Thomas A. Smith (University of San Diego School of Law) has posted The Death of Judgment? Artificial Agency and the Surrender of Human Agency on SSRN. Here is the abstract:
Most discussions of artificial intelligence in decision-making concern machines that err. This essay takes up the harder case, in which a machine’s advice is reliably better than the decision-maker’s own. Deference to such advice is often rational, and the literature on epistemic and practical authority supplies strong reasons for it. Yet rational deference has cumulative effects. Repeated across a career, and built into institutions that require justification only from those who depart from a recommendation, it may erode the capacities on which meaningful human control depends. Kant’s account of immaturity does not fit this case. The dependence arises not from a want of courage but from a sequence of decisions each of which was sound, and for that reason exhortation will not cure it.
The essay distinguishes two grounds for preserving human judgment. Where a person decides his own affairs, the value of his participation is a reason he may weigh for himself. Where an official decides for others, the ground is the affected party’s claim to a decision by an answerable office. That claim belongs to the litigant, not the judge. It resolves the apparent conflict between accuracy and judicial independence: a judge must adopt a machine’s analysis, not merely sign it. The essay also identifies a danger that outcome-based review cannot detect. Systems that prepare the record can settle which questions are presented before any officer exercises judgment. It concludes that preserving judgment is a problem of institutional design, and that its benefits will grow harder to measure as machine judgment improves.
Highly Recommended!
To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum