Document Type
Article
Publication Title
Arizona State Law Journal
Volume
58
Publication Date
2026
Abstract
It is time to reorient the debate over the right to a human decision. Within that debate, the strongest arguments in favor of human decisionmakers are Arguments from Explanation: AI technology is increasingly and necessarily opaque, and so machines should not replace human decisionmakers because machines cannot give the required explanations. Or so the humanist argument goes. Meanwhile, machinists argue that most humanist principles have been deflated by the Better Decision Argument, which reframes such principles as grounding not a right to a human decision, but merely to a “better” decision— whether by human or machine.
This Article turns that debate on its head. First, it offers a reason to doubt Arguments from Explanation: human judges sometimes don’t know what they ought to do. They have what is called “normative uncertainty.” But if they respond to that uncertainty rationally, they will not provide the kind of explanation demanded of machines. Thus, Arguments from Explanation do not count in favor of human judges, the quintessential example of decisions where an explanation is owed.
But while normative uncertainty gives machinists one advantage, it also undermines the machinists’ Better Decision Argument. Machines may offer the illusion of an idealized AI decisionmaker like Dworkin’s Hercules. But building a machine usually requires hardwiring objectives. Such hardwiring can preclude rational consideration of normative uncertainty, making it harder for machines to aim at a “better” decision. As a result, normative uncertainty also deflates the machinists’ Better Decision Argument.
The best arguments from both camps having been thus undermined, a new question emerges, sharpening the concerns at the heart of the debate. We gain language to diagnose the lingering worry—the pit of dread in one’s stomach— that persists in the face of Better Decision Arguments. And we enable lawmakers and regulators to better understand the problem that engineers face but struggle to articulate. As engineers also suspect, hardwiring Hercules is dangerous if you are uncertain what he should do.
Recommended Citation
Courtney M. Cox,
Hardwiring Hercules?, 58 Ariz. St. L. J. 79
(2026)
Available at: https://ir.lawnet.fordham.edu/faculty_scholarship/1438
