In this article, we discuss and outline a research agenda for social science research on artificial intelligence. We present four overlapping building blocks that we see as keys for developing a perspective on AI able to unpack the rich complexities of sociotechnical settings. First, the interaction between humans and machines must be studied in its broader societal context. Second, technological and human actors must be seen as social actors on equal terms. Third, we must consider the broader discursive settings in which AI is socially constructed as a phenomenon with related hopes and fears. Fourth, we argue that constant and critical reflection is needed over how AI, algorithms and datafication affect social science research objects and methods. This article serves as the introduction to this JDSR special issue about social science perspectives on AI.