Artificial intelligence does not replace lawyers; it exposes what legal education has at times ceased to teach. When information overflows, scarcity no longer lies in access to texts but in the quality of judgment. The task is not to choose between memory and reasoning, but to rebalance their use in the age of AI.
Acknowledging the useful inheritance
Let us start with what the classical model does best. Leading law schools transmit an intellectual architecture rather than an inventory of details that quickly perish. As Philippe Stoffel-Munck puts it, “what is essential … is the intellectual architecture of a subject … like the load-bearing beams of a structure.” This demand for structure rests on discipline in language: clear, concise, precise. A crisp style is not cosmetic; it orders the mind. Because practice is steeped in ambiguity, we cultivate methodological doubt (questioning, shifting perspective) without sinking into a relativism that dissolves truth. Casuistry and dialectic have always been the lawyer’s training ground.
A new balance between memory and judgment
What changes today is not the nature of law, but the balance between memorisation and discernment. Even in the past, programmes valued reasoning, yet the scarcity of information gave disproportionate weight to memory. Now that information is abundant, we will continue to memorise the essentials but much less, and we will focus on teaching how to address the right questions . This shift centres the capacity to judge and to act on what one knows.
Human vs. AI: the decisive difference
To rethink the curriculum, we must clarify the boundary between what AI does exceedingly well (searching vast bodies of decisions, summarising, detecting patterns, drafting initial analytical frames) and what remains irreducibly human: deciding what is just, arbitrating between legality and equity, and assuming responsibility before clients, courts and society. The decisive difference is this: a well-trained lawyer knows how to formulate the right questions and then answer them methodically; a model does not always know how to pose the right question. It may reduce hallucinations with more time, data and safeguards, but it does not, on its own, see where the real question begins. The goal is not an all-automatic system, but a “Lawyer-in-the-Loop”: a professional who knows how to use, audit and supervise tools while retaining control over qualification, strategy and proportionality.
Three curricular shifts
First, an epistemology of law: not only what “the law says,” but how legal knowledge is constructed, validated and contested, even when algorithmic models propose answers that are plausible yet fragile. Second, judgment under constraint: deciding when multiple solutions can coexist, recognising when “lawful” and “just” diverge, and giving reasons for choices made. Third, a systemic reading of regulation: understanding how texts create incentives, where enforcement distorts outcomes, and how overlapping regimes generate unintended effects. Here, ethics is not a capstone module; it is the architecture that guides problem-framing and sets thresholds for non-automation.
Academic echoes
This evolution resonates across major traditions. At the Sorbonne, robust intellectual architectures are still transmitted. At Harvard, Holmes reminds us that law is a practice rooted in experience and responsibility. In India, N. R. Madhava Menon, father of modern legal education, championed clinical teaching and learning-by-doing as the matrix of professional judgment.
Institutional consequences
Educating for this “augmented” law requires recruiting faculty who can transmit systems thinking and discernment; assessing less on recall and more on reasoned decision-making under uncertainty; and weaving technology, ethics and governance throughout the curriculum. Employers already automate research and parts of drafting, but they do not delegate clarity within complex systems, the arbitration between lawful and just, or the professional posture that sustains trust when algorithms fall silent.
Conclusion: memorise the essential to better question
The question is not whether AI can “do law,” but whether our institutions can realign education with the right scarcity. As in the past, we will memorise; from now on, we memorise the essential in order to better judge the essential. Schools that place reasoning and the right questions at the core, while equipping students to audit and supervise AI, will produce rare lawyers, able to unite intellectual architecture, experience, ethics and responsibility.
About the author Zied Miled is an engineer-lawyer specializing in network industries, regulation, and AI. The founder of Aidalet and Doudy.ai, He designs native-AI legal and educational technologies that enhance human judgment. He advises institutions across Europe, the MENA region, and Türkiye on regulatory transformation and AI ethics.ıı