| Sumario: | This article offers a critical response to Kate Crawford’s discussion of AI classification. In her celebrated book Atlas of AI (2021), Crawford examines the ethical consequences of biases, reifications, and oversimplifications that emerge from the decision-making behind AI taxonomies. I argue that there is an ambiguity in Crawford’s work about whether she believes that the problems we face with AI are solvable or not, in other words, whether an ethics of AI classification is possible. I investigate the source of this ambiguity by outlining multiple trunks and branches of arguments for and against the claim that, however complicated a solution may be, AI classification distortions are fixable. By uncovering a total of five arguments within her classification chapter, my aim is to clarify and develop Crawford’s thesis about the encoding of power while also reflecting on why this question of an ethics is unresolved in her work.
|
|---|