Episode transcript
AI Lawyers, Judges & Juries: Can AI Advocates Deliver Real Justice?
Full transcript of the published episode, with timestamps. It was generated automatically, so expect the occasional misheard word or speaker label.
0:00Cold open: AI and access to justice
David Crump
0:00
Can individuals use AI to help them develop their case who are not lawyers? Absolutely. You know, there are there are more dangerous enterprises out there trying to take away access to uh the law and justice for people.
Unidentified speaker
0:21
Okay.
David Crump
0:22
And I don't think those are machines.
Unidentified speaker
0:24
That's right.
David Crump
0:25
You know, consider this. Um many small law firms now with the power of AI can compete with much larger law firms, right? Because we're sort of all having a resource now. Um that uh let's face it, if if if if a law firm with 50 lawyers has the same AI as one with two, they've got the same access to resources. They may not have the same amount of money, but they've got the same effective smarts.
Viet Nguyen
1:11
Hi, I'm Vienna.
Natalie Chantal Luke
1:13
And I'm Natalie, and we're the hosts of Volume Control and my podcast. We're here today with our guest speaker, David Crump.
Viet Nguyen
1:23
Hi, Natalie. Hi Deal. Hi, David. So before we get into today's podcast, and thanks for joining us uh uh here today, um, can you tell us a little bit about uh uh your you know your line of work and you know what you what you do uh uh with in your practice? Um I understand you are an attorney and uh tell us a little bit about some of the cases that you handle.
1:41Meet David Crump
David Crump
1:46
Well, I've been practicing law for 42 years, and uh I've seen a lot. Uh primarily I'm a plaintiff's personal injury attorney. Often that requires uh trial practice. Uh what I specialize in doing is representing people who've been injured because of negligence on the part of individuals or entities, corporations, governments, businesses, and trying to uh set the tables uh back to status quo.
Natalie Chantal Luke
2:17
I hear that you've recently become the dean of the Keenan Trial Institute. Congratulations.
David Crump
2:21
Thank you.
Natalie Chantal Luke
2:22
At the Keenan Trial Institute, you're a part of its AI group, am I correct?
David Crump
2:26
Uh I have presented to the AI group, and I do attend uh the monthly presentations that they have.
Natalie Chantal Luke
2:32
What are lawyers most concerned about when it comes to AI and what are they actually using it for?
David Crump
2:38
Well, I think my experience and my observation of other lawyers is they're most concerned about uh it helping them do their work more easily and more accurately and ultimately to the benefit of the client. Uh now there are a lot of different ways that my lawyer colleagues are using AI. They're using it to summarize documents, voluminous documents. They're using it to create scripts for direct and cross-examinations that we would have in trial or deposition, deposition outlines. Uh that's the nuts and bolts. But many of them are using it uh the cowork function, for example, to do a lot of their administrative tasks or their marketing tasks, tasks, advertising, um, engagement uh with their emails and letters and such. So I haven't gotten to that point yet, but uh it's astounding, really. There's so much.
Viet Nguyen
3:39
So um actually, what so what what does the the the institute uh actually do? Are you are you training lawyers or or what Yeah?
3:42Inside the Keenan Trial Institute
David Crump
3:46
The uh the Keenan Trial Institute's mission is to train trial lawyers to be the best trial lawyers for the people. And we uh have uh a rather large curriculum of classes that focus on every element of a case, a case preparation and trial, um, client development, preparation of clients. Uh and that curriculum is developed, enhanced, refined, and presented uh throughout the year, some some colleges or courses many times, um, and we uh essentially train new lawyers and and some not so new lawyers to be the best they can be.
Viet Nguyen
4:33
Okay. All right. So um, you know, in this institute then, and you've uh and as the dean, you you probably have uh a certain role in being able to pace the adoption of AI, right, for uh through the institute. Um so would you consider yourself um, you know, where are you personally with AI right now? Are you a are you an early adopter, a cautious user, skeptic?
4:47An Early Adopter in the Legal Profession
David Crump
4:54
No, I'm I'm definitely an early adopter. There's no question about it. Um, as I uh I think people know I I've said this. I wish I had adopted it earlier. Um, although I'm I'm heartened to know that I'm actually pretty early in this game uh for for lawyers anyway. Um and I just wanted to say one other thing about the Keenan Trial Institute. We have uh over 6,000 people that have attended uh Keenan College and colleges. So uh it's it's rather robust. And and I'd like to, as the dean, my job is to keep keep it growing, keep students coming, keep the uh curriculum in uh enhanced and the teaching teaching teaching good.
Natalie Chantal Luke
5:38
Awesome. So you've been doing a lot of work with AI. What can a lawyer safely trust AI to do and what things actually really need a human still?
David Crump
5:49
Well, I don't think a lawyer can really safely trust AI to do anything without uh robust oversight. Which when you think about it is no different than uh a paralegal, uh legal assistant, associate attorney. You're always going to have to be the person where the buck stops if you're using AI. And so uh always, always continue to ensure that your product is as accurate as it possibly can be, as you can make it.
Natalie Chantal Luke
6:23
Yeah, I'm sure people are kind of worried about whether you're using AI to do their case, and if so, why do they need you if AI is doing it? Where does the human and AI intersection occur?
David Crump
6:38
Well, that's not uh something I've uh encountered personally yet, but it's coming. I have seen potential clients send me case summaries that were clearly prepared by AI. In some ways you like that because it's easier to read and review and get an understanding about the case rather than have to interview the individual. But uh for the most part, I think our clients know we're using AI. In fact, my fee agreement says I'm going to be using AI. And uh that may require documents of yours to be on a platform like ChatGPT or uh Claude, and there are security and privacy concerns, and they give me a privacy waiver allowing me to do that for the purpose uh that it will be more helpful to their case.
6:42How Lawyers Are Using AI Today
Viet Nguyen
7:33
So, I mean, given the legal field, um there are obviously some additional requirements, uh, probably more so than you know than than than many other industries in regards to how uh AI is is utilized, you know, so it's proper use and what's improper use. And so, you know, at what point do you think um AI stops being about you know uh sort of innovating, you know, its use within within the legal space? And when when does it sort of cross over into professional negligence, right? Like like when is when is it just uh uh an issue?
David Crump
8:11
Well, I think if you rely on it and it's wrong and it is to the detriment of your client, then that's probably legal negligence. Or at the very least, the judge will sanction you if you've put a product in front of the judge that has an incorrect case site, or in these frightful cases that I've I I've heard about, but thankfully have not experienced, making up cases out of you know, whole cloth.
8:19When AI Mistakes Become Legal Negligence
Unidentified speaker
8:35
Yeah.
David Crump
8:35
Uh and it's easy because uh it's easy to let that slide because uh otherwise you have to read every case and make sure that every pin site is accurate and that takes time.
Viet Nguyen
8:47
Yeah, yeah. And and obviously uh some of those older models, uh some of the legacy models have uh you know had higher hallucination rates and things.
David Crump
8:56
Yeah, and personally, I only see this, these products getting better to the point where those problems will be eliminated for all intents and purposes.
Natalie Chantal Luke
9:10
Let's go back for a second. You said that you had a client or clients that have used AI to present to you their thoughts on their cases. What happens when they go in and they say, XYZ has happened to me? And then, say, ChatGPT or an LLM tells them, you absolutely have a case. This case is worth millions of dollars. And suddenly you're competing against the ideas of AI and what you really think a case is worth or where it should go.
David Crump
9:45
Well, that requires a conversation. Now, for the most part, I've only had potential clients reach out to me who have presented me with AI summaries of their potential case. Uh I didn't see that there was a case that fit my uh case matrix, so I referred them to someone else. Um but it's it's a fair question and a fair area for them to say, I've done this work and this is what I think, but ultimately my years of experience allow me to have some judgment that they don't and AI doesn't yet.
Viet Nguyen
10:21
And so when they when they come, does that end up being confrontational sometimes, or is it managing expectations with the potential client? Uh how how do you sort of uh address that, right? I mean, they they may be coming in there with a stack of paper or you know, a printout or something like that saying, oh, hey, this is what ChatGPT told me. PowerPoint, yeah, exactly.
David Crump
10:45
Yeah. Um it is always a case of managing expectations with clients. If there's a confrontational aspect to it, it's a case I wouldn't take. Uh, and if there's a confrontational aspect mid-case with a client, I would tread very, very carefully to make sure that uh everybody moves on from that. Um without uh there being a lot of heartburn on both sides. And that could mean saying, I can't represent you any longer because uh you see things differently than I do. And ultimately I have to be the captain of the ship.
11:20Clients Are Already Using AI
Unidentified speaker
11:22
Right.
David Crump
11:23
I mean, to be fair, the client ultim ultimately will make a decision whether or not to accept a settlement, whether or not to go to trial. But when it comes to tactical decisions, uh yes, there may be a conversation with the client, but uh I prefer that I be the one who makes those decisions.
Viet Nguyen
11:40
Right. It's your name, it's your name on those court documents and and and you are as much a party to everybody has a different uh aptitude as well.
David Crump
11:53
So some clients may present their case and be more rational and realistic and and accurate. Uh has a client ever told me I'm wrong? Probably. And if I've been wrong, that's probably happened. Yeah. Yeah. We're none of us perfect.
Natalie Chantal Luke
12:11
Should a client, though, if they think something has happened to them, let's say I got into a crash or something happened, should I go online and type in the details as far as I know and ask AI whether or not I have a case? Is that a good first step?
David Crump
12:29
Well, I think it is. I think anybody who should have some agency over their own life, their own choices. I think I'm most pleased if people will be most concerned about getting well. I want their first thought to be, what can I do to heal from whatever this injury is, whether it's physically or emotionally or whatever? Uh I prefer my clients putting all of their energies there. Um, however, I think it's entirely fair, particularly in areas of that are complicated, medical malpractice, for example, where somebody says, I would like to know what is the standard of care for this particular procedure and what it was that happened to me. Was the standard of care followed or met by the physicians or the hospital or or was it not? Now, AI will not be able to give an accurate, 100% accurate answer, but it may give the person some guidance that they can rely on and then say, you know, I think I'll go talk to an attorney who's experienced in that field.
Viet Nguyen
13:34
That's uh that's that's real interesting because I I I would imagine that when you're talking to these uh these clients of yours, they many of them are in emotional states of distress, crisis, you know, physical states of distress, and it becomes very difficult to articulate um what it is that they're needing um in in many cases. And you know, I I was you know, I think back last year when I when I had a kitchen remodel that I was uh going through and you know working with the insurance company, even that, even that uh emotional distress, you know, working you know with the insurance company, yeah. Um AI was actually really helpful in that in that instance because it it got a chance to remove me from that you know that emotional state and actually allow my uh me to articulate better uh what I needed uh in an email.
13:44Should You Ask AI Before Calling a Lawyer?
David Crump
14:25
Right. It acted as attorney intelligence.
Natalie Chantal Luke
14:31
Attorney intelligence is quite clever.
Viet Nguyen
14:33
Yeah, yeah. So so um, you know, AI can can already do uh a lot of work traditionally, right? I mean, we're we're in this age of agentic AI, uh, which is different than where we were with just generative AI, you know, over the last couple of years. And so it now it's it's got the ability to do work um that maybe some of your, you know, some of the junior lawyers, maybe some of the paralegals on your team uh might uh have done in the past or might, you know, might be doing today. Um, you know, things like research and summaries and drafting and document review and things like that. Um, are we, I mean, just square up with me. Are we at risk of automating away uh the training ground here for the more seasoned lawyers?
David Crump
15:18
I think so. I see this particular field being uh at risk in terms of the human element, the the individual people who have been doing that work all along. I think that once we get AI to be you know nearly accurate all the time, no hallucinations, accurate citations, uh accurate uh research, uh it's going to be very difficult to have a human come back and do that kind of work for for so many reasons. Um you've got the instant instantaneous access of that AI engine. You've got it working 24 hours a day for you. Uh it doesn't uh ask for days off, you know, it doesn't ask for um leave, it uh it doesn't have political issues with other people in the office. Um and it's certainly less expensive.
Natalie Chantal Luke
16:18
So are you saying that just the middle's gonna be gone of legal things like paralegals, junior attorneys? Or what are we talking about here? Is the entire industry going to change?
David Crump
16:31
Well, it's certainly, in my opinion, going to become much more efficient than it ever has.
16:35Will AI Replace Legal Jobs?
Natalie Chantal Luke
16:36
I think that's one of the- You kind of dodged the question.
David Crump
16:39
Did I? Yeah. I I to answer your question, yes, I think that it's going to uh at both um the I think at all levels, quite honestly, uh at uh the paralegal and and uh law clerk level, at the mid-partner and mid-associate level, and at the senior level. I really do.
Natalie Chantal Luke
17:01
Yeah, I've seen some friends post on LinkedIn that they actually used AI to do things at work that they would have normally had their internal legal department do. And they said they got a lot quicker results, they were able to do it themselves. Doesn't that put us at risk, though? A lot of these cases are really complicated, and there's no such thing as a perfect case, from what I understand. Right.
David Crump
17:27
Well, yes, ultimately, there's an arbiter, there's a judge or a jury, what we call the finder of fact. Ultimately, there is a human who is going to have to make decisions, trial court, the jury, appellate court, supreme court, and there are going to be presentations by human beings to those finders of fact. So we're still going to have that until someday comes when people agree to allow AIs to present cases to AIs.
Natalie Chantal Luke
18:02
Oh my gosh. So there are many AI plaintiffs.
David Crump
18:07
Right.
Natalie Chantal Luke
18:07
Well, I think the way I see it attorney intelligence and a defendant's attorney intelligence, and they're going to be in a Zoom court or virtual court?
David Crump
18:20
Virtual court. I think of it this way.
Natalie Chantal Luke
18:22
Uh will they have an avatar?
David Crump
18:24
They could. That certainly. They could have their lawyer be an avatar, the judge be an avatar, but here's how I see it. Uh, what you started talking about was can individuals use AI to help them develop their case who are not lawyers? Absolutely. If they can do that, there's certainly other people out there smart enough to know that they can collect those individuals on both sides, have them sign a contract to resolve the case by some AI arbiter finder effect.
Natalie Chantal Luke
18:58
Aren't you supposed to have a jury of your peers, though? People who are in your community that help to establish whether justice is being served. How can I know that justice is being served when the community is a virtual community that's based in an entire system that's the world?
David Crump
19:17
Well, many cases don't have juries. For example, divorce cases don't have juries. Many arbitration cases are tried to a single arbitrator or a panel of three arbitrators. Uh, those arbitrators, being human, are loaded with biases. So someday I believe that an AI engine could be developed with minimal to no bias.
Viet Nguyen
19:46
So how do you how do you then address the issue that I mean, who do you blame, right? If something goes wrong during that arbitration process, you know, when there's when there's a human arbitrator uh in the middle there, um, if a you know, if that judgment comes down, you know, and you don't and you don't agree with it, and obviously you have a contract that's already signed, but there's still there's still what we refer to as a throat to choke. Right. It's somebody that you can complain to, somebody that that that uh has the um the authority uh to be able to um uh to respond uh to to that. How do you how do you adjust for that if it's just AI-based?
David Crump
20:34
I don't know. I don't know. I don't think we're far enough down the road on that yet. But let me tell a little story. Uh friends of mine recently uh well have spent the last two years developing a rather significant medical malpractice case in North Carolina. The client was a little girl uh who got um a disease that ultimately left her paralyzed as a as essentially as an infant. Um this case received no offer of settlement from the defense attorneys. They went to trial. They spent two weeks in North Carolina just picking a jury. Now in Washington State, we get about an hour apiece. Uh two weeks of time, right off the all the lawyers' resources doing that. Then they went into the evidence portion of the case, starting in the third week. They got into openings, objections were flying from one party and the other party, the judge was becoming overwhelmed, didn't know medical malpractice law. After the first expert came on, there were more and more objections. The judge finally said, you know, I don't understand medical malpractice law. I don't think I can try this case. This was after tens, if not hundreds, of thousand dollars had been spent. It's so inefficient that how could there be anything but A system like this for the average individual's claim, small auto cases or uh you know, products cases or contracts cases.
21:20Can AI Advocates Deliver Real Justice?
Natalie Chantal Luke
22:11
That that sounds all very logical and wonderful, but I don't think that's actually very feasible because of the following thing. AI, especially the way it is right now, it tries to give you an answer based on a statistical curve. So it tries to find the middle. What's the middle? That's where we're starting to see everything be the same.
David Crump
22:33
That could be fundamental fairness without bias.
Natalie Chantal Luke
22:38
How could that be fair? You need somebody to go to bat for you.
David Crump
22:42
Well, you need an advocate. And so your AI advocate can go to bat for you, and their AI advocate can go to bat for you, and then you aggregate those into a finder of the then program some kind of combat AI. Somebody would much smarter than me would do that. Yeah.
Natalie Chantal Luke
22:59
I see. So there would be really a difference in the programming for, say, a plaintiff's attorney intelligence or AI. Sure. And a defense. But but the problem with that is right now, am I incorrect about this? But the defense has the deeper pockets.
David Crump
23:20
They have deeper pockets, that doesn't always mean that they're right.
Natalie Chantal Luke
23:28
Sure. But if you're trying to go for justice, isn't that the problem?
David Crump
23:32
Well, that's always going to be a problem.
Natalie Chantal Luke
23:35
So then if people are trying to do the same and the two AIs are trying to get to the same point, that seems like an impossibility.
David Crump
23:46
Well, there's usually in the area of an injury that's preventable, some element of negligence there that the law um that the that the law has rules about. Okay. We have to prevent those harms that are foreseeable. Okay. If somebody doesn't take the reasonable measures to protect the harms that are foreseeable, then that's negligence if they owe a duty. Um so I'm not sure I understand quite what the concern is.
24:21AI Attorneys, Judges, and Virtual Courtrooms
Natalie Chantal Luke
24:25
Well, I'm worried most about spin. Like I understand what you're saying.
David Crump
24:30
Or is that good advocacy?
Natalie Chantal Luke
24:31
I understand what you're saying, possibly. But what I'm concerned about is if one side spins it from their direction and the other side spins it from their direction, you won't actually ever have any resolution.
David Crump
24:48
Well, I think what you're saying is that if we have perfect spin on each side, that the ultimate decision maker would freeze and not be able to make a decision. Okay. Well, I don't think so because I think there are elements of you know fundamental right and wrong. And what is the definition of negligence? And I don't believe that you can argue everything away.
Natalie Chantal Luke
25:08
But you think that AI is going to determine what's right and wrong. So I think this is a wrong.
Viet Nguyen
25:16
I think this is a really interesting conversation because there's actually a precedence for this um uh in machine learning. Um uh there's this concept of uh of adversarial uh AI. And um what happens in in machine learning is you actually have one algorithm that tries to move along a path, right, as it's kind of trying to discover the answer. And then you actually match it up against what's a a discriminator um AI that actually tries to oppose it. And it's actually the uh where those two land up uh that sort of provides the the result at the end. So it's really fascinating though that you bring that up because you know, in in in in the world of of AI, something like that does already exist, right? It is a fundamental um uh learning uh uh capability. So it's really interesting.
David Crump
26:03
And you bring up the point that I hadn't thought about uh during this conversation, which is it doesn't have to be a centralized decision maker, it can be a panel of of different ones. You could have ChatGPT, you could have Claude, you could have Google. Uh a council. Yes, a council, as they call it.
Viet Nguyen
26:20
Yeah, model, yeah, actually model council, right? Yeah. So you actually have uh you know different different models looking at uh the same problem, and and maybe this one model over here is gonna hallucinate, and this one is gonna hallucinate, but it's gonna hallucinate in a different way, right? And then and then maybe you got a third one that comes in and uh and where they disagree or where they agree um tells you a lot about um not only their biases, but also where the weaknesses and the strengths are from what you with what you end up with.
David Crump
26:51
Right. And as long as the people who have the dispute agree to this, and they're going to agree to it because it's so much less expensive than uh a court trial might be, the evidence will be heard fractionally within within minutes or hours. A result will be rendered quickly. It will be uh hopefully not biased by the country town judge who's in favor of the people that he sees at the lunch counter all the time, or the the lawyer, the judge who was hired by insurance companies his whole life and doesn't like plaintiff lawyers. So I think there's ways that they'll get around those biases.
Natalie Chantal Luke
27:32
It does smack a lot of minority report, though. It has a very scary overtone because I understand what you're saying, but it very much feels like justice would be by the bots, for the bots, of the bots.
David Crump
27:51
And please don't misunderstand me. I'm not advocating for this. Okay. I'm not, I don't have a business model that's creating this out there. But think about this every day on television, on those channels that are higher numbers, somewhere you're going to see Texas Justice, Maine Justice, um, Judge Judy. Those are all cases where people have agreed to submit their dispute to a TV judge. Uh, and do you think that TV judge uh doesn't have biases? Uh, would treat everybody equally, is going to be as learned or fair and even-handed as are you telling me that Judge Judy is not infallible? That's what I'm saying.
Natalie Chantal Luke
28:35
Or it hasn't been prescription.
Viet Nguyen
28:37
Sorry, Judge Judy.
David Crump
28:38
My apologies to Texas Justice because I'm a good friend of Larry Joe Dorby.
Natalie Chantal Luke
28:43
But the the whole thing about that is I think only certain cases will fall under this umbrella. The type of cases where, for example, you get involved, these are specific cases that have, they're not like something happened, like I ran into Viet's car on the way to the studio, and we need to resolve that. I didn't do that.
David Crump
29:06
There are lots of there are there are lots of questions of fact, and questions of fact generally need to be decided by a jury or a judge. Questions of law are always decided by a judge. Uh, I think an AI could determine all of those, given enough information to be as equally confident as a human being can be.
29:23The Future of AI Decision-Making
Viet Nguyen
29:33
Well, certainly AI is very confident, right? You know, it's often confidently wrong. Just like me, yeah. Well, at least that's what I'm told.
Natalie Chantal Luke
29:41
It would be very strange to have a human jury trying to interpret what an AI, judge, uh, plaintiff's attorney, defense attorney were saying, and then you would have a human jury in between there. Um, we know already from Mythos that manipulation is something that would happen. That's something that becomes very scary then.
David Crump
30:09
All valid points.
Viet Nguyen
30:14
Okay, so so in your world then, you're you're you're seeing you're seeing this situation where um caseloads could probably increase dramatically, right? Because they would be resolved quicker, more efficiency in the in in the in the production and the resolution of these cases, caseloads going up, and then the efficiency with smaller firms. Right, with smaller firms. And um the benefit then, who wins and who loses then in in that very efficient world, right? Do you how do you see where do you see the benefit sort of falling?
David Crump
30:49
Well, from a business standpoint, the efficiencies would would tend to favor both the business and the client, because presumably the client is getting a better product product at a better price. Um we're talking about justice, which is a very, very difficult thing to measure here. We're not talking about you know manufacturing. Um but who wins and loses, I think let me say this the more accurate information is out there, whether it be uh medical papers and journals or information about uh people, what their prior, what their past testimonies have been, whether it be about behaviors of entities and corporations, uh statistics about past jury verdicts, the more the information is, I think the better for all of us in this business. We are an information business. We're also a people business, but uh it is the packaging of the information that that matters. And so the right best information should win.
31:20The Future of the Legal Profession
Viet Nguyen
31:56
Yeah. So so the firm benefits, the client benefits, probably the courts in some way, you know, less uh uh, you know, will will benefit. Absolutely, they will. Paralegals now.
David Crump
32:08
No, and uh and the courts are using it too.
Natalie Chantal Luke
32:11
Well, that's so interesting because we started out by talking about the fact that you were the dean of the Keenan Trial Institute. If you can't become a junior law associate or a paralegal, what is the point of having an institute?
David Crump
32:30
Well, that's an excellent question. That only time will tell, and uh millions and millions of steps forward by every individual in this business, we don't know. We don't know.
Unidentified speaker
32:44
And I think that's fair to say for all industries medicine, law, accounting, design, uh Hollywood, you name it.
Viet Nguyen
32:59
I would imagine, I would imagine with the increase in you know being able to handle potential caseloads, right? That there is um, you know, as the as the the the the law firm uh becomes more efficient and and gains like that means that there's more resources than potentially to go around to to to uh you know allow it to be spread elsewhere, right? And so maybe that maybe that role for the paralegal doesn't go away, but maybe it changes into something else.
David Crump
33:25
It's entirely possible if that can evolve. And that is what tends to happen in these innovative cycles, in these disruptive cycles. We we do have displacements, but the innovation continues, people evolve and find new ways to uh be productive and creative. And hopefully, you know, the benefit to society marches on. It's my hope.
Natalie Chantal Luke
33:50
What can a young attorney or a person who's interested in the law do in order to prepare themselves? If I'm 18 years old, let's imagine. And I'm interested in the law, especially because something maybe happened to me in my own personal life where a good decision made a difference in my life. One of my parents was in a car wreck or had a medical malpractice case and it turned out positively, and I want to do something like that for everybody else. How do I pursue that?
David Crump
34:27
Well, uh, if you're bright and people like you and want to hire you, you certainly would want to look ahead and see what the landscape might look like five, ten, twenty, or more years down the road and adapt. Maybe what you're doing now won't be around anymore uh in in in a way that's gonna be meaningful for you. Uh my advice is always for for younger lawyers who tend to be in smaller firms or on their own is keep your caseload small. Let AI uh essentially manufacture time for you. All right. Uh AI has done that for me. It's allowed me to do much uh significant amounts of work in shorter amounts of time, which has freed me to have what I like to call thinking time. And I think the thinking time is priceless for me to consider always what's coming on, what's going to happen tomorrow and the next day, not only for me, but for the world. Uh, my advice is to just keep doing what you're doing, but always be ready, willing, and able to evolve.
Viet Nguyen
35:47
So it could provide opportunities then for um, you know, that time to allow for the attorney to express themselves in different ways, you know, maybe a thought leadership type of, you know, like a blog post or thought leadership, or or maybe even to take on a role of dean of uh of an institute. Right.
David Crump
36:07
Right. And and this is uh this is what I've been reading about AI in general, is that it seems to promise us uh the ability for it to let us have more time for the things that we want to do for ourselves. So uh I think that's a good thing, as long as we can make a a living for ourselves and our families and uh continue to have jobs uh to go to. But you know, consider this um many small law firms now with the power of AI can compete with much larger law firms, right? Because we're sort of all having a resource now that uh let's face it, if if if a law firm with 50 lawyers has the same AI as one with two, they've got the same access to resources. They may not have the same amount of money, but they've got the same effective smarts. Yeah, yeah.
36:13Who Wins When Law Firms Become More Efficient?
Natalie Chantal Luke
37:04
Can you talk about experts for a moment and the role of AI with experts?
David Crump
37:11
Well, it's certainly helpful. You know, experts is a very important part of personal injury law and in fact much of law in general. And experts are important because they uh explain subject matter to judges and juries in ways that we're not qualified to do. Uh experts, however, require, rely on we lawyers, to give them all of the information for them to review. Now, frequently that can be 10, 15, 20,000 pages of documents, which unless you want to pay them ungodly sums of money an hour, and I mean ungodly, um, you can create summaries for them, or you can have your agentic AI um review that body of information and help you determine what's important to provide to the experts. So yeah, uh, as far as experts go, it's no different. They can benefit just like we can.
Natalie Chantal Luke
38:06
Do you think that they will go away? And at some point in time, you will simply ask AI, what would your expert opinion be in this case?
David Crump
38:16
Well, we're doing that now initially to vet our cases. I may have a potential case come to me. It's in an area of medicine I don't know anything about, but I take the facts that I have, I run it through this engine and that engine, and I say, what's the standard of care based on these facts? And I get a very uh easy to read, uh simplified report, because I always say, give it to me in plain English, in terms of fifth grader can understand. Um so before I ever even speak to an expert, I've got what I think comes close to an expert level analysis. Then when I begin to vet individual experts, I can explain to them this is what my research has shown. Do you agree generally or disagree? So that's very helpful and efficient. Yeah.
Viet Nguyen
39:04
Do you do you find then as you go through legal documents? I mean, legal legal ease is can be a really difficult challenge. It's not easy. It's not easy. Huge blocks of text and they're super boring. Super boring. Uh and so uh, you know, as a as an attorney, being able to uh have AI do that first pass of hey, summarize this for me, what are what are the red flags? Um even the cognitive load that and uh that you know you're able to take on a much, much larger amount of data sort of coming in.
David Crump
39:36
Well, I can take on a much larger amount of data. What I've those of us who've been practicing for a long, long time know is that whenever you prepare a case and you go to trial, a very small fraction of all of that data ever matters to the ultimate decision maker. Okay. Uh particularly given that we think many of them are deciding at sort of a an emotional level anyway. Um so I believe that the the AI can distill to the essence what are the data points that I need to really understand and communicate to the jury, and in what uh uh timeline fashion and in what uh PowerPoint type uh visual can I present to them? And so having it do that, and of course I don't ask it to do it once, I ask it to do it many, many times in lots of different ways, um, and always to explain it to me in the simplest possible terms.
39:49AI Expert Witnesses and Case Evaluation
Natalie Chantal Luke
40:40
Aaron Powell You slid underneath there, though, that decisions were being made in trials on an emotional basis.
40:46David's Experiment With 1,200 Virtual Jurors
David Crump
40:48
Well, uh I'm not a social scientist, but I've read enough social scientists and uh including Nobel laureates to know that there is ample evidence that suggests that most human decision-making behavior is done uh at a non-conscious, likely emotional level before the rational mind gets involved.
Natalie Chantal Luke
41:14
Well, so time out for a second. Because you thought or you implied that it could be possible that we have AI attorneys and an AI jury with an AI judge making decisions about what justice is. But you just said that a lot of that is based on emotion.
David Crump
41:36
In human beings, yes. However, I have been I created a model where I asked an engine to uh assemble twelve hundred virtual jurors from my jurisdiction here in Seattle, in King County, based on all of those demographics and factors that go into uh those people who can be a juror in King County. Um I then said break this down into a hundred groups of twelve. Take all of those different personalities and have them debate among themselves. Don't just give me what they're bent about the facts is, have them actually deliberate. Um it was fascinating. Now, is it something that I can rely on? No, but what it does do is it spots issues. It's wonderful at saying this juror didn't like this and that juror did like that. So again, it's a tool. Um I'm just now beginning to open up, you know, the book.
Viet Nguyen
42:45
I I get this mental picture of like Sims jury, and you've got like these like little avatars walking around and little thought bubbles popping up, and they're and they're arguing with each other and maybe well, you know, end up uh getting into a fist fight.
Natalie Chantal Luke
42:57
Well, and it's it's really scary that you're giving AI emotions or you're asking it to be emotional, because that's contrary to what you were saying earlier, which was that it's more unbiased. And so all of a sudden, if you add in the emotion, then it's once again, they they seem to be contradictory.
David Crump
43:20
Well, for the tool that I'm using it for, I want it to assume that it's human and emotional, that it has the same, or that it's biased. Often we say not so much emotions, but it's biased because it's this race or this sex or this job or this political place on the spectrum or this age or whatever. So um now, those people who want to create a AI judge or jury might want it to be not emotional, just purely factual in a mathematic way, let's say. Um, so I'm asking it to uh adopt these to to virtually uh create these personas and have them go at it.
Viet Nguyen
44:09
I want to I want to see your your your Sims jury where it's like modeling like the current mental state of this of this juror where it's like um I was at uh 3 a.m uh you know last night drinking heavily and I have a huge hangover and I'm super grumpy. Right.
Natalie Chantal Luke
44:26
And uh mostly jurors don't want to be there. That's the biggest thing is jurors tend to not want to show up or be there and even if they're called they won't show up. So it seems like you might get a larger segment but I worry I very much still worry about the justice for the individual who is trying to get some restitution for a wrong that's been done to them.
44:48Who Really Threatens Access to Justice?
Unidentified speaker
44:53
Well I don't my personal belief is that AI is not going to be what ultimately takes away the delivery of justice for the little guy and gal.
Viet Nguyen
45:09
I think there are other more uh what's the word I'm looking for uh I don't know what's uh what's chat GPT well uh you know there are there are uh more dangerous uh enterprises out there uh trying to take away access to uh the law and justice for people okay that's fair and I don't think those are machines that's right so when it comes to the legal profession and AI what would you turn the volume up on I would turn the volume up on diving in jumping in uh getting at least two if not three of the most robust engines use them robustly don't be afraid to burn a few credits on their higher level um versions uh and uh be as creative as you possibly can with it and if you run into uh a wall talk to it it may have an answer it's weird like how do you how does that even work like you know uh under a billable hour system you're like you know my agentic AI has been running for 24 straight hours like you know how do I bill this i don't I'm not on a billable hour uh model I'm on a contingent fee model but you're right the billable hour uh model has all kinds of of traps for the unwary there and you're absolutely right I mean you could be a flying on a plane and you're billing for your time while your engine is turning away doing work for you on another case um for the most part double billing is discouraged yeah there's an old joke about that well so so do do we end up going to like token based billing for for uh AI firms I mean or for AI law firms uh for me a token is still the same as money well so now where where do you turn the volume down on then uh you know in this uh you know I mean AI seems to be touching everything right so uh turn it down on on uh over reliance you know it's so easy to get this product and go oh this is really great and uh but if you look at it carefully under a microscope it isn't um so always always double check your sources uh which can be very difficult to do when you're asking it to summarize 12000 pages of medical records for example um but I will say this no human being could do the job as accurately in this in the amount of time and for the cost well david um that is you know fantastic you've you've given us a lot of insight in terms of how this affects the uh your uh your work in in in in you know law uh I'm sure that as this situation evolves with uh with AI in this particular industry uh we're gonna see a lot more uh uh of this of this conversation right it's uh this is we're just at the very beginning of uh of of this revolution so um that's it's it's really exciting and really appreciate you taking the time out today thank you we like that you coined attorney intelligence in this podcast well some people think that's an oxymoron what about what about sim's jury like of course no I think that might be trademark it was it was so good you mentioned it three times well thank you for joining us today for having things a pleasure thank you thanks by take down the hype on AI exerting Volume Control
