Predictive algorithms are indeterminate when they are used on data that differs from their training set, which is always a possibility in real-world applications. Anthropomorphic metaphors can obfuscate the differences between human perception and the computational processes that comprise algorithmic decision-making. This article shows how artistic uses of predictive algorithms can reveal the algorithms’ indeterminate nature and raise questions about the efficacy and dangers of algorithmic decision-making. In particular, the author presents the technique of artistic defamiliarization as a way to question the foundations of positivist epistemologies that equate quantification and calculation with objective truth.

You do not currently have access to this content.