Abstract

Watson, Mökander, and Floridi (Watson et al., AI Soc, 2024) propose that contemporary discussions in AI ethics revolve around two competing narratives, which they term sociotechnical dogmatism and sociotechnical skepticism. As an alternative to this dichotomy, they introduce sociotechnical pragmatism, which they describe as a constructive synthesis of these opposing views. However, the distinction they draw between these narratives conflates their epistemological and ethical dimensions, making it difficult to classify possible alternative positions. As a result, sociotechnical pragmatism, as formulated by Watson et al., does not fully account for the potential of pragmatism as an approach to inquiry, regulation, and governance of technology. Rather than treating regulation as a reactive measure, pragmatism seeks to integrate inquiry into the development process itself, ensuring that both the consequences of new technologies and their ethical implications are critically examined in advance. In this manner, pragmatism as sociotechnical fallibilism and meliorism provides an alternative to a large-scale societal experimentation with technology which treats harm as an inevitable cost of progress.