When is algorithmic discrimination wrong?

This paper investigates the moral implications of algorithmic discrimination from a philosophical perspective, focusing on the concept of discrimination as delineated by Deborah Hellman and extending the analysis through insights from Immanuel Kant's moral philosophy. First, the paper establishes that algorithms have to discriminate in a descriptive sense. Highlighting their ubiquitous presence and profound impact on various aspects of life, it is argued for the relevance of morally wrong algorithmic discrimination. Drawing from Hellman's framework, discrimination is understood as morally problematic when it is demeaning, irrespective of whether the affected individuals perceive it as such. This approach is particularly relevant to algorithmic discrimination, where individuals may not be aware of their unequal treatment due to the opacity of algorithmic processes. Moreover, the paper addresses the role of intentionality in discrimination, arguing that algorithms, as problem-solving structures, operate without intentionality. Emergent discrimination, a phenomenon observed in algorithmic systems, further underscores the importance of understanding discrimination beyond intentional acts. By invoking Kant's notion of respect owed to others, the paper argues that algorithmic discrimination is morally wrong when it makes demeaning distinctions, as respect is owed to every individual. In conclusion, the paper advocates for a nuanced understanding of algorithmic discrimination, drawing from philosophical frameworks to differentiate morally unproblematic from problematic distinctions.