognition in public spaces, and chatbots in citizen services. These deployments create obligations of transparency and accountability that are distinct from, and more demanding than, those that apply to the private sector, because the power of the State is involved.
The lesson of Robodebt: the risk of automating social decisions
The Robodebt case in Australia, where an automated system generated incorrect debt claims against hundreds of thousands of welfare recipients, is a warning directly applicable to benefits management systems in Latin America. The elements that made the Australian disaster possible (automated decisions without real human oversight, the absence of effective challenge mechanisms, the opacity of the system for those affected) are present in several Latin American public programmes. The lesson is clear: automating decisions that affect social rights requires real human oversight, not nominal oversight.
