Sean Tibor: Hello and welcome to Teaching Python. This is episode 130, and today we're going to be discussing the provocative statement, is coding dead in the world of AI? My name is Sean Tyber. I'm a coder who teaches. Kelly Schuster-Paredes: And my name is Kelly Shuster Peredes. And I'm a teacher who codes. Sean Tibor: Kelly, we've got quite the full house compared to our usual episode. We've got Michael Kennedy here from Talk Python and Python Bytes. Welcome, Michael. Michael Kennedy: Hey, all. I'm excited to be here. Thanks for having me. Sean Tibor: And we've also got Blake Rayfield here, who's a professor at Northern Arizona University and a friend of ours that we met at Pycon last year at the education summit. Welcome, Blake. Blake Rayfield: Hello, everybody. Happy to be here as well. Sean Tibor: Nice brief winds of the week this week and talk a little bit about some good things that have happened. We'll go fast because I know we all want to jump into the conversation. And Blake, I'm going to make you go first this week. Blake Rayfield: Oh, geez. Okay, I'll go two wins. Okay. A personal one. I am a father of a six month old, so we had a whole night of sleep, so that's definitely a win. It really is. And then some of my students won a research competition, so I'll plug them. Go, nau. Sean Tibor: Here. Kelly Schuster-Paredes: That's awesome. Congratulations. Brian Okken: Thank you. Sean Tibor: All right, Michael, over to you wins. Michael Kennedy: My daughter also slept through the night, but she's in high school. The youngest ones, it's a different type of deal. Kelly Schuster-Paredes: You mean she woke up? She woke up. Michael Kennedy: She woke up to go to school. That's the kind of win we get. Blake Rayfield: That's the dream. Michael Kennedy: Yeah, exactly. It is the dream. You'll get there. You'll get there. Yeah. I've had a ton of things I could call out as wins. I'm always super busy, but a couple of my projects that I've been working on for six months came to a close, and I'm going to call those wins. I created two new courses over at talk Python. One building an AI audio app with Python, which is super cool, and another one that had been in the works for a long time. I was just able to finish it up for Python type information as well. So tons of production go into those things, even if they only turn out to be four or 5 hours long. So that's. Those are my wins. Getting those out, for sure. Sean Tibor: I think people rarely understand how much work goes into creative product like that. It's a lot of planning and processing and redoing, and what they see is the final product, but they don't see all the takes in between. Michael Kennedy: Yeah, that's right. It's probably full time. Probably a month to six weeks for a four hour course, something like that. Sean Tibor: Yeah, that's how you make them good. Michael Kennedy: Yeah, I can make it quicker. You can one out quicker if you had a different idea of what a done was. Right? That's right. All right, Kelly. Sean Tibor: Go ahead, Kelly. Kelly Schuster-Paredes: So, well, I'll give my 6th graders a shout out. So my third quarter 6th graders, we actually did a new project for me for the second app. It was Python Turtle. We always use turtle library, but I had kids wanted to do an almost an I stop motion comic strip, so we had some, a lot of little moving parts going around again, shouting out to Steven Grupetta, you gotta love him with this turtle watching all his animations. So the kids were really excited. We had some water molecules evaporating with the sun. We had some lacrosse players catching a ball. It was a fun project, and they pushed the boundaries for 6th grade. And, yeah, it was a good win. They were fun to grade, at least for me. Michael Kennedy: That's awesome. Sean Tibor: All right. For me, it was also sleeping through the night, but not because of a six month old. I'm just rolling off of a two week long intensive project to migrate a bunch of files from several different companies. So we migrated over 3 million files from three different companies over the course of two weekends. And the tricky part was not getting the files from one place to another. The tricky part was taking all of the permissions that people had on their file shares about who could see what and who couldn't see things, and getting them over to our company and remapping all of them. So I got to use some python for a bunch of it. But I also got to learn Powershell on the fly, which was not the most stress free experience. But I got to finally sleep last night. We shipped everything. It's all working. Users are logged in. I wouldn't say it went perfectly, but it was definitely one of those examples of never giving up until it's done. Until it's done right. And we got there. Kelly Schuster-Paredes: Yeah. And then my win on the side is you might answer my emails and my texts. Sean Tibor: Yeah. Poor Kelly's been suffering because I have been totally missing in action for the last two weeks. Kelly Schuster-Paredes: That's funny. Sean Tibor: Why don't we jump right in and talk about this main topic? I saw this a few weeks ago, and Kelly and I were looking at it together, saying, oh, my gosh, have you seen this? Do you think this is real because we've devoted so much of our career to both coding and teaching. And we've got this provocative statement that came out from the CEO of Nvidia that said, coding as we know it is dead and no one will need to code anymore because we now live in a world of AI and that will take care of the coding for us. I thought, we definitely need to talk about this, and we definitely need to get some friends involved in the conversation to talk about this from different angles and talk about, especially given that similar provocative statements about the role of coding in the world have come out over the years, what do we think? Is this true? Is coding really dead? Or do we see a long lived future for code? Kelly Schuster-Paredes: I'll go for that. For you know what? The first shock statement as a teacher I have to add to that is, wow. Oh, but we have 15 years at least. We gotta wait till the colleges catch up because we're still pushing other coding languages as it is in the high school. So for me, I was like, oh, we got a little bit of time to process that, but it's hard to convince a kid if they hear it. Another reason is, why am I in this classroom? Totally. Sean Tibor: So, Michael Blake, did you see that quote also? What did you think? What was your first reaction? Michael Kennedy: Optimistic. It's optimistic, I think, from, I mean, it depends on who you're asking, if they see that as optimism or pessimism. But I think Jensen, the CEO of Nvidia, meant it optimistically is what I would say. So I don't really know what to think. So I just asked. I just asked the mistral LLM whether or not I should still learn programming. In fact, it says yes, it still makes sense to learn programming. And it gave me five reasons. Understanding the fundamentals, creative and problem solving, adaptability, career opportunities and personal growth. When Jensen said that, I feel like I don't know his background, I don't know how much coding he does, and certainly as a CEO, probably not that much. But I get the sense that maybe plays around with code but doesn't have to create applications. Like this thing that you spoke about, Sean, right? This 3 million documents, you move in the mall. And I'm not sure I would want to have that created by an LLM, maybe supported by an LLM or some sort of AI, but just completely out of the blue. Imagine this. Imagine your job was to create that. And you're like, we don't need programmers anymore. So I just asked chat GPT or GPT engineer or whatever it was to make this for me. And so I gave it a really good description, gave it some of the descriptions people told me to make, and then it gave me some code. I ran out one document. It seemed like it worked. So now what? If you don't know programming, how do you assess whether you should run it and destroy 3 million documents of a bunch of companies or not? Sean Tibor: Yeah. And that was actually how it worked. I was learning Powershell on the fly because I was asking chat GPT questions about how to solve problems using Powershell. I knew it in my head from how I would do it in python. How would I do this in Powershell? And so as we went into this, I think we really like. And as we started thinking about this, yes, you could, like, the code itself can be generated, but I can't replace the problem solving approach, and I can't replace the thought process and the iterative approach of, I solved one problem that uncovered another one, and now I have to solve that one. And how does that affect the ones before that? Maybe, for it works really well, but I still think that whether I'm physically typing the code myself or whether I am using an AI to generate some of that, how well can I think through and solve the problem, right? Kelly Schuster-Paredes: I think that's the scary part. And I'm going to let Blake answer this further. It's scarier. It's scarier for teachers because most of our teaching is at a fundamental level, and we try to say that we're doing problem solving and logical thinking, but really, my kids are making. My kids are making little stick figures walk around and holding a ball. Blake, you got a little bit further. Blake Rayfield: That's fundamental, for sure. When I heard the statement and read it, okay, part of me is in a subject discipline, right? And that subject discipline expert tells me, wow, this is really exciting. I can take that statement on face value from a finance perspective. I already know colleagues who are using this, who've never programmed before, and it's just now all of a sudden, they can do great. They can get some things going. I can focus more on the actual content. Right. The finance side from a teacher who codes. I'll say, it is scary. Boy, does it sometimes make mistakes. And I'll go back to both my colleagues and my students. I've seen things that, just like you're saying, you wouldn't want to run that, that kind of thing in production. Yeah, it's scary, but fun. Michael Kennedy: The program says we should buy this company for a billion dollars. Oh, wait, it was wrong, actually. We should have never bought that nothing could go wrong in finance, I'm sure. Blake Rayfield: No, no, nothing ever goes wrong, especially when real money's involved. Sean Tibor: And so that makes me think about this a little bit differently. Between the both of your answers, the provocative statement, the one that makes the headlines, is coding as dead. But the one that might be more real and less newsworthy is coding just got a lot more accessible. Coding became something that, with AI, now so many more people can participate, so many more people can use code as a tool to solve problems or to create things or to be creative. That before might have been very difficult for them to do. Right, but that doesn't sell ads next to online articles. Right. That's not an exciting article to publish. Kelly Schuster-Paredes: But it definitely, I think, totally lowers that sort of entry barrier for the kids. Right? So 6th grade, we have foundation. 7th grade, we start introducing them to a few things. 8th grade, we have a couple of child safe coding places. And, wow, the sky's the limit. I always tell them you're. I'm going to give you something that's completely above your pay grade. It's above my pay grade. And we're going to dream up what we can do with this library. And then the kids start plugging into this generative AI, and it's saying, oh, go download this library. And I said, do you know what that library is? What are you downloading and what are you putting onto your computer? And they're making so many things. We have a guest enter, a guest, surprise guest coming into the show. Blake Rayfield: Kelly, while you're on that, though, I want to jump on this. I think, again, from a teacher perspective, I've always been taught, we got to start the fundamentals. Hey, Brian. Brian Okken: Sorry for being late. Sean Tibor: No worries. We're happy to have you. So, Brian Aachen has just entered the call. Welcome, Brian, from the Python testing podcast. So excited to have you join. It's muscle memory now at this point, Brian, to say, test encode. So I have to overcome that. Brian Okken: It's okay. Thanks. Sean Tibor: I'm glad you're here because we're just talking about how, as we are exploring this world of coding and AI and this idea that coding is dead, what we're also talking about is how coding just became a lot more accessible to everyone, with all the perils and pitfalls of what AI can generate. So, talking about mistakes that AI makes and how do people even know or new programs that people are writing, where they're downloading libraries to solve problems, but they have no idea what a library is or what it does, or whether that code is even safe to run. So you're entering at the perfect time because we would love to have some of your knowledge and expertise around. How do you know that the code that you're running is worth running? And will it work? Brian Okken: Actually, yeah. You ready for me to jump in? Sean Tibor: Do it. Let's go. Brian Okken: Code verification is one of the hardest parts to know if your code is doing what you want it to do. And actually, that's one of the things that a lot of people have asked, tried to figure out if we can have AI generate our test code for us because people don't like to write their test code. And if we're having AI generate the code and the test code, who knows what it's doing, then I don't think that's a good idea. I actually think part of the verification part is just seeing, is the code good? Is it maintainable, good readable code? And I'm actually excited about the ease of which people can go from zero to something working now with the help of AI. And so I'm thinking, like, especially in teaching and teaching new people to get from beginner to expert, maybe instead of teaching them how to get something done, we can focus on how to make, how to get something done well, how to code good, how to write quality code, not just how to get something up on the screen. Kelly Schuster-Paredes: I was, I had, oh God, I had a whole bunch of things. I'm going to let you speak on this because I'm going to, I'm not going to go teacher, ease on this before, Michael, thank goodness. Blake Rayfield: So Brian kind of said this, but as a teacher, we harp on fundamentals so much, right? And of course we know they're important and I think they still are, but there's this like fuzzy area in between learning hello world and syntax versus getting to stick figures running around the screen. And so I think AI just encapsulated that middle fuzzy world and just pushed it all together so we can go from syntax straight to stick figures running around the screen without having all those, that extra time. That's one of my perspectives. Kelly Schuster-Paredes: 100%. It's like I was just talking to 6th graders today and they're coming up to me and all I've taught them today, literally, this is the second day of the quarter we installed new editor. On the first day I showed them print statement, and today I showed them comments, doc, strings, variables and data types. And the kids coming up to me. Yeah. So if I want to ask somebody, and I said, oh, okay, that's like lesson three. Yeah, but then I want to check to see if they've said yes or no. And I'm thinking, you know what? If we could just get them past that level and go straight to the cool stuff? These 6th graders would be creating amazing products right away. But your comment, Brian, triggered a thought. Wouldn't it be cool if they created these programs and then had to figure out how to test them to make sure they, they actually do what they want and then edit it? And it's that context and the content of knowing what they want to create. That, I think, is why teachers teeter between, oh wait, the fundamentals. Oh wait, the creativity. Oh, okay, just throw it off in the air and see what happens. Brian Okken: Especially when I'm teaching around testing, there's a lot of coders out there that can get something done. But when you push back and say, what is it that you're trying to do? What is the code supposed to be doing and how do you verify that it's really doing that? I get a lot of blank stares because people are used to, like, I don't know, looking up some, looking up some code off the Internet and pasting it down. So AI is giving us a more efficient something that can look something off the Internet and paste it. But we don't have, we do. We still need somebody to think about what is it that you're really trying to do? And is that thing really solving the customer problem or your problem? Not just, it did what I told it to do. Sean Tibor: Yeah, I was thinking about that a little bit too with, when Michael was talking earlier about Jensen and maybe how much coding he actually does. And I see this a lot with the world that I'm in, where there's not that many of my colleagues across the entire company that write code. So what they see from tool like mistrol or from chatgbt is I tell it, I want some python code that does something, and I get something that looks like python code. And so it looks good, right? It looks like code, and I think it would work. But how do I know that it's the right code to solve the problem? And that's where some of that experience comes into play, but also the ability to have that process that we want to teach, which is, I have a problem, I have a well defined problem, I have a design to solve that. I implement it, but then I test it and verify it, that it's actually solving the problem that I started off in the first place. How do I know that it actually worked? And maybe that's the missing link here is that cycle that we go through when we work with another coder or an AI model? How do I define the problem really well so that I can solve it? But then how do I know that it was solved when I finished? And I don't think a lot of people know how to do that other than I'm going to just try a few different combinations and see if that works and then I'll walk away. Right, if that. Blake Rayfield: Well, that's where I think we're really tacking on how much coding do we want to learn now versus how much content or subject specific material do we really want to focus on? Brian already said it. If we're serving the customer, we want to know what the customer does. Right? Michael Kennedy: I have a couple analogies, one from programming and one from teaching for you all to think about. We teach kids. I don't teach kids. You all teach kids and other older people, but at a young age you teach them to do math without calculators, right? They're told you cannot use a calculator, you're basically cheating. And yet calculators exist. Despite the comment that maybe many middle school teachers said you'll not, you're not always going to have a calculator. You know what? You pretty much always have a calculator now, because we have phones. Both want to have them understand that, but you would never expect a professional civil engineer to build a bridge and not let them use some kind of computing machine, a calculator, an excel type thing or something like that. That would be insane. And I think this. Do you teach fundamentals versus do you ignore them because of llms existing? Not really. So that brings. My other analogy is I've had people build mobile apps for me for talk, Python for the courses, and that's been a great experience. I've had two rounds of that and they were completely unrelated to each other. Basically. One was a from scratch reimagining, right? And in the end, I have code bases that were created by professional software developers. That is pretty nice. Code is shipped in the App Store and it's still quite inaccessible to me. It's not like I can just dive in there and completely change it. Oh, I've got this idea for that. No, it was built by someone else in a technology that I'm not a super expert in. So I can poke at the edges, but I can't really understand it in a way to make big changes. And I think that is what we'll end up with if everybody believes you don't need to learn coding. You just use llms. What we're going to get is here's a great big pile of a very specific application. You have no idea what it does. You don't have no idea how it was built. It's probably complicated and there's little nuances. It's going to be hard for you to work with. Good luck. And so if you want to quickly come up with something and go, look what I made. This is awesome. I'm here. Look. It does the thing. You're like, wow. It does the thing. We just asked for it. But if it is something that actually matters over time, or like Blake was saying, that answers really matter because you need to understand it. It's not as big of a help as. You can't skip that step. You can't take kind of ownership of that and make it part of your education, part of your job, part of your business, without understanding it. Would you expect people to never write again because you're in literature? So I'm in literature. We just read. We never write. I just read stuff. That's all I do is read. No, that would be crazy. That seems what half of literature was writing about, the stuff you were writing and under. I think that same thing is here. I think you still got to teach people to take it and make it theirs, but you won't get stuck as often and you won't worry about, oh, you really need to learn how to make that regular expression, because someday you're going to need. No, I'm not. I'm just going to ask llm how I do it, and it's going to tell me. Or you need to know this little detail. Like, the little tiny details you get hung up on won't matter, but, like, knowing the language and knowing how to use it in general will still be super important. That's where I feel like we're gonna be. Kelly Schuster-Paredes: As you're talking, I'm thinking of this, and I completely, 100% agree. A couple of things that come to mind, and I think this is one of the issues with AI and teachers. Definitely in a k twelve, believe it or not, k twelve setting. And I'm sure the colleges are handling this as well. It's. You're stuck with kids that aren't ready or aren't mature enough to say, I'm learning this for me. I'm learning this for the. Aha. I'm learning this for the. I remember when I was learning Python, I was like, frustrated, frustrated. And Shawn was like, you just solved it. And I'm like, I just solved it and I'm so excited and I get goosebumps. And then when a kid solves something, you see that aha. It's hard for the kids to really put that struggle in. I always tell them, stop riding the bikes for somebody, stop touching their computer, let them ride their own bike, let them fall and make mistakes. It's hard for them to make mistakes when they have an instantaneous tutor there answering the generic problems that we are giving them in a classroom. And so it's pushing those boundaries. If they have a calculator in math class, would they use it? Heck yeah. You think they're going to scratch out their division problems? No, but because they're forced without it, they're not going to use it. So I'm stuck as a teacher now doing silly things like lockdown browser, where they can't go, the Internet, where before class challenges I was like, google it. You can do anything you want. You just can't talk to somebody. Now you can't even talk to somebody. You can't even google it, guys, no stack overflow for you. You have to do it all by memory. And it's on lockdown browser. And I feel like a fake because people don't code that way. And it's hard. Michael Kennedy: I think they used to in the nineties. That's how we did it. It wasn't that fun, though. Brian Okken: I had reference books that I had bookmarked. Michael Kennedy: Yeah, but that was like a thousand page book and you had to go through it. That's not the same as chat GPT. Sean Tibor: I had the big PHP read book and my sequel and everything, and that's how I learned it. And it's funny cause I, when I talked to my junior engineers, I remind them, like what? Google came out my sophomore year of college before that. And even when it first came out, everything was book based or a course or something like that. We learned it in a different way. That doesn't mean that this is a bad way of learning with all of the resources that you have at your disposal, but it does mean that you have to be very mindful about where you're taking in information from and how you're using it. I give people Brian's course, your courses, Michael, all the time from my team to say, go watch this course. I know it's all video content, and I know that you can go to Coursera or YouTube and you can get all the video content you want. But I also know that Michael has spent the last six weeks building this four hour course. So that it has consistency all the way through, or that when you are going to Brian's course, you're going to see the content that he produces and it's going to match what's in his book because he's thought about this from end to end. And I, one of the things that I'm a little bit concerned with the AA model is I can't sometimes get it to reproduce the same thought process twice in a row. So it's not the same as having that curated, thoughtful content that helps them get there. And some of this focus that we need to be able to help them really learn things well and learn them deeply and get to that quality code that you talked about where it's like really good stuff and thought out, like we need some way of making it, having a through line on there, like a thread that connects the beginning to the end of what they're learning. Blake Rayfield: Sean, you mentioned something. What about learning in the age of llms? What does that look like now? Sean Tibor: It's a great question, right? Is that I know that I use it to learn. Other, my other developers use it to learn. I know students are using it to learn, right. And it's, it reminds me a little bit like of the early days of Wikipedia where it was like teachers would say, oh no, you can't use Wikipedia because it's not an authentic source or it's not authorized. It became one of the best sources. And yes, there's still some bias there and there's some, maybe some issues, but it's a far better resource than what came before. And as these, as learning evolves with the AI models, I think we're going to get to a place where we find the right place for it in the process. It doesn't replace the thoughtful work that we're doing to develop courses, whether that's online or in a classroom, and having that connective tissue with all of it. But maybe it has a place where it acts as that AI tutor or it helps smush that fuzzy area between, I know some fundamentals to I'm solving real problems. There's a place for it here. But I think we're all struggling with figuring out what is that role in the learning process. Kelly Schuster-Paredes: It's funny because I shifted an assignment, the evolution of my course this year in 8th grade. 8th grade, they, we play a little bit with libraries, do a little bit of requests from going off of your weather request outfit thing, which is great, and then we do a little bit of matplotlib and a little bit talking about data science and big data. So this quarter that just passed, I said, you know what? We're going to bring in beautiful soup. I told the kids, go use our child safe. It's a kid friendly kind of chat sheep tea that we house our own data in a, in our own storage. And I said, go learn everything you can about beautiful soup. We're having a quiz tomorrow. They said, what? And I went to chat GBT. And I said, I need a good code that scrapes this website about kid jokes. And it comes up with all these classes that kids can't even understand because they don't have that knowledge yet. So I'm like, make it simpler. And it kept coming back with other things and make it basic and finally pared down to a really nice code that looked spaghetti like, but it had nothing but conditions, a couple of functions, and I gave it to the kids the next day. And surprisingly enough, they were able to interpret the code. So it was a win and it was a different skill. Would they have been able to code with beautiful, beautiful soup library? Not on their own, probably not very well with documentation because they had to sift through some of the tags in order to get that joke out. But the fact that they could understand the code, it felt like a win. And I think when we start using AI, maybe even as the developers, that's a huge skill that's shifted, being able to read the code, understand what it's doing and knowing where to fix it and where to tweak it. Brian Okken: Yeah, that's where I actually want education to go more towards is how to evaluate code and fix it, because that's actually more of the job. There's very little of my day to day job that's generating new code. It's maintaining, it's adding features, it's fixing bugs, it's porting to a new, new library or something. And all of that is I. Maybe you can get a llm to do that, but I don't know how. And there's like things like just today I looked up, there's two pack, two packages I use. One use case was broken if you used this particular version of this package with this particular API of that package. And I'm trying to find out what the current best workaround for that problem is. That's going to be time limited. It's going to change when the version gets fixed. So how do you get that into an LLM? I don't think it's going to be there, but I want to jump on Michael's calculator thing a little bit, because I actually think that we should stop teaching all the basics in math, and I think we should start teaching people what to do with math and instead of how to do long division and integrals and stuff. And I think the same could be done with software, is instead of teaching people the basics so much, be able to tell good writing from bad writing. But we don't need everybody to be a novelist, and we don't need everybody to be able to code a website from scratch, but we need more people to be able to use software to solve more problems. And I think that, and even in different realms. So, like, I might be really good in communication systems, but if I want to help out my local restaurant with their website or something, I'd like to be able to get up to speed quickly on that so I can help somebody out, even though I'm not trained in that field or somebody's finances or something else. And I think if we've got more people solving more problems with the help of AI, that would be awesome. But I think we can only get there if we try to tell people how to use software to solve problems, not just how to build software. Sean Tibor: I saw that a bunch with statistics, right? So when you're teaching statistics traditionally and visualizing data, a lot of that is done by hand plotting. Right. In traditional education, it's often once you get to the college level that they're worried, introduce a software package to you to do this. But the idea of using software to help visualize things faster. Here's a data set. Let's look at the distribution of that. Okay, that's three lines of code with Matplotlib to get it from a data frame into a visualization, suddenly that process of engaging and interacting with the data becomes much more enjoyable, because I'm not wasting time hand plotting things out. I'm focusing on the subject matter to Blake's point that I actually care about. How does this move around? And so I like the metaphor of the calculator here, because if we do this well, hopefully we get rid of the boring parts that are repetitive and time consuming and remove from the actual insights that that students can get when they see how what they're doing has an outcome or a relationship, or they can make a better connection when the code is shortening the gap from. Here's where I started to. Here's something I can actually respond to. Michael Kennedy: Yeah, I want to follow up with something on. Kelly said. You know how it. The kids have a really hard time motivating themselves to say power through the details until they can get to that aha moment. And I think a lot of academic education, college for sure, also quite a bit of k through twelve at least high school stuff is taught backwards. We're going to start with the most fine grain detail and we're going to build up sediment layers until you get to the very top where it's actually interesting. And that's the thing you actually want. Why can you do that? Because you have a captive audience for at least a semester. So you've got 16 weeks, they can't bail. And by the end to 16 weeks, you end up with something pretty cool. But I think you should have those wins and those interesting moments really early and go, dang, this is cool. I want to know more about that. Oh, you want to know more? Let's talk about the details. Let's dive in and let's. I think it almost should be taught in reverse and I think programming and llms can help that happen because you can see these cool problems and explore them and then you can like, okay, let's get into the theory and let's get into what the steps are. And like, why did the LLM have you do that? And I think that would inspire a lot of kids to have a genuine interest rather than a grade based interest. Kelly Schuster-Paredes: 100%. Sorry, Blake was like, yes, I was. Blake Rayfield: Going to say sometimes they bail, sometimes they're physically there, but they definitely bailed. Michael Kennedy: Yeah, yeah, no, but having like online courses, they can literally just leave and never come back versus they're at least required at least through high school to physically be present even if they're not trying to engage. Kelly Schuster-Paredes: I think, and I always tell cause the english teachers are having the heck of a time with AI and it's just killing them because blah, blah, blah. But there's always those two ends of the spectrum, right? You have the kid that would never use AI because that kid loves to hack through and try to solve the problem. And then you have the kid that's always going to find a way to cheat, whether it's with AI, some tutor doing their homework, or copying off of someone else. So I always throw those two anomalies out. But I do agree that hooking them in to get the fun and to do really cool is a huge game changer because they always find out what could I do and how can I do that? I don't know. Go ask Chad. GPT, go figure it out is a fun question right now. And I think that's the one thing that teachers are missing out on those that are afraid of using AI. They're missing out on harvesting that creative mindset from these kids. I had a kid who does not code well at all, but so creative, he. His first project was to take all these basic concepts that we learned in 6th and 7th, that. Here's the list. It's on a rubric. You have to have these components in your code and go make something that's important to you. Give you one guess what this kid did. It's a boy. He made a Fortnite tracker connect, collecting all of his whatever shots and kills and who he killed and what skins and how much money. And he was like, and every time, and, look, I can save it as a file on my computer. I'm like, do you know where that file is? He's like, no clue, but he made it. So then we got into the whole conversation about pass and where this is getting saved locally on this computer, and how do you find it and how do you export it, whatever. And it was a learning moment that an 8th grader would never have had in computer science. So it does push those boundaries. Sean Tibor: Blake, are you seeing that at the college level, too? Blake Rayfield: I am. First off, the origin story is always video games, I think, especially when it comes to programming. But I think they really. I like the calculator example. I know that's already come off, and we debated whether or not that's great or not, but programming is a tool, right? So, like, when I'm teaching finance and things like that, I want them to be able to use the tool. I happen to teach in the backwards method, where it's, let's do something fun with finance, and then we can figure out the details later. And, yeah, I want them to be able to use the tool, and they're definitely able to go a lot farther than I've ever seen before. I'm able to give them a question or give them something and just say, hey, go figure it out. Doesn't always come back. Michael Kennedy: Right? Blake Rayfield: But it never did in the past, either. Now it just goes a lot farther. Brian Okken: I actually don't. Don't know why we keep. I don't. I haven't seen a math classroom in. In decades, but look the same. Kelly Schuster-Paredes: Exactly the same. Unfortunately. Unfortunately. Unfortunately. Brian Okken: Okay. Do we still have kids plotting functions by. By hand? Kelly Schuster-Paredes: Yes. Brian Okken: Like, why? Nobody does that in the real world? Kelly Schuster-Paredes: Yes. Michael Kennedy: Okay, Brian, we've done it for 600 years. There's no time to change. Kelly Schuster-Paredes: I need to keep my job, Brian. Yes. Sean Tibor: And don't get me wrong, there are some cases where you would want to show, like, the trivial example, we're going to plot a few things on a graph, or we're going to do the things so that you can slow down the learning process and see what's happening. I love running things through Python tutor from philip guo because you can step through it one by one, step at a time, and see how the process is happening, how the algorithm is being executed or the list is being built. There's value to that. But then again, we have tools that let us do that one step at a time instead of having to draw, bring out graph paper and a ruler and plot it on an axe, right? We can do that same sort of thing in other ways. Blake Rayfield: This was really the promise of python, right? It's simple, it's easy to read. We're supposed to be able to debug it. We're supposed to be able to go from, like, start to something cool really fast. So now you can write Python using an LLM even easier, right? I don't know. I think that's really neat. I think llms did python better. Maybe that's. That might be a bombshell, but I. Sean Tibor: Do feel like we're repeating a conversation that might have happened ten years ago with, like, low code, no code solutions, right? Oh, you don't need to code anymore because you can just drag blocks onto the screen and create your program that way. Or when we went to the first higher order programming languages, right? And all the people that were writing Cobol and everything said, oh, it'll never work because it's too abstract and can't trust it. Michael Kennedy: You can't trust it, Sean. Sean Tibor: You don't know it's going to work until you hand code the registers yourself. So I do feel like we're repeating a conversation here, but what feels different this time is that maybe those earlier conversations were like, you know, we ten x'ed each one of them. We got ten times more productive by going from lower level coding languages to higher level coding languages. Or we made it more accessible when it became no code. But this feels like we just went 1000 x, right? We just spread it way out to the point where now so many people can use an LLM to write code and you can write so much more with it. That, to me, feels like the different part is like, the scale of this feels a lot larger and a lot more accessible. And I know that's where I'm grappling with it personally. Kelly Schuster-Paredes: And I can't help but think of the Netflix movie. What does it leave the world behind and our over reliance not only on, on all tech, but now on this AI. And I don't know if you've seen this movie, but the whole premise of the movie is all of a sudden the satellites have fallen out of the sky. There's nothing. TVs, am Fm, don't even work. Teslas are driving and crashing into each other. All of a sudden the world is going to self implode because they have no tech, no tv, and they don't even. The guy gets lost going back to the airport because he has no clue how to drive there because he's been relying on maps to get to where he's going. And I see this already. Within a year, every time chat GPT goes out and it's like this sorry, server not working thing, I'm like, God, okay, Gemini, please work for me. I need something to go. So this over reliance already on AI, is incredible. Within a year and a half. So I feel like somehow we have to have this balanced approach of taking a step back, using our abacus back in the math class, and figuring out how to solve problems. But what are you going to do? Michael Kennedy: I can't believe you called it. I'm moving to Alaska next month. That has to live in the woods now. Sean Tibor: I'll have some advice for you, Michael. Where to live, where not to live. Michael Kennedy: Yeah, maybe I'm not doing maybe a vacation, not living there. Brian Okken: It. Michael Kennedy: I think you're right, Sean. I think it is a thousand x very easily. Brian Okken: It's crazy. Michael Kennedy: I had a guy on the talk python show, and his name is Fred. Awesome guy, our age, and he'd been in music, like producer type stuff, and he had been in recording studios and in marketing. And a year and a half ago, he decided he wanted to learn coding. And now he didn't really learn coding, I wouldn't say. But he learned to control computers. So he built what is a functioning website, 100% with chat CPT. It does a whole bunch of cool AI stuff. It has e commerce built in, it has customers, it's collecting money. He was on the show. We were talking about stripe. I told him about some other payment processor the next week. He's like, yeah, I swapped out the payment processor and I asked him, like, are you learning the code? As he's. I just keep asking chat GPT, and he just keeps giving me answers. Really? Yeah, no, I just keep giving it this stuff until it works, and it just keeps giving it back. That's not like someone gave you a calculator, so you just all of a sudden whoops, I built a bridge. You still, there's a lot of steps and I have a calculator. I just can do it quicker. And this is something else. Blake Rayfield: Depends on how many times. Michael Kennedy: Yeah. Brian Okken: And we can have like, individuals solving problems that wouldn't, that aren't financially viable to have one person spend time solving it, or it would take ten people before. And you can't build a company that's viable because it doesn't make that much. There's a lot of problems out there that can be solved that wouldn't be solved before because it was just too expensive and now it's cheaper and that's awesome. The part that can frustrate to me is the, I don't know if we're going to get to something that's like good code faster if people don't know how to read it, because, because if I get a problem that sort of works, how do I determine the difference between works and actually works? And one of the videos I saw, like, yeah, the system, like, generated code all by itself and it's 18% of the time it works. Like 18% of the time. There's no way I could get hired as a software developer if 18% of the code I wrote worked. Just. That's terrible. I need 80% or 98% to work closer to 100%. How do we get there? And if you don't know, how do you remember when you could read HTML? A lot of stuff that's generating HTML now and we can't read it. And there's a lot of people pushing to try to make HTML readable again because you can swap out tools. It's a, you should be able to, it was designed to be readable, but we use it in a non readable fashion. And I hope that python doesn't become that, like that the python code out there or all the c code or whatever is just stuff that's generated by a computer and nobody can read anymore. I think that's where we're going to be gridlocked. And when chat GPT goes down, we can't do anything with it. I don't want to get there. Sean Tibor: It's interesting. Does it need to generate human readable language? If it doesn't, why can't it just generate the bytecode? I think that's the same thing. What's the point? Point of writing code from an AI, unless people are going to read it or be able to understand it or be able to diagnose what's going on, if all you're doing is generating code. Do we really even need languages anymore? Blake Rayfield: That's the real question, I think. Or is Python the right question or the right language? Kelly Schuster-Paredes: I know that was just. Yeah, no, I was just thinking that right now is so can you imagine for me actually, if it started spewing out whatever c or c sharp, and I can try look at JavaScript. Okay, no problem. And it would have to increase with do Python because most people can actually read it. And it has all the nice little. I don't know how much chat GBT code you've produced. It's got pep. Eight rules, really nice and neat and their variables are so well defined and all the comments in there. So it almost feels like AI knows that Python has to be the language of the people who don't know how to code. I don't know unless there's a new language coming. Brian Okken: One of the things I want. Sorry. Michael Kennedy: Go, Brian. Brian Okken: Oh, one of the things I didn't know. I don't know if we've already covered this, but I wanted to make sure we hit was there's thoughts in my mind and others is to like, should we, if you're in high school right now or entering college or in your college, should you consider a CS degree? Is programming a decent thing to learn for the future? And one of the, because of all of this, and I can't remember the name, person's name that brought this up recently, said, really, coding is a creative act and it takes, you have to examine stuff and you have to detail oriented and it takes a lot of creativity and analytics. Any other job. If once software is not something that we have to hire people for anymore, there aren't any other careers that are safe. The skills involved in needing to write software match the skills in everything. In writing copy and doing art, in teaching what's. What is left? I don't know if there's anything left. If we can take out the software. Michael Kennedy: People, I think there's manual labor left. I built a, I did roughing construction for houses one summer. Yeah, I would have loved to ask chat GPT to do that. Brian Okken: I don't know. You don't think we're going to have robots that can do that for us? Kelly Schuster-Paredes: Robots can't really climb a roof. Really. I think maybe. Michael Kennedy: Maybe they could, yeah. Another key important part I learned about building houses is have the senior people insult you constantly. And I'm not sure robots are good at that. Blake Rayfield: Chatty Bt is good. Kelly Schuster-Paredes: That was a fun, that was a funny thing because Jensen Young, Navidia's CEO, he said, if I was telling college students right now what their major should be, he said, sciences, which I find really interesting because we're always going to have some need to discover and find something new that has not been yet trained in an AI makes sense to me. You're on the, you're going to be the discoverer. You need to have that person that is going to be looking, persevering, trying to find what's not known. And maybe it's only in the sciences. Sean Tibor: But I guess another question is, do we feel like everything is known in computer science? Because I don't, right? That's the exciting thing for me. I feel like there's always new things to learn and create and discover. I don't think we've solved every algorithm or every problem problem out there. And if we're expecting AI to do that, I think that's a completely wrong assumption. Brian Okken: Right. Sean Tibor: There are new frontiers all the time in computing that get opened up. The fact that we're even talking about AI as a replacement for coding is a great example of that, because it wasn't AI that created AI. Computer scientists created AI. Jensen's company created AI. Right. I still feel like there is a lot of future ahead of us when it comes to solving problems with code, because there's a lot of problems out there that can be solved, and not all of them are going to come from an LLM generating what it's learned from other people's code. Michael Kennedy: I think the definition of what to be a coder changes a little bit. Kelly spoke earlier about having the kids learn beautiful soup and requests and using these things. And you could make a real argument that you shouldn't use all of these half a million really handy libraries to build your python apps. You should learn how to write that stuff from scratch. Like how many people do job interviews where they ask them to sort something? Like, I'll tell you, I've never in my life had to manually sort a thing, right? That goes, you call a function, right? You work with some libraries, and I think it's like that. But another step change. Do I need to really think about, okay, here's how I create the structure to safely open a file and read the text. Or do I just ask the LLC, I want to open this file and read it in this format. Boom, here you go. Okay, so my building blocks become bigger pieces, like working with the libraries, then working with line by line, character by character. I'd start, you might need to start to think in, like, larger utility chunks. That you click together. More lego, less sewing. I don't know, let's fine grain stuff. Right? Exactly. Exactly. More duplo. It's going to be the huge Duplos that like the adult Duplos. Well, and I think it's going to be something like that, but that's still a skill that is not that different. Sean Tibor: Yeah, and you made me think of this. Look, there's a lot of programming jobs that suck, right? They're just, they're not interesting. They're not solving something that is incredibly valuable, but they still need to be done because we don't know any other way of doing them. Michael Kennedy: This is. Sean Tibor: I have a lot of data that needs to be cleaned and it's not particularly interesting data. I don't really care about it, but it needs to be done and they're paying me to do it right. Can we free up the people that are tied up in those sorts of jobs to go do things that are more interesting, more higher level stuff that is more valuable to themselves, even they find more intrinsic value in it. That to me would be a big win of AI. But in order to get there, to Brian's point, we still need to do that work and we need to make sure that it's done right. How do we get there right? Brian Okken: It's a bummer that a lot of the companies aren't doing that, though. They're not shifting people to do more interesting problems, they're just letting them go. Blake Rayfield: I wish you had said finance. Kelly Schuster-Paredes: So I have. So I have a question for all of you. As a k twelve teacher person generates code, complete creative idea, 100% could have not. It could have been something they've seen, but they've generated, they spent the time, they've manipulated chat GPT. They said, I wanted to do this, I want it to do x. Is it theirs or is it plagiarized? Michael Kennedy: That's a deep question. Brian Okken: I don't think the answer has changed. Blake Rayfield: I'll jump first. Okay, so putting on my like, university assessment hat, right? I don't like it because now I have to try to judge intent or like motivation or even trying. Right? Do I grade it? Is it theirs? I don't know. How much did they make from the LLM? How much not. I tend to grade on better ideas already anyway. Before LLM, I don't know. I would give them credit if it was really something cool. Brian Okken: We've had to like in job interview stuff, we've had to have code samples and have, and try to determine if somebody actually did it. Or if they had somebody else do it for them. And the best way is to have them talk about it, ask them questions about it, and, and if they can't describe what the code is doing, then they probably didn't write it. Michael Kennedy: I like that answer a lot, Brian. I totally agree with that. Kelly Schuster-Paredes: I'll let Sean say his thought, and then I want to. I'll give you some examples. Sean Tibor: I was thinking about it more from a business perspective. Right. So you start a business. The guy who makes his website, he asked chat, GPT to generate all the code for it. Is it his intellectual property? Yeah, it is, because he. Michael Kennedy: But it gets, it gets interesting, because so much of the code that it was trained on for the ability to write that was GPL. So maybe it's his, but maybe it is subject to other licenses and legal constraints that. So do these ideas flow? Are they conveyed through GPT and other AI's? If the stuff that went in has these restrictions, the stuff that comes out, is it just laundering? Like money laundering, but for rights and legal constraints and stuff, or is it not? I honestly, I think we're gonna have to. We're gonna see something at the Supreme Court at some point. Somebody saying, that billion dollar thing you made, that was off of my hobby project. I own that. Blake Rayfield: Yeah. Michael Kennedy: Copyright on. Blake Rayfield: Then we get to the question of, like, how much is distilled from books and ourselves. Right. I read a book. It inspires me to do something. That's a hard question. It's a totally different question. Michael Kennedy: Yeah. The people are on the side that it's okay. I think they quote, like, fair use. Like, I read a book or I looked at a picture and I drew another picture. I didn't steal the picture. I was inspired by the picture. So can I. Inspired? We'll see. Sean Tibor: Legally speaking, what's interesting is that I've been investigating using AWS codewhisper, which is AWS's version of this. And one of the features that it has is it will warn you if the code that you're writing looks like open source code that it was trained on. So it actually helps you avoid some of those pitfalls. Right. So maybe there's some of this that could be built into the tools that we have that say, hey, the code that you're writing looks a lot like this stuff that was already out there or is plagiarized. Maybe you should look at this differently or try to solve it yourself. Kelly Schuster-Paredes: But what about collective intelligence? So if a student turns it in and says, listen, this is my idea with chat, GPT code. It's just like having a tutor or using stack overflow. How many people cite the stack overflow website saying, I got this from Johnny off of stack overflow, these 20 lines of codes I used. Sean Tibor: And that's a little bit different than plagiarism versus intellectual property and licensing, because the expectation is if you're publishing it on stack, you are making it available for other people to use to solve their problems. Kelly Schuster-Paredes: So, and they say that they got most of their data, a lot of their data from open source GitHub and stack overflow. That was where I read somewhere. I don't know how legit that is. Michael Kennedy: I think you're probably on the plagiarism thing. Blake Rayfield: I think it really comes down to what you're trying to get the student to learn, right? Are you really just trying to get them, if you're trying to get them to use a tool coding to do something in a domain, right, like finance or biology or whatever else it might be, if they come up with something neat, even if they use an LLM, it's hard not to say that is an achievement. Michael Kennedy: So it's interesting. Well, I saw someone getting a sorry really quick, saw somebody get either an f or some kind of administrative punishment at a universe, a high respected university in the US for using Grammarly because hey, they're using AI on this paper they wrote pretty big stretch. Kelly Schuster-Paredes: There's a lot of websites and educator websites that say, and they're, when you submit an article, have you used any AI? And I find that kind of funny because we've been using spell checkers for a long time and you don't want any of my papers going public. Sean probably laughed at them all the time. My papers are gotten great now because I flipping through Grammarly, I've got. The Grammarly has even gotten better. I don't know, it's hard. I do think, though, going back to interview question that you were saying, brian, people can remarkably, eventually figure out what's happening in the code, because I can dump a code in and have it explain every single line by line. And that is one of the things that the kids have learned to do, because I tell them, okay, you can make this in chat GBT, but you're standing up tomorrow and you're going to explain. Explain it. Show what it does. Tell me what's happening line by line. The hardest part for them to explain, though, is when they are reassigning a function that has a return value, they get lost. Because sometimes chat DBT does these funny things where it starts doing this convoluted renaming, reassigning, pointing, whatever. And the kids just go, this is like calling the function. I said, no, it's not. Brian Okken: So one of the things I'd really love to see show up in schools in, at least to the college level, but probably high school also is not write some new code, but that's going to be there. But here's some code. Here's the defect report we got. Here's an english description of what the problem is. Can you generate a test or a series of tests that reproduces the problem and then fix the problem? That's 80% of my job, and I never, we never taught, didn't get taught that even through a master's degree. We didn't cover that stuff. But it needs to be covered because that's what the real work is, changing existing code. So being able to look through somebody else's code and find out what the. Blake Rayfield: Interesting chat is, really good at that as well. Right. So if you've ever taken some code and said, oh, hey, this isn't giving me the output I expect, or this is giving me this error, I'm sure your students have used it. It's really good at that, actually. Brian Okken: Okay, I should use it more. Kelly Schuster-Paredes: 100%. You have to flip back and forth, though, between Gemini co pilot, and you. Blake Rayfield: Have to use the whole, yeah, I wouldn't be agnostic. Sean Tibor: Yes, I do feel that skill and that approach does cross disciplines, though. And I think it's a really interesting thing to do to demonstrate understanding is that context of, hey, here's something that is we think is wrong or we know is wrong. What's wrong with it? Why? How would you fix that? How would you approach it? And I find that I did that a lot in teaching. I think they treated me as like a human debugger. Why isn't my code running? And it's often because they're missing, like a parentheses. But I got really good at spotting that when I'm doing code reviews for other engineers right now, most of my job is reviewing this. And even if it does work, am I, are they being complete and thorough and everything? Are we being secure in the way we approach this? And I could see that in finance as well. What's wrong with this analysis? Or is this model correct or in biology? So taking that same sort of approach that you brought up, Brian, around, how do we think about the problems that we're trying to solve? Some of the problems are analyzing other things that are not working right or have a flaw, or there's an issue with it and being able to see beyond what chat GPTCs, right? And apply our experience, our context, all of our, even our intuition and emotion to it, to say, is this, how could we solve this? Maybe the code is beautiful, but it's biased, right? Like it has a significant bias built into it. And only we would know that as humans, because of our context and culture and background and everything. Could you teach a large language model to spot racial or gender bias in code? Is that even possible? Kelly Schuster-Paredes: So that leads me to a question again for all of you guys. Most valuable skill for future developers, big problem solving. Michael Kennedy: Learning to solve problems precisely with code, right? Not just kind of we'll talk about, but concrete steps, almost like you would in geometry, right? You have your axioms, and these are the four steps I took to get to the answer. That is, say yes, these angles are equal to right. The building blocks will be bigger, but the thinking process is not that different. Okay, what are the core things going on? What is independent? What comes before the other thing? What can I do first? And just that kind of thinking, I think that's, I think that's transferable. And those are the kinds of things you might ask chat GPT about. Give me this section. I don't know this. What's this step here? I can't forget. But knowing that even there is a step is the first step in solving it. Brian Okken: Most important skill for a software developer is communicating with people, listening to people, paying attention to their emotional state. That I know it's not a hard skill, but it's a very hard skill to communicate well and listen to people. And it's one I'm still trying to get better at. Blake Rayfield: I like all of theirs, but I'm going to go with creativity. Okay, so it's something that's always been in demand, but, and I'm going to go to domain specific creativity as well. So we know these GPT models, they can really help you save time on the development side, as Python really did. So I think something for future developers is really to use some of that, save time to go deeper into some of these other areas, whether it be biology, hopefully it's finance, or whether it be one of these other fields that are specific. I really think getting creative in those areas and that way you can push the code a little further and things like that, that would be really valuable. Sean Tibor: Sean, I think it really is about getting, it's still about getting really good at learning, right. Learning new things, learning new areas. The quote that always comes to mind, to me, or the axiom is that what makes me an expert is being really good at being a beginner and learning things quickly. And I think if you can do that, you can accomplish pretty much any task that you want to tackle. Kelly Schuster-Paredes: Yeah. And I guess for me to add to the list of things that are already amazing that you guys said, believe it or not, I'm going to say efficiency, because you can be efficient if you think and you spend a couple of moments just trying to problem solve before you go in and say, code this for me in chat GBT. Because if you don't even think about how, what you want, where do you want to go, how you're going to test it, you're going to spend a lot of time going, no, that's not what I want. No, that's not what I want. No. But I want this. And I know that error is not working and you've end up going down a rabbit hole of really bad generative AI code, and you're not being efficient, even though you think you are, because it's writing code for you. So for me, being able to use all the tools to make yourself a more efficient person is going to be the win win situation for the kids of the future workers. Brian Okken: So those are awesome answers. I'm glad we planned that ahead of time. Kelly Schuster-Paredes: I know. I mean, we're approaching 720. We're like on a roll here. Sean Tibor: John, I have to say, I think we've hit a lot of great topics today, and obviously there's more we could discuss and a lot more we could go into, but I'm already thinking about how much I'm going to have to do post production on this and try to clean it up a little bit. But honestly, this has been a really great discussion, and I'm glad that you were all able to join us and give your thoughts to this. And I know that as it evolves, we'll continue to work with it and figure it out and find ways to apply these new tools to the work that we do. So thank you all for joining us today. I wanted to just give a shout out to Blake at NAU. If you're looking at colleges, programs where you want to learn finance, consider Nau. It's a great place to be, and you'd get a chance to learn from Blake and learn with Blake. So big shout out there, of course, for michael and Brian. Each with your respective podcasts. Talk Python Python testing, and then, of course, your Python Bytes podcast that you do together always makes a rotation through my car's audio speakers as I'm driving the kids to and from school. Books courses just can't plug you guys all enough for the work that you're doing to help. Help all of us learn. And I know that I have been very appreciative. Brian, my copy of Python testing with Pytest is right here on my bookshelf, so I can grab it whenever I need it. Brian Okken: Awesome. Kelly Schuster-Paredes: And you can't forget to plug Pycon and educational summit. You guys are going to make an appearance and say hi to all these educators. Are you flying in early enough? Thursday, May 6? Michael Kennedy: I will be there Thursday, but yeah, I'll be there Thursday. I'll try to drop by if it's on Thursday. Kelly Schuster-Paredes: It is on Thursday, and it's all day on Thursday. In the morning. Python anywhere, python everywhere kind of theme. So sky's the limit. It's going to be a great summit. I think we're going to have a big turnout. It's going to be praying for a bigger turnout than the past. And in the afternoon, we're going to have some sort of birds of the feather little workshops. I know we have some good friends from anaconda going to come talk about Pyscript and one of the workshops in the afternoon. And so it's just exciting. So that was a little bit of hint to what's going to happen. Michael Kennedy: Excellent. You all should do some open spaces around python and education as well. Kelly Schuster-Paredes: We do. We're going to have a whole bunch of open space at the time. Michael Kennedy: Awesome. Kelly Schuster-Paredes: Maybe a gathering in the evening. Those are always fun. Sean Tibor: So I guess that wraps us up for this week. If you have questions, thoughts, comments, you can always reach us on social media. We're still on Twitter or X or whatever it's called at teaching Python. But we also love getting emails from our listeners. You can go and send us a note through our website at Teachingpython FM, and that should do it for this week. Kelly, am I forgetting anything else? Kelly Schuster-Paredes: That's it. You can check us out on LinkedIn. I'm on a goal. Sean Tibor: That's right. Kelly is building our LinkedIn community. So if you prefer LinkedIn, go over there, join the fun. She's posting all the time, and we'd love to have you join us over there. Kelly Schuster-Paredes: Cool. Sean Tibor: All right, so for teaching Python, this. Kelly Schuster-Paredes: Is Sean and this is Kelly signing off.