Datafied Religion: How Well do Commercial Image Recognition Services Recognize Religion?

Abstract This article explores the relationship between technology and religion by empirically analysing how the image recognition services provided by three leading US tech companies and machine vision developers—Google, Amazon, and Microsoft—classify religious images and what implications follow from their classifications. The data were collected from Google Images: they consist of images related to Christianity, Islam, Hinduism, Buddhism, and Shintoism. Methodologically and theoretically, our interdisciplinary and exploratory article combines computational social science techniques with religious and media studies theories and qualitative research methods. The main result of our research is that in the context of religion, image recognition services produce secular, Christian, commercial, and racist biases.