Algorithms Do Not Decide: How Decisions Are Made

This paper argues that algorithms do not make decisions; they generate outputs that become decisions only through human practices. What is commonly described as “algorithmic decision-making” obscures a fundamental distinction between output and decision, thereby mislocating both authority and responsibility. Drawing on Kant’s account of judgment (Kant 1790) and Wittgenstein’s analysis of rule-following as practice (Wittgenstein 1953, §§198– 202), the paper maintains that decisions require a capacity that cannot be reduced to rule- following or automated processes. Outputs, taken in isolation, are neither decisions nor risks; responsibility arises only at the point where outputs are enacted. What has been described as a “responsibility gap” (Matthias 2004) is therefore a category mistake. The paper introduces the concepts of authority illusion and ethical withdrawal (Enser 2026a; Enser 2026b; Enser 2026c) to explain how outputs become actions while responsibility is simultaneously displaced. Drawing on Arendt’s account of the suspension of thinking (Arendt 1963), the paper concludes that the problem is not ignorance, but the continuation of action without reflective interruption.