Orli Oren-Kolbinger (Oregon) presents Timing Matters: Sequencing AI in Tax Education, 24 Pitt. Tax Rev. ___ (2026) (AI and Teaching Taxation), today at Pepperdine as part of its Tax Policy Workshop Series hosted by Deanna Newton:
Artificial intelligence (AI) now plays a growing role in legal education. Instructors use AI technology to develop classroom exercises, discussion questions, simulations, visual organizers, and formative assessment questions. Students also use AI to explain doctrine, summarize readings, and work through legal problems. This shift raises a deeper question: whether AI strengthens legal reasoning or weakens the cognitive processes necessary to develop it. Legal educators are increasingly confronting this question in practice, as law schools adopt varying approaches to generative AI use in teaching and assessment.
This question is particularly significant in tax law, which requires students to work through layered statutory structures, classify transactions, apply rules in multi-step analysis, and connect doctrine to policy consequences. Students must move beyond memorization and learn to work through complex statutory structures and policy trade-offs. They do not learn tax law by listening alone. They learn it by working through problems, making classification choices, working through multi-step reasoning, and seeing how one doctrinal move affects the next.
Because learning tax depends on this kind of active engagement, structured active learning already produces strong learning outcomes in tax law without AI. AI does not replace doctrinal instruction or analytical reasoning. Its strongest contribution comes later, when instructors or students translate substantive understanding into exercises, simulations, and structured classroom activities.
My point is straightforward: AI works best as a tool for working with analytical structures rather than as a substitute for the learning process itself. The problem addressed in this paper is not that AI created a new weakness in legal education, but rather that students have always struggled with complex doctrinal reasoning, multi-step analysis, and the development of professional judgment, which requires multiple rounds of trial and error and can improve over time. The concern is that AI may intensify those challenges when students rely on outputs before they develop the ability to evaluate, reproduce, and modify the underlying reasoning. More specifically, AI is most useful when students can assess whether an output is doctrinally sound, adapt it to fit the learning objectives, and independently reproduce the underlying reasoning. The sequencing question is therefore not simply whether students have used AI, but whether they have developed sufficient competence to evaluate AI’s output.
The remainder of the article proceeds as follows. In Part I, I identify structured active learning as the baseline architecture of effective tax pedagogy. In Part II, I develop a sequencing-based model of AI integration in which AI becomes a factor only after doctrinal understanding has formed. In Part III, I apply the sequencing model across doctrinal tax courses and seminar-based instruction. In Part IV, I consider potential objections to the sequencing framework and clarify the model’s limitations. Finally, I conclude by returning to the importance of timing and sequencing in the pedagogical use of AI.
