Sean Tibor: So, hello and welcome to Teaching Python. This is episode 151 and today we're talking with Prates Patel about AI, ethical issues, the corporate world, a little bit of everything with Prates. My name is Shawn Tyber. I'm a coder who teaches, and my. Kelly Schuster-Paredes: Name is Kelly Schuster Paredes and I'm a teacher who codes a little bit. Just kidding. Sean Tibor: So, Prates, welcome to the show. It's a pleasure to have you here and we're really excited to meet you. Pritesh Patel: Yeah, definitely. I've been on a couple podcasts, but I have to tell you, this is going to be the most exciting one for me. It's a python podcast. Wow. Kelly Schuster-Paredes: It was so great meeting Pritesh. We met at the Fisher Phillips. I always have to catch myself, so I don't say Fisher Price. I don't know why that's like the common name. At the Fisher. At the Fisher Phillips conference this summer, it was a really fun conference, not to mention that we were in the Willard. So that was really nice. Pritesh Patel: Yeah, that was a good time. We got some time together to speak about stuff. So I'm glad I got a chance to meet your old table, too. Kelly Schuster-Paredes: Yep. Very cool. Sean Tibor: Why don't we start where we usually do with the winds of the week? So something good that's happened inside or outside of the classroom, workplace, wherever it is, and Pritesh will have you go first. Pritesh Patel: So outside of the workplace too, right? Is that what. Sean Tibor: Yeah, any. Anything you want. Pritesh Patel: Well, one of the things is my parents, they don't run a frozen yogurt shop. And it just so turned out we had our best month this past month. And it was something that they've been working on for decades now. And it was just something really positive to see. And I was chatting with them about it and they felt very excited. The work has paid off. So it's a lot of work running a restaurant. So that was a very positive thing for me. Sean Tibor: Did they have any reason like that they could point to and say, this is what did it, or was it just everything happened to come together in the perfect way? Pritesh Patel: Well, the restaurant businesses, you have to be very consistent where you offer a good product and then people have to know about it. So if you're there consistently giving a good experience, giving a good product, eventually people start to take catch on. So it's more of that. It's the consistent work over a long period of time where you have a clean store, you have a great product, you have great service with your employees, which takes a lot of work. To do. And that's where that comes from. Kelly Schuster-Paredes: So. So the real question is, does anything part of that business have code or AI in it? Pritesh Patel: Oh, yeah. Oh, yeah. When we jumped into this, we started to realize how difficult it is to run any operation at the end of the day. Right. And then you have to be there every single day. So I've done a lot of automations there. Some of the marketing that I've done there is automated now, where it sticks to the brand, purpose and message. It's still human in the loop, where we're still approving it, but we're not doing all the work setting everything up, which saves us hours a week. And most of the time we wouldn't even get to it because we're just so busy running the store. So some of that stuff I've automated, like generating a profit and loss statement. The next thing that I'm working on is I'm trying to use AI to help us detect where are we going off in our expenses. So it detects it as it's going, and it tells us during this month, hey, if you change this decision, your total potential profit will be better by the end of the year. So that's the next thing I'm working on. Kelly Schuster-Paredes: That's pretty cool. Sean always tries to tell me, find a problem and once you have a problem, just solve it with a code or something. Pritesh Patel: Yeah, definitely. Kelly Schuster-Paredes: So who's going first? Flip the coin. Sean Tibor: Go ahead, Kelly, it's all you. Kelly Schuster-Paredes: You know what? I was gonna say I had a win a week, but this is actually the win that just happened like 30 minutes ago. My kid is taking Comp Sci AP and this is not necessarily a win for him, but he's banging on the table because he's trying to get this stupid dog to do this loop. And I was trying to tell him, but he would not tell him how to do it, but how to think through the problem. And he's banging on the table. And I laughed and I was like, who does that remind you of? Sean Tibor: I wonder where he gets that from. Kelly Schuster-Paredes: So when I was first learning how to code, Sean had just got hired. And I was just like, I can't do this. I can't teach python. I can't. I don't understand it. And I would just bang. He's like, kelly, just walk away. But he. It was funny because I was trying. He doesn't like to listen to his mother. So that's another fun win for me. But I was just like, work through it. Brute force it to make it do what you want step by step. And then look for where it has repetition and then add the for loop. I don't know. JavaScript. I kind of read it because I had to do a little bit for the boot camp with Georgia Tech, but to think through the stupid dog walking along the thing. So it's funny, I'm just waiting to see how long it takes before he goes to AI and tries to cheat his way through it. But right now, he was still banging on the table. That was my win. Pritesh Patel: You went on a boot camp at Georgia Tech? Kelly Schuster-Paredes: Yes, I did that. Like, was it two years ago? Pritesh Patel: Oh, amazing. That's my alma mater. That's where I went to school. It's a great. Kelly Schuster-Paredes: Yeah, it was good. I did six months of hellacious trying to understand data analytics and the Python part was the easy part. And all my group members were like, okay, we'll do all the machine learning and whatever you do all the Python. And I was like, done. And so I'm good with that. I can play with pandas all day long, but once you try to make me analyze stuff, it's a little bit different. Thinking skills. Sean Tibor: Look at that. We've gone from banging on the keyboard to she's the Python queen, right? Pritesh Patel: Exactly. Sometimes you have to do that. Kelly Schuster-Paredes: Okay, Sean, your win. Sean Tibor: There's a lot, surprisingly. And I wouldn't have expected this when the week before last started because I had a pretty rough week. I had a small bout of identity theft. Pritesh Patel: Oh, no. Sean Tibor: Which makes it sound like I had a cold or something. No. Someone took out a PPP loan in my name three years ago, and I finally found out about it, so I had to go deal with that. And that's just extra work. And my car got T boned and totaled and I had just put a bunch of money into repairing it. So I was like having this really bad week. But then I guess the win is just the resilience and the persistence and the confidence that it can't be like this forever, that things will get better and you just have to stick your way through it. So over the course of the following week, I flew to Chicago. I presented to AWS about my company's use of AWS with infrastructure as code and AI coding assistance and the tremendous progress we've been able to make over the last three and a half years. And it was just a really great moment because I was presenting to this very technical community inside the company about our journey together. And when I say our journey, I was privileged to be the person presenting it. But I tried to put throughout the presentation photos of the team and accomplishments that we had done together, because I got to share the journey that we had done together and all of the great work that the team's been doing. So it was this really powerful moment, and it was very well received. And we got a lot of questions and a lot of, surely that can't be possible when I said, I am serious and please don't call me Shirley. But it really did. It really was this great moment. And then I got a replacement car, and I'm happy with that. And things are progressing more at work. So the win is not so much the individual items, it's the being persistent and bouncing back and realizing that, like, one bad week does not make a bad month or a bad year or anything like that, and that anything that happens, you can get through. And I'm incredibly grateful and lucky because all of those things were fine. The bad things were not that bad. Although it sucks to have your car totaled. I walked away, she walked away. Both of us were fine. Everything else is just a mechanics problem. So that was really. The win was just. I'm really glad to be here. I'm grateful for the opportunity to do these things. I'm grateful for the opportunity to bounce back from having a bad week and. Kelly Schuster-Paredes: We get to be together with Prachesh. Sean Tibor: See, it keeps going. That's the amazing thing. Pritesh Patel: That's the worst part. Kelly Schuster-Paredes: Oh, so funny. So Pritesh briefly introduced the fact that we met at this conference. It was, for me, I don't know if you realize, it was weird because normally at a tech conference that I present at, I always wear very casual or T shirt, a crazy jacket. And I brought my crazy jacket and I could see my colleagues kind of like. And then I looked around a little bit and I saw everybody in black and brown talking, like very mute lawyer colors. And then I gravitated to the few people that happened to be tech people. So I was really excited that I met the crowd. Everyone was great. But you had a lot to say on the couple of the webinars before. So I thought it was interesting just to share a little bit of your background, of how you got to where you were. You went to Georgia Tech and you're a techie person, a data person. Pritesh Patel: Yeah, yeah. I'm a computer scientist. Yeah, I went to school and I studied computer science. I specialize in AI. I did AI research for a little while. Back then it was called meta learning. So reinforcement learning is what it kind of became, but it was meta learning. And then I was doing also nlp, which is natural language processing, which is now the precursors of all these language models. But yeah, I was doing that work and been in industry about 20 years now. Being a tech guy, most of my experience comes from like non tech first companies. I was at Turner Broadcasting, I was at Walmart, I was at ge. So these are what you would consider old startups that started decades ago, but digital technology was in their first thing. It's a support function. So I've had a good amount of experience in how to teach people how to use this effectively in already functioning workflows. And yeah, here I am at the law firm. I ended up joining the law firm because I was working at Walmart. I was helping them automate their supply chain, which I can talk about if that's also interesting. It's very interesting how Walmart gets all their products to where they get at the right time so they're not out of stock, which is a very analytical project. But they wanted me to move to Bentonville, Arkansas, the headquarters, I live in Atlanta and I was flying there every week for a little while. So eventually I could just couldn't do that anymore. So I was looking around and the law firm came around and just started doing some research on what's going on in the knowledge based industry. It's like, okay, this could be pretty fun. So here I am. Kelly Schuster-Paredes: I love that terminology, knowledge based industry. That's pretty much teaching in school. That's a pretty cool thing. Sean, you want to add? Sean Tibor: I wanted to just ask. I work for a company that supplies Walmart, right? And we're part of the supply chain and everything. You're right, it's a fascinating problem and it's large scale and far bigger than anybody can really comprehend. All kinds of nuances to it. When you were looking at the law firm, I guess maybe we'll work backwards from your current experience into some of the other things. But what was appealing about that? Where did you see the opportunities? What was ripe for transformation? What was the stuff that was like, oh my God, we could totally make this amazing if we just make it work. Or if I were there, I could do this. What led you to that? Pritesh Patel: The main thing that kind of led me primarily the culture. There's a cultural difference also that was in this phase of my career where I've gained all this experience and I'd love to be able to use it somewhere where it's actually useful for folks because when you're working a large company they usually narrow it down because there's so many folks and that for a good reason. I guess you're focused on one small thing. But I wanted to expand more. So I saw over here that we can build for bone solutions, Greenfield, blue field, whatever you want to call it. So that was the exciting part. The secondary thing is like the culture was really good here. So just in my experience in whenever you're trying to innovate in any industry, the places where people are open and sharing knowledge with each other and what they know and they mean, obviously they're different backgrounds, but if they're sharing, that's when you can really build and innovate new things. I saw that openness here and that's why I came here. So it was more of that opportunity that got me excited about it. And then the third thing is obviously there's a lot of room for cool things we can do here since AI is taking off and it's going to affect knowledge based industries. This happens to be a firm that wants to innovate from the inside out as much as possible instead of getting disrupted from the outside. Kelly Schuster-Paredes: I find it really interesting, I think as an educator first to think about coding because most people, I'm not going to talk about my middle school kids, but my kids that are now in their junior and senior year that still come bother me and teach me everything because they're way smarter than me. But the idea that the first thing they think about in coding is, oh, I've got to go into machine learning or I have to be a developer or I need to do some sort of finance and coding. I don't think many people really think about the fact that yeah, all these big business businesses, even your parents shop, they. They have aspects of code that can really optimize the output. Did you always think that way or were you also like a software person, a data science? When did you like, oh, go, oh and say, oh, I'm gonna work on the back end with somebody and really put tech into their world? Pritesh Patel: Yeah, that's a great question. I never thought I'd go in the technical field. I always was wanting to either go on the medical side of things or business. My family struggled financially growing up and I was the eldest of three, so I was always motivated. Let me see if I can help them. So first thing you think about in us is like, we can go into business. So that's what I wanted to do. But when I was graduating high school in 2000, there was programming and programmers earning like 300 bucks an hour. Wow. I don't have to go to school for 11 years as a medical professional and I can maybe four years done, earn 300 bucks an hour. So that's why I went into it. That was the crux. And I found it interesting too. I'm just. Generally, it's interesting stuff. But what I really learned over the years is in any engineering or science, there's science and engineering, and science informs the engineering. So I went into computer science, which is slightly different from software engineering in the sense that this is the science of computation that informs how to engineer things. So just like if you're a civil engineer who's building a bridge, they rely on the science of material science to understand the strength of material. How will that hold up to certain weather conditions? So you pick the right material and then they'll engineer it. Same thing as in computation. There's the science and then the engineering parts. The science comes first. Once I started to understand that, it was just like, I'm going to go deep into the science, start and understand the fundamentals. And once you start understanding the fundamentals of this stuff, any new thing that comes around, it's just easier to understand and pick up at the end of the day. So I think there's a key difference, like science and engineering. Understand the science first and you'll just be in a much better foundation. Sean Tibor: Are there any technologies or breakthroughs or developments that you can think of from the science side? I think one of the obvious ones is the kind of rapid emergence of Gen. But anything that struck you as something that was surprising, where it wasn't so much, wow, this is easy to understand, it was, this is big and this is transformative. And then looking back on it, realizing how transformative it was. We're going to talk a lot about AI, but are there other things that you can point to as examples of that transition and relationship between science and engineering? Pritesh Patel: Not the specific comes to mind, but one of the things that did really take off, which was surprising to the community, is when they started to, you know how we learn language the way we learn language in elementary school? As you. I'm talking to teachers. You all know this, right? You'll write us in the cat in the blank hat is the answer. And then you'll say the blank in the hat. So if the teacher writes that and shows that enough times to the student, they'll understand, okay, these are how words work. And then you take the test and there's a blank. They know how to fill it out. That's exactly how you train these models. They just Basically give it all these text and they have blanks in there and all this. As a human brain, our hardware is brains which has intuition and feelings and all that beyond just the pattern recognition. Because we'll have an emotional attachment to also cat and what it means to us. Right? So it has deeper meaning. But for a machine it's just math. It's just capturing patterns and statistics. So if it sees trillions and trillions of examples, eventually start to learn, oh, to write a good sentence, if you have these words, this is the next word that's most probable to make it a good sentence. When they were doing that, they figured out that if you give all these neural networks all these words and it's figuring that out. If you took the word the number representation, which is actually called a vector, it's just a number representation of the word king and you subtract it from the vector representation or just think of it as a number. As a from boy it would approximately equal queen minus girl. So meaning. So if you just think about meaning, meaning is how things are related to each other. Just the fact that you're throwing all this stuff at it. The pattern recognition was starting to understand which words seem to be similar to each other. So that's what really took off all this early on. Like, oh wow, this can be really useful in terms of storing knowledge early on. And then there was a bunch of other seminal papers that came out and blew this up. But that was one of the surprising things I think I also remember thinking like, wow, this is cool. Kelly Schuster-Paredes: I was thinking, I can't avoid it. I'm sorry. I've been trying to avoid talking about AI, but that's all I live for. So I'm thinking about this knowledge based industry, this knowledge economy and I'm thinking about the people in tech that are have been constantly problem solvers. Just working with my son tonight, trying to make him work through a problem without giving him the answer. Then I'm thinking about how your employment has embraced this culture. How does it differ from where you were before? Where? Because first of all, gen AI wasn't really populating. But were you in an environment where people were like, yes, let's go and automate it, let's go and turn it. Let's get rid of that big 40 page spreadsheet where I'm manually entering in my formulas. And how does that kind of relate to the industry and what you're doing? Does that make sense? What I'm trying to say? It's like a huge question bringing it. Pritesh Patel: Down no, it makes sense. People get used to doing things the way they do it because they become very good at it. So changing behavior and showing them that there might be something that can help them, it does take time and that's okay. My approach and that was the case also at Walmart, it still is the case here, even though it's a very tech forward firm. People are used to doing a certain thing a certain way, but you have to teach them because everything's still through people. So you got to convince them since they're still accountable for the work, they gotta trust that this will work for them. My approach has always been just educate them on the intuition. Like what is this? So this example I gave about how we learn in elementary school, it's the same way I explain it to them so that now it doesn't seem like such a big thing. Foreign thing like, okay, this makes sense. I can see why this works. And once they have the intuition, they can make their decisions themselves. Like why it could be useful for their particular function and workflow more than others. The way I talk about it though with AI is you can delegate the responsibility, but you can't delegate the accountability. So let it do stuff. You can assign it stuff, but you're still verifying. You need to always be verifying this stuff. If you think about work in general, verification is easier than creation. So let it create and do all the stuff. Give it instructions on whatever flow in this mechanism you created. But you're always verifying at the end of the day and that usually saves you a lot of time. Even though if you have to iterate multiple times, it'll save you a lot of time. How I talk about it, are you. Sean Tibor: Finding it easier now to get people to adopt these things or at least be willing to try things out? Given the consumer market for Gen AI and what's happening in that space, it's. Pritesh Patel: Generally because now it's the public. Right. So that influences everybody. Like, oh wow, I probably should pay attention to this. So that does help. Every day there's some new thing that the foundational model developers are putting out there. So it's so fast it is getting easier for folks to pay attention to this and you got to be intentional about it at the end of the day and spend some time on it. Yeah, I have noticed in even the last few months it is starting to get a little easier. So I think that's gonna change really fast. Eventually everybody's just gonna use it. Kelly Schuster-Paredes: Yeah, there's a lot to relate for a lot of Teachers out there in the schools and stuff especially like my new role now has morphed since I came back from Washington. We came back from Washington. All of a sudden they're like, oh, I think we need to do a trainer. And I'm like, sure, I'll do it, no problem. I know everything about AI. I played with natural language processing. I can totally rewrite Martin Luther King's speech into ten lines, no problem. So it's been a really interesting thing and I'm the only one that was really always talking about this and everything. And I can say that it has gotten a lot of backing from not just the admin because you know, that is the place that helps the change happen. If you have group of really thought provoking leaders who say, listen, we're innovating. That's everyone's hard spot. If it doesn't happen at the top, luckily. But it's. The rest of the teachers were like, oh wow, it's not really that hard. You're right. If you explain a little bit of things, here's what we can do, here's where you can learn and you go from there. My biggest problem is what you were just saying about that intentionality. Everyone goes, I want to do this. I said, okay, let's just pause, let's just wait. How are you dealing with that? It's good because you're in a business where you have to. But how are you dealing with it on a. Maybe a smaller scale? Pritesh Patel: So far it's always started with the education aspect. Still, I have a feeling that'll start to change maybe by next year, I don't know, or maybe not. Maybe there'll be new advancements also. Now you got to keep talking about this is what this means and how you use it. That'll probably still be there. But a lot of it's just education in the beginning of and that, I mean you guys are teaching. If you actually educate somebody on some how certain things work, that's empowerment and they'll make the decisions themselves. So that's usually been how I've been approaching that. My philosophy always has been you guys are still the experts in whatever you do. No AI person is going to come here and say they're the expert all of a sudden because they know AI there's a certain function. It's AI is still just a tool for whatever you're trying to produce as an outcome. So you are the expert. I'm just here to show you a new tool that can help you and the best way to help you do that is you understand how to use it, what it means, and all that kind of stuff, and you'll be the best person able to figure out how to use it for yourself. So that's how I have approached it so far. Kelly Schuster-Paredes: What are your, like, key fun lessons that you like for adults? Pritesh Patel: It still goes back to the Batman voice on chat GPT. I. I go and I actually have these things where I go to chat GPT and say, speak in the Batman voice and sing Twinkle, Twinkle Little Star. Still gets a laugh. It's amazing. Kelly Schuster-Paredes: And how is that useful in law? I know it'll come up some way. Pritesh Patel: Yeah. All these lawyers are so serious. Usually all that. But when I start going up and I show them, like, they're all laughing. Okay, this is working. Kelly Schuster-Paredes: My favorite one is give me critical feedback like Simon Cowell. Pritesh Patel: I love that. Kelly Schuster-Paredes: It's really good. That one is that one or Simon Sinek. When I'm trying to curate a presentation, I'm always like, but where's the why I need Simon Sinek fix my presentation. Pritesh Patel: Wait, so Simon Sinek, not Simon Cowell? Kelly Schuster-Paredes: Both. I use both. Oh, oh, Both my Simons. Simon says Simon Cowell when I need really harsh feedback and Simon Sinek when I'm making a presentation. Pritesh Patel: Kelly, are you singing to these AIs? Kelly Schuster-Paredes: No. Rarely do the voice. Only when my brain is working too fast do I do the voice thing. But I feel it's like the sketch noting with a notepad something about the writing for me, because I come up with crazy ideas when I. Especially when I leave in the morning from the gym. It's like, crazy, crazy ideas. And I'm like, oh, you know, coding's like riding a bicycle. That was one that I did. Or, oh, what if this was matching with this? And how would it go? And if I did all this and then I talk it in or I type it into to chat, and then it, like, formulates this web of craziness. Pritesh Patel: I love it. That's fun stuff. Kelly Schuster-Paredes: Yeah. What about you, Sean? What's your favorite Persona? Sean Tibor: I go back to the one I think Mike Kennedy shared this one on Python Bytes, like, years ago. And it works really well for people who are familiar with code. But you have it solve a problem like, hey, write a recursive algorithm for generating a Fibonacci sequence or something like that. So you come up with this fairly familiar algorithm, and then, okay, now explain it to me. Okay. So they explain the algorithm and then you say, okay, now explain it to me like a pirate. And every time it's like yar, matey. We're gonna delve it. We're gonna. It always says delve. We're gonna delve into the mysteries of the Fibonacci sequence. It cracks me up every time. Kelly Schuster-Paredes: You would really enjoy my Sheldon Bot that I made the GPT. I'll have to share that one with you. He's quite funny. You would nerd out. I can't believe he didn't do some sort of gaming. Gaming character. Sean Tibor: I. I don't know. Kelly Schuster-Paredes: Married like a Mario. Sean Tibor: The accent with the pirate always gets me. So. Pritesh Patel: We can still be kids and have fun with and learn. This is still amazing. Sean Tibor: Well, and that's one of the things that I really like about this recent iteration, is that there's a certain amount of playfulness to it. This is really important. When we play, we engage. We learn better. We learn faster when we're playing. So I ask my engineers, go play with this. Go try this out. Go explore. Go experiment. You don't have to come up with something that's immediately valuable to the business or immediately solves a problem. I want you to play. And that play is really important. Kelly, you and I saw this in the classroom a lot when we could create that state of play within the classroom, the level of progress that students would make. And this applies equally to adult learners and child learners. The level of progress that's made when you're playing is dramatic and it's very different. And there's a room for serious learning as well. But playful learning, we forget about that as adults. We forget about how powerful that is because I walk into my job and I'm Mr. Serious and I've got to be the boss and I've got to know my stuff. There's a certain amount of vulnerability that you have to accept in order to be playful, and that's difficult to do in these big, serious adult corporate jobs. That's something that I take from my teaching experience back to that world, is that there's a way to be playful and there's a way to be fun and engaged and also demonstrate the value of that playfulness as well. So that's something that I've been reflecting on with this, is that this is almost tailor made for that because it is so playful and so creative. And I love even just doing things like I generate stickers for my laptop using ChatGPT because it's fun, you can make it and why not, right? Pritesh Patel: Yeah, definitely. Kelly Schuster-Paredes: We need to make a new teaching python sticker out of that there, Sean. Give you something to do. Sean Tibor: Yep, I will open up another tab and we'll get right on that yet another tab. Kelly Schuster-Paredes: Sorry, Pradesh, I didn't mean to cut you off. Pritesh Patel: Oh, no, no, it's all good. This voice stuff has made it much more even accessible because it has personalities. Now. One of the things you'll see it. We're doing a webinar soon. They haven't announced it yet. I think we're doing it October 2nd. But I built a little tool just for fun where it's actually like all these AI personalities that I created will debate a human for running for president. So we're gonna do a webinar with. Kelly Schuster-Paredes: Nothing controversial there, huh? Pritesh Patel: I made it presidential debate. Just going to use it for debating like how legislation works with some of our exports and stuff, which is going to be interesting. But I have like a New York City comedian, cab driver personality. I have this Britney Sparkles made up name of social media influencer. I have a Southern speaking governor from Texas, all that kind of stuff. So it's going to be. It's just fun trying to do all these things and how people. Kelly Schuster-Paredes: What are you doing with that? How are you building it? You're just. Do you have. Are you. Did you build it all out? Are you using your own models or how are you doing. Get some secrets, share something. Pritesh Patel: I'll share the code with you. I'll put it up online too. But it's simple. It's just OpenAI's real time API. Kelly Schuster-Paredes: Okay. Pritesh Patel: You can actually prime it for a certain personality. You can tell it, I want this personality. You have to be very specific. You have to give it the prompts that I want a southern personality. And then you tell it examples like these are the types of words and sentences they say and all that kind of stuff. And it'll actually prime itself to speak like that. Then I just created an interface through Streamlit that triggers each personally. You can click a button and it just triggers. I want this personality now. And I click a button saying, okay, let's start speaking. Let them hear the human answer and then it'll respond so it understands the human answer and then it responds. So it's pretty simple. And I use cursor and it's all written in Python so anybody can do this. It took me like maybe three, four hours to do this in the evenings. It's so amazing how fast we can do this now. Kelly Schuster-Paredes: So it is pretty cool. But there is a lot of fun things. I've been playing around with a little bit of a llama following some of the things that they did. But there's so much. You need a lot of little dedication, focus, time to get these things going. If you had to, if you had to tell educators training what would be like a fun but useful thing that they could do easily with little. With little coding knowledge. Pritesh Patel: With little coding knowledge. But you want to develop something. Kelly Schuster-Paredes: I don't know, I mean, vibe coding. We could vibe. Let's put that aside because you can do that with Canva and Chat GPT. But what would you do? What would you do for your lawyer friends who are like, hey, I want to do something, I want to learn, but I don't know any code. Pritesh Patel: Oh, got it. One of the things that used quite often though is the Notebook lm. People like using Notebook lm. They seem to be like, just continue to build out new features. It's like they're just going and going. So that's been a cool thing. Just whenever I showed people that that's an easy thing to do. Oh, Suno. Suno.com. you've probably heard of this one. So this one's really fun because you just make songs in a minute. And so every time I presented this song, I always get messages later that this weekend it was my kid's birthday party and I created a song. It was a hit. So those are fun things to do. Kelly Schuster-Paredes: Notebook seems to be like a push. I actually doing a training on Friday. Four of them, I should say. Not a. They time me out all day on them, but that's a fun one. The one thing I like about Notebook, and this is probably really handy for lawyers, probably why they love it, is I can take those really long scientific. Pritesh Patel: Yeah. Kelly Schuster-Paredes: Documents that are like 25 and I. When I start to read, I skim down to the outcome. Instead I just put it in the notebook. And I did this the other day with critical thinking. This is what I'm going to do for a presentation. I'm sharing people because people keep asking me what I'm doing, but I'm ta. I took a notebook and I put in a lot of studies about just regular critical thinking and then a couple of articles about how we need to focus on critical thinking in the age of AI. And then by asking questions, you get pushed back to a lot of the skills needed for teaching just critical thinking and you wipe away the. Oh, we gotta focus. This is about AI. No, it's still about critical thinking. So that's my hopes of coming out with this presentation because NotebookLM is pretty cool and it gives you that site. So instead of Having to skim through it, you're like, oh, it was here. And it pulls up the location of that large document. So, yeah, it's fun. I haven't quite fallen in love with the movie. It's real fun. But it's always the same thing. Like, here's this one, two, three. And then. And I told Sean when we did the podcast, the female kind of sounded dumb. And I was like, I hope that's not how I sound. She giggles a lot. Pritesh Patel: Yeah. Sean Tibor: Because Kelly never giggles on her podcast. Kelly Schuster-Paredes: Never giggles. Pritesh Patel: Yeah. It's a terrible thing to giggle, right? Yeah. Sean Tibor: I'm curious, Pritesh. One of the things that is really important about this and going from that transition, because I know I just talked about being playful and everything, but we also want to be productive, and we want to solve problems or create things, build things that didn't exist. And I'm seeing the power of being able to use LLMs as a building block for something. I can use a model to generate text, or I can create an agent to go do something, or there's all of these new things that are coming together. And I feel like we're creating and continuously building these building blocks that can be used and assembled in such a way to solve interesting problems. Some of them are still playful. One of the things that I saw that was really cool was Halloween decorations. And someone took a Halloween skeleton with a speaker box in it and wired it up to an LLM with a camera so that, as trick or treaters would show up, the skeleton could respond to that costume that. That the kid was wearing. And, like, you're such a cute little spaceman or whatever, you can prompt that into a really interesting character. But it's the combination of the two. If they were just, like, on a laptop doing that, it wouldn't be interesting. But putting it out in your front yard is pretty awesome. So have you seen or have you experienced ways in which we can help turn this from being something that we engage with on a webpage or in an app on our phone into something that's now a component component or a part of a tool that we're building or a problem that we're solving beyond just the browser, beyond the chat interface? Pritesh Patel: Yeah, 100. Because they're talking about that too. Like, we're spending so much with the chat, but how are we integrating into real workflows? And that's where the work really is from that. You do have to realize that AI is still just a tool for something. One of the things that I've Always used from a product and technology innovation perspective is this concept called jobs to be done. If you ever heard of jobs to be done is at the end of the day, you still need to know what your function is, the workflow and problem you're trying to solve. Like a classic example is if somebody's going to a hardware store and buying a drill, they're not really in the market to buy a drill. What they're really in the market for is a hole in the wall. So that's the outcome they're looking for. And they're just hiring the drill because that seems to be the best option to produce that outcome. So you just need to understand whatever you're trying to solve, there's a certain outcome you're trying to produce. Recognize that's separate from the solution. Technology is still just a solution to that. Once you figure that out, then it's going to become a little bit more obvious where we can integrate. Because the actual outcome is this technology, something that can produce a better outcome for that. So I approach it that way. I've approached it pretty much after I started to learn how to use that. I approach that anywhere and it's always worked. And that's what we're approaching here from as well. Or else you just get into this problem of a solution, looking for a problem. It just starts from the design, like, what are you trying to do, Kelly? Sean Tibor: It's interesting because combining this with what Pritesh was sharing earlier, I see this kind of crystallizing for a lot of teachers and not necessarily like CS teachers, but earlier Pritesh, you mentioned it was about process. Teaching is a process and there are teachers who have become extraordinarily well versed in the process of helping learners learn, helping students, whether that's children, adults, whomever, learn in a way that is very effective. Sometimes we as teachers, we get focused on the process and lose sight of the outcome. So for a teacher who say, teaching English, is it really about the process, process of reading the book, the discussion that goes with that, and the papers that we write about it, and the grammatical analysis and looking at the metaphors and allegories and things like that that are contained within the story? Pritesh Patel: Right. Sean Tibor: Or is it about the outcome that we want, which is to have people, humans that are more intuitive, more empathic, more analytical, more flexible in their thinking, that have the ability to read, understand and apply what they've read to other parts of their life. That's that I would all describe as outcomes that we want. The challenging thing for a Lot of teachers is that they're so invested in the process that trying to incorporate something, or worse still, feeling like it's being forced upon them. Pritesh Patel: Right. Sean Tibor: Is disrupting the process, and they're having trouble linking that to making a better outcome or achieving that outcome in a more effective way. And I'm curious to know if that's fitting with what you're talking about or what you're thinking, if you're seeing that in business and in law, if it's. If you're seeing similar paths. Because I think that's where a lot of educators are still struggling is the fixation on the process and struggling with linking it to the outcome. Pritesh Patel: That is very well articulated. That is 100% what this is about. So the point of the jobs you've done framework is it holds you disciplined to the fact that what is the outcome? So you don't get bogged down in your current process, because your current process, it may have worked whenever it was created, and it was the most optimal thing at that time. But if you're just stuck in looking at the new tools through the current process, you might be missing a full opportunity of the actual outcome that you want to do. So even when I came to this firm, I told them, I don't want to know anything about the process yet. I want to know how legal and law firms are supposed to idealistically function. First, let me determine, what are the outcomes we actually want for clients? What do the clients really want? What do the firm want? Map all that out. Then we'll start digging into the process, because now you'll see the gaps, and then now you can. When you see the gaps, you can actually determine, okay, we have these. All these new tools. How do we bridge that gap now? And you completely now go above the process, you may end up having a better process. So I 100% agree what you're talking about, that you got to understand and map out the functional outcomes first. And then you work backwards from there, and then you'll. And you can look at your process, but you'll see gaps there. And then you can correctly assess. How can this tool help there? Kelly Schuster-Paredes: That. That's the thing. And I love all this thing. And here comes the teacher side of me. Sean's, like, ready. He's like, I know what Kelly's going with this. So this is the hardest part for me with AI in education, and I think this is the hardest part for a lot of. I'm gonna make an assumption for a lot of people who are trying to bring AI into education, maybe the consultants or whatever, whoever's out there doing this work to get this in. So going back to Sean's thing about the teachers going through the step. I know if I'm writing an essay, then I'm going to have them brainstorm, and then I'm going to have them write an outline or. And then I'm going to. Whatever. I'm going to have them write a thesis sentence, and then I'm going to have them outline. They're used to this. But if they were focused on that hole in the wall. Like the hole in the wall is to get better thinkers, to get kids that can make those connections between, I don't know, whatever the name of the book that my son's reading, some Plato, Socrates book, and, you know, make him make connections of how this is so important to his life right now. I think that is a different story. And going back to that process versus output, the hardest part is to say to a teacher, what is that job that you need to get done? Because there's. Well, they have to write an essay. Well, no AI can do that. That's not an outcome. And P. And English teachers hate that. When you say the AI is gonna. I. I tell them all the time, I don't need a spell. You should see how I talk to Chat gbt. It doesn't care how I spell. And I type. I type really fast without looking. I'm like, yeah, yeah, yeah. And those words are all jumbled. And it still has my idea. This applies to, like, everything that's going on with this. Is that getting. I don't want to say layman, but a person that's not a coder or not a techie to understand that, you have to know you have to have that problem that's well stated in order to work backwards to make that project. That's huge. So I like, I really do. And this isn't a question I really do like the idea of people don't go in to buy drills. They go in to buy something that's going to put a hole in the wall. Right? Pritesh Patel: Yeah. Kelly Schuster-Paredes: And whether that be a hammer, a fist, a drill, a nail, or whatever. It just depends on what type of hole that you want to build or what do you want to do with that hole if more people are teaching AI with that philosophy. And that's what I loved about your conference, because just that whole thing of that philosophy of really, we're not just trying to use a tool to do a problem. We're trying to take the problem and fit it with the right tool to make the learning happen. Yeah, that was not a question to. Pritesh Patel: That accord anybody's ever interested. The guy who created the Jobs To Be Done framework, his name is Tony Ulwich and he actually has a company called Strategen where they just focus on teaching people that this is how innovation is done. So that they even have a playbook. You can actually. It's for free. You can download it, go to their website. You can download jobs, job maps and Job To Be Done framework and it teaches you how do you actually formulate this and how to think about this. There's guide you step by step. It took me a little time to figure that out because I was always like stuck in the process. Being an engineer and a scientist, there are tools for that and that I'm just sharing that helped me a great deal. It's exactly what I helped me. Sean Tibor: I'm curious to see or to understand if you're seeing this with your work. One of the reasons why people can be resistant to that kind of change to process is that they've defined their sense of self based on their ability to execute that process. Like, I'm a good teacher because I do the process better than everyone of teaching. I give better lectures, I have better assignments. I've invested in my own mastery of that process craft. And it puts that sense of self at risk to disrupt the process or to revise it. So I guess my question is, are you seeing examples where people are willing to change, willing to undergo that metamorphosis of saying, I defined myself as a lawyer who was really good at the process of being a lawyer, but now I'm turning into something else because I'm willing to adopt and transform the process and redefine to me what it means to be a good lawyer. Are you seeing that happen? What are the characteristics of people who are willing to do that that maybe we can transfer or identify for teachers? Pritesh Patel: Yeah, and that's not solved yet. Even in this place. What I've noticed is the folks who are curious, just naturally more curious, are going to be much more. They have been more open to things. There are certain folks who are not as curious that you're always going to have that. And so for them, it becomes more of an influence once all their peers are doing it, and then they'll start to catch on. So I have seen a bit of that. That probably is from personality. But what I try to tell them is if you're really good at a process, it's not that this is going away for you, it's just shifting now. It's shifting to something slightly different. But that also is going to have a process. So if you're really good at following a process, you're just going to be that much better than everybody else in this new process that is producing even better outcomes. So sometimes people are really good at. I've had friends who are like this as well. There's very good discipline at certain things and they're good because they know it's producing an outcome. They're good at doing that. But you can still do that. It's just with new tools and it's a new process. Now I'm working on that type of way of convincing folks, but I have noticed a difference. The folks who are more curious are much more capable of getting onto that now. Not capable, but they're just much more interested in getting into it. Now. Kelly Schuster-Paredes: This is probably another question. Well, at least I know I have my own question. So I can't even imagine how many people work at your law firm across the world. What is the goal? Is it for a certain number of people to be AI trained or AI aware? And I know you guys do webinars. You did your conference. I'm assuming you have training inside of the office and probably train the trainer kind of aspect where you like have to. There's only one of you. So you have to push out. Like, what is your goal? What is the goal of a company that is truly or a business that is truly AI aware? What is a goal? What is a number? Like, where would schools go to say, oh yes, our school is. All of our staff are. And how do you rate that? Like, how do you know at your business of whether they're aware and capable? Pritesh Patel: It's a good question. I don't know how to measure that in terms of just the goal. And this could probably similar for any company is just being more efficient and more productive. That's the business goal. But for me personally gets an opportunity if we're intentional about it. Where work becomes even more fun. Because here's the thing, in general, whenever we're working on things, if we do it over and over, it gets boring. Like it's like we've done it so many times now. Maybe some people like that, I don't know. But I just feel like after a while it just gets boring. So if once it's more deterministic and these are the steps you're supposed to follow, if you delegate that off, it leaves more thinking power for you to do all the Creative, more curious things that you're. You were curious about, which essentially translates to being more. It'll be more fun. So I'm motivated by that. From a personal level, I've noticed that even in terms of my coding skills, I'm just getting to more stuff that I could never do before. Like, I have all these ideas, but I can never get it done in the time period that I have. So I think that's there. But the business goals is be more efficient and they produce better outcomes as well for whatever function we're working on for. So they want to produce better outcomes for their clients, which is going to be there. Be more efficient in the way we're doing it. So essentially, in the long run, it's going to get cheaper. And to me, that's a great thing. Like, access to the legal services is supposed to be for everybody right now. It is, in my view, it's very expensive. Like, if you have more money, you have the more resources you have. This has the potential transform where a law is more accessible to people. People. So that's what drives me then, from the personal side, it's the fact that whatever work you're doing, you're going to start to align more towards what your real personality is and having more fun and really use a creative thinking that we actually naturally have for whatever we're trying to work on. Kelly Schuster-Paredes: That's very cool. And it's something for educators to think about, is that we're in a. We're in a cool spot right now because everybody's in that, oh, crap, we're all using AI mode. So it's like this moment where education is meeting the real world when it comes in with AI. So I think it's something for educators to look towards the places like what you're doing in their work and seeing how you are rolling out AI. I know that a lot of the things that were presented at the conference, I was like, yeah, I told. I was like, tell it T Talking, tapping on my director's shoulders. Like, we do that. I. I know that I got this. She's like, I know, Kelly, I know you have a podcast. No, but it's really important that education goes out and says, just because it's not a school doesn't mean that everyone's not in a place of learning. Because I think we're all in this place of learning and growing, even in the businesses. So I'm going to let Sean take though, maybe if he has any other questions, because we're trying to wrap it up because he has nightly duties that he has to do. Sean Tibor: I just have one final question which is I guess it's two parts. What's one thing technology, tool, platform that you are playing with right now that is blowing your mind, that you think everyone else should check out and see what they're thinking? And then what's one thing that you're keeping an eye on to see how it develops in the future, looking for that you think has potential to be interesting, transformative, etc. So what's one thing now and what's that one thing in the future? Pritesh Patel: One thing now is, really, is just I don't think people know too much. It doesn't seem like people know much. It's like if you look at like the APIs for OpenAI versus what anthropic has and all that, OpenAI is just much more open. I don't think you people realize like you can't really take an anthropic model and fine tune it. They don't really enable that in the API, but in OpenAI they're open about that. You take it and they're trying to make it easier for you to fine tune and stuff. So the opportunities there is really cool. Like you can really customize things. If you have data, you can customize a lot of stuff. And I think that's fun. Even the open source models where you don't have to pay for every transaction call, they're getting so much better where you can run that locally if there's privacy issues and all that, which is what we're doing. It's not really fully aware that people like that API is actually really good. The area that I'm also really interested in seeing where it goes is this whole reinforcement learning piece. Because if you think about these models, they've been trained on all our public information and they've learned. But as humans, how do we learn? We learn from books and stuff, but then we start learning by how we interact with the world. How we interact with the world. So the next phase really is this reinforcement learning where it's actually integrated into workholes and it's learning as it works. So that's the piece that I'm keeping a close eye on. I'm building some things for that that's going to get better and better as well. So the reinforcement learning piece is going to be very interesting. Sean Tibor: It's interesting about applying a well known idea back to something that's brand new and combining those in an interesting way is a pretty powerful idea. Pritesh Patel: Yeah. Kelly Schuster-Paredes: And I'm teaching APIs next week. Pritesh Patel: Oh, that's awesome. Kelly Schuster-Paredes: This is the one unit I always talk about with 8th grade. I never get rid of that in the music one because I. Now, it's totally cool that we say to the kids, like, when you get older and when you're allowed to, you can use an API to bring in your models. And everyone's like, what? What? Yes, but first you have to know how to get the key and value out of. Pritesh Patel: Yeah, exactly. Sean Tibor: Yeah, exactly. Kelly Schuster-Paredes: Cool. Sean Tibor: All right, well, we're kind of out of time. Pritesh. This has been fantastic and just really enjoyed the conversation with you, and I hope we get a chance to do it again. And especially as the world continues to march along and evolve and everything. I think there's plenty to talk about. Pritesh Patel: Absolutely. I appreciate you guys inviting me, Kelly. You're the ones who I spoke to the most at this conference, just talking about stuff. So I was really excited about this. So thank you for inviting me. This is great. Kelly Schuster-Paredes: I love being around smart people. I always try to avoid the room where I'm the smartest one, so I always fill the room with super smart people. I constantly want to learn. So I appreciate you for coming around. That's really great. If people want to reach out to you or if you. They're going to pick your brains or is there a way that they can reach out to you on LinkedIn or. Pritesh Patel: Yeah, yeah, absolutely. Help. You can hit me up on LinkedIn. I can put my email address out there, too. Either way. But LinkedIn is usually the easiest way. Happy to talk to people. Kelly Schuster-Paredes: Perfect. Pritesh Patel: Yeah, sounds great. Sean Tibor: Kelly, any other announcements this week? Kelly Schuster-Paredes: No. I'm going to Poland. I'm going to Poland. I'm going to go to Poland. And going to Poland during the coldest time. Did I say I'm going to Poland? I'm going December 1st and 2nd to the technology Readiness Council. Wolf, we have to get this out before the next month, Sean. Not like, you know, November 20, but Wolfgang invited me to come speak. I am leading a workshop. That's interesting. I'm leading two workshops. One that's taking us through scenarios of looking at policy and what trigger points and talking about how policy is usually written from August 1st or July, whatever, in the summer. And then in the meantime, you have this guy switching models, this one doing something else. This person, teacher found a new shadow AI. And so we don't really evaluate the policy. So it's just getting people to think about what are the triggers. And maybe we should set up a different cadence of how we do policy and then the second one stemmed off of our Fisher Phillips conference of what ifs. So we do have this policy now. What if something illegal happens and what if we blame somebody for cheating and we reprimand them? How does our policy align with how we reinforce things and how the policy is good and how we reinforce those things? So I'm really excited about this. There's a lot of conversations going on about governance, which is a hard and boring conversation for some, but you know me, everything that's boring and requires extra thought I really like. So I like mapping with matplotlib. Alright, so that's what's happening. That's December 1st. So if anyone's interested in TRC, come check it out. Sean Tibor: Sounds good. Kelly Schuster-Paredes: Yeah. Sean Tibor: All right, well, Pritesh, thank you for joining. We're going to wrap up here and sign off. So for teaching Python, this is Sean. Kelly Schuster-Paredes: And this is Kelly signing off. Pritesh Patel: Sam.