Moral Entropy: Auditing Bias and Uncertainty in Moral Judgment

Excerpt

arXiv:2609.21992v1 Announce Type: new Abstract: Most work in computational ethics treats annotator disagreement on moral content as noise to be voted away, collapsed into majority vote or the more permissive any-annotator rule the moment a single annotator flags an item. We argue this uncertainty should instead be modeled and learned from. We introduce Moral Entropy, a Bayesian framework that keeps a full posterior over the true label and decomposes its entropy into aleatoric uncertainty (irredu