[Originally posted to LinkedIn on August 13, 2025 and reposted here for archival purposes.]
This is a law/legal profession/law schools and legal education post, prompted by my conversation yesterday with a group of incoming law students who are going through my law school’s orientation week. Every single one of them asked some version of “what’s happening with AI and law?,” always with a spark of optimism and enthusiasm.
The short version of my response:
Every lawyer in America, regardless of type of practice, is dealing with AI systems in one respect or another, even if (in a handful of cases that I know of), “dealing with” means “keeping AI affirmatively at arms’ length for one good reason or another.” When I was a law student, “Westlaw” and “Lexis/Nexis” were the exotic interlopers into law study and practice. By the end of the 1980s, they were part of the furniture both in law offices (where the cost curve took some time to come down) and in law schools. AI has become a default setting in professional practice worlds far more rapidly than that.
In law schools today, we are still, collectively, far from comfortable with AI. At most, AI is accepted as “a tool” of the contemporary lawyer. That’s an early stage adaptation; the phrase “it’s just a tool” simplifies and domesticates a complex and rapidly changing technology whose directions, functions, and impacts are only very gradually coming into focus. Eventually, the profession (in whatever size and shape it continues to exist) will see AI/human combinations in law the way that researchers in socio-technical studies have understood machine/human combinations for a couple of generations: as systems with overlapping domains of autonomy and dependence. That’s a broad description; the specifics are to be determined. But I am certain that in time, AI in practice will be accepted as much more than a tool.
The longer version, which I had no reason to share over lunch:
As AI swallows whole big chunks of law (along with professional services and expertise of many sorts), again we hear calls for law schools to innovate and reform at more or less warp speed in order to produce law graduates who are prepared to succeed – and to meet both client and community expectations – in the AI era. Reform, die, or fade into irrelevance. And again we hear responses from inside the walls and halls of academia: not so fast. System-level innovation faces system-level headwinds. We’re innovating as fast as we can. Let’s not forget the values and ethics that built systems of law and that (we believe) clients depend on.
Is there a right way to reconcile the two, seemingly contradictory views?
No. But I have notes from my seat, as I start year 28 of full-time teaching:
System-level change in legal education really is difficult largely because law schools, like the universities of which most of them are parts, are built not to change. Metaphorically, they are like the brick house made by the third little pig. The Big Bad Wolf can’t blow it down, meaning that performance expectations for all constituencies are clear and consistent, internally and externally, across the decades (yay). But it is stiff and brittle and very, very difficult and expensive to renovate (boo).
That’s not to ignore the fact that accreditation requirements are unnecessarily onerous (hello, the ABA Section on Legal Ed), nor the fact that as a group, law schools have learned to tune their faculty recruitment and curricula to the demands of “the bar exam” (hello, state Supreme Courts), nor the fact that essentially all US law schools operate as minor curricular variations on a single model (they do, hello both of the above, path dependence, and a supreme lack of imagination). But turn law schools today loose to do whatever they might like to do, and odds are high that we would still have a three-year JD, a suite of mandatory, uniform first-year courses, a bundle of electives in the upper level, and clinical and legal writing attachments.
The fixation with, well, fixation, implies that as with a lot of buildings built of brick and of concrete, the (unformed but sure to arise) needs of the future are often sacrificed to favor the (known) demands of the present. Deans typically are recruited and rewarded for short-term performance (fundraising metrics, bar exam performance, and “looking good, Billy Ray!; “Feeling good, Louis!” (cue Eddie Murphy and Dan Ackroyd in the film “Trading Places,” which is a fair if entirely different sort of parable for the fate of established, corrupted incumbents confronted with nimble upstarts who know where the system is flawed); faculty usually are recruited and rewarded for plugging into the curricular and field-specific system as it has been, rather than “the law” as it is becoming. Throwing around labels like “Langdell” and “the Socratic method” as diagnostics has only limited value, because those terms are so broadly accurate that they mean lots of different things among different audiences. What matters is that the Provost usually wants to see bar passage numbers, and the faculty wants to hire someone who will cover Torts while publishing whatever the collective faculty expects in student-edited law reviews.
There is no single solution to all of that, no single way to “unstick” a “stuck” system, but the starting point for (somehow) aligning “the law” and “AI” in education, without simply building metaphorical houses of straw and sticks, has to be looking at the extensive literature on system-level and organization-level change management.
In one nutshell:
Every organization, including every law school, has people on the team who are already innovating, often (but not always) in productive and forward-looking ways. Hello, positive deviants; hello to the work of John Kotter. A good Dean will find and support those people. Celebrate them, support them, amplify what they’re doing, and shield them from resistance and resentment from colleagues who are comfortable in the existing system, and from influential alumni who venerate the grand old days. All of this is easier said than done, both on the ground and in the Dean’s suite, but some law professors and their Deans have been doing that, in some cases for years. Hello, Suffolk University Law School and its Dean Andrew Perlman Dennis Kennedy Cat Moon Northwestern University Pritzker School of Law and various Deans, including Daniel B. Rodriguez. Hello Daniel W. Linna Jr. Outside the US, hello Dan Hunter at King’s College London. I see many others in and next to US legal education, sort of like Miss Mary Ann on Romper Room. I’m sorry to omit some great people!
Notably, an enormous number of new and potentially useful and powerful things are being done in law schools today by energetic and imaginative people, things that are often consolidated into durable systems, despite (the ABA Section on Legal Ed) (state Supreme Courts) and (institutionalized inertia). Give it time, one might say. It is easy to get and to be impatient, especially if you are on the outside of the educational system, looking in. (It is also easy to be impatient on the inside, but insiders should have different expectations.) I yield to Denzel Washington, in the film “Remember the Titans”: “I run six plays, split veer, like Novocaine. Just give it time. It always works.”
In a different nutshell:
If you don’t like dealing with the house of bricks, then go around it. “To change the rules, change the tools,” as the computer pioneer Lee Felstenstein once said. The change management literature helps in a different way.
In my opinion there is no good reason to expect that “the law school” in the US, as the profession and academia have refined the model since 1870, will or should continue to be the singular, dominant training mode for “law,” let alone for “lawyers.” Hello to the work of Clayton Christensen and the Clayton Christensen Institute. I am not a fan of the rhetoric of “disruptive innovation”; I focus instead on the insight that incumbent organizations tend to innovate (if at all) by incremental adaptation, creating space for upstarts to meet social needs (sometimes market needs) by operating outside the bounds dictated by pre-existing bureaucratic imperatives.
In that spirit, what might it look like to build “law training” from the ground up, independent of ABA and AALS histories, rules, and expectations? The “why?” – the social need – is relatively easy; look at the one legal services sector that is actually thriving right now and explore the numbers of people without JDs running and working in those companies. Would more JDs turn these into better companies? I am skeptical. Would more legal training turn them into better companies? I am optimistic.
I don’t know what the end product of a “build a new training system” exploration looks like; this is entirely speculative, although I wrote up an “invitation” to this conversation back in 2018, before “AI” became the prompt of the moment. Some recent law schools have tried this approach, or claimed to (hello, University of California, Irvine School of Law), but have abandoned it and abandoned it rather quickly. I would love to explore the topic with anyone who sees the legal world through that set of goggles. I would especially enjoy exploring it with anyone who is willing to step away from the assumption that ABA compliance is a foundational prerequisite for building institutional pathways to legal expertise, away from the assumption that qualifying graduates for bar membership is a foundational prerequisite, and away from the assumption that one-to-one client service (whether the client “one” is an individual or an organization) is the default be-all-and-end-all of delivering legal expertise to society.
In short, if this is your nutshell, then let us start over, as “society” collectively did between 1870 (the founding of the modern Harvard Law program) and the early years of the 20th century, more or less. That’s when we got full-time law professors, “law” as a systematized body of knowledge and analytic techniques in their modern senses, and the kernel of the client-centric model of the lawyer. There is a baby in that bathwater, but let us not assume that “training people to be lawyers in the AI era” is the contemporary goal. Instead, assume that experts are needed to build and operate systems of law in complex and divided social settings defined significantly by systems of information production and circulation. That means data and algorithms and AI, but not only data and algorithms and AI, and it means blends of material and immaterial wealth, power, community, and opportunity that only dimly resemble their 20th century forebears. What institution(s) will produce that expertise? What does “law” have to do with that expertise? Where will those institutions be built? Who will found them? Lead them? Staff them? Pay for them? What do “systems of law” look like? I mentioned values and ethics a while back; they do not go away.
No one knows what the end point of any of this might be, “this” being “AI and law and training the next generations of professionals.” AI is moving too fast, and higher education moves too slowly. We do need the willingness to try things, to build things, watch them succeed – and not – and to iterate to make them better.
Last but by no means anyone’s final word:
Like a lot of law professors I know, I have been trying things for a long time, tweaking and iterating mostly on my own. In multiple respects, in my professional life at my university though only indirectly at my law school, I spend a ton of time thinking about AI and collaborating with colleagues on AI-themed projects. The conceptual and practice divorce between “my effort at my law school” and “my effort at my university” has gotten pronounced in recent years. At my law school, I am mostly still teaching “conventional” legal subjects, in part, though I have long since abandoned the classic, “Langdellian” single end-of-semester exam (or exams of any sort) as a measure of student learning or performance.
But the most fun I have as a law teacher, the thing that I do that is now most consistently in demand among my students, and my primary effort to maximize the effect of my teaching on the future careers of our graduates, is offering a course in “Leadership.” I introduce students to a collection of skills that – I hope – will prepare them to thrive as professionals and as people, come what may. The syllabus and readings are here. Almost everything there is open and free to all; no Canvas or LMS for me.
It turns out that highlighting AI and technological change in law has provoked me into emphasizing the humanity of my future colleagues most of all. If there is a key to AI, law, and education (“one ring to rule them all,” as Tolkien might have written), in my view, that is it.
Have at it.
The post Brief Thoughts on AI, Law, and Education appeared first on Madisonian: On Governance and More.