Episode 016: Can an AI Judge Train You Better Than the Real One?

SPEAKER_00: Welcome to AI Tools for Practicing Lawyers. Practical, no-nonsense guidance to help attorneys put AI power to work in their practice right now. SPEAKER_02: For generations, lawyers learned advocacy the hard way. You drafted a brief, you argued a motion, and then a judge kind of knocked you down a little bit and told you why you were wrong.

But today, AI is introducing something very new. The ability to test arguments against a stimulated judge even before you file the brief. Is that the future of legal training? Or are we fooling ourselves into thinking AI can replicate judicial reasoning?

Today we're joined by Chris Ryan to talk about bench stimulations, AI judges, and what happens when lawyers can practice in front of a virtual bench. I'm Ron Drescher. I'm Heather Gardner. And I'm really excited to introduce Christopher Ryan.

Chris, how did you find us and why are we here? SPEAKER_03: I found you because I'm a listener and a fan. It's an honor to be here, and I really appreciate the opportunity to be talking to you both. SPEAKER_02: Chris, tell us a little bit about yourself.

I understand you're you're in big law. What firm are you in? And what's a little bit of your legal background? SPEAKER_03: Yeah, so I'm a litigation partner at Taft and started out like a lot of litigators, running between courtrooms, arguing a lot of motions, uh, you know, weekly and and sometimes daily.

And uh ever since COVID, that's all sort of changed, but uh I'm still doing a a ton of litigation, uh, mostly commercial litigation, and I do a lot in the healthcare space. What did COVID change? So it it didn't really change the motions, it just changed how the court was deciding them. So I think during COVID, uh, judges learned that they can rule on a lot on the papers.

And if they want oral argument, a lot of times it happens via Zoom. And um that that's certainly true, but there's just less in in-court hearings. And so what that means if you're a young lawyer just coming out of law school is less time to practice, less reps uh at the podium, so to speak. And for for the more senior attorneys, I don't think they're immune either.

We get less in-person, in-court opportunities to argue. And it seems like the uh instances when the judge grants oral argument in person, it's for the big, the most important motions. And unfortunately, sometimes you get out of practice if you're not uh if you're not doing it on the regular. SPEAKER_02: So this is a little bit of a conundrum to me, which is the idea that there are fewer opportunities for young lawyers to cut their teeth in front of a real life judge.

But then from a client standpoint, why would you ever want a young lawyer cutting their teeth on your matter in front of a judge? How did that work in the old days? SPEAKER_03: Yeah, so in the old days, right, um, young lawyers would be sent on motions where the stakes were low. Um, they might be things like motions to compel or motions to adjourn, and you know, at least where I litigated a lot.

You might be in a court with 50 other lawyers waiting for their case to be called. And while you were waiting, you would see really good advocates present their case in front of a judge. You'd see some maybe things that you shouldn't do in front of a judge, and you learned and you could argue, and the the stakes weren't as high. SPEAKER_02: Just being in a courtroom, I think, had an impact on developing lawyers.

Because then you you're right, you got to see a whole bunch of things. So, what's the solution? How did how did you come up with a way to get lawyers a little better trained? SPEAKER_03: Yeah, so you're right.

I think that the uh the the best training is doing, uh, second best would be watching other people do it. And you can get a little bit of that in Zoom and and on YouTube. But some some colleagues and I had talked over a number of years and some uh people at other firms as well about this kind of reps problem, especially when it comes to young lawyers and training young lawyers. And so I was taking a AI masterclass actually about AI, and during it, I sort of had this idea where I thought, wouldn't it be neat if you could upload your briefs that you're actually going to argue, opposing counsel's brief, and have have an AI model learn from that and act as the judge.

And you could argue out loud in real time against an AI judge, and you know, they might push back, they might ask questions just like a real judge would, and and is a way to get the reps. And it's it's not a perfect one-to-one substitute for being in court and being on your feet, but it's the closest I've been able to come up with. And so I started started coding it and started building the platform. And uh that's kind of how Bench Sim uh was born.

SPEAKER_02: Now the website is benchsimai.com. Is that right? That's right.

So, did you have any coding background before you tried this labor of love? SPEAKER_03: If you asked a layperson, they would say yes. If you asked somebody with real coding background, they would say no, he didn't have any coding background. I my my coding background is about 20 years old in obsolete technology.

So I in that regard, no. But I had enough knowledge of the way that these platforms and systems are built and structured that I was able to use AI coding, vibe coding, essentially, and use that to help build and do the actual hard part of coding. SPEAKER_02: I I think when most lawyers sit down at their desks to try to advance their AI skills, they realize they've got to input a lot more data and other information than they may have thought. Did you have that experience?

SPEAKER_03: Yeah, I did. Here's the thing about when you're vibe coding. I think if you've ever tried it before, it will produce something fairly quick. Refining it, though, is where the time and the effort and the fortitude has to come in.

Uh the refining it and getting it the way you want it and uh making sure that it's tight from a security perspective and and all of that. Uh, but the initial product comes out quicker than you might expect. SPEAKER_01: Heather, have you done some vibe coding? Uh no, I've just started working with Claude Code recently, working on a software project.

SPEAKER_02: But you created a bunch of GPTs once upon a time. So what do you what do you think, Chris? Is creating your own GPTs or your own gems or your own Claude skills more or less the same as vibe coding? SPEAKER_03: Yeah, I mean, I think you can sort of use the idea of vibe coding to create skills, right?

If you go in and you tell Claude, I want you to create me a skill and maybe create me a markdown file uh for that skill that does X, Y, and Z, it's really the same process that you think about with vibe coding. I mean, it may not be coding in Java or it may not be coding in a coding language that you might think of, but it's the same process, right? It's building something based on your input. SPEAKER_02: I took this turn down that digression road because vibe coding's become a hot phrase.

And I think a lot of people really don't have an idea of what it really is. And to me, all it really is is working with an AI tool to accomplish an output that you're probably going to want to use again for some other purpose down the road. Does that sound right to you, or do you do you think you'd add or change something? SPEAKER_03: No, I think that that's right.

And I think that that's a good way to think about generative AI in general is you're you're talking to it like you would the person who would otherwise solve the problem. So if you're talking about vibe coding, if you were talking to a software developer and you had this idea in your mind for a piece of software, what would you say to them? You'd say, well, here's how I want it to look, and here's how I want it to function, and here's how I want the data to be stored. And that's what you're doing when you're vibe coding only.

You're telling Claude Code or Codex or whatever the platform is that you happen to be using to vibecode. And you're talking to it just like you would a software engineer. And the legal AI is no different. If I'm prompting legal AI, I'm gonna talk to it like I would maybe a paralegal or an associate or whatever function that generative AI product is doing.

I'm gonna give it context, I'm gonna tell it what I want, I'm gonna try and be specific. But really, it's a good way to approach generative AI in general, is just talk to it like you would the person who would otherwise be carrying out that function if it weren't generative AI. SPEAKER_01: And that's that's a great point that I harp on a lot to people now is that you need to learn to interact with your AI and collaborate with your AI and give it better context rather than trying to talk to people about prompting anymore.

Just collaborate. SPEAKER_02: Right. I think the more comfortable you are interacting with the AI over a particular topic as it develops during the chat, I think the better outputs you're gonna get. SPEAKER_03: Yeah, and and recognizing that it's different.

You know, I think most lawyers look at they they try to approach generative AI like they would running a legal search. This isn't terms and connectors. Uh, this is different than running your typical legal search. And so uh the way I always try and tell people if I'm if I'm encouraging them to just kind of dip their toe in the water is just talk to it like you would if you were giving the associate the assignment.

You know, if you were asking an associate to summarize a deposition, would you just say summarize this deposition? You might, and you might they might summarize it from the wrong angle, or they might focus on the wrong things. And so if there's particular things that you would ask the associate to focus on, tell your generative AI product of choice what to focus on, and it will focus on it. SPEAKER_02: So let's talk about Ben SIM for a minute, or or uh several minutes.

What does it look like when a user launches the doc? Is it a standalone app or is it a web app, or how how is it packaged? SPEAKER_03: Yep, it's a web-based app. And essentially you log in, you upload your briefs, uh, you pick the judge that matches the temperament that you want.

Maybe you know your judge's temperament, or maybe you don't and you want a more neutral. So there's the three temperaments are there's the quiet judge that doesn't ask a lot of questions, won't interrupt you. There's a neutral that's kind of a nice sort of medium, and then there's the hot bench that will interrupt you and cut you off and ask tough questions. And then you argue.

The the AI model learns your brief. Uh, it doesn't train on your data, but it for that session, it it argues based on the arguments in your brief and for the opposing counsel's brief. And then you argue out loud in real time. So the judge will, the AI judge will listen to what you say, will respond, will push back if appropriate, will come back to points that you made earlier, follows up on weak points, asks hard questions.

And then after the argument is done, you get a transcript and a structured feedback report that you can review. It'll tell you where things went right, where things kind of went wrong, uh, things that you can maybe do differently next time. Um, it obviously isn't going to tell you if you won or lost because uh, you know, it's only gonna hear your your our side of the argument. And uh, you know, it isn't a real judge.

But the the goal of this is is not to figure out whether you would win or lose, it's to refine your argument so that when you do get in front of the judge, you can make the best case possible for the client. SPEAKER_02: This is really an interesting and exciting product. How do the other partners in your firm feel about it? SPEAKER_03: I think there's a lot of excitement.

Uh all the feedback I've had so far has been very positive. It's been very exciting. I think, you know, everybody that I've I've talked to has said we've known that this is a problem, right? Especially for the young associates, but certainly for more experienced lawyers as well, that that it's a problem and how do we train younger associates?

I kept saying before I built this product that you can have YouTube on in the background and at least watch some motion calls, watch some arguments so you get the flow of how a motion call or a live oral argument will go. I've told that to a number of young associates. I don't know how many of them uh took me up on it, but I think it was as good as you could get to sitting in the courtroom waiting for your case to be called and seeing good lawyers and, you know, maybe less than good lawyers argue things.

And so I think there's been a lot of excitement, and I think uh it hopefully will help to solve a problem in the litigation field. SPEAKER_02: Is there a sense that, well, we don't have to use senior partners at$900 an hour to train junior associates to do this when we have a software program, an AI program that will do it considerably less expensive. Is that conversation happening? SPEAKER_03: Yeah, so I know there, especially in bigger firms, um, you know, you'll have you'll moot cases, right?

You'll have a a panel of three lawyers who will read up on it and before the associate or even a partner on, especially at an appellate level, does an argument, you know, they'll take time out, they'll and you know, kind of uh grade the argument, so to speak. So not only is it just the time lost on hearing the argument and giving the feedback, but if you're gonna really do a kind of a moot oral argument for a case, you you gotta read the briefs, read the lawyer's brief that that is doing the argument, but also the opposition to figure out what questions is the judge gonna ask.

All of that takes time out of a partner schedule. And if a litigation associate is like I was when I first came out, I was nervous to go to court and I wanted more practice. I can remember before my first big, you know, dispositive motion videoed myself actually giving the argument and then watched it back. And I I look back, that was 15 years ago, but I look back and go, God, I would have loved a product like Benjamin where I could have practiced the night before and kind of gotten that nervous energy out and rehearsed it before I actually went in for the first time.

SPEAKER_02: Do you have a recommended configuration for how the lawyer should interact with the product? For example, is the lawyer just going to be sitting at their desk talking to their computer? Or do you ever have it set up where the lawyer is in this big old corporate boardroom and there's a projector at the head of the table, and the projector is the computer, and the lawyer is talking to a microphone to interact? Are are you able to set it up like that?

SPEAKER_03: I mean, you could set it up, you could set up your computer wherever you want. The more realistic you can make it, the better. I think in practice, a lot of people will probably just do it at their desk at their computer, but uh you can anywhere you can access it, and you could do it on your cell phone if you uh really wanted to. You could you could do it on the iPhone.

It's all audio back and forth argument. So you just upload your brief, an opposing counsel's brief, and it hit enter the courtroom, and you can argue from anywhere. SPEAKER_02: All right. So there's no visual judge.

It's just it's just an audio experience back and forth. SPEAKER_03: Yeah, the uh visual judge, I I think the the problem with doing that and having the the video, like an AI video rendering of a judge, um, is it would cause a latency problem at this point. The technology is getting there, and maybe eventually we'll be able to add that. But I think if we tried to do that now, it would diminish the usefulness of the product only because you'd have maybe a 10 or 15 second delay between the lawyer's argument and then the next question from the judge, which would really detract from that sort of rapid fire kind of on your feet thinking that comes sometimes with arguing motion.

SPEAKER_02: Let me ask you this. How do you code a personality of a judge into your product? How do you how do you do that? SPEAKER_01: You could upload, I mean, well, never mind.

Go ahead, Chris. Sorry. SPEAKER_02: I wanted to I want to hear, I want to hear your version also, Heather. SPEAKER_01: Yeah, but I'd look to your research version.

SPEAKER_03: What would you do, Heather? I think that's a good idea. SPEAKER_01: I would pull, I will depending on what type of law, if you were doing it for litigation, I would find, you know, go back five years or whatever you think would be adequate, find transcripts for some cases that you feel had maybe more complex, maybe simpler, different levels of complexity and upload those transcripts into the AI so that it could study what those judges were saying, what their responses were, and get an idea of how to behave and react and respond like a judge.

unknown: Yeah. SPEAKER_01: I love that. SPEAKER_03: Is that what you did, Chris? So not with that much depth.

What we did is basically we we've instructed the judge to develop a lot of the counterpoints based on the opposing counsel's brief, which in practice is where a lot of this comes from. The way it typically works is the judge will have read both parties' briefs before the motion, and then we'll have the oral argument. We'll ask you about the points raised in your opposition's brief. And that's really what these AI judges are gonna do.

They're gonna focus in on those core arguments that that the judge might have for you. And then in terms of the temperament, you know, we we had three temperaments, we gave them uh little personas, they're they have different names, and the the judges are, you know, the especially the hot bench is is programmed to interrupt uh if certain criteria are met. For example, if the user is talking in circles and not answering the question, then it will actually interrupt the user. It will cut the microphone off and actually interrupt with a question and say, you know, counsel, you're not answering my question.

I'm gonna ask it again. My question was, you know, whatever the question was. SPEAKER_02: So, how many lawyers in your firm, how many sessions have you actually had with this product, with the lawyers at Taft? SPEAKER_03: Good question.

I should look it up. Bench Sim is going through SOC2 certification, which is a very uh arduous process to ensure that the security standards are up to what they need to be in order to really be able to market this to law firms. Uh, once that starts, then we'll really be able to approach law firms about having this available to their associates. So does it hallucinate?

Does Ben Sim hallucinate? I haven't had it hallucinate yet. Now, again, we designed it so that it uses the briefs that you're submitting, right? So it it's not going out and researching law.

So I suppose if you or your opposing counsel submitted a brief with a hallucinated case in it, it might hallucinate that back to you. Um, but barring something like that, I haven't seen it yet. SPEAKER_01: So now it go ahead, Heather. So this this brings up so like I love what you've done.

I think this is an amazing tool that you've built. I and we we don't talk about on this show, of course, the gap in young associate training or even legal support staff training. Do you see AI simulation legal-based training expanding? Like at some point, this could be for uh, you know, law students preparing for the bar, things like that.

Like you could create multiple of the simulation training models. SPEAKER_03: Absolutely. I think that um generative AI has given people who who are trying to learn a new skill so much of an advantage. In fact, I I have a sixth grader who, when he was studying for exams, I I showed him, look, you can take a picture of your study guide and ask Chat GPT to create you a practice test so that you can practice the types of math problems.

So you're not just learning what the answer is, but it's it's giving you a new math problem based on the type of problems that you're gonna see on your exam. And then you can it will give you the answer key. And I mean all of those things. So yeah, I think that that the generative AI is absolutely a game changer for anyone learning how to to acquire a new skill.

SPEAKER_02: So when you have experienced litigators using your tool, do they miss things? Do you do they make mistakes? SPEAKER_03: Yeah, I mean, I can't tell you the number of I've I've argued hundreds and hundreds, if not thousands, it seems like, of cases over 15 years. And I I I ran hundreds of simulations through bench sim.

And uh absolutely it will point out afterwards things that I that I missed that I could have followed up on better. And we see experienced litigators in court all the time that, you know, all the lawyers in the back of the room might cringe if they say the wrong thing that they know isn't going to resonate with a particular judge. I mean, we see it all the time. SPEAKER_02: How about newer lawyers?

Are you seeing patterns of misses that are more common amongst younger lawyers than the more experienced attorneys? SPEAKER_03: So I I will say I don't see that kind of stuff through bench sim because I don't see the, I don't see the output. I don't see the feedback reports, I don't see the transcripts, I don't, I don't see any of that. You know, it's all secure on the the back end.

Certainly in practice, I I see a lot of younger associates that might make the same kind of error. And it's usually just a lack of experience. But I also think that human judges, unlike perhaps an AI judge, may uh sympathize a little bit because those judges were young lawyers at one point. Whether they remember that feeling or not, most judges remember having their first argument, being nervous, being overly formal, maybe, or maybe not being formal enough, depending on the judge.

SPEAKER_01: Um, and do you do you feel that this tool could help younger attorneys gain confidence more and be be able to go into court better prepared? SPEAKER_03: Oh, absolutely. And that's that's one of the advantages, right? It's not only just honing the skill and and flexing that muscle, but it's getting out that nervous energy and and rehearsing the argument and refining it and gaining a little bit of confidence that, hey, look, my score was a 52 out of 100 the first time I ran this through bench stamp.

I practiced it four times and now I'm at a 75 out of 100. It's improved. I know what the the key areas to focus on are gonna be. Uh, you know, they've thrown some curveballs at me and I was able to respond okay.

Yeah, I think it's uh it could be a big confidence boost. SPEAKER_02: So when you get your SOC2 certification, how are you going to go broad with your marketing? What's your plan? SPEAKER_03: Yeah, I think the I think the the real uh pitch is to law firms, right, that have associates that have this training gap.

And you know, we've we'll maybe be in contact with uh there's a a couple companies that uh may may partner with. We'll see, we'll see what the future holds there. But uh yeah, I think it's it's it's a it's a a broad market. I think there's a big need uh not only for the young associates, but like I said, I think experienced litigators uh can get a lot out of it as well.

SPEAKER_02: So do you have a feature roadmap, things that you want to do with Ben Sim down the road, or as you gain more experience, or as the AI gets more powerful? SPEAKER_03: Yeah, I mean we've had some some really interesting suggestions. Um certainly when the video technology catches up, I think uh we'll we'll try to implement that. And and moving into other areas of litigation.

Um so you know, perhaps there's an opening statement or uh, you know, wadier, or you know, other other aspects, uh, you know, direct examinations and and all that kind of stuff. The possibilities uh for expansion are are pretty endless. SPEAKER_02: I it feels that way. It feels like an amazing product.

Uh so speaking generally, what do you think is the most overrated AI feature? SPEAKER_04: Ooh, overrated AI feature. You know, I I don't know. SPEAKER_03: I it that's hard to say.

I I I don't I don't know that I could identify an overrated AI feature. I have one. SPEAKER_02: I think the idea that AI is going to save a whole bunch of time. Heather and I have an ongoing debate about it.

I don't think that I do not think that AI in the long run, we have a thing called an airport test that I'm sure you've heard because you're a listener. And the airport test and the airport test is if it would be quicker to drive, then drive to the airport and park the car and check your luggage and go through security and wait at the gate, et cetera, et cetera, et cetera. If then you should have just driven. And I and I wonder if there's the idea that AI is going to replace all of these personnel in a law firm.

I don't think it's going to happen that way. Uh, because I honestly, as much as I love AI and I think it is an amazing tool, I don't think it's like the best thing since sliced bread. I you know, uh, I do. SPEAKER_03: Yeah.

I so so you you have you so you you raised two points. I'm gonna agree with you on one. I don't think it is going to slash a whole bunch of personnel from Waffern. I agree with you on that.

But I'm gonna side with Heather on I do think it will make things faster. I I like to say when it comes to technology, the the litmus test is does it make things better and does it make things faster? And if it can do one of those two things, it's got a benefit. I have found that a lot of these AI tools can do both.

Uh, it can it can make something better. If I write something and I put it into one of our approved products and say, you know, improve this uh paragraph or improve my writing here, more often than not, I will incorporate something from from that improvement. And it's made it better. As to making it faster, it might not make it won't make every task faster, but I think uh a lot of tasks it can make faster, especially when when you get into some of the large volume document review and things like that, right?

Which take tens, if not hundreds, of hours of time. When that gets refined and it's we're, you know, thing the the the industry is working on it, that that will be a huge time saver. SPEAKER_01: And and it depends on how you're using it also, Ron. That is another, that's something we don't really often talk about.

Right. I have Cloud Code building me software every night before I go to sleep. I put in more, and I have it building while I'm asleep. It is saving me tons of time.

SPEAKER_02: Heather, we're gonna have to get you on the show to actually talk about how you're doing that. Uh it's very it sounds impressive and mysterious. SPEAKER_03: Well, you know, I mean, look, I mean, to Heather's point, right? I mean, I I started coding or I started started this endeavor with Bench Sim in April of 2026.

And and so, like two months ago. It if you had said 10 years ago a litigation partner with basically no coding experience is going to come up with this idea and be able to launch the platform within about six weeks and have a working product, there's no way that was unbelievable. There's no way I mean you would you would have to invest so much capital in in causing that timeline, right? Not to, and if you weren't gonna invest the capital, it would take it would take years, like like Heather said.

SPEAKER_01: Yes, it would be both, even potentially, even with capital, you'd probably be looking at at least a year. And without, you might be looking at three to five if you had to really, you know, bootstrap it yourself. The tools are so powerful and they're so capable. And I don't believe they're even all of these capabilities are going to be as accessible and affordable as they are now.

So I'm doing as much with it as I can while it is. SPEAKER_03: Yeah, it's uh it's the it's the big shift that the industry seems to be hinting at toward a consumption-based model. SPEAKER_01: Right. Yeah, right.

SPEAKER_03: As a lot of people in in illegal AI very uh apprehensive. Yeah. SPEAKER_02: You know, we're we're all talking about the price per gallon of gasoline. Uh at some point we're gonna talk about the price per token of AI usage.

unknown: Yeah. SPEAKER_02: And we're gonna get there, and I don't think that's gonna be a pretty conversation, but let's let's wait and see. Uh, Chris, what's the most underrated AI feature? SPEAKER_03: I don't know if it's an underrated feature, but I think that there's an underutilization problem.

I think that people could be utilizing it more than they really are, and I think I'd like to see see that people using it more. SPEAKER_02: What's the AI tool you're using every day? SPEAKER_03: Uh right now we're in the process of doing proof of concept and and testing for both Harvey and Lagora. And so I I touch those in some form or fashion almost every day.

And then, you know, on the uh bench sim side, I use Android Claud. And then, you know, I just use it as part of my kind of day-to-day personal life, right? So uh extracting calendar entries from emails that I get for my kids' school. I've got a an Claude agent that basically scrapes my email on my Gmail once a day and pulls out emails from my kids' schools.

It knows that I've got a first grader at this school and sixth grader at this other school, and it it sends me a text message and says, here's the calendar entries that we think we need to add to your calendar and your wife's calendar. Should I add them? And I can say yes to one, two, and four, and no to one and five or or whatever, and it will do it. You know, little little things like that.

SPEAKER_02: So do you have a recent wow moments that you could share with us? SPEAKER_03: Yeah, you know, um this isn't like the most Jetsons level thing I I could come up with, but I I used it recently. I I drafted the complaint, and you know how sometimes you get in this advocacy mode and you you lose a little bit of the objectivity. And so I said, look, I drafted this complaint.

I want you to play the role of the defense attorney. And you are dead set on filing a motion to dismiss this complaint. What arguments are you making to dismiss the complaint? And it gave me some arguments.

A couple of them I thought were a little far-fetched. I wasn't too concerned about facing those arguments on a motion to dismiss, but a couple of them I said, you know what? I could plead that better. And it would basically negate that.

And so I used it as an adversary, not just as uh, you know, AI tends to tell you what you want to hear. And I was specifically saying, tell me, you know, grade my paper, tell me what I did wrong so that I can correct it before it becomes a problem. SPEAKER_02: That's a powerful use for litigators, no question about it. SPEAKER_01: Do you think it makes you think better?

I mean, I've I say this to people a lot, and from what you're describing, it feels like to me you're using it in that way, which is a really unique and awesome way to be utilizing AI. SPEAKER_03: Yeah, I mean, it's a sounding board, right? And if you tell it, don't just tell me what I want to hear, right? It will listen.

It will do that. So use it to brainstorm. Absolutely. I might say, I might have an idea.

For example, if I get a complaint and I'm gonna file a motion to dismiss, I might have a pretty good idea of where I'm gonna go with that motion, but I may not tell, tell it yet, right? I may say, you you tell me what five arguments you'd make if you were the you, you know, a tenacious senior associate who wanted to really impress their partner and come up with creative ways to file a motion to dismiss, what would you do? And, you know, I may say, well, three of those are really kind of that's never gonna work.

But one or two of them, I might go, you know, I never even thought about that. SPEAKER_02: Hmm. You know, Chris, when you get your SOC 2 compliant, I'm definitely going to want to do a workflow options episode about Bench SIM. So you haven't between now and then to come up with a great soundbite to make our list really want to demo that product.

So for Bench Sim AI, are we looking at 2026? SPEAKER_03: Yes, that's my hope. SPEAKER_02: All right, that's exciting. We're looking forward to that.

And this gives me an opportunity to segue into our practice signal segment where we talk about uh information we get from lawyers in the wild, about problems or issues they're having in the legal world, and maybe can there be an AI solution? This one I've been holding in abeyance because I knew we were scheduling you, because it has to do with mentorship in a law firm, and it's a little dicey. This was a Reddit thread by uh a young woman who, I think she's a young woman, who just finished her first year of law school.

She was a paralegal, and nothing like this has ever happened in her old firm. So she's she's working at a mid-sized firm, and one of the founding partners is in his late 70s or early 80s. And on her second day, he called her into the office for a meeting towards the end of the day and rambled on for an hour about his life story in golf. All right.

He also told the story of a woman employee who broke water in the office, which is odd and probably just on this side of being appropriate to talk about an office. But then things get really dicey. On a Friday, he comes up to her cubicle, so she's really a captive audience at that point, and he starts describing the movie star great body of one of his clients who testified in a trial. And he described how she appeared after removing a sweater, and he's concerned that the jury is going to see her, going to see that she's an attractive woman, and she's going to get a lower damage award because the jury is going to think that she can just get another husband because her last husband has died.

He didn't describe the substantive law of the case or anything that can be remotely like coaching a future attorney, other than describing her body. So we're all listening to this and it's super cringy and it's totally wrong, and you don't have conversations like this with young lawyers in your office. But it's a form of mentorship, right? He's giving her a trial litigation strategy.

So besides all of us agreeing that this is just inappropriate, is there a better way for this particular older lawyer to have interacted with this younger law student using AI to accomplish the same kind of mentorship? SPEAKER_04: Wow, it sounds like uh the MPRE, right? SPEAKER_03: The the professional responsibility exam. It sounds like one of those questions.

Right. Right. Minus the AI. I, you know, look, I guess if I were the older lawyer and wanted to, if there was a point to all of that, to what do an object a legitimate objective to what he was trying to convey, and he didn't know how to convey it in a way that didn't come off that way.

I suppose you could use AI as a sounding board on, look, I want to convey this. I understand it, it's coming off the wrong way. You could ask AI, how can I convey the message and the the takeaway from the litigation perspective in a way that doesn't have the obvious inappropriate uh connotation? SPEAKER_02: But isn't isn't this interaction in the same family as advising clients how they should dress when they go into court?

Have you have you done that? SPEAKER_03: Do you think I can't say I've done that. Um well, I mean, other than you know, telling them you know, you need to dress appropriate, wear a suit, you know, tie, things like that. But um, you know, again, I I think using AI as a as a sounding board, as a way to kind of uh come up with different ideas and brainstorm is a is a good way to use it.

All right. SPEAKER_02: Heather, what do you think about that whole crazy story? SPEAKER_01: I have so many questions. I think that if I were that young associate, I would wonder if this senior attorney had come up with this assessment based on his own experience.

I would ask him that. I would say, have you had clients that were awarded lower monetary judgments based on their appearance? I just didn't know this was a thing. I I do think AI could have helped if he cared to go to AI and say, I'd like to tell this new young attorney, this is my concern.

But I don't want to come across so cringy. The problem is, I have zero doubt that he thought anything he said was cringy because we've all experienced that, or many of us have experienced that in our careers. And unfortunately, it is somewhat common. SPEAKER_02: In a light most favorable to that older founding partner, I do think he was trying to mentor her.

I do think he was trying to communicate ways to be more effective in trial based upon the perception of how a jury is going to think about what it's seeing and and considering. SPEAKER_01: I I 100% agree. I also just liken it completely to, I can't tell you how many times over the years in firms, someone came up to my office door and told me I would be happier if I smiled more. And I feel like that's what he was telling her.

SPEAKER_02: That's right. All right. So I think we all agree this is a totally cringy kind of practice signal, but it it impacted mentorship, which I think is at the bottom of bench sim AI. SPEAKER_01: Could you just build uh a great legal AI mentor for associates, which you kind of have in the way the education, but you could also do something in the mentor all fields of law, need more mentors.

SPEAKER_03: Yeah, that's that's an interesting one. SPEAKER_01: Take that your your next project, Chris. SPEAKER_02: Yeah. All right.

So let's let's run away from this practice signal and and talk about uh the Flintstones, Simpsons, Jetsons, and up leveling segments. And what I came up with was also in line with Bench Sim AI was using AI to stress test your brief. If you're on the Flintstones level, go ahead, take the brief you're working on, upload it to your AI tool of choice and prompt, summarize my argument, and tell me whether it is persuasive. Now, we're all sitting here thinking that you could do so much better with that prompt than that, but you're still at the Flintstones level.

We've somehow managed to get you to stick your foot into the AI pool. And I I think that's that's a kind of a safe way for an inexperienced AI user to ask that question. I think for a Simpsons lawyer, you upload the brief and you ask, act as opposing counsel, what are the three strongest arguments against my position and how would you attack this brief? Which is really what Chris, what you were talking about just a little while ago, uh using AI for that purpose.

Uh now you're using AI as an adversary instead of just a cheerleader. Finally, if you're on the Jetsons level, you upload the brief and you might prompt, act as the judge assigned to this case, conduct a bench simulation, there we go, or we're bringing in the product. Identify the issues most likely to drive the outcome, the weaknesses in both sides' positions, the questions you would ask at oral arguments, and your tentative ruling with supporting reasoning. What do you think, Chris?

Or are are we are we hitting the right spots on the Flintstone Simpsons Jetson spectrum? SPEAKER_03: Yeah, absolutely. Yeah, I think I think that's exactly right. I think the the best advice I can give to the Flintstones lawyer is to just start using it, right?

And start it with simple things that are low, low stakes. Summarize this case that just came out from the Court of Appeals that everybody seems to be talking about. Or summarize this deposition, especially if you were already at the deposition. You know what happened.

Ask it to summarize it. You know, the the Simpsons lawyers, the best, best I can say is like use it as a thinking partner. You know, you don't need to ask it to draft, you don't need to give it some big prompt. Just say, look, here's here's the complaint, here's what I'm thinking about in terms of you know responding to it.

What do you think the pros and cons are? You know, use it as a thinking partner. SPEAKER_01: Heather, what do you think? I I think, yeah, for in today's episode, Flintstones, uh, CLEs, PDFs, shadowing, you know, trial and error, attorneys just listening and trying to get it that way.

I think most attorneys are at the Simpsons level. I I'm getting there at this point. I think that more and more attorneys are getting to using it. And they're using it more like a sometimes partner, uh a training assistant, if they are smart enough to use it that way.

But the bench sim and what Mr. Ryan has done, this is Jetson's level. This is using AI to create a realistic legal training environment. And I think that's next level.

You know, if there's something beyond Jetson's, we could put him in that category. SPEAKER_02: For generations, lawyers learned advocacy through experience, mentorship, and occasionally painful feedback from judges. AI introduces a new possibility: practicing before you ever step into the courtroom. Whether Benjamin ultimately becomes indispensable or merely useful, tools like that point towards something bigger.

The shift from AI as a drafting tool to AI as a judgment training tool. Christopher Ryan, thank you so much for being here. Uh, we what a great episode. We learned a lot.

Uh, thank you. I'm Ron Drescher. Thank you. Thank you very much.

I'm Red Garner. Thank you for having me. And we'll see you next time on AI Tools for Practicing Lawyers. SPEAKER_00: That's it for today's episode of AI Tools for Practicing Lawyers.

Thanks for listening. We'll see you next time.