The University of Saskatchewan College of Law is beginning its search for a new dean. We’ve been invited to think about the main challenges facing the lucky winner. And there are a few things one could mention - every law school, large or small, has its issues. But, right now, at this moment in history, there’s really only one challenge worth talking about. Because it’s existential, and it’s an issue for all law schools: AI.
I’m not talking about the use of AI by students to cheat - though, as I noted in my last post, that is a troubling problem in itself. Without belaboring the point, a law school that cannot, practically speaking, require its students to write - and therefore learn to write better, more persuasively, more elegantly - is not a law school as many of us understand it. A law school that cannot depend on its students to read large quantities of material, digesting it, synthesizing it - i.e., because they can now outsource the hard work to a machine - is not able to perform one of the basic functions traditionally expected of a law school. We can stop assigning take-home written assignments - research papers, memos, (sigh) reflective journals - and replace those modes of assessment with hand-written, in-class examinations invigilated by a platoon of heavies. That, however, is not victory. It is an admission of failure; an admission that we can no longer get future lawyers to learn some of the basic skills of lawyering.
(Oral examinations are great, but they are not replacements for written communication skills. They are, at best, supplements for written work. And there is a reason they fell out of vogue: they are intimidating for many students, and (more important) they are not easily scalable.)
Nor am I talking about the problem of academics using AI to produce and peer review “scholarship”. Again, this is a real, profound issue. The idea of law professors using AI to churn out papers they didn’t write, which are then reviewed by AI and added to a database of AI-generated papers used to generate still more… well, it’s just a soul-crushing thought.
These are big concerns for the entire academy, not just law schools. But in many ways they are symptoms of a malaise that had long afflicted the legal academy before AI. Wherever did students get the idea that it was okay not to do the assigned reading, or create their own notes on the material? Could it have something to do with the fact that law schools have, for the last 30 years, steadily diminished use of closed-book exams (which would require a sustained focus on memorizing and internalizing knowledge), while simultaneously reducing assigned readings to a fraction of their former level, and also turning a blind eye to the circulation of “CANs” that offer a distillation of previous iterations of the same course taught by the same professor, and also inflating grades to ever-higher levels?
Or could it also have something to do with the legal academy’s ever-increasing emphasis on “work-life balance” - on treating the study of law as something that not only can but must be confined to a fixed number of hours per subject per week - perversely sending the message to students that they should actively try to minimize the amount of time they spend wrestling with a given subject, or researching a paper, or trying to grasp a doctrine or idea that is elusive; rather than treat the pursuit of knowledge as an end in itself, a beautiful rabbit-hole down which one might happily fall forever?
And could all that, in turn, have something to do with the wildly increased emphasis - at least since I was a law student - on law school as an institution principally (if not exclusively) dedicated to the acquisition of ‘skills’ necessary to become practicing lawyers; an emphasis that would always create the impression that the work being done in classrooms is, well, not really all that important next to the hands-on (“experiential”) learning that could be obtained elsewhere?
Could the use of AI by faculty, likewise, have something to do with the fact that law schools have increasingly hired people with little interest in law as an area of knowledge that is worth pursuing - worth being genuinely, deeply curious about - in its own right? And could even those who want to indulge their curiosity (“let a thousand flowers bloom”) be discouraged or outright thwarted by student editors of law journals who relentlessly demand to know - because we have trained them to think this way - what immediate practical, concrete purpose is served by this research?
We created many of the cultural problems in the legal academy ourselves. They would be there if AI had not emerged on the scene. Had it not, we would have just continued to ignore them; continued to rot slowly from the inside.
But I actually want to focus on what is, I think, an even greater, practical threat to the legal academy as an institution.
The central challenge for the dean of any law school in 2026 is, quite simply, to justify its continued existence in an era of artificial intelligence. Law schools, for better or for worse, principally hold themselves out as places to train future lawyers. And we may already be at a point when higher-end AI products, particularly those pitched at a professional legal market, can do much of the work expected of a junior- or associate-level lawyer in a law firm. The emergence of agentic AI allows lawyers to leave much of the work of drafting, writing, and analysis to a machine. The technology is only going to get better.
(Yes, I have heard about The Hallucinations. These stories tell us more about the lawyers - and, often, unrepresented litigants - who filed the submissions, than about the state of the technology at its best. Sure, if a person uses the basic version of ChatGPT to make their legal submissions, they will get what they get. But there are vastly better products out there if one is prepared to pay for them - and they are certainly cheaper than hiring lawyers.)
This technology is, I want to emphasize, a very good thing from some perspectives. From an access-to-justice point of view, the kind of assistance that agentic AI could provide to unrepresented and indigent ‘clients’ – at a small fraction of the cost of even lower- to middle-tier lawyers – would be a godsend. It should, moreover, drastically reduce the costs absorbed by the state in the form of legal aid, duty counsel, Rowbotham/G (J) applications, staffing costs for the Crown and judiciary, and more. It is no exaggeration to say that the existence, and widespread availability, of high-quality agentic legal AI could make the administration of justice truly scalable for the first time.
More good news: There is little reason to believe that AI could replace human lawyers altogether. There will, I suspect, always be a place for human judgment, human contact, human relationships, and human accountability. For those who can afford the very best (human) legal representation and services, there will always be a lucrative market for lawyers who have the legal understanding, skill, knowledge, acumen, and judgment to critically scrutinize and analyze the work produced by AI in light of their on-the-ground experience. There will always be a market for those lawyers who can vouch for that work (or not), ‘translate’ it into accessible and practically useful terms for laypersons and other actors in the legal system, understand and explain the limitations of a rarefied legal analysis in light of local conditions, build relationships with clients and ‘repeat players’ in the justice system, and generally be the accountable human in the room. (I often think about Hollis Robbins’ discussion of ‘the last mile’.)
So that’s the good news. But notice that all of the things that law schools can effectively teach at scale – i.e., to a large number of fee-paying students at once, thereby making content-delivery practically achievable – are things that high-quality legal AI can already do or will be able to do better and much more efficiently than many or most of the law students we graduate. Insofar as what we are teaching is sufficiently general that it can be - and has been - delivered in a more or less standardized fashion across a range of casebooks and textbooks and across a range of law schools, it is something that AI can also assimilate and deliver with vastly greater efficiency.
Observe too that the required courses in law school have a shared core content, as reflected in the syllabi, that renders them more or less interchangeable across schools. Different professors might have different ‘takes’ on this or that topic, but the core of what we teach is not so very different. (This is precisely why law professors often roll their eyes when students complain about having to adjust their studies for this or that individual professor’s proclivities. We know very well that the subject itself does not vary too much across different instructors.) This interchangeability and generality of much of what we teach suggests that AI can do it at least as well - and faster and cheaper.
Notice, too, that all of the things that people will want human lawyers for – human judgment, human relationships, human accountability – are things that law schools cannot effectively teach at scale. For the knowledge that AI cannot assimilate pertains to matters that are not general; that are profoundly local in some sense. The knowledge might be local to an extremely niche sub-sub-sub-field of law, involving a very small number of repeat players. Preparing students for that field of law involves familiarity with the highly bespoke practical and legal problems faced by participants in that specific context, and may also involve acquiring some familiarity with the participants so that one can understand why and how certain legal solutions are attractive to them.
The knowledge might be, to some significant degree, geographically local: immigration practice in Chinatown in downtown Toronto; or a legal practice in a given rural community in Saskatchewan. What do people want in that community? Who are they? What customs and practices and understandings inform their approach to the legal questions they confront on a day-to-day basis? How are they resolved? Alternatively, it might be knowledge local to a given corporation’s in-house counsel.
A lawyer who possesses this local knowledge is highly desirable (to someone), and cannot be replaced by AI. But a law school cannot really deliver that - not, anyway, to the vast bulk of enrollees. Every law school offers experiential learning opportunities; indeed, the Federation of Law Societies requires law schools to offer them. I chose my example of a rural legal practice in Saskatchewan because the University of Saskatchewan College of Law literally offers an externship for a rural legal practice in Saskatchewan. Most if not all Canadian law schools provide opportunities like these.
I, however, want to stress several points. First, precisely because they are local, experiential learning opportunities are resource-intensive and exert real pressures on general program requirements. They require special arrangements and dispensations. There is, I submit, a reason why the FLSC only requires law schools to offer opportunities for experiential learning; requiring every student to participate in them would likely be cost-prohibitive.
More importantly, though, they impose special pressures on the relationship between the student and the law school. Even if a full-time professor could drop what he or she is doing for a term and spend it in rural Saskatchewan running a practice with a student, or working in-house for a corporation, what possible use would that be? The local knowledge that a student needs is almost certainly not possessed by the professor, but by a lawyer who regularly works in that community. Experiential learning opportunities of the kind I’m talking about are therefore, first and foremost, run by practicing lawyers who liaise with a member of faculty. That faculty member’s role is, mainly, to ensure that the experiential ‘course’ has ‘academic’ content - meaning some role for critical reflection on the experiences, possibly through the assignment of a paper or reflective journals.
I don’t want to speak ill of this, since I think critical reflection is an important thing for universities and law schools to provide to students. But I want to stress that the critical reflection may be the part of the ‘course’ least valued by the students - simply the price of admission for doing the sort of experiential learning that they would do without any academic mediation at all if they had the opportunity to do so. It may also be the part of the course least valued by the profession and the local community ostensibly served by students participating in the experiential learning opportunity. And insofar as the ‘critical’ dimension draws on general ideas about law, ethics and professional responsibility, it is the part that can most effectively be offloaded to AI. The value added by the course - the “last mile” - is not the critical dimension, but the uncritical exposure to local knowledge.
There are other kinds of local knowledge. It might be the local knowledge of a discipline that only a person who has done serious, committed research in that field his or her entire life could acquire: the legal historian who has done first-hand archival research and turned up fascinating new tidbits to which AI could not have access; the legal philosopher who engages with debates about the nature of law by looking to an unjustly neglected thinker from another philosophical tradition; the legal anthropologist working within a community, showing students first-hand how anthropological methods might be applied to it; academics bringing students into the debates that they have with other academics in their field, introducing them to the personalities and idiosyncracies of leading minds.
This is deeply honourable work for an academic. I would feel much better about the state of the legal academy if more of us were doing this sort of intellectual labour of love. But again I want to stress that this is not scalable: a tidbit is not new forever; a thinker is neglected until she isn’t. This requires active, ongoing work by academics deeply committed to knowledge for its own sake, working with students who also have that sort of commitment. And it is precisely the sort of work that most law schools under-value, and that most law schools - especially smaller, regional law schools - cannot afford to emphasize. Realistically, it is only the ‘elite’ law schools that could do it; the schools that, not coincidentally, are most regularly criticized for being impractical and insufficiently concerned with the inculcation of legal ‘skills’.
What about values? I know that many of my colleagues across the legal academy believe that this is the thing we are here to teach to students. I have doubts that this kind of answer will satisfy our critics. Leave aside the fact, for now, that questions about what a lawyer ought to be trying to do simply because he or she is a lawyer are deeply contested. (Look at the arguments over the Notwithstanding Clause, the controversy in Ontario over the Indigenous cultural training course, DEI/EDI initiatives, the TWU controversy a few years ago, the COVID-19 measures and response to the convoy protest, and much, much, much more. These reflect deep divisions over the nature and significance of the rule of law, its relationship to democracy and other values, and what it means to be committed to law.) Leave aside that it is not a good look for a university to even appear to be indoctrinating students on controversial questions. Leave aside even that legal academics, given the public pronouncements that many of them make on social media, are not obvious role models for the bar and bench. The fact remains that many law professors simply lack the background in legal, moral, and social philosophy to engage with students on the meaning and significance of law at more than a superficial level; i.e., to engage in much more than vacuous sloganeering. (Some do, and I’m sure everyone will choose to think of themselves as the exception.) Even if AI could not produce passable philosophical papers on these issues - and we know it can - the idea that many law professors can offer instruction to a roomful of law students that travels ‘the last mile’ on those issues is, with the greatest respect, implausible.
What about just doing what we have always done, but doing it better, more rigorously? Continuing to focus on the teaching of general legal doctrines, concepts, methods, techniques, but insisting that students learn them at an exceedingly high level - such that every law graduate could critically assess the output of an AI-generated brief or memorandum? Perhaps that kind of high-level mastery would amount to a kind of local knowledge. Maybe. But even if it does, that kind of mastery cannot be delivered in anything like the way we deliver legal education at present. It cannot be massified. Though it is not a popular thing to say, we know full well that most law students do not have a deep understanding of the material we teach; many lack even a superficial understanding. To provide the sort of deep understanding of the law that will be in demand would require vastly more intense, curated, one-to-one, ongoing mentoring, with iterative assessment on every topic. A single professor could not possibly do this with every student we currently enroll. It would not be scalable and we know it.
Back to my central theme. If AI can do all the things we are capable of teaching, and law schools are not capable of teaching much (if anything) else, then what is the case for law schools? A dean must be able to answer that question when it is inevitably raised by external stakeholders. (Yes, it is inevitable. Don’t kid yourselves.) The problem is especially sticky because, for several decades, law schools (like universities more broadly) have defended their existence principally on the basis of the marketable skills and credentials they confer upon graduates. Law schools are unused to defending knowledge as a worthwhile pursuit in itself. To do so now would invite unwelcome comparisons to humanities departments that are themselves in deep trouble.
And, to be clear, deans do not just confront a public relations exercise. Whatever it is that law schools are expected to provide, the dean must ensure that we provide it. I anticipate that there will be a profound level of disruption to legal education – and sooner than people want to believe.
Perhaps there is room for optimism. I don’t want my next dean to indulge in despair. But a would-be dean who tells me I have nothing to worry about - well, I can’t take that person seriously.
Comments welcome.
