Josh and Aaron discuss the introduction of AI into the public space and the implications it has for the legal field.
Listen here or read the transcript below. FVF’s Summary Judgment podcast is available wherever you listen to podcasts including Apple Podcasts, Spotify, iHeart Radio, and more.
0:00:18.2 Aaron Von Flatern: Hey Josh.
0:00:19.1 Josh Fogelman: Hey Aaron. I heard about this super cool program this weekend that I’m gonna check out.
0:00:23.8 AF: Oh, what’s it called?
0:00:25.0 JF: It’s called Skynet.
0:00:27.6 AF: What is Skynet?
0:00:28.9 JF: Well, apparently we’re gonna have armies of artificially intelligent soldiers that are gonna be taking over the world and ending humanity.
0:00:45.5 AF: Oh. Yeah. How is that relevant to our business?
0:00:48.5 JF: Oh, we’ll stop existing.
0:00:50.6 AF: Okay. Got it. Yeah. Got it. Got it.
0:00:52.7 JF: Yeah, we should make a movie about it and call it the Terminator.
0:01:01.2 AF: Hey Josh.
0:01:01.3 JF: Hey Aaron.
0:01:02.1 AF: Do you like ChatGPT?
0:01:06.5 JF: I think it’s incredible.
0:01:08.6 AF: I’m in a relationship with ChatGPT myself. I think it likes me.
0:01:14.4 JF: That’s perfect.
0:01:14.8 AF: And we talk a lot.
0:01:14.9 JF: That’s Great.
0:01:15.8 AF: Do you know what it believes our tagline should be for FVF law?
0:01:19.2 JF: I’d love to know.
0:01:19.9 AF: FVF Law, future Forward Legal Care which…
0:01:26.4 JF: It’s actually pretty good.
0:01:26.5 AF: It’s not bad.
0:01:27.7 JF: It’s pretty good.
0:01:28.7 AF: I like that it called it legal care which really speaks to me. I’m like, that’s what we get. It’s almost like… It sounds almost medical.
0:01:35.1 JF: It does make sense.
0:01:38.2 AF: But, yeah, GPT is a lot smarter than both of us put together. Except it doesn’t actually do any thinking, is what I understand about this natural language processing AI world that we’re in.
0:01:48.7 JF: It’s an interesting time to be alive right now with what’s going on in the artificial intelligence world. Are you a Bard fan or a ChatGPT fan?
0:02:05.1 AF: Great question. I love them all. I’ve been using Bard, ChatGPT, I’ve used claude.ai. If you haven’t heard of that, that one’s known as a constitutional AI, meaning that it has like, not like the United States Constitution. It’s like it has an internal constitution, which is, because the Anthropic Corporation that made it is trying to make it more like ethical, so that when you ask it a question, it doesn’t just give you internet garbage regurgitated. ’cause the internet, unfortunately, is like a floating cesspool.
0:02:41.3 JF: Speaking of ethics, does ChatGPT know that you’re cheating on it with both Bard and Claude?
0:02:51.3 AF: I’m sure it does. Yeah. Yeah. Luckily.
0:02:54.9 JF: Okay. All knowing.
0:02:55.2 AF: It’s a robot, but we should probably tell our listeners why we’re talking about it and where we get all this amazing knowledge that we have about AI.
0:03:06.8 JF: Yeah. So tell them.
0:03:08.4 AF: Well, we’re MIT students. No, Josh is taking an AI class through one of the executive… It’s not the real MIT. We shouldn’t like fool our viewers into thinking you’re an MIT scholar, but you go to MIT.
0:03:22.4 JF: We’re learning about artificial intelligence and its implications for business strategy, and it’s been a really interesting road so far for a couple of different reasons. One, I’m well past the age of being a student and having to study and learn new things is fun and exciting and a lot harder than it was when I was 20 years old. Right. But we’ve decided sort of as an organization and as people in this society that it’s a good idea to ride this wave that is artificial intelligence because it seems as though it’s going to fundamentally change a lot of things about a lot of things in a lot of ways and…
0:04:21.1 AF: Exciting times.
0:04:21.6 JF: Very exciting times that we wanna be a part of…
0:04:26.1 AF: Terrifying.
0:04:26.3 JF: Yeah. In some ways. Terrifying.
0:04:30.9 AF: I’m just gonna keep throwing out stuff.
0:04:33.3 JF: But since the beginning of our organization, you and I, I mean when we first started this business, you and I have always been thinking about ways to do things better. And that’s not just from a business point of view. Well, that’s part of it. But from a client interface point of view, from a client education point of view, from an efficiency point of view, from a results point of view, from a team improvement point of view, improving their quality of lives, giving them the best opportunities. It’s an important part of who you and I are. And it’s an important part of the ethos of our business. So that was sort of the impetus why we decided to go down this road and try to develop and improve our understanding of artificial intelligence, how it’s gonna impact our industry in both the short term and the long term, and how we can use it in our business to just make things better.
0:05:33.9 AF: So maybe we should kind of review for the skeptics out there ’cause I mean, I was pretty skeptical when I heard that this thing was doing what it was doing, and I thought, well, it’s probably just like a kind of parlor trick. And to some degree it is. I think we’ve learned that it doesn’t actually think it’s more of a prediction machine based on context. But that being said, we’ve seen some pretty powerful applications already, especially when this technology, the natural language processing, large language model, whatever you wanna call these ChatGPT type things, when these things are married with the machine learning that’s been kind of building for the last 10 years in the tech world, as well as married with software that like CRM software that’s already out there, businesses are starting to become capable of doing some really intense things really quickly, really cheaply, and that has some big implications. So what are some of the things in the legal world that you think could be coming?
0:06:35.5 JF: Yeah, so I think what would be great to talk about two different… When you say the business world, that’s a big sphere. I wanna kind of break that down. I think from our business point of view, there’s I think two things worth talking about here. One is business in terms of client interface, client access to education and information. And then there’s also from an internal workings point of view, the practice of law, and there’s also the management of an organization. We can kind of share on those. But I think from an overarching point of view, what interests me the most is how can these artificial intelligence models be utilized as tools to improve the way that we practice law. Not just from a, Hey, we’ve got a lawsuit that we need to file, rather than having someone in our organization spend hours upon hours, literally drafting the lawsuit document, but having this tool available to them where they can prompt it what the lawsuit is about, and it will generate that petition that they can then review and tweak, saving those hours so that they can focus those hours on client care or client management, anything from that to actually drafting discovery questions and drafting discovery responses. There’s so many opportunities it seems to use these models as an incredible way to be more efficient as a lawyer.
0:08:25.9 AF: Yeah. And I think we should kind of catch people up a little bit because if I’m sitting there listening to this, I’m like, well, I heard a lawyer filed a brief with the court in New York, and it was a huge embarrassment and the guy almost got disbarred. And that honestly was legitimately terrible what this lawyer had done. He begged for mercy of the court but his excuses were kind of even worse ’cause it was like, well, I just assumed this was gonna have good… It was gonna check the citations and all that. I mean, this machine was making up court cases and even making up what they were about. And they, the court somehow allowed this guy to keep his license. But what we should tell people is since that moment, even since then, which was very recent, they have made strides in terms of how to use this thing in a responsible manner.
0:09:27.0 AF: There are different software out there now that can, keep all of the universe of information that you want to talk about with this machine in a HIPAA compliant box and can limit the machine’s ability to “hallucinate” to come up with these facts. It can, you can sort of start to put parameters on it that prevent it from doing that. And I think the biggest thing that turned me, into a believer that this stuff could actually work and help law firms and help clients ultimately is the fact that you can feed it your own work. So instead of it drafting a petition, just sort of guessing what a car crash petition should look like, you feed it one of your own or 10 of your own, or a hundred of your own, it then starts to, it’ll read the police report, it’ll take what the way you write lawsuits based on facts and make it sound like your own voice. And of course you gotta edit it. And it, I don’t think we’re ever gonna get to a point where a lawyer is not supposed to look over every document, but, what does this mean for the client?
0:10:36.4 AF: It’s like, well, does that mean I’m paying for the lawyer to not even lawyer? And I think what the answer is, is it’s freeing up the lawyer’s time for the real work, because that’s busy work to us. Putting a petition together, deciding on the objections that go into discovery or drafting discovery request to a corporation, these are things that we have to do, but they’re not the creative part. I’d rather get 50 discovery requests drafted for me to a trucking company, and then have me come behind that and think of five more that are really interesting and specific that the machine might not have picked up on so that I can drill down into the corporation’s practices and policies. So I think there’s a huge potential there to make legal services faster, more reliable, and ultimately cheaper.
0:11:26.9 JF: Yeah. We’re certainly not at the point now where the AI can be relied upon to do comprehensive legal research and write legal briefs, legal arguments and things of that nature. It’s funny, I had a dinner with a an old colleague of mine, a friend of mine. She was telling me the story about how her law partner, they had a case, she is a criminal lawyer. They had a case where there was a question about whether the police had lawfully searched the car. So her partner goes into ChatGPT, types in the facts scenario and says, give me a case that says this is illegal. And sure enough, ChatGPT spat out the perfect case, Jones v State, Texas Supreme Court case right on point. He was so excited. He was like, I got it. I got the right case. Then he of course went into LexisNexis and did the legal research and typed in that case so that he could read it and make sure that no case had overturned it. Turns out the case didn’t exist. ChatGPT made it up.
0:12:40.6 JF: And that’s of course a big problem and a big limiting factor for the extent to which it can be used in the practice of law and a great reminder of why you still have to have oversight of the work that you’re doing and why you still want a practicing licensed attorney human being doing the legal work itself. But yes, to your point, if you can utilize this tool to do some of the heavy lifting that’s a little bit less analytical, but more kind of hands-on document drafting, creating shells of things for you, that you can then just review and go out, even if you can free up an hour or two of your day, that’s an hour or two of your day that you can then turn and use towards either doing additional legal research or more importantly, working with the client and doing things with the client to help advance their case forward in some meaningful way as efficiently and effectively as you possibly can.
0:13:40.3 AF: Yeah, And I think you had asked me earlier about Bard versus Claude versus ChatGPT, and there’s a ton of other ones out there, by the way. Those are the three that I have the most experience with. But, I wanna point out that Bard is probably the least capable when it comes to say, drafting a document, drafting a story to say, if you said, write a story about palm trees and a tiger, and it would probably do the, it would have the least capabilities in that department. But when you ask it something that where it needs to go out and check the internet, it’s the only one of those three that has a direct pipeline to the regular internet. And so if you asked it, who won the Chiefs game yesterday, it would actually be able to tell you, whereas ChatGPT would say, well, I’m stuck in 2021, all my data and my training data is in 2021, so I don’t know the answer and I can’t ever give that to you. There is still a problem though.
0:14:41.2 AF: And so my point is that we’re only seeing the tip of the iceberg here. Eventually, all these machines will be connected with a live feed to the internet and will have a way of blending its creative abilities with actual current knowledge, as well as all the contextual powers that it has. So it’s… This is more of a conversation about what’s possible, not so much like, Hey, everybody run out and use this machine right now to run your legal case. But in the present moment, it is possible for people to get some answers to some questions using these machines. Like if you had a question like, what should I do if I get in a car accident? What’s the first thing I should do? Or what’s the first 10 things I should do? These machines are pretty good about spitting out some pretty solid advice, and they’re not really coming up with the advice. What they’re doing is reading 10000 blog articles and papers that have been written and synthesizing the most common items from there.
0:15:47.6 AF: So first thing is call the police, make sure you document the scene if you have photos, if you have the ability to take photos, take ’em, get names and witness information. So it’ll give you all that as if it was a lawyer. Why is that important? Because from our perspective, we get to start counseling clients as soon as they call us, but sometimes people take a year before they call us. And so it’s nice to think that they might actually be getting some decent guidance along the way.
0:16:14.1 JF: Hoping that of course, that guidance is true. And I think there is still of course some question about the accuracy of the information that you’re getting. You and I have also talked quite a bit about how is artificial intelligence going to potentially change the game in how people go about finding the attorney that they wanna hire. What are your thoughts on that?
0:16:44.7 AF: Yeah, I’m encouraged. I would say in the near term, I’m terrified like everyone else that all of our marketing strategies might not be as relevant as they were before this stuff came along. But in the long term, what I see is a system that’s capable of understanding at a deeper level than Google currently understands us, understanding us, right? Like, I think it’s likely, I mean, this is a self-serving bias here, but I think we’re a lot more substance than the average law firm that does what we do. And I think Google, when it goes to find us, when a customer, a potential new client is searching for us, they’re just sort of comparing surface items. And that goes back to a prior podcast where we talked about the numbers game, putting big numbers on your websites to get people to call you.
0:17:32.5 AF: You kind of almost have to do that because the customer will never see you if Google doesn’t see you. And Google is only looking for like these certain categories. These new artificial intelligence models are capable of a much deeper dive into who you are, what results you get, why people like you and why you’ve earned the five stars that Google says you have. They’re going to weed out fake reviews. There’s a lot of coming where if you are a superficial organization, I think you’re gonna crumble in that environment. But if you have, like you say about a house that got good bones, we’ve got good bones. And so when we get reviewed by these supercomputers, I think they’re gonna find that there’s a lot of substance there and there’s a lot of reason to direct clients to us.
0:18:27.6 JF: Yeah. Behind closed doors when we were talking about our marketing, all of our reviews, five star reviews, those are earned, those are real. We could point to the human being that we served, that wrote each and every single one of those.
0:18:48.8 AF: And they’re heartfelt.
0:18:50.3 JF: And they’re heartfelt. And that’s a lot of years of blood, sweat, and tears from our organization to get to that point, to achieve that success. And that’s our mission. That was what we set out to do from the very beginning. And it’s amazing to see it have come to fruition almost 10 years later now. So behind closed doors, when we’re talking about marketing strategies, it’s so frustrating when you’re looking at who your competition is in the community, and you know those folks that are out there paying for Google reviews and it makes you want to pound your head against the wall because you know that, that success is bought and paid for. And one of the things that we have talked about is, well, right now it doesn’t seem like Google cares. It seems like in some ways they actually reward that type of behavior, which is, I mean, if you’re paying a non-client to write a review for services they never received in order to induce somebody to hire you, that’s bad.
0:19:57.8 AF: Real bad. Yeah.
0:20:00.8 JF: It’s bad. Yeah. And Google…
0:20:02.8 AF: Bad behavior on a lawyer’s part in particular.
0:20:03.7 JF: It’s bad. And I mean, it’s beyond unethical. It might be illegal, I don’t know. But our hope and thought is, ’cause right now Google doesn’t seem to be differentiating between real reviews and reviews that it ought to know are fake. And one of the hopes that we have, kind of going back to what I was talking about earlier, how can this benefit the client or the potential client in the future? Well, when they’re out there searching for a lawyer, we’re hoping that to your point, these artificial intelligence processors will have the ability to differentiate what’s good from what’s bad, what’s genuine from what’s fake, and use that as part of its synthesis of information to make a recommendation or point a person to to a website, to help answer their question or link them with the service provider that might be best suited for their line of work. So I mean, if you’ve got a website that’s willing to pay for fake reviews, how high can the quality of their content really be? It calls into question everything about it, and that’s terrible for the consumer.
0:21:25.0 AF: Well, and the same goes for firms that may have a really easy way to give someone five star reviews, but it’s kind of contrived, like maybe they’re giving it in the middle of the case, maybe there’s an incentive behind that, that’s especially large. Who knows how these reviews are being generated? But, and we’re not pointing anyone out specifically, but we’ve definitely seen instances where someone has jumped by more than 500 reviews in a week, and it just seems like, man, that probably isn’t real. And Google’s not picking up on it. So this isn’t a complaint so much as a way to say, here’s something good that’s happening. There is gonna be better grading of reviews by these machines. They’re gonna be able to identify what’s real. They’re gonna be able to quickly cross reference who a person is. Is this a real person or… What’s their Facebook account? And kind of figure out everything about them online to make sure that they’re real, that they’re actually in the same geographic area as the business that’s being reviewed. That’s a big one. So we’ll see.
0:22:30.3 AF: I think from the client’s perspective, they’ve got a lot that they can learn from these machines right now. I would just caution, that you should please call us. We have gone to great lengths to make sure that everyone who calls us is in a zero pressure environment. If you call us, there’s no danger. We’re gonna grab you and make you sign a contract. We are adamant that people get the information they need, get educated and make informed decisions. And if we’re involved and they wanna hire us, great. If not, we just are glad to be of service. So, if you’re out there using these services to get some advice, I would just say make sure you call us, FVF Law.