Machine Learning and the Re‐Enchantment of the Administrative State

Abstract

Machine learning algorithms present substantial promise for more effective decision‐making by administrative agencies. However, some of these algorithms are inscrutable, namely, they produce predictions that humans cannot understand or explain. This trait is in tension with the emphasis on reason‐giving in administrative law. The article explores this tension, advancing two interrelated arguments. First, providing adequate reasons is a significant facet of respecting individuals’ agency. Incorporating inscrutable algorithmic predictions into administrative decision‐making compromises this normative ideal. Second, as a long‐term concern, the use of inscrutable algorithms by administrative agencies may generate systemic effects by gradually reducing the realm of the humanly explainable in public life, a phenomenon Max Weber termed ‘re‐enchantment’. As a result, the use of inscrutable machine learning algorithms might trigger a special kind of re‐enchantment, making us comprehend less rather than more of shared human experience, and consequently altering the way we understand the administrative state and experience public life.