When (and How) Should Courts Use AI?

Andrew Coan

 When and How Should Courts Use AI?

(coauthored by Harry Surden)

The debate between constitutional formalists and realists has largely focused on the kinds of controversial questions that come before the Supreme Court. In such cases, there are nearly always plausible legal arguments on both sides. Constitutional formalists think judges should decide between those arguments on the basis of original public meaning—or, at any rate, on the basis of some criterion other than their own moral and political judgment. Constitutional realists doubt this is possible and, at any rate, think the moral and political judgment of judges is at least some of the time normatively superior to the various criteria defended by formalists. The stakes of this debate are very high because the resolution of such cases by the Supreme Court shapes public policy on vitally important questions for the entire country.

In all of these respects, the constitutional questions that come before the Supreme Court are exceptional, rather than normal. The kinds of constitutional questions most often posed in the federal district courts—and quite often in the federal courts of appeals—generally have clear or fairly clear answers that most or all judges applying any mainstream interpretive approach would agree on. The same is true for many, if not most, of the constitutional questions that never make their way to court. Questions arising at the lower levels of the federal judicial system—and completely outside it—also tend to have lower stakes for the legal system as a whole. Decisions of federal district courts have no precedential effect, and the decisions of federal courts of appeals govern particular geographic regions, rather than the whole country.

These courts also have far larger caseloads and far fewer resources to devote to the decision of each case than does the Supreme Court. The same is generally true for government officials and government institutions grappling with constitutional questions outside of court. For all of these reasons, the plausibility and attractiveness of using AI to generate clear legal answers with maximum speed and efficiency is significantly greater and less controversial outside the rarefied realm of the Supreme Court.

For a fuller explanation, see our new paper, “Artificial Intelligence and Constitutional Interpretation.”