The Lawyer’s Swan Lake: Generative AI, Legal Epistemology, and the Nondelegable Duty to Think

Authors

• Angie Zinzilieta

Abstract

Generative artificial intelligence (“GAI”) has entered law wearing the costume of competence: fluent, confident, and fast. But legal practice is not a memory sport despite the bar exam. Lawyers are paid to create, test, and refine ideas (arguments, narratives, and theories of liability) inside an adversarial system that punishes error, rewards precision, and demands accountability. This article argues that the central risk of GAI in legal work is not merely “hallucinations” or confidentiality breaches; it is epistemic atrophy, an outsourcing of the intellectual labor by which lawyers understand where law comes from, how it changes, and why it applies. Using sanction orders involving fabricated citations as case studies, and synthesizing Illinois’s policy on artificial intelligence in the courts, Illinois judicial utilization guidance, the ABA’s Formal Opinion 512, and Missouri’s Informal Opinion 2024-11, this article proposes a contemplation-first framework for ethical AI use, especially in plaintiff-side litigation. The solution is not a ban. It is a discipline: (1) task-and-data classification; (2) tool selection proportional to risk; (3) signature-level verification protocols; (4) documentation and client-centered transparency; and (5) evidence-aware safeguards against deepfakes. In short, you cannot seek the output of Swan Lake without ever taking a ballet class.

First Page

73

Recommended Citation

Angie Zinzilieta, The Lawyer’s Swan Lake: Generative AI, Legal Epistemology, and the Nondelegable Duty to Think, 10 Bus. Entrepreneurship & Tax L. Rev. 73 (2026).
Available at: https://scholarship.law.missouri.edu/betr/vol10/iss1/6