Sean Tibor: Hello and welcome to teaching Python. This is episode 135, and today we're going to be talking about Python for law students. My name is Sean Tyber. I'm a coder who teaches. Kelly Schuster-Paredes: And my name is Kelly Schuster Peredes, and I'm a teacher who codes. Sean Tibor: This week, we're joined by Morgan Gray and Wes Oliver, both from the Duquesne University School of Law. Welcome, Wes. Welcome, Morgan. It's great to have you on the show. Morgan Gray: Thanks for having us. Sean Tibor: So we met Wes and Morgan at Pycon during the education summit. They presented on their efforts to teach more law students about computer science and coding and Python as a way to engender a lot of different benefits for their students as they pursue the law. So we wanted to dig into that more and bring what they're doing to a little bit larger audience around the world. So thank you for joining us and looking forward to getting into the conversation a little bit later. So let's start where we normally do with the winds of the week. Kelly, would you like to pick our first contestant this week? Kelly Schuster-Paredes: I'm gonna have to pick Morgan. Cause I talked to Morgan more at Pycon, I think, than wes, and I think he's just, like, ready to go. He looks beaming with the win of the week. So, Tagore. Morgan Gray: Well, let's think. There are just so many. Wes Oliver: You bought a house? Morgan Gray: Yeah, I just bought a house. Kelly Schuster-Paredes: Wow. Sean Tibor: Oh, congratulations. Morgan Gray: I just thought the python win of the week is I just started really putting together the materials for a class that I'm going to teach in the fall. That's an extension of coding for lawyers, and it's natural language lawyering, and that's where I'm going to take law students through what's going on. Natural language processing, where that's going in the legal field, how it impacts them. My hope that by the time they walk out of there, there will be at least comfortable understanding at a moderately detailed level how an LLM works. Sean Tibor: Nice. Kelly Schuster-Paredes: That's really cool. Feel like you should test it out on me. And then I can just learn on the side like a university student, but for free. Sean Tibor: All right, well, Wes, over to you. How about one of the weekends? Wes Oliver: Sure. Well, I spoke on Friday at Calicon in Seattle, and you always come away from those kind of things. A little bit of a tidbit of some sort. And I learned about Pyqt, which I'm just now starting to get in. I mean, truly just. I've opened the file, I opened it up, I've looked at it, and I think it could be very useful for teaching students how to build graphical use during. And that's actually been a thing that we have nothing focused on because neither of us were very new to this. Morgan just finished all the coursework for his PhD. We've done a whole bunch of work. We both been coding at what, four or five years? No more than that. And so there's a lot we have obtained and done and graphic that we haven't focused on at all. I taught a statistics and machine learning class and we did a lot of graphics there. And what I found in teaching is that when you can do, when you can make things pretty and you can get students to make things that are pretty, then they become attached to the process. So I'm very excited about what I might be able to figure out with IQT for helping learn how to build graphical user interfaces. Kelly Schuster-Paredes: Sorry, I know I'm supposed to go onto the once a week, but that's one of the things. When you guys were presenting at Pycon, I was like, these guys are like us, everything you talk about. So we'll talk more about that. About your win. That's awesome. Sean, are you gonna do your win? Sean Tibor: Sure. So my win was something that didn't work out that well, but it was pretty amazing that I got it as far as I did. My in laws were cleaning out a closet in their house and we discovered boxes of old cassette tapes. A lot of them are recordings of like classical music and things on the radio. And I found a copy of the Hanukkah song that I guess my wife recorded at one point from Adam Sandler. But we found some recordings in Hebrew from overseas relatives that were sent over on tape just prior to the ability to have long international phone calls prior to the Internet and all of these things. I was telling Kelly before the show, we found a really old recording and I don't understand any of it, it's all in Hebrew. So my job though was to hook up the deck to the audio input on my laptop and digitize everything. I was able to get a decent digital copy of it, clean up some of the noise, things like that. But then I was like, I wonder if I could transcribe this? I've only looked for english transcription, but I'm sure that there's a way to transcribe this from Hebrew. And so I did, and I used like AssemblyAI, I uploaded it and two minutes later I had a bunch of Hebrew that I still can't read. But then I could put that into a translation model and translate it into English. Of course, as you can imagine, by the time you go through several rounds of taking a really old audio tape, cleaning it up, trying to transcribe it, it's very conversational and not necessarily formal speech, and then translate that from Hebrew into English, it was basically just unintelligible by the end. But I still thought about how cool it was that we could do this like that. This whole chain of going from I found a tape in a closet to here's a translation in English on it. We're probably getting to a point where, you know, maybe that whole chain isn't great right now, but we're maybe a year or two away or just, I need to find a better model to get it to something that actually would be useful. So I thought that was pretty cool. Wes Oliver: Yeah, that's really cool. Kelly Schuster-Paredes: Very typical of Sean. He's like, oh, I'm just going to do this, and start, starts doing stuff. Well, mine's kind of cool, but not so accomplished like most of my projects lately. But when I was taking my data science course, I started playing around with flask and I was trying to build an app just to see if it functions and playing around the activities. And then after speaking with Jay, I was like, I wonder if I can do this in general. Django. I absolutely have zero knowledge of django. So I put it all into GPT and didn't know that pretty much Django has everything built out for you and it has a sequel light thing already in there and authentication for like sign in and cybersecurity. And I, within less than two days this week, not even a full two days, I have a working app that loads data from a form into sqlite. Then I'm trying to mess around with the HTML. I'm trying to do this completely with chat GPT and see how far I can go. But my webpage is very grayed out. So I'm trying to find out what I did and put a mask on it somewhere. But it was cool just to see. And now I'm trying to figure out actually if I can learn Django because it's pretty intense with all the packaging that it makes. I don't know if anybody's seen it, but it makes my package my app and then the main app, then you have all these files. It might just be a chat GPT thing, but it was pretty fun. So that was a huge learning win. Morgan Gray: I will say, Kelly, when I want to do like rapid development, if there's a project that I need done and I need to get it done quick. My editor in one screen and g four in the other. Shamefully. And so whenever I know that I come up across something, maybe it'll take me ten minutes to think through. I'm just like, you know what? I'm going to be lazy this time. Kelly Schuster-Paredes: Well, that's what we're teaching. That's what we're teaching the 8th graders like, we show them Matplotlib. And then I said, okay, go for it. Other libraries that make cool graphs. Here's the other libraries where you can do this. And I had a kid, I was on a podcast last weekend, I was saying, this one kid didn't do any of his homework at all. But then he came to me and said, my grades really crummy. Can I get a boost? Here's what I've been learning. And he shows me how he was taking about, I don't know, 800,000 lines of data about horse racing. And he was plugging it in and learning tensorflow and machine learning with Chad. Gpthen said, okay, can you write a summary about that? And I had shown him a couple of things, and I showed him anaconda's data science whatever, and he comes up to me and asked me this math form. I was like, you don't have to know that math. It's just know if it's one or negative one and it's the model's good or not. I don't know what's going on with that math problem, but it's crazy. Anyways, I digress. Sean Tibor: I think it's interesting, like you said, shamefully having GPT four there, I don't think there's any shame in that at all. Morgan. I think if you showed someone from 1985, one of our modern encoding editors, where it does autocomplete and linting, and you hover over and it tells you exactly what the problem they'd be like, that's cheating. We're going to be very close to a point where a few years from now, like, it's going to be so commonplace that nobody even thinks about it, right? Morgan Gray: Yeah, it's invaluable. Um, if you're just having a bad day and you're just tired and you could figure it out otherwise, but you just really need to get through it sped me along many, many, many times. Kelly Schuster-Paredes: It's also great when you're given a little pop quiz and you generate some code and then you say, no, make it basic. No, don't use functions, no, don't do this. And then you give it to the kids, and you say, okay, I can prove to you that you've learned Python with me. Read this code. And they're like, oh, yeah. So it's a great little tool for that. Wes Oliver: Absolutely. Sean Tibor: Why don't we jump into the main topic of discussion and start talking about Python for lawyers and proto lawyers, Wes and Morgan, could you introduce yourselves, talk a little bit about where you work now, what you're interested in, and how you came to be teaching Python? Wes Oliver: Yeah. So I had no background in this as a kid. I was really fascinated by computers, and we didn't get one in my family, and it really wasn't taught in school. We had a basic class. I remember in 11th grade, there was no real in depth sense of using math and computers, although I always remembered my 11th grade teacher talking about how computer programs were using some of the things that we were talking about. And I was fascinated in math class in 11th grade programming, programming classes in high school were in basic, and were. And that was a great description of what we were doing. And I really got into them. And I was a history major in college and went to law school, and I always wanted to find my way back to math because I always found math just fascinating. I never really knew how this would happen. So every year I would teach Crimpro, and there's a legal standard called probable cause, and a lesser one called reasonable suspicion. Probable cause allows a search or an arrest, and reasonable suspicion allows someone to be detained while the officer conducts some kind of an investigation. And I would teach these standards every year, and I would tell my students, look, I can't tell you what satisfies it, because I can only show you cases where it has been satisfied and has not been satisfied. The court can consider absolutely anything. So anytime you change any one fact, it's up for Grant, what a court's gonna do. And I'm teaching you how to argue. I'm not teaching you how to arrive at an answer. And that was a terribly unsatisfying end, for although much of the law is undetermined, this one's getting particularly so. So we set out who. I set out to try to find an answer to this. I was watching 60 minutes one night. They were talking about how AI was being used to read medical journals, because no human could read all the medical journals in the world. What if a computer could read them and find connections between medical journals? Maybe research on malaria could help us with research on aid in ways that no one would anticipate, because no one would think these things would be connected in any way. And I thought, isn't that probable cause? You've got a whole bunch of cases out there. Is it possible for a computer to do this? We started talking to computer scientists about solving the problem. They said it was doable. They started playing around a little bit with it, Morgan and I. Morgan was my research assistant. We both started working on this. We both became unfathomable, a spot, not knowing what they were talking about. And we decided we need to learn it for ourselves. One thing led to another. He starts getting his PhD, and we realized other lawyers could benefit from what we know. And python for lawyers was born at Newcaine. Kelly Schuster-Paredes: I have so many thoughts on that. You want to go first, Sean? Sean Tibor: Morgan, was this just something like, wes came to you one day and was like, hey, I'm thinking about this, and you're like, let's do this. Let's get into it. Did you have a background or a story? Morgan Gray: That's a fun story. I'm thrilled anytime we get to tell it. Our relationship begun because I told him I didn't like how he taught. As an excellent person as he is, he took it in stride and asked me why, which started the conversation about computers, computer science. And I had admitted I'd taken, like, one class on it in college, and it was just like an upper level statistics class. I had to use some sort of statistical software. I don't even know if you'd call it a programming language or something like that. I had been interested, and I remember really looking back on it fondly. So that's how looks from my perspective, on how we got onto it. And I will say it was really, truly driven by. We were just tired of not. It was really hard to tell whether and not that anybody who we were working with was nefarious at all. They were extremely helpful. But you need to know these things. You need to understand what's going on. What do you mean when you say. I think for both of us, it was just we were getting unsatisfying answers a lot of the time, and we wanted to be able to do it ourselves. One of the things that I did want to mention is that heart were like both tinkerers. We can get into something. We start tinkering with it and playing with it, toying with it, figuring it out, and that sort of really kept it going, as you know, with programming. Literally always something to tinker with. We find all our students are that. Wes Oliver: Way, too, and some of them aren't. That's actually fascinating. I've kind of stopped advertising for the class because I really encouraged a whole bunch of people to take it. The first year, that was great, and a lot of them whittled down after the first week, and we were left with a lot of great people. Second year, I encouraged a lot of people to take it, and there was no whittling, and there was a certain amount of discontent in the people who were taking it. Almost like, you know, the story of. You mentioned your hebrew text, the story of Gideons army, how they let the guy go, they feared was drinking the water a certain way, and they said, lend everybody else home. I kind of feel that our programming classes are like that. I've stopped advertising for them and stopped really encouraging people to take it because I think people will find their way to us. We certainly don't hide what we're doing. The first time we went around to all the classes and said, look, you really need to know this for your future. I absolutely believe that's true. If you don't know this stuff in five years, you're not going to have a job. I might be exaggerating a bit, but not much. But I also think the dynamic of the room is so much better when you have people who are excited about it. Also, the dynamic of the room gets boiled a little bit when you have people who have a constant frustration, who are not tinkering. So I've been a little lighter in the way that I've gone about trying to recruit, and I think that created a smaller and more committed core. Kelly Schuster-Paredes: There's so many things I was saying earlier in the start of the show. There's so many things when you guys are talking that I'm just like, yes, that's me. Yes, that's me. I teach middle school computer science. They're forced to take it. They don't want to be there. We have about five people at the beginning who are just in it. They've been coding and scratch. These are little 6th graders. It's this. How do I get them to understand and love it and feel the passion that I feel as a non computer science person? It's all these things that you were talking about that for you guys, getting into it, that kind of just in time learning things that Shawn and I say all the time. You have a project. That's why Python comes across, or that's why we need to learn how to code, or this real world application. I'm particularly interested in that aspect of getting into the python, and I fully believe less than five years, when you start talking to those law students. It's gonna be less than five years if they don't have some sort of computer science. I wanna go back to that whole approach to integrating the python into your law curriculum. You guys are non computer science majors, so, Morgan, you've had the experience of computer science teachers now at a doctorate level, right? So you understand they go. Morgan Gray: Went past. No, maybe twice, maybe 400, and then wound myself up to graduate level. Cs is where I started. For anybody out there, I would not recommend coming in on top floor. You can avoid it. I couldn't avoid it. Kelly Schuster-Paredes: So, you know, like how a computer. Morgan Gray: Science kind of go. Kelly Schuster-Paredes: How does this differ? I think it's best that you describe it in your terms. How does it differ from traditional computer science courses? I'll let you explain, because I know how I differ. Morgan Gray: Yeah. I'm trying to think if I can put this eloquently, maybe even colloquially, in a funny manner, in coding for Williams, you're met with a loving embrace, and you just don't get that in computer science class. It's very cold, and I hate to use this word, it's cold science. If it's not precise, then it might not be anything at all. You know, sort of, kind of just throw that out the window. Not very welcoming. Our first go into this is we turn that on its head entirely. We want you to be there. We're glad that you're there. We're absolutely welcoming to the nth degree. We also. The one thing that we say is we do understand this, because we've been where you've been before. We remember what it. Wes Oliver: Like a year ago. Morgan Gray: Yeah, yeah. Not that long ago. Wes Oliver: Right. Morgan Gray: Adding up. No, that's how. Wes Oliver: That's true. Morgan Gray: We show like we have been here before. We know the. This doesn't feel nice all of the time, but it feels great some of the time. Yeah, it does feel great some of the time, but especially when you're starting, it just doesn't feel nice. It's truly like learning a different language. I made the remark at Pycon. It'd be no different if I started spewing German or something to you from the lecture. You're not going to get it if you don't know it already. So we take the traditional sort of Cs attitude toward things and throw that right out the window. We want it to be as welcoming and inclusive as possible. I don't want to say extreme as, and it's bad, but in terms of what you would get from a standard course, we do go to extremes. Extended office hours were available combined, like 20 hours a week or something like that. For the coding? Wes Oliver: Yeah. Actually, I was 20 hours on my own. In my office. We've got a giant monitor. Well, they can come in with a laptop and plug it in. I've got a word that we plug into their computer so I can type along with them. Group can sit around, giant monitor, and we just try to create a space where it's very comfortable for people to do this. Morgan Gray: Yeah. And we say, if a student does bad in the class, probably because you just didn't come see it, that's saying if you were confused on something, I guarantee we were always there to take your question, work through it with you. I've had students who have, if anybody's taught you've seen this an hour before the projects do a panicked email. And I literally got on Zoom on Saturday night at 09:00 and worked through something with them. And we just know that they're not coming from a CS background, knowing that this is hard to get into, that never leave the front of our mind. So treating somebody in that situation like they are and being able to respond. Wes Oliver: One of the questions that we had to answer when we proposed doing this is since this is essentially an undergraduate class in Python, or from what you're saying in 8th grade class in Python, why are you teaching a law school? Why not just have them walk across campus and teach it there? The answer we gave to that is there are all kinds of impediments for people to cross register. I do. Part of that is if the professor is a computer science professor and not a law professor, if the other students are computer scientists and not or people who are taking sten classes and that's what their focus is, there's something intimidating about that, where there's something much less intimidating than taking that from me, who learned this stuff entirely online just a couple years ago. My background is in history. I sound like a law history guy. Whenever I'm describing a future iterating variable, I'll slow down. Wait a minute. You might not know what an iterating variable is. That's the thing that's going to keep picking off and increasing. One time, every time you go through there, I'll remember to say that in a way that a person who does that for a living for years and only does that, might not remember. Sean Tibor: I'm going to make an assumption and tell me how valid you feel this is not being a lawyer, but knowing lawyers and having very good friends that are lawyers and just observing the way they think the way they approach the world. And this is a sweeping generalization, but my perception is that law is one of the purest forms of information processing. It is the practice of being able to take in information, synthesize, arrange, organize, and then present or share. So you have this idea of being able to take what you've learned, what you know, what you can prove with evidence or gather, and turn that into some sort of other information. It's transforming it from one state to another. To me, that feels a lot like what we do in related fields to computer science, like information systems and data management and information processing in the library and everything. How much of the way that you've structured your course makes it more tailored towards the aspects of coding that pertain to information processing and retrieval of information, storing things, explaining things in that way, whether it's consciously or subconsciously. Like. I think of computer science more as the study of computing things, and it sounds like you're teaching more from the perspective of, here's how to gather, process, and use information with computer science like tools. Morgan Gray: Yeah, that's it. I might quote you on that in terms of what lawyers do. So I'll say, well, so my asset, I've never thought about it that way, but my assessment would be, after the fourth week of the course, which is like, in the fourth week, we're just getting an iteration. And so by the time you've seen basic syntax variables, conditionals, by the time we're getting to the fourth week where we're starting iteration, that's what I would say we do. We inject data into the process, and that's what they're doing. I mean, our homework assignments, in class assignments are, we've given you some blob of something, whatever it is, whether it's Kentucky Derby data or something having to do with sifting through legal opinions to find a particular citation, something like that, it's very much take that here's, use your programming knowledge and then output something useful with it. And I think, Sean, that's definitely fair comparison. Lawyer writing a brief or an argument or something like that certainly would have an entire record, an entire 200 years plus of law in front of them. Got to take that and boil it down into a little package that somebody can do something. Wes Oliver: And the conference I just spoke at, David Colorusso from Suffolk, who teaches computer science stuff to lawyers there. He said there was a famous lawyer who tried to square the circle and what it's squaring the circle. You take a circle and you try to create a square that has exactly the same area. We all know that's actually mathematically and theoretically pretty easy to do. But how do you actually make it with just a straight edge and a pencil and a protractor? He goes through and talks about how this was Lincoln. Lincoln was fascinated with it. He spent two days after trying to literally do it, and Lincoln was fascinated with Euclid. If you've seen the movie Lincoln with Daniel Day Lewis, he quotes Euclid in the telegraph office. That one day, and I've often thought so, lost, learned, and more in the past than the present. But still we were praying based on some form of the socratic method. And I've often thought that lawyers are following the wrong greek. I feel like we ought to be doing more Euclid than we are Socrates. I think Socrates invites certain forms of analogical reasoning where you trace the analogy on down, and all of a sudden you're doing things in the analogous circumstance that might not actually fit back in the original circumstance. It's more about trickery, whereas I feel like can be and I feel like Euclid. We actually did what Lincoln did and studied more about logical mathematical proof, that very few things in life will get a mathematical proof. But if that's the gold standard rather than the credit method, we might have tighter. Kelly Schuster-Paredes: I love this going back. I wanted to break that down for me, because I like to put it into my head and analyze it. So as a lawyer, you teach the python for law, and you're like, the fourth week also in my lower levels, when they have something to apply, when they're able to use that critical thinking, that method of deduction, that brain power that's connected to their law, and they're learning Python, they're able to make those connections. Like I can almost foresee. Here's Python for doctors. I just feel like we have different levels of understanding, and when we are able to make those connections, those neural synapses, big idea pictures connect, it makes Python become more useful, more personal. So I want to think about shifting a little bit gears. Can you think of someone that was like, oh, my gosh, I know, you gave us a situation of a girl who hated it at Pycon. You explained her, that was awesome. Can you think of someone also when you saw that other aha. Moment, like they made a connection and maybe what helped them connect more with python and law? Or you can use the same example, because I love that story. Wes Oliver: Yeah, this was great. We had this student, and she was one of the people who was very frustrated early on. I tell her told the story earlier about how ive been less Pt. Barnum liked in my effort to get students because im one of the smaller, more committed group. I wonder if she would have come in if we hadnt hyped it so much the first year. So if we lose people like her, then that would be really sad. She was a fascinating case. The first week or so shed come into office hours or shed raise her hand in class and be incredibly frustrated, just like part angry, part upset whine. It was just awful. She finally comes to me, four or five, and she walks in my office and she just mad at me. I said, let's get back to basic. Why did you sign up for this class at all? She said, I thought it would be fun. Let's get back to trying to make this fun for you. She still left the office mad at me. So then it might have been the next week or so. She raised her hand like we have a period of during the class when they work in groups and she would always ask really basic questions and we didn't really think she was following it. The light bulb just hadn't turned on for yet. She raised her hand. I'd always try to ignore her. She was always going to be at times suck, whatever. But I'm like, I've got to go talk to her like I do every week. I went over and looked at I don't understand why this isn't working. She written some crazy complicated thing that had a minor syntax here, like was missing an AMA or something. He said, put a comma there. Not only does this solve the problem, the far more elegant solution than we were teaching the light bulb had come on. So at the end of class we do object oriented programming as like one of the last modules. Morgan said to the class he was teaching that day and he said, don't try this at home. We don't want you to do this on your projects because we wanted to introduce it to you so you'd know what it was. But we don't expect anyone to use it in the final product. So she gets this panicked look on her face. What's she worried about now? And she said, I'm three quarters of the way done with the project. I realized I couldn't solve this without object because I googled how to solve this problem. They said, I need to do object, but I went ahead and learned objects on my own. And do I have to take them out? Absolutely not. If you can do that, go for it. And her final project was absolutely brilliant. So yeah, there are these people who have a fear, have a reluctance, and some of them will never get over it. And there are some like her, who discover a whole new world. Kelly Schuster-Paredes: I want to go in because this is what is connecting to your Euclid kind of Socrates thing. And I'm thinking the law students are more ethical and philosophical and they're. Isn't that so? More. And you're now trying to twist it on them, right. Say, okay, now we need to like think of reasoning and systems, chronological order. How does that work? I don't know anything really about the law degree. I'm a bio person. So do you think maybe that has a different way, a different. Wes Oliver: They read statutes better? Kelly Schuster-Paredes: Yeah, that's what I'm thinking. Morgan Gray: Yeah. Wes Oliver: I'm now taking a class on Java. And when you look at Java programming, it's like if you really want to understand, like jurisdiction, best job. Can you go there? Can you go there? Look at what the entry code says. Then if you look at substructure, like with things being triggered there, that law has fuzzy qualities to it, but there's also just a basic core structure, particularly to a statute. You go here you go here. You go here. And there were fuzzy words there. You got to go to the right place. Morgan Gray: There's a underlying principle of logic where in law a lot of stuff is conditional. If this happens, then we'll do this. If this doesn't happen, we'll do that. Even when you file a legal complaint against somebody, you can plead the alternative, which means this fails. And we're also saying this or that or the other. Could be any one of these sponsoring. We're alleging that there's a really good paper, it's more to do with AI, but it talks about how judges will resolve issues before them as a decision tree structure. We were presented with this issue and on part of it we found this. So that means we split the analysis here and we go somewhere else. So you see that sort of logic popping up all the time in law and as programmers, that's like, what, 75% of it? Kelly Schuster-Paredes: Funny story. So I'll give you a lesson that you can attempt. Because I gave up with our humanities teacher, I tried to code the Bill of rights. There's a lot of conditions. There's a lot of conditions in there. And she's teaching me the Bill of rights. She's like, oh, that depends. Then you have to consider the ethical and all this stuff. So you see a lot of nested conditionals. So there's a fun little activity. Give them amendment and have them go. Wes Oliver: Exactly. Actually, that you mentioned one of my next project, I want to develop a tech version of criminal law, teach criminal law, where they, you can opt into this. And I want to teach it like I want to move criminal law to the year for folks who do this. So they take quoting first year and then take the tech version criminal so that they would actually write a code so that you could give the program, they've written a, a set of facts and have who they've written take the exam, rather than how do you do this right now? People saying we'll just throw everything into GPT chat. Well, no, no, obviously that's going to give you problems, but what you can do is, like with the Bill of rights, there are certain things that are conditional. So if you're talking about you have the right to counsel for it, there are certain binary things that are true. If it's a civil case, the answer is absolutely not. If it's a criminal case and you're not going to be sentenced to more than six months at jail. No. If there's a possibility of you getting more than a month at jail, then yes, you absolutely do have a right to counsel. Now, what does it mean to be a counsel? No, that's very fussy. You might send that to some large language model, but recognize the answer, get out of that may or may not be correct, but will be a first draft from what you might then start to play around and work. But there are certain aspects to the law that are clear. Yes, no conditional. And then there are things that improve the performance of large language models by fine tuning, but then recognizing those aren't going to necessarily be correct. There are things that will most always be correct. Sean Tibor: This is why there's so much connection here and why I was so fascinated, because this is the same sort of thing that I was getting at my undergraduate degree and graduate degree was all in information systems. And yes, there's a ton of computing that needs to happen, but a lot of it is based on this information that we're processing. What is the right output that we want to have, or what's the structure of this? How do we flow from one state to another? And the more I'm hearing about this, the more I'm seeing all of these connections, because it's not necessarily who's better at math. It's more about how do you organize your thinking? How do you structure the flow of the information through whatever system you're building? This leads me to my actual question, which is, Morgan, you talked about how judges are making decisions using that flowchart in front of them, whether that's explicitly defined or implicit based on their knowledge of the law, they keep it in their head. You addressed this question at Pycon, and I wanted to bring it up. Are you seeing students who take your courses being more effective in the other non coding courses that they take in law school? And what are the outcomes that you're seeing as a result of this to maybe make them better lawyers? Morgan Gray: Yeah, we've gotten great feedback about what it's doing in terms of the law students who go on from coding, and then now they're on to something else, like drafting. They're in a simulation course where they're drafting documents on behalf of client. What we're told is that coding students are heads above their peers in terms of quality of the writing, in terms of how they handle issues. So theyre presented with something. Its presented in a clear, clean cut format. Their writing has certainly improved. Their way of thinking, has improved our class, that if you step into coding for lawyers every day for the hour and 15 minutes that were together, that sort of thinking from start to finish, were seeing the benefits pour over in through classes. Certainly, we're hopeful that other faculty members give us that nice feedback. Kelly Schuster-Paredes: It's nice. Do you think that their analytical skills and everything, critical thinking, that whole analysis is changing? Morgan Gray: Yeah, I actually wanted to remark about that, and that was the story that I thought of. Wes was talking about our one student who had been frustrated by the end of it. So it's this roller coaster thing. They're sort of down. They're like, this will never amount to anything. What are these nuts teaching us? And then they're up there, oh, my gosh. If then statements can solve absolutely everything, and then they're back down again. They're like, oh, my gosh, this is really freaking hard. After they come back down again on, wow, this problem is really hard. They all had a little bit of enlightenment, and they see everything and now understand they're realistic. Oh, this is computationally expensive. This is going to be difficult. We can't account for all these. One thing. The project that came up in is taking the language of statutes, taking the language of police complain, and then trying to give in a police complaint what crime was committed. We gave the students, like, a really naive way to start it. Maybe you can think of just keywords with just a bag of words approach. Start looking for shared word, cross documents. If there's a high match, then output that crime or whatever. Then one of the students who was one of the best that we had. He came to me and he was like, I'm not going to solve this, just bag of words, am I? I said, no, absolutely you're not. We started to have a deeper conversation, and my PhD research is all NLP. We start to really have this conversation about the intricacies of natural language processing. Language model, how we model language, why it's ten times harder in the legal domain. He really had that moment where you could tell if he were to go out and pursue this or do something forward, he would be competent to handle what's in front of him because he would know the challenges involved, at least be able to think through it, understand problems that he was encountered with on top of analytical power. As lawyers, they know the precision that they have to get in to in front of a judge or a jury, and so they're able to translate that. And that script that I just write might not even 10% of the way. Wes Oliver: And he went on to take our statistics machine learning class in the law school, but he also took class on data structures and algorithms in computer science. So what we really are starting to see is that folks are opening up to the idea of now going back and further developing these skills. We've had some conversations with folks like Bloomberg law about hiring our students as software developers in the law, because they can get through data structure and algorithm. We can teach them the rest what they need to know. This combined skillset is something that we haven't found. Kelly Schuster-Paredes: I'm really excited for you with also that NLP. You were saying earlier that you're going to be adding some more in the second course. Morgan Gray: Yeah. Kelly Schuster-Paredes: And you're going to do like legal analysis or research and all these other. Morgan Gray: Automating or machine learning and natural language processing. After I learned Python, they became near and dear to my heart, and I absolutely love and adore those areas of study and those fields of discipline. When I got into NLP, at first, I would feel there are all these models out there that can handle all these amazing tasks and maybe dealing with news or summarizing news stories that was popular back then, or something that could do all these things. But then you read any of the literature on how these sorts of things go in law, and it's like the plane lost both engines midair and it's heading down. So I became really passionate about, well, geez, how do you even go about taking tools that exist in natural language processes? How can you do anything even remotely useful in the legal domain? I mean, even basic classification is hard. So what I'm going to do is take them through the very bare bones basic of natural language processing to understand what goes behind it, why we're thinking that way, why we do these certain things, and then along the way pull in relevant literature from like, okay, well, let's say something super simple like, here's a bag of words, vectorization, representation of words. Okay, here's how somebody tries that maybe 15 years ago, and here's how it massively failed. And here's why go through in the legal domain where it becomes challenging. I hope it'll be sobering for them as lawyers to realize if they're confronted with AI tools based in NLP, why you just, you totally cannot trust whatever comes out of there. Even if it's a well built system, you should know what's going on behind the scenes, and I hope they'll be able to interface with that. The class is going to be structured that somebody who hasn't gone through coding will be able to come and listen to lecture and pick up what they're dealing with. Then I'm doing another part of the class for students who have had coding, and that's where we're going to have the fun and take legal documents and corpus and break it down, do some cool things with it. By the end of it, I'll have them assemble and fine tune their own transformer model like Bert or something like that, and put it to a task. Kelly Schuster-Paredes: Have you heard of Aness Montani? She was on our show. You need to check her out. She's the CEO of explosion and Spacey. She's NLP. That's what I was thinking of. Was it prodigy or something? She made some sort of annotation tool that helped. Sean Tibor: Yeah, I don't think she's looking specifically at the law space, but I bet you'd have a fascinating conversation with her and be able to dig into some interesting applications of both what you're doing and what she's working on. I wanted to wrap us up a little bit and talk about the future. You mentioned that you've been speaking with Bloomberg law, and they're very interested in finding this blending of skill sets between software development and law students. Are you seeing a lot of demand from employers of different kinds, whether that's law firms or public defenders offices or the typical sources of jobs for law school graduates. Are they seeing the value in this? Are they asking for people with these kinds of skills, or do they even know that this is possible yet? Wes Oliver: The latter, yeah, that was what was so fascinating. Like the Bloomberg people didn't know we were out there and lawyers don't. Because I think there are seven different schools that are teaching some form of python. Law schools now. Eight. The University of Washington. No, no, they're not. They're doing art at the University of Washington. So there are seven that are teaching some form of python. The numbers are really, really small. And what's the, what's fascinating to me is the legal tech people don't even recognize the need to find lawyers who have this hybrid skill who are actually building this stuff. We've talked to a few companies that like, oh, yeah, no, lawyers aren't at the level where they could do this. Maybe you ought to talk to us a little more. We can actually do so. Bloomberg expressed a real interest and there's a firm in London, or London based firm, and they have a legal tech research and development department. So we hope to start to placing a couple students as summer associate with that London based firm. Reed Smith at one point had a tech summer associate program, but I think that's gone by the wayside. I really think that it's an issue of timing. My wife worked for an investment banking firm on Wall street. On Wall street that years ago. And about 20 years ago her boss was really fascinated with electric cars and charging stations. There was an investor who wanted them at charging station for electric cars and his business failed miserably because if there are no cars, charging stations don't have any. I'm wondering whether we're at point where we have enough cars and enough electric charging station, enough lost to the hybrid and another folks out there who recognize and can use tyward fog. Bloomberg is one of our start. The London firm is another possibility. We got to start. We've all got to find each other. And that's going to be the real trick. Sean Tibor: Yeah, it feels like it's like a super cool liquid. All it's going to take is one hard shake and it's all going to crystallize much faster than any of us expect. Wes Oliver: Exactly. Kelly Schuster-Paredes: I have a totally off topic. You know me, so Pycon's very unique experience. I don't know if this was your first Pycon this year, so I wanted to ask you as lawyers, because I remember my first Pycon as a non coder. Really? How did you share your experience and what did you think about being a bunch with a bunch of nerdy people like Sean? Morgan Gray: My initial reaction was coming from bias towards education and academics is like, why wasn't everybody who had a stake in some company or something like that there. Listening to what's going on at the education summit, I was in charge. Everybody there would have been mandatory attendance that we all would have came and sat and listened to the education summit. Because a lot of times I find, especially in research, something really cool will come out of the way and then industry has absolutely no use for it. Okay. Or industry really needs something. Nobody in research cares. I just guess I was thinking, wow, even here there's a divide. I would say the first thing that I noticed is everybody at the education summit were just like ten out of ten jab about Python. There was a really invigorating environment. It was cool to nerd out with somebody about something other than something legal. Wes Oliver: I was amazed at how interested people were in what we were doing because we've come at this from a totally different angle. You had a lot of folks like Kelly, who were teaching in secondary school, who were there, and here we are, graduate schools completely unrelated to computer science. I went back the second day and with the math, I figure hard to identify people. A couple people came up to me just to talk about my talk the previous day. So I was actually really fascinated with how many people were interested in what we had to say. Kelly Schuster-Paredes: I'll tell you why, at least for me, we sit there in 8th grade and 9th grade and 10th grade, 11th grade. They're like, why do I have to take this course? I'm like, everybody needs to learn how to code. I'm not going to need to learn how to code. I'm going to be a lawyer. And now I'm like, yeah, right. So I think as, as K twelve teachers, we try to predict to these kids of what's going to happen in law school. A lot of degrees now are requiring at least some sort of computer science for at least people like me being able to say, oh, yeah, you don't think you need coding? This is how it's going to help you. So I think that's like a rewarding thing for us to see that you are passionate about it as professors of law and a university, and that you're coming in and saying, hey, look, coding is great for the brain. It is going to be useful for your job, and here's why. So kudos to you guys for that, and thank you for coming to Education Summit coast. Wes Oliver: Absolutely. Kelly Schuster-Paredes: It was great. Sean Tibor: I have, my off topic question is for anyone who's thinking about coming next year because Pycon will again be in Pittsburgh, because we go two years in a row. Oh, yeah. So the Pycon pattern is to stay in the same city for two years, because by the time you make all the connections and you organize everything, it makes sense to use that investment. Over two years, I spent most of Pycon with my friends from all over the world, showing them how Duquesne does not pronounce the way it's spelled right, based on my experience as a long ago CMU student. So I wanted to know, for people who are coming back next year to Pittsburgh, what's one thing from each of you that you would recommend they check out in Pittsburgh? That's one of your favorite things that maybe is not downtown or not in the convention site center. Wes Oliver: Yeah, downtown. So your fan architecture. I think Pittsburgh has some of the most gorgeous houses of worship of any city in the country. Boston might rival at London. Yeah. But beyond those two, so there's a temple right across the street from the Mister Rogers studio. And then there are just multiple massive churches. And outside of that, like just architecture, generally downtown, there is the. What's that? Union? Is it the union trust. Union trust building? Well, I think that there are some. There's some architecture in this city. I think churches are to come stand out for the architecture in the city. Then even within the downtown, the union trust building is really worth like from the outside looking at. Get us a step back. And there are these two little like apartment houses, things on top of the skyd. And then when you go in it, there's wonderful pre art deco, almost gilded age style skyscraper. So I would say various aspects of our art. Morgan Gray: Yeah. Going off, maybe not exactly off of that. I would say if anybody there also interested in history, Pittsburgh has been around for a really, really long time. And most people don't really like recognize that. So I would say anybody, especially who's really interested in revolutionary history, there are some interesting sites around the city. There's the well preserved fort, maybe an hour drive from the city. If somebody wanted to do that, anybody who has it specifically like revolutionary history, would be at least moderately reserved and a pirate fan. Wes Oliver: Because the. Not so much of the team, but because it's one of the prettiest parks in the country. Not particularly stored itself, but wonderful view across the river of downtown. It almost looked like it's not really real. It's a very neat, unique experience. Kelly Schuster-Paredes: Yeah, I can. Sean took us there when we did our corset CMU for robotics. I agree. That was a good view. That was fun. Sean Tibor: And you can still, if I remember correctly, they still have a marker in Oakland for the original Pittsburgh ball field before three Rivers. There's just a ton of those little hidden gems around Pittsburgh where you can go and see something cool and see something historic. I went to the high History museum for the first time while I was in Pittsburgh for Pycon and really just enjoyed walking through all the different exhibits. And they have a bunch of things. They're just like, here's our archives on display. We have to store them somewhere, so we're going to put them in a glass case so you can at least look at them while they're in storage. It's really cool to be able to see all of that. They had, of all things, they had one of the two Tommy guns that the Pittsburgh Police Department bought back in the 1920s. Random things that are cool. That's one of the things I love the most about Pittsburgh, is just there's a lot of history there and a lot of cool things to go see. Wes Oliver: And the Natural History museum one day, it's phenomenal. I think that's got to be one of the large dinosaur bone collection I'd ever seen in one blade. Kelly Schuster-Paredes: Very cool. Hopefully I get to go back next year. I'm going to put that in early and hopefully you guys get to come back and join us. Sean Tibor: I think we're already at time or over time, so I'm going to wrap us up here by saying thank you to Wes and Morgan for joining us. Thanks for sharing more about the work that you're doing at Duquesne and the School of Law. And if people want to find out more about the programs of the Duquesne University School of Law, specifically around the coding for lawyers course, we will put links to that in the show notes so you can refer to that and get in touch with Wes and Morgan to learn more. Wes Oliver: Absolutely. Morgan Gray: If somebody's interested in implementing something similar, we're always very willing to help. Sean Tibor: Yeah, that would be pretty cool. If you got your first incoming law student who came because there was coding for lawyers in the program, I think that'd be a pretty great win also. Wes Oliver: It would be all right. Sean Tibor: So for teaching Python, this is Sean. Kelly Schuster-Paredes: And this is Kelly signing on.