Creating Certainty Where There Is None: (Mis)representing Artificial Intelligence

Recent developments in the field of research on Artificial Intelligence (AI) have given rise to a growing number of conversations concerning the technology’s potentials and pitfalls. These discussions have tended to pit AI ‘catastrophists’ against AI ‘enthusiasts’, with the latter hyping the positive transformative powers of AI and the former exposing its dangers. In this paper, we analyse these discourses to identify the spoken and unspoken assumptions they make about AI, and consider their political implications. Approaching AI as a political concept, we identify a series of anthropological, agential, economic and temporal themes and assumptions associated with contemporary AI discourse. We then evaluate these assumptions and show that they are technically contestable because they create certainty concerning the level of advancement of AI, when its capabilities are in fact more uncertain. Discursively, they are also problematic because they are based on unquestioned but dubious assumptions with far-reaching political and social implications.