“Computer Says No”: Artificial Intelligence, Gender Bias, and Epistemic Injustice

Ever since its beginnings, the field of Artificial Intelligence (AI) has been plagued with the possibility of perpetuating a range of depressingly familiar kinds of injustice, roughly because the biases and prejudices of the programmers can be, literally, codified. But several recent controversies about biased machine translation and automated CV-evaluation highlight a novel set of concerns that are simultaneously both ethical and epistemological, and which stem from AI’s most recent developments; we don’t fully understand the machines we’ve built, they’re faster, more powerful, and more complex than us, but we’re growing to trust them preferentially nonetheless. This chapter examines some of the ways in which Miranda Fricker’s concept(s) of “epistemic injustice” can highlight such problems, and concludes with suggestions about re-conceiving human-AI interaction-along the model of collaboration rather than competition-that might avoid them.