AI Is Looking for a Job

Thanks a lot, Daniel. Great to be here.

Sure. It's it's a good question and actually something I think about a lot. There's a lot of differences. I mean, I felt really lucky to kind of find my way into the beginning of that whole industry when it really was nothing and turned into an industry and got to be part of it.

I didn't think that would ever happen again in my career, but fortunately, you know, we have what's going on now, the sort of AI revolution, which is a lot bigger and a lot more pervasive throughout society at a business level, at a social level, at, you know, the Pope's writing about it. So it's obviously a bigger impact. If I look at one of the biggest differences for me, and I think that this has a lot of impacts, is that the eDiscovery industry arose out of a very specific problem that was identified and needed to be solved, was dealing with the mass amount of data and discovery and the rise of digital data and the unwieldy nature of the data and how to deal with it.

AI is kind of an amazing revolutionary solution without a problem. So now I see everybody trying to figure out what to do with it, trying this, trying that. There's some, you know, applications that seem to be useful, maybe some that don't, but it's really more of a solution in search of a problem. So by nature, I think that puts everything in a kind of a different frame.

Yeah. It's another good question. I think when you said that, it's very true. It's it's one of the things I actually enjoy the most.

And when I speculate about how I sort of got into that or why I like it, before law school, was a musician and I played in bands and had a lot of fun and enjoyed the camaraderie and sort of people working together. So I feel like I'm maybe subconsciously trying to recreate that all the time. You know, working with people and building teams is a lot of fun, and it also putting together people in a different way, I think, allows you to come up with ideas and things that weren't there before. So I think it's hard to come up with a completely new way of doing things or new technology or new approaches with all of the same people who are doing whatever you did before.

You definitely have to sell people on a vision, which is a lot like what you do in business when you're trying to get investment. You have to have a coherent enough idea that people get it. They have to see a need or you have to explain the need to them. And the things to me that where it's been most successful and I think a critical element is hopefully directly, but either directly or indirectly, you have to get buy in from the top because at any institution, whether it's a law firm or a business that you're trying to change from the inside, there will be people who don't want it, who are threatened by it, who don't get it, who are you know, there's a million different reasons people can want you to fail, let's just say.

So if you have high level support and other layers, you know, slightly below that, you can use that to combat whatever kind of adversarial pushback you get. So getting support, getting the right people, and I think being really clear about letting people know what you're doing, what your successes are, what your goals are, in addition to the vision, how you're actually achieving them.

That's a good point, and I think the latter is what I agree with. I see people coming up with solutions or trying to use AI and legal for things that I don't think it's ready for right now. I think there are certainly a lot of problems and a lot of things that AI can be used for, but there's a level and you see this at, you know, very prominent institutions where they take a lot of safeguards, but still things go wrong. So there's a level in legal where I to me, at the moment, that if you keep it at a level that's not the highest level of thinking or trust, you can make use out of it.

But if you try to push that too far, you can wind up crossing a line, and things evolve. So what I see the uses are, whether it's for me doing legal work that I do or just other things, is you never just rely on it. You have to really vet everything. So that that creates a point of diminishing returns.

If you really have to do everything over to check it, what is it really saving you? But if you wanna gather things and look at them and come up with some kind of rough ideas or rough drafts, I think it's useful. So, again, to me, there's kind of a theme of what problem are you trying to solve? If you can't define the problem really clearly and make sure that the tool that you're using matches it, you're gonna run into trouble, and that's what I see, whether it's law firms or corporations doing sometimes when they go astray, is they're maybe trying to do too much, not really knowing what they're trying to solve, and depending too much on this thing that they really don't understand.

Yeah. Well, one thing that strikes me as interesting when I read, and there's certainly a lot of experimentation going on. There seems to be a lot of smart people talking about it and working on it. Law firms are wonderful places. You know, I've I've built my career at them, and there's things that they're really great at and things that they're not as good in. And I, you know, personally don't think law firms are normally the greatest place to use new technology and innovate and build things.

It's hard anyway, but and law firms are fabulous at what they're doing. I think they're more successful than ever. Everyone's always predicting the downfall of the legal profession and the billable hour and the pyramid revenue model. And, yeah, when I look at it every year, they seem to be doing better than ever despite AI and despite everything.

So I don't know that they're the greatest place to really figure out what technology is because you have a whole bunch of people who are typically, you know, the firms I've worked at, really good at what they do. They're very successful at it. And and so you're trying to, like, break into that again. If they perceive that nothing's broken, it's gonna be like, oh, why why are we fixing this?

It works for us. So I think there's maybe a lack of clear insight into what the gains should be, what we're trying to build. I think that can be done, and I think it can be done effectively, but I see a lot of scattershot approach to things. And just assuming because I'm skeptical a lot about the marketing and the way AI is sold, so I think there's a lot of hype that doesn't really match reality and a lot of false expectations.

Yeah. I think that's an important question, and part of it goes to understanding what a law firm by when I say law firm, there's all different sorts of law firms. There's a one person law firm up to, you know, multi thousand. I've spent that part of my career in large global law firms, so that's kind of the frame of reference I'm talking about.

Although, I think what I'm gonna say will apply to a lot of them. Law firms are a series of different businesses that are put together and work together. Sometimes they split apart. They're like separate businesses and leaders of each thing.

And even within that, the leaders of different groups are kinda parallel. But, really, it's different profit centers, different groups that have their own business, that do their own thing. And they work together and they collaborate, and that helps them sometimes. Sometimes they conflict.

So it's hard to think of solutions or building things for that whole thing, unless maybe you're talking about infrastructure or back office kind of issues, which, you know, we can talk about. But, you know, to me, the more exciting stuff is how the people are actually doing what they do to make money, is give legal advice and analyze legal problems and help solve them. So that's the background. Where I've had success, and you mentioned Paul Hastings where we actually built technology and sold it, is by focusing on a couple of individual subgroups of the law firm as a whole.

We didn't try to say, hey, everybody. Here's innovation. You know? Let's try it. I went around the whole firm actually and talked to a bunch of people and talked about different problems they had and where they might think that they need help.

So we weren't talking about AI. I would just say, what's going on that makes your job harder? You know, what would be, well, helpful? Some of the things made sense for technology.

Some of them had nothing to do with technology. Some would be things that might be possible in ten years from now or something.

Figuring out what their actual pain points were, then sitting back and analyzing and saying, where could technology possibly help with some of those? We picked a couple where it really made sense. And then we built prototypes, tried them out, got really good feedback, went back and forth. It was people who thought it was a good idea, thought they had a problem, were willing to work with us, and then we solved those specific problems. So there might be lots of different ways to do it, but that really worked for me. And I found that that cut through a lot of the political stuff, and I dealt with those particular leaders of those groups with, again, the support of the leadership of the firm itself.

I think it's hard for lawyers to most lawyers, not everybody, but a lot of them to step back and see what they do in terms of a kind of a business process, which is what we were were talking about then. I think it's maybe more acceptable nowadays to think of that for certain practices. And I know law firms often have hired project managers or business analysts and try to organize what they do in that way. Some people are probably more open to others.

What I think you're identifying is how that now is intersecting with this incredible technological explosion that's going on with agents and, you know, to go from, you know, cobbling shoes to using teams of AI agents who are working with each other and collaborating, that's it's quite a leap. So I think when I start thinking about that and agents in general and vibe coding, I guess I think more about the risks and things that can go wrong because you're also potentially putting very powerful, somewhat, you could argue, uncontrollable technology in the hands of people who, again, are super bright and skilled at what they do but don't really understand that technology.

The, you know, the risk light flashes to me when we start talking about things like that rather than, you know, how great of a workflow can we come up with.

Yeah. That's an important point. I think there's a number of reasons. People go to law school for different reasons, and some of them have business backgrounds or they get an MBA when they're going to business.

When they're going to law school, can currently get an MBA. So I don't want to paint a total broad brush with that. But generally, people go to law school because they're interested in the law and they're interested in practicing law, which is not sort of natively thought of as a business. They obviously if they go into private practice or corporate law or whatever kind of stuff we've done, they obviously are dealing with businesses.

Some of them develop expertise in those areas if you're doing deals and so on, but it's not exactly why people go to law school. And I think importantly, although things are changing, I imagine, you know, some law schools have different programs that they're implementing, but if you go to the sort of the upper tier of and maybe more innovation is at the lower tier. The upper tier of law schools have these people who teach at them that are brilliant. You know, it's fantastic stuff.

I think it's a good idea and I think, you know, law students, people who go to law school and are successful are not smart. They're good learners, so I think forcing people to think about things differently. And there has been a move even before AI. I remember even the early days when I started practicing law at a big firm, they started bringing in accountants and people who understood finance.

And before that, it was evidently before I was doing it, but it was really hit or miss. And so there's been a general move, I think, clearly can be tracked of running law firms more like businesses. That's that's definitely happened. And maybe now with all this technology that's going on, it'll even accelerate that.

There's a few, you know, sort of, I think, identify themselves as native AI law firms, and there's a lot of interesting stuff going on that wasn't before. So I think things are evolving. And to the extent they're successful, that'll be looked at at the law firms. I just I think unless things change despite the sort of predictions of their demise, they're doing fabulously well.

And doing fantastically well is never a huge motivator to change. You could argue it's the opposite, and that's just human nature. Like, why would you try some radical change when things are going incredibly well? It might be the law firms that are upcoming or these new law firms that are gonna force change, but if you're if you're doing better than you ever did and someone says, hey.

Let's change everything. I don't think that's a natural response.

I don't think that. I think if you can make the business case and show that it works, lawyers typically are not, you know, wanting to be the first movers in something. They wanna see if something works and then do it. Again, it goes to maybe learning about managing risk and sort of that's the role with with companies.

Sometimes you tell them they can't do something or, you know, to be cautious. So lawyers are not with exceptions, individual exceptions, obviously, but they're generally by training, massive risk takers. They're more risk managers. But I think if you can show a case to the right group that, you know, you can save this much money and not have to deal with that, I think a lot of those business aspects are hard for law firms.

They're challenging. They cost something.

But I think if you can show that we can do this more efficiently or, you know, this is the way things are going or the market's moving that way, that they would be open to listening to it. You have to find the right ones, and I think it's definitely possible. And I think there is a level where things start changing and lawyers at big firms and are aware of what's going on and they're smart and they see that there's change. So there could be an inflection point where things start changing enough where they well, it's why why all these big law firms are dealing with in one way or the other trying to deal with AI.

Right? Because I don't think they were sitting around in their meetings and saying, you know, we really need AI in our firm just because of a natural technological need. There's things that we're not really getting done that we could. They think we need more independent agents that can go off and work on their own.

And really think everyone in the law firm should start coding because we need more, like, spontaneous software code within our infrastructure. So I think it's more of a reaction to all of the, you know, the PR and the publicity about what's going on and clients asking and so on, very responsive. Clients saying, hey. What are you doing with AI?

So they say, well, we need to do something with AI. So it's a bit reactive, but I wouldn't close the door on it.

Yeah. Well, my experience is when they think they need AI, it's more of, like, we should have AI. It's not that they have any actual idea of what it will do for them. And, again, that maybe is coming from client pressure or the industry that they work in.

So, typically, it's a business problem that they have. They might wanna automate things or they think things are not getting done efficiently or it's taking too much time or too many people to do something. So when it typically makes sense is when there's an actual again, I go back to what's their problem. You know, there's an actual business need.

And then we can say, oh, know, AI might be useful for this part of that. Like, you know, you have this issue, and this technology can help you. Maybe it'd be good to have a kind of a human language interface with that. Do you think it would help the opportunity just to talk to this thing and tell it what you're trying to do?

Know, actual instances of where AI can actually do something or maybe analyzing a whole bunch of data and pulling out things that you need for you to look at and analyze. So the AI part where it makes sense usually is elements of a larger problem, and we can present that to people and explain it. That's usually very graspable and makes sense to folks rather than, like, here's a bunch of AI, you know, it's gonna take over your firm.

One thing for lawyers, they they are receptive to discussions about risk and where things can go wrong, and they're mindful of that because if they have a firm, you know, they're concerned about the reputation. And certainly, we've all read about situations where firms have problems and none of them want to be that firm. So I think they're pretty receptive to risk. There's a whole industry coming up with, you know, dealing with risk of AI, and I'm actually advising a company doing that now.

And so I talk about what they do and the kind of issues that they're dealing with, which are AI can introduce a lot of things into an organization that the people in the organization may or may not be aware of that can create problems. So how do you prevent that? How do you actually analyze code that you're introducing into the firm to see? Have to introduce them to the ideas like where the problems can occur and how you prevent them and whether they can do it themselves or they need help in helping manage the risk with technology and sort of advisory work.

Yeah. So I think that the idea that technology, as you said, it's not that's not the core problem. And I do think the problem is more what you're alluding to, which is you cannot apply technology or AI or anything to a problem if you can't define it and break it down into what its steps are and what it's composed of. And I don't think people who are not in technology maybe understand that in the same way.

But in our times, we're more the users of technology. So I think you have to figure out what the process of what you're trying to apply technology to is. And if you can't articulate that or document it, I learned that from a person Eric and I worked with who was a double MIT grad, who was a technologist who we happened to deal with be in charge of information technology and so on at at a law firm we worked at. And he was a big proponent of you can't automate something.

You can't really apply technology to it unless you can really define it. So we used to sort of build the process first without technology to the extent we could, and then say, How do we apply technology to that? So I think there's a lot of wisdom in that. Things have gotten faster, and, you know, it's a lot easier to do things now with technology that exists that didn't exist twenty years ago.

But I think those principles still apply.

Yeah. There's there's a lot of lot of stuff there. One thing I don't think works is trying to get lawyers who are not so inclined to become technologists or technology experts. Now some are really interested in it, and certain IP lawyers have technical backgrounds, and they're they'll be all over this stuff in that way.

But one of the things I think that led to our success at Paul Hastings was we made the technology invisible to the lawyers using it. We told them what we were doing, we got their buy in, and we showed them results. We got them excited, but we didn't try to get them. They wanted more information.

We give them as much as they want, and we developed ways that we could testify about it if we had to and really explain it. So for the first part, we just said, you're doing this now. We can help you do this, and here's what you do. You just ask us a bunch of questions about the dataset, and then we'll give you some samples of documents that we think answer those questions, and you tell us if they do or not.

So they weren't thinking about AI. They were just thinking, oh, I needed that information. I wasn't getting it from search terms or from document review. Now I'm getting it.

Like, so now now they're big advocates of what we're doing. So in some ways, if people are not inclined to get into the technology, have them not worry about it. Like, they want to solve problems. And going to your other point about people being really busy, that's true, and there's deadlines.

But lawyers also learn from the first day of law school, if you use a citation and back then you had to shepardize it or look it up in a book and make sure it's really still valid. They learned that from those days that if you don't do that, you can quickly get into trouble and you can have you know, potentially get sanctioned and lose a case and have your license taken away in extreme cases. I think lawyers understand fear, and they understand risk. Are there gonna be people who just, you know, ride the wave and don't care?

Like, sure. That's always gonna happen. And I think those in the same way it happened in eDiscovery where there was a couple of really famous cases where people got sanctioned millions and millions of dollars and their names were all over the place for the wrong reasons. I thought that was a pretty good motivator to get people to try to do things in a defensible process oriented way.

So I think, you know, for better or worse, we have a lot of good examples that we can use.

I think you're you're right in the sense that there's new releases and new things coming out. I always have try to have a really good hype and PR filter because I think it's you know, there's a lot of, you know, mind boggling money being made and at stake in this whole AI innovation wave and, you know, some really huge powerful companies of scope and size we probably weren't dealing with before, and the amount at stake between them is enormous. You know, they're the size of countries. So I also try to filter back whatever I'm being told and deal with actually getting, you know, my hands dirty in the applications and seeing what they can do.

And I'm often, you know, impressed and amazed by the things that they can do, but I personally don't feel like we're, like, rushing towards artificial superintelligence or even artificial general intelligence. So and I'm often impressed, and I'm often like, wow. That's that's really not a very good result, or that completely misses the point. So I maintain a level of skepticism, but it's exciting to be part of it and try things.

We're doing something which is incredibly sophisticated pattern matching and statistical analysis to take huge datasets and replicate things that sound like what people would say. I'm very much mindful, I've written about it, that that's not what I don't think what intelligence is to us. There might be some aspects of pattern matching in what we do as humans or or animals even, but that's not everything we do, and I have yet to see any model that really gets to that level because I think it's a much more complicated problem than just pattern matching. Not to say you can't do amazing things, but there's a lot of stuff we do when it comes to judgment and creativity and spontaneity that we can kind of computers can kind of imitate it, but they aren't there yet.

Great. Thanks a lot. Really great talking to you guys. Hope to do it again.