This paper reflects on proliferating AI for Social Good (AI4SG) initiatives, with an eye to public health and health equity. It notes that many AI4SG initiatives are shaped by the same corporate entities that incubate AI technologies, beyond democratic control, and stand to profit monetarily from their deployment. Such initiatives often pre-frame systemic social and environmental problems in tech-centric ways, while suggesting that addressing such problems hinges on more or better data. They thereby perpetuate incomplete, distorted models of social change that claim to be 'datadriven'. In the process, AI4SG initiatives may obscure or 'ethics wash' all the other uses of big data analytics and AI that more routinely serve private interests and exacerbate social inequalities. As a case in point, it discusses the prominence of health-related applications in AI and big data fields, alongside the politics of more 'upstream' versus 'downstream' health interventions.