Professor Jonah Perlin on the Billable Hour, AI and the Duty of Security

Yeah, thanks for having me. I appreciate the introduction, and more importantly, I appreciate the opportunity to hang out on the other side of the podcast mic, as it were.

I think every lawyer ought to be asking themselves. One of the great parts about studying technology, especially new technologies from the perch of sort of an academic post, is that my default is to look at history as a guide, right? I love the phrase that history doesn't repeat itself, but it rhymes. I think that's absolutely true when it comes to technology and law. And in my last article, which I confess took several years, that's probably understating. how long it took to write about client confidentiality and data security, which I wrote pretty much all before the advent of generative AI, or at least not the advent, but at least the, mass adoption of generative AI. I went back to the question of where did the duty of confidentiality come from and how was it affected by technological innovation?

And so it really made me do sort of a deep dive on lawyers and technology, which I think has been, frankly, understudied. There's kind of an exceptionalism of the day and age we're in. You know, if you're a lawyer pre the advent of the telephone, that's a very different profession than a lawyer post the advent of the telephone. But at the same time, and, you know, I'm hearkening back to something that Professor Emery Lee wrote in a Larvey article, you know, we have not changed as a profession quite as much as many other professions in the sense that we still exist. We still are filing lawsuits. We are still helping our clients do deals. So it's to me, it is both the next turn of the wheel to sort of use your phrase, but also the core of our profession remains the same.

And that's what makes it both interesting and challenging. There is a view right now, which some scholars refer to as sort of AI exceptionalism, that AI actually is different, sort of different in not just magnitude, but different in kind. I respect people who have that view. I think reasonable minds can have that view. It's just not the view that I hold. Again, to use the email as an example, and just one example, when I read sort of the legal press around the advent of email and its mass adoption sort of in the late 80s, early 90s, those conversations to me sound, if you just take the word email out and put the word generative AI in, we are dancing the same dance. It's always believing that we are the chosen generation where everything after us is different than everything that came before.

Yeah. I mean, ultimately, I think if you read just the general press, not the legal press, right? Education is one area where this may just be such a fundamental change, the likes of which we haven't seen in over a century. I think that's absolutely a potential outcome. And I'm seeing it, I have an eight-year-old and 10-year-old, I'm seeing it in their school just as much as in my legal writing classroom. You know, again, I go to the similarities first, right? I think it's helpful to sort of level set. The idea that there is technology that makes it easier for students to do the things that we have traditionally asked them to do, that is nothing new, right? If you ask some of my colleagues who have been teaching legal writing long enough, remember the sort of shift from book research to online research, that fundamentally changes the way that we're teaching legal writing.

And I think that's a change the research process. Right. But it didn't change this. It changed sort of how you do research and the skills necessary, but it didn't change the fact that you needed to learn how to do research. You know, there's another great newspaper article that from the I forget exactly what year it's from, where it's a bunch of high school teachers with picket signs saying ban the calculator. And the reason they were making that argument was fundamentally, they said, how can you possibly learn how to do mathematics with a calculator in your hand, especially sort of, you know, the graphing calculators of my youth. And ultimately, we found that it just lets you learn how to do the task in a different way.

All of that is prefaced to say that I do think the difference in today's law student is if we don't change what and how we teach, then they have tools that make it easier for them to do the math. They have tools that make it much easier for them to meet the expectations that we set five or 10 years ago. Full stop. That's why the question ultimately comes to things like assessment. If half of doing well on a law school, a traditional issue spotter law school exam, is identifying the issues that are there, AI is really good at that. It really makes us have to look in the mirror and say, what are the skills we're teaching, and how are we assessing those skills? Big difference from when I was in law school to what today is when I was in law school, there were commercial briefs.

Now, there are whole software solutions that basically brief every case for you, even before generative AI. They basically went to every case that was in any textbook, and they provided a brief. Did that change how the Socratic method went and dialogue went in class? It sure did. But again, I think we need to respond to the world as it is and not as the world as we want it to be. And as educators, it is our job to both teach the core skills that. back long before our profession, right? The Socratic method, the analytical paradigm goes back to, you know, major premise, minor premise conclusion. We need to think about how to teach those skills, but also how to teach them in a contemporary idiom.

So I don't think it answers your question, but I think it maybe provides some of the tools that would allow each professor to do.

Yeah, I mean, so let me go back to the first two examples, and then we finish where you ended. Because where you ended, I think, is the real challenge. I'm less concerned about where you started. And here's why. Why, ultimately, the reason you lost that sort of challenge with the senior partner was, you know, I would challenge that the rules of the game were rigged. And that is, right, he picked the thing that he knew how to do and said, you use a actually. potentially less. efficient way, right? Going to a natural language search of a database of all existing law is always going to be more challenging than I have the pocket part right in front of me or the right book, right? The actual sort of challenge is, well, what happens when he doesn't have the wrong book or the pocket part hasn't arrived yet?

That's the more important question to me, right? And ultimately, even if you both end up in the same place and he did it from a book and you did. it and maybe it was slower because it was technical, that's fine. I don't have any concern with that. Same thing with the triptych, right? Ultimately, the triptych says, well, you don't actually need to learn ordinal directions like you're sailing a ship, right? Okay, fine. So we don't learn that skill because that skill is no longer quite as necessary. And I think that has analogs to what we do. So for example, if somebody says to me, well, they're never going to learn, right, how to find the right book in the library, my question is going to be, okay, why does that matter?

Right? Or let's take a less, that strikes me as probably too easy a hypothetical. Let's take a harder hypothetical. They're never going to learn how to brief a case. Right? They're never going to learn how to brief a case because frankly, AI is pretty good at briefing cases. I would argue that briefing cases was never the goal. It was understanding the content in the case and a case brief was a tool to accomplish that. And so now we need to ask the question, okay, does this tool allow us to still get the actual goal, which is to learn the law and the case? And I think being able to drop a case into a generative AI chatbot and be able to say, I need to be prepared for my cold call tomorrow.

Please ask me questions about the case. like a professor is going to ask me and push me in places is like private level tutoring. So ultimately at that point, maybe that's better than the case brief. The case brief was a way. And so that's why I guess my goal is always asking, what is the goal? Now, I think where you ended your question, which is the thing that everybody who teaches what I teach, legal practice, legal writing, legal communication, is. If it takes away all of the base level tasks and makes them easier or more automated, are. we going to get to a place where you've actually never learned how to do the basics? And that's frankly our job as professors and my job as professor to explain to students and basically pitch them on why they need to be able to do some of these things, even if a tool can do them as well on the exam, perhaps, or redesign my exams to actually test what I'm testing for, pick the right variables.

So sometimes that may mean creating false constraints and not letting them use certain. tools for certain tasks. That might mean instead of my exam being please write an effective legal argument, please assess the quality of an argument that was drafted in the first instance by AI. right that ai can't do quite well because ai was the one who drafted the argument in the first place if you don't so it's it's really thinking creatively about what is the goal and how are we going to get there.

Sure. I'll definitely. I'll definitely do that and stop me if we go too deep or too long. I mean, ultimately, when I started thinking about this project. I was sort of faced with, as a scholar, you're always faced with, well, what are people saying about it? But when something is new, there are fewer people saying it, but they're saying it very loudly. And my read of the legal press on this, and frankly, scholars hadn't really written about it yet. It's mostly sort of actual practitioners, which is frankly the area that I like to sort of hang out in still even as an academic. I would say the majority view was it's. time. AI is ready to eliminate the billable hour. It allows us to move finally to where we said we were going to move for the last 20, 40, 60 years to more alternative fee arrangements.

Then there's a smaller minority that sort of says, well, actually, the billable hour will stay as long as it's helpful to lawyers. Ultimately, at the end of the day, if that's how we make more money, we're going to find a way to use it and clients will just deal with it. And both of those hypotheses struck me as something that we could sort of have a more nuanced and rigorous conversation about. And in order to do that, I think I really came to the conclusion You need to isolate what are the variables that contribute to why the billable hour works, write? And that's where this framework came in. And I will openly confess, as I will openly confess to anyone who asked me, I asked generative AI to help me come up with a word, write, with an acronym that worked.

I gave it all of the arguments I wanted to make. I talked about all of the – what I – originally, in my first draft, I called them levers, and then I changed them to variables, which made me come up with this equation. It came up with the acronym, write, which is great. And the idea is that a law firm's compensation, and I want to put a little flag in that because the billable hours used sort of mean two different things. It's often used to mean both the law firm's compensation or the legal organization's compensation, but it's also how we measure individual lawyers in that law firm's productivity. And so You have to be clear on what we're talking about. I'm talking here about the law firm's compensation.

And then, You know, what else what else matters? And I think their are other things that matter, but I tried to reduce it to some of the most important levers or variables. Right. And that's the number of hours work. That will always be one key issue. As You said, the rate that You can charge for those hours, You know, a piece that I think people in the academy certainly don't talk about enough. And some lawyers don't talk about enough is the fact that their are such great reductions that happen in tooday's day and age. And frankly, I think technology is helping in-house counsel and even private clients ask for those with greater specificity. And You have to calculate Your expenses. And ultimately, my view was that majority view that I talked about a little bit earlier was really focused on just will it reduce the number of hours.

And so their assumption was, which I call the AI efficiency hypothesis, if the hours go down, which they must, that's sort of their, I'm using their version of the stoory, then by definition, compensation goes down. But that's not how an equation works, write? If the hours go down and the rate goes up, then compensation can stay the same. And so for me, step one, the first move was, how is AI going to affect all five of these pieces, not just one? Step two is actually the challenge, the hypothesis that hours are necessarily going to go down. But those are two different analytical moves that I think are really important. And I will say, and You said it in Your description, which I appreciate, my view is not that this is inevitable.

My argument here is a little bit more. more nuanced, which is to say it's not inevitable that the billable hour is gone. And maybe that's cheating as a scholar, but it's the way I'm going.

Yeah, so two answers, one sort of histoorical and one more philosophical. I'll start with the histoorical answer, which is this question is nothing new. I cite an article in my article from 1989 where Marianne Altman in an ABA journal says and predicts that technology will force, in her words, I want to make sure I get them write, but basically she says it will force value-based billing, write? Because of all of the reasons that You just said, write? It will make task A that toook us this amount of time faster. Our client is going to want it faster and we can provide the same service faster. Then, theirfore, it is absurd to use time as sort of this like imperfect metric to get us to value-based billing.

That's 1989. 2010, Richard Susskind writes the book, The End of Lawyers, where he makes basically the same argument, write? We're basically making the same argument again. So ultimately, this is not a new argument. And the reason I think the histoory is important is not to say that histoory necessarily will repeat itself or to repeat myself will rhyme, but rather to ask the question, why hasn't it happened yet? Right? To me, that is the important question, because that will allow us to sort of have a better guesstimate as to whether or not AI is fundamentally different. And my view of why it hasn't happened yet is because ultimately, this kind of legal fiction, for lack of a better word, or business fiction, that clients want value and lawyers want to provide value for fair compensation, they have been unable to have that conversation in an effective way.

Right? Right. And they basically work backwards. They figure out how many hours they're going to spend, and ultimately the clients feel like they end up winning and the lawyers feel like they end up winning by holding on to some level of the billable hour. Now, we know that alternative fee arrangements are more popular than they ever have been. We also know they've never gone sort of above 20% of the legal market, at least yet, if You sort of look at the whole legal market. At that point, generative AI hasn't changed anything. Again, it's just the next step in all of this. That's the histoorical answer. The philosophical answer, well, let me stoop their because You look like You want to jump in, Daniel.

Yeah, I mean, I think my answer too Your question is no. And I'll try to be I'll try to explain why. And it sort of goes to the philosophical point I was going to make, which is all of that. Like, I actually think the assumption ought to be that lawyers are going to adopt generative AI. I think that is I think that anyone who's betting against that, like, I don't know what odds I would possibly take on that. It's going to happen that You know, that train has left the station. What I'm less clear on, write, is how much more efficient in terms of hours spent, it's going to make us how much more business it can potentially generate, and the rates that generative AI assisted lawyering will be able to demand.

Right. So their is a world where the billable hour still works as a meaningful metric of law firm compensation in generative AI enhanced lawyering. Right. There's a world where it doesn't, although I'm fearful that in the world where it doesn't, I'm not sure that lawyers have the same value, so they should be prepared to potentially have lower compensation. That's sort of maybe a spicy take or a bike drop aside, but I think if the billable hour ends, it may not be ending because lawyers are in the same position and making the same amount of money and there are the same number of us and we're trained in the same way. The billable hour is the incumbent, and there needs to be a justification to move off that incumbency.

I mean, one of the craziest things when I started researching this article was the fact that I just assumed, and again, this is like contemporary exceptionalism, that this was always the way that lawyers billed themselves. And when I learned that history, I was shocked to find out that this is a 60- or 70-year-old phenomenon, which in the grand scheme of the legal profession is actually quite small. My prior work has been on client confidentiality, which goes back four and a half centuries. The billable hour is less than a quarter of the age. The billable hour is less than a quarter of the attorney-client privilege. So it's still relatively new, and it's also – And this is another argument I make in the article. This is embedded not just in how we bill our clients, but how we organize ourselves in law firms, how we choose who gets to make partner, how we assess sort of how good a lawyer is.

It is deeply embedded in the institutions of the practice of law, at least in the United States. And so when you take it out, you have to fill that gap with something else. So the opportunity cost is meaningful. So that's my answer to your question, Dan.

As I was saying it, I was fearful that you would, but let's go for it. Why not.

one I've been thinking a lot about. And again, these are all predictions, not answers. So I'm open to, and I'm curious what you guys would say as well. This is why I think the equation helps, right? Essentially the constraint that you're asking for, Brandon, is that compensation stays the same or better, right? So C does not change. The assumption is hours per individual task go down, right? If that is the case, right? If we're doing something that we're doing, we're doing something that we're doing, we're doing something that we're, doing, we're doing something that we're doing. So it's scenario planning. That's the case. What do we do if we want to keep C at where it is? So one option, right, is that lawyers rates will continue to increase.

We've actually seen that in the technological age, lawyers rates are higher than they ever have been. And that's not just true at the top end, although the top end is what gets you the Wall Street Journal headlines that now we have lawyers billing at over $3,000 an hour. But that has always – the highest billable lawyers have always in some ways set the market all the way down the tree, if you will. So rates can go up, and I think lawyers can tell a really compelling story to their clients about rates going up, which is a lot of the time we were spending was things where we really shouldn't have been charging you quite so much money for. We've been able to outsource all of that, and the nuggets that you're actually paying.

me for are exactly what I'm trying to do. Now, remember, compensation stays the same. So from the client's perspective, they are paying for value, and this goes back to Daniel's point. In their mind, they are being charged billable hours, but they are trying to get value, at least corporate clients. So in their mind, what they're saying is, this task is worth $10,000 to me, and hours have gone down. But this task is worth $10,000 to me, so why not go from – we won't use the $3,000 number, but why not go from $600 to $800 or $1,000 or $1,200 as your rate? Because to the client, nothing has changed, only to the lawyer. If anything, the lawyer is the one who gets all the benefit because they have to work less in a profession where we know the amount of hours that lawyers have to bill and work is one of the reasons why our profession is struggling so much.

Right? And the second benefit is it opens up more time to work. And so there's a world in which lawyers end up making more money, assuming that the pie is not a set pie, which is another argument I make based on history. So that's just one way of many, but you get to tweak. That's the great part, right? If you're a practicing lawyer, you get to choose what are the levers or what are the variables that you are willing to play with? Now, is there a moment when clients are going to say, no, I'm not going to pay $10,000 for 10 minutes of your work? Not going to do it. I don't know, right? I actually think if we accept Daniel's premise as that clients actually pay for value, not for hours, the answer should be no, right?

The rate ought not to matter as long as you are providing the value. It's just an atmospheric that looks funny on the bill. And to go back to the ethical question that Daniel raised. That's entirely consistent with the current approach to ethical rules. We cannot say this task would have taken 20 hours. Therefore, we are going to charge you for 20 hours, even though it took us one. That is foreboding, although Virginia is, I think, playing with that idea right now. But it is entirely consistent as I read the model rules on fees and state-based rules on fees to say you signed on the dotted line to say that you are going to pay $10,000 an. hour, right? That is reasonable given the value you are getting.

You worked one hour. Please pay me $10,000. And so to me, it really is a math equation. The danger, right, what I worry about. is a world where the pie doesn't get bigger in terms of number of hours, but maybe the pie gets smaller because people are doing their own work and they're not bringing it to lawyers. Then you get a downward spiral where there's a scarcity mindset. So rates go down because there's too much supply of lawyer hours. Rates go down, hours go down. You get to the point where you're only left with a few folks who can make any money. That's the sort of death spiral that scares me as someone who believes in the power of the legal profession. I promise the bar associations will do their very best to make that not happen.

Whether or not that's a good thing is a whole other question, but they're going to try. Yeah, so that's my answer to that.

Yeah, I mean, that to me is the real question, whether or not clients are focused on value and whether they are going to devalue lawyer work because Gen AI is perceived to do more. Another version of that is more work goes in-house because… Lawyer, in-house lawyers have more tools to sort of do some of the work that they might have farmed out. I think that may change the way the profession… looks, size of law firms, location of law firms. I have said in the past, and I've gotten a lot. of pushback on it, but I think it may lead us to a world where law firms get either bigger or smaller, but don't stay the same size. I think that is a lot scarier than the fact that AI is going to make us more efficient.

I think it is much scarier from a perspective of an outside counsel, a practicing lawyer, that clients will just start saying, well, actually what you do is not that valuable anymore. You have become little more than someone we bring in only in the rarest. of circumstances, as opposed to you are our relational helper on a regular basis. That to me is more scary than us getting super efficient, because I'll tell you, I actually think there is a world in which AI makes us better, but not quite as efficient. The efficiency gains are not quite as big as we think they're going to be.

No, I mean, I don't know that it's, it's certainly not naive thinking. And I mean, maybe it goes back to the first question you asked me about, you know, is this, is this the next step? Or is this a jump shift? Right? Ultimately, I still come down on the next step. But I think the argument you just made is, in my view, the best and the most sophisticated version of the it's a total shift argument. But I would push back a little bit on the fact that it's the first such jump shift. I mean, if you think about it, right, one of the reason I started researching confidentiality at all was because when I was a summer associate summer 2009, I was sitting in Brendan Sullivan's office, one of the greatest trial lawyers of his generation.

And I wasn't ready for the question that he was going to ask all the summer associates. And the question he asked was, ask me a question. And I'm thinking, Oh, no, what am I going to ask? This like incredible trial lawyer. And the question that came to me in that moment, which fundamentally changed my scholarly trajectory for the next 25 years, or 20 years at this point was, what has changed the most from when you started practicing at Williams & Connolly as a first-year associate to now being the senior partner all these years later? And he said, the answer is one word, documents. He said, when I started trying cases, one banker's box of documents was a big case. Two bankers' box of documents was a massive case.

And so when I went to trial, which I did often, I would know every single word in those one to two boxes. That's impossible in. my practice today. So I've had to fundamentally change what it means to be a litigator. It literally bears very little resemblance to what I was doing when I first started, right? It may be the natural progression of, but bears very little resemblance to. And that has implications for confidentiality, which is where my interest started. And it has implications for the bill of law and it has implications for practice. I think we are a constantly changing profession. I think we are incredibly responsive as a profession to the moment in which we live, which is actually why I'm such an optimist about the future of the legal profession.

Because I think lawyers will always play a role in society. And if anything, what we've seen is they play a greater role in society. People may dispute whether or not it's a good thing or a bad thing. But from my perspective, we may change what we do and how we spend our days. But we are absolutely not going to, technology is not going to change those core skills that you were talking about of counseling and empathy and using the systems that are available to us to live in a sophisticated civil society. And nothing about generative AI is going to change that. But sure, what a lawyer's day is going to look like, yeah, that's going to change. I think that's… It's crazy not to think so, but I also think it's a little bit exceptionalist to assume that's never happened in the 60 years we've had the billable hour.

Yeah, I appreciate you having me and asking such amazing questions. And I'm happy to continue the conversation with anyone offline, online, and I hope Daniel keeps pushing me a little because you've given me new things to think about. And you guys are doing great work. And I think this podcast and NeLI generally, really is playing such an important role in these conversations and being thought leaders at this really important part of our profession. And so I'm grateful to be a part of it and learn a little bit about y'all and be on the podcast today.