Abstract
This paper examines whether sentencing algorithms – machine-learning-based tools for assessing the likelihood that a convicted individual will commit further offenses if released on parole – are consistent with Peter Singer's preference utilitarianism and the principle of equal consideration of interests. It begins by explaining the functioning and ethical challenges of such algorithms, especially the challenge of individualized sentencing. The paper then explores how these algorithms align with Singer's preference utilitarianism, particularly his principle of equal consideration of interests. Analyzing the key elements of this principle – maximizing and equally weighing interests, impartiality, and rejection of irrelevant group memberships – reveals how critics might use it to oppose the implementation of sentencing algorithms. A contextually more sensitive reading of Singer's views suggests that the same principle, in fact, supports the use of these algorithms. The paper concludes that sentencing algorithms are not only consistent with Singer's position but are, in many respects, reinforced by it.