Brent Warner 0:08
This is the HigherEdTech Podcast season four episode 17. A Student Perspective on ChatGPT with Austin Wagner.

Tim Van Norman 0:28
Welcome to today’s HigherEdTech Podcast. I’m Tim Van Norman, the instructional technologist here at Irvine Valley College.

Brent Warner 0:35
And I’m Brent Warner, Professor of ESL here at IVC. We both enjoy integrating technology into the classroom, which is what this show is all about.

Tim Van Norman 0:42
Welcome. We’re glad you’re here with us. So spring breaks over. And now back to work.

Brent Warner 0:50
Was there was there a break, Tim? Really?

Tim Van Norman 0:53
Hey, I worked. I got one day off. Yes. But I worked. Actually, I loved it. I actually got to get into some other tools that we have, but haven’t really been implementing or have access to or some or might have access to. And so it was kind of neat to actually have the time and not worry about emergency phone calls. Yeah.

Topics for future shows, too. So that’s good.

Brent Warner 1:18
Oh, it’s gonna be a lot of different things going on.

Cool. Yeah, so I was at the CUE conference. And it was great. I did, I did a couple sessions on chat. GPT. So we’re continuing that conversation a little bit more today. packed out people spilling out into the hallways, we waiting, just hungry for information about it, right. And so the conversation is just going and going and going. And so it was really great. There weren’t a lot of sessions because the way that the application door closed at a weird time and before chat, GPT was announced, and then everything happened. And it’s a teacher’s conference. But But anyways, we figured it out, got a got a session in. And there were a few sessions, but every single one and in all the hallways, like AI is just the conversation everywhere in education right now. So so we will, we’re gonna continue that conversation. And today is part of that too. So I think we’re gonna is gonna have like fairly regular sessions throughout throughout the seasons in the shows for the foreseeable future, Tim, because there’s just so much to continue to explore here. We’re not trying to make every single show about that. But we could easily lead to seeing how much and how quickly things are coming out. That is not really what we’re doing ultimately, with the show, but we will have them on a semi regular basis, kind of talking about what’s going on with AI. And so today, we’re gonna jump over and get into our interview. So let’s jump over now. All right, we are here with IVC student Austin Wagner. Austin. Welcome and joining us.

Austin Wagner 3:02
Hi, thanks for having me on the show. I’m glad I got a chance to talk with you guys about AI. I mean, I’m really excited talking about ChatGPT does in general, and because of the uses it provides for students and for everyone in general.

Brent Warner 3:14
Yeah, we’re excited to have you here. We wanted to get a student perspective on things. And then of course, you and I were introduced by Professor Liu, Emily Liu, who kind of sent you essentially in my direction, because she knew that I was talking about this stuff. And you were talking about this stuff. And then we’ve had a couple of really good conversations, some insights, you’ve opened my eyes to some of the thinking around understanding chat, GBT and all of these things. Really a lot of like, how to prompt it. And we’ll talk about that a little bit later. But there are there are quite a few parts to this. So let’s get let’s get started in an Austin, can you kind of introduce yourself and tell us a little bit about you so that we kind of know where your perspective is and where you’re coming from?

Austin Wagner 4:00
Yeah, I’d love to. I’m a student at IVC. Right now, I plan on transferring UCI for Language Sciences. So that’s combination of like linguistics and computer science. Because I have a big interest in just the language and how it evolves and the AI aspect of it. Now it’s learning from us and how we’re teaching it these languages and how to respond was better and how to communicate. So I have a lot of passion with AI just in general chat GBT and how to use it for my my own writing but also how to present it to people so that they can use it to better themselves and reach these new goals reach better writing better understanding, learn about how to process applications outside of school, but just in real life, how to apply for a business loan, how to apply for, you know, citizenship, or food stamps, even just the end. The possibilities are endless, which activities I really love bringing into people and talking to them about it and encouraging him to use it.

Tim Van Norman 4:51
Excellent. So as you’re, as you’re looking at this, you mentioned a couple of times using it. Are you finding uses for it within the within your classes? Are you taking classes that you’re finding it useful for? Or? Or is this still brand new since this is less than six months old at this point?

Austin Wagner 5:11
I mean just right from the get go. As soon as I found ChatGPT I found uses for it within my coding and writing classes. So I took Intro to film. And it was just nice for certain parts of the movie that I didn’t understand. I could ask it like, Hey, why did this person do this in this movie, you know, what kind of theme does that represent? What kind of imagery does that symbolize when they use this kind of camera angle. Additionally, for coding, you can take Java, and you can copy and paste parts of your code that you understand into chat to the tee. And it will tell you exactly, hey, here’s what this error means. Here’s how you can revise it. Or here’s an alternative way to do what you want them to do. You can tell it what you want to do with the code, and it’ll give you direction or paste code for you. It’s not correct all the time, you do have to do a lot of error working back and forth. But you can write a full program using chat to PT, essentially, without any real prior coding experience. And it’s always just helpful for classes when you get stuck on a part. You don’t have the professor there with you all the time. Like you have chat GBT. So it’s almost like this private tutor. Give an answer you pushed you just pass out the obstacle, you get stuck, kind of answer the little questions. You don’t have to spend time researching, going into forums trying to figure out this little solution. When chat TV T has it right there consolidated for you.

Brent Warner 6:26
Yeah, I think this is part of what I really appreciate about your perspective, Austin is that it’s, it’s not just for one field, right? I think this is, this is what a lot of people are, oh, it’s like, Oh, I’m just gonna use it to help me write emails, right, or, you know, like, like, some little thing. And you and I were talking a little earlier, you were in my office earlier today, we were having a chat about this. And it’s like, you know, kind of this idea that people haven’t really latched on to yet that you can ask it anything? Like, I mean, really anything? And And yeah, if you could share a little bit about your thoughts around around how people are and maybe haven’t seen yet the potential for how they might use it.

Austin Wagner 7:12
I think when people look for use for chat to be at the thick of it very like narrowly like for a single use. And so that is something I tried to kind of work on is how do we branch it into other things? How do we let people know that you can really use it for anything, you know, if you want to come up with a a workout regime or a food plan, you can just do the things you asked it to do it anything you want. And it will do it for you how to write an essay or poem, how to even financial advice, it can provide you just simple information on here’s what you can do with the information that you have. So you’re able to put in your specific information and have it give specific results, which I think is the biggest step up from what we’ve had with the internet and search engines in the past is it’s all based on, you know, hoping that you can encounter someone else with a similar situation that’s posted a forum but which actually be T is anything live and accessible. It can cater it to your specific situation. So you can give me your own prompt your own business, your own city where you live. And they will understand that and give you results based on that very specific information. And you can go back and forth ask for clarification and other information that you want. Humanistic response, I

Brent Warner 8:17
want to I want to kind of fine tune part of what you said, which is like, Oh, you can ask simple things. But you can also ask it very complex things. Because when you’re when you mentioned financial stuff, I was like, oh, yeah, like, that’s kind of wild, right? But, but I actually saw someone who had chat GPT do their taxes for them this year. Like they went in and figured it out. And then they aligned it with their own filing of the taxes. And they’re like, it was pretty much right on right. And I think they’re using chat GPT for right, the more more advanced one but like that is a crazy thing to be able to just say, Hey, I don’t need my tax guy anymore. Or you know, like I can figure out money, money things partly through here. But I know that Tim is gonna want to jump in on this a little bit around accuracies. I’m gonna let you go for it.

Tim Van Norman 9:07
So I don’t have a chat GPT account, but I’ve been using Google Bard, which is Google’s similar, not going to equivalent, but it’s a similar package. And one of the things that I noticed is that when you ask it questions, for instance, I asked how do I contact Irvine Valley College technical support. It gave really good information that was wrong. It you could tell it found information on maybe old websites or something like that. But even at that, like phone numbers that it gave, were close but not quite. Maybe they were a satellite phone number instead of an Irvine Valley College phone number or something like that. But it’s things like that, that I noticed. The accuracy wasn’t quite there. I asked it to, to explain something, tell me how this happened. And it got all the way down to second paragraph from the bottom before it started making mistakes. And it was naming people wrong and stuff like that. And I’ve heard that over and over again, in earlier versions, at least check GPT that. If you ask it to do something that’s too long, 20 paragraphs long, it often starts degrading with its the accuracy of what it says, Are you experiencing that? Are you seeing I mean, because AI is just going to continue to improve. But that’s one of the things that I’ve had. I’ve actually had people who were I’ll call it plagiarism, because it technically is they copied and pasted their ID from what they got out of chat GPT and submitted that as their work. And the way it was easily found is the fact that the information, they cited people that weren’t in the book that they were supposed to have read, for instance. So what do you find?

Austin Wagner 11:22
I have noticed kind of some some what you’re talking about with the inaccuracies in TBT. One instance was an essay I gave it to like test it out, and I gave it a word requirement. And the 1300 words is what I told it within allowance, like 100 words, but my next and it gave me back an essay with 800 words. And I told her and I was like this is only 200 words, you know, the Oh, my apologies. And it. So it, we did it and it gave me another essay with 1000 words only. And so it’s kind of like little things like that, where I even told that I was like your account for your textbooks limitation, because you can only pay certain parts, it was like post it in sections and tell me I’ll tell you when to continue. So I think which adds up to you really do have to break it down into these little pieces. And you have to kind of fine tune it. Because it it still has a long ways to go. I think before it’s really festival to give accurate information all the time. I mean, the database who was trained on did have a lot of inaccuracies, I think. And that’s stuff that the team at open AI is working on filtering out and refining the AI from. But yeah, there have been instances where you tell him to do something, and it acts very confident it’s very convincing in the way he presents the data. But it’s not correct. Or like you said it’ll kind of mix up names of characters, or I had a post asked for the chapters of Dr. Jekyll and Hyde. And it posted the first chapter just multiple times. And it would just change this head on the next time. So unfortunately, it’s there are some issues with it. So I think you have to very be very specific with what you want and keep it short on the shorter end. But I didn’t do it like one paragraph at a time, a couple paragraphs at a time. And when you input the instructions that you want to give it, you have to make sure you go with specific instructions for what you want to do and keep it consolidated to pieces.

Brent Warner 13:05
Okay, so we’re gonna we’re gonna get into that part in a bit, Austin, because Because definitely, there’s some good insight onto how to how to get answers out of all of these things. But before we do that, I kinda want to talk a little bit more just about that, that student side of things. So have you seen, and you don’t have to name any names or anything like that, but but have you seen your classmates using it? Or have you had conversations with other students about? How are students talking about this right now? And in the real world, right? I mean, right now, teachers are just kind of like, oh, are students using it? And the answer is like, yeah, yeah, they are for sure. But like, but between students like, what are the conversations going on?

Austin Wagner 13:48
I think there is a lot of that kind of the fear and the unknown of how the professors are going to react. So I think students are using it. And if you there seems to be that either students just haven’t gotten into it yet. Or they’re using it for everything. Because it’s, it’s just so handy. You can once you got to start using it, it’s almost inadvertent that you do use it for every little piece of your life, because it makes it so easy to use. And I was telling my friend about, about how I was bringing you just kind of examples of the prompts I’ve done, and I’ve worked on different writing just to test it out. And she was concerned that, you know, like, don’t give his his professors you know, don’t let him know that we’re using it for stuff like, you know, for writing for assignments or for essays or for helping with our homework. And so I think the big issue is that the students are using it like they’re using it to help with their homework, they’re plugging questions they don’t understand, you know, they’re using it as a private tutor. But they’re still afraid to kind of let the professor know or to bridge that gap of, you know, here’s what we’re using it for. The one thing I kind of don’t, I guess we haven’t really delved into, I guess, in my classmates is, whether they’re just kind of copy and pasting the prompts into it, or whether they are refining it more Couple times, that seems when I talk to people, they’re kind of more on the beginning end of it, where they are just in that period of copy and pasting prompts into it, and just seeing what it can do at the very beginning, without realizing the full customization that is possible with putting your own writing in as a pattern, or the strategy to follow and use an example and it’s raining.

Brent Warner 15:20
Yeah. And I think it’s an interesting point that you bring up, and I hope the teachers listening or paying attention to this, because it’s a, it’s unfortunate that the students have fear around talking to their professors, right. And it’s also unfortunate that the professors have fear around talking to their students. And it’s like, we can’t really step forward with this conversation unless we’re willing to have open and honest conversations, right? And that might mean things like, hey, oh, I kind of think I plagiarized, or I kind of play like, or, you know, when I was working with this thing, and I thought I was doing my own, but maybe the teacher sees that as a cheat, right, or something like that. And so, like, the conversation can go in lots of different ways. And including, I feel like I learned a lot by interacting with chat GPT, getting feedback and then refining the results. It could go the other way, too, right? Oh, I didn’t really spend much time on it. And I didn’t learn anything. But I got I turned it in and I got a grade. Right. And so I think all of these parts are going to be part of the major shift of education. Right? And but it’s also going to be a question of, and this is a part that teachers have a hard time grasping is, what are the students going to want and need out of these services? And how is that going to affect their future, you know, going forward from here, because maybe, and Austin, you can tell me if this is right or wrong, the values of learning are going to be different to the next generation of students. The things that maybe you need to learn or the or the skill sets that you need to have might be totally different than what me and Tim came up with needing to have for our success in our world. Right. I don’t know if you have any thoughts around that.

Austin Wagner 17:05
I think with just how much information there is available. Now it’s there’s almost a question of you know, why spend time learning something, when you could just have an AI presented to you present everything more than you could ever learn and summarize a book perfectly so you can understand it. And it kind of begs the question of whether this is a bra, you know, whether it’s like a painter’s brush, enhancing people’s abilities and tools and enhancing their learning, or they just gonna use it to skip the learning entirely. And it’s, I mean, it’s a hard thing to kind of figure out in this early stage of the AI, on what we’re going to do with it, and where these, like whether the students are going to use it to increase their learning ability new, they can learn more than ever before. I mean, I use different coding, and I found myself learning the languages, just at a record pace, because anytime I got stuck, boom, private tutor right there, it will be able to answer my questions presented to me.

Tim Van Norman 17:57
Yeah, okay. So with that question, from your perspective, is using AI ethical?

Austin Wagner 18:09
I think that’s one of the parts, we’re still figuring out. It’s like, how much of an essay do right I mean, if you turn in a good essay, but it’s not written by you, and the teacher doesn’t know, you know, still is unethical writers still committing academic dishonesty, essentially. But as they how much of the essay, if you go through chat GPT and read an essay where it takes you two hours, you’re revising it, you’re going back and forth, you’re reading each section. And you know, you tell it hey, let’s work on the intro. Here’s the kind of topic sentence one, here’s the kind of evidence I want. And here’s the style I want. To what point is the amount of effort you put into it? Make it yours versus plagiarism? Make it the AI is, you know, when is it yours? When is not yours? That’s a very big gray area that I think we have to kind of bridge and I think the only way to do that is through this openness and communication to let the students know, and kind of find out how they are using it. And how people are really just skipping assignments? Are they just taking the easy way out? Or are they actually learning from it? And using this as their new style of learning? And how to is using AI going to be our future way of learning, right? The same way you use a Google or search engines now. That’s just a future exploratory tool that we’re using to enhance our ability to gain information.

Tim Van Norman 19:24
So when you think about that, what are the areas that you feel conflicted about? Where do you feel that this understand that in six months that might be a completely different area? But right now as you’re looking at it, where is it that you’re feeling Wait a minute, this I can see this going too far? This is not this is past the limits of ethics.

Austin Wagner 19:50
I guess one instance would be had a film class and at the end of the film class, it was a final essay. And it was a few write it on a movie and theoretically you could destroy If you could do the whole essay, have it write it on a movie you never saw, never analyzed, do come up with the themes all the information in an hour and tested it out, it wrote 200 word essay in an hour. And at the same time, I didn’t watch the movie and I didn’t learn the material necessarily. And so at that point, I will consider that a plagiarism because I didn’t learn from it. But at the same time, I did spend time revising, and going back and forth and making the personalized essay and I still understood the prompt that the teacher wanted. And so I would say when the students instead of learning to start using it, to just fulfill assignments in order to just complete the assignments is when it’s plagiarism, but it’s not really something you can standardize, or that puts the burden on the students themselves.

Brent Warner 20:54
Yeah. And so often, I’d like to ask you this, because this has been one of the conversations that I’ve been having with teachers around the, you know, to me, I kind of feel like, there could be endless plagiarism, right, that conversation is kind of the door is open. And so it’s like, okay, there’s not really a lot of ways to check for that, whatever else it is. And so in my thinking, I’m going around and talking to other teachers and saying, Well, maybe it’s time to stop having one doll assignments that are that aren’t engaging students, right. I mean, that’s, hopefully, hopefully more people are doing that. But also, like trying to build relationships with the students and show that you care about the work and that you care about what they’re doing and what they’re learning. Because, and, you know, you might be one person’s perspective, but you might be able to say, like, Speak on, you know, other friends behalves like to me, I would think that hopefully, if I have a better relationship with my students, and they would feel more compelled to do better quality work, or want to figure out how to do the learning, or you know, whatever else it is inside of there. And I’ve had other teachers who push back on that and said, No, students are just going to cheat no matter what if they get the chance, they’re gonna cheat, right? And so I’m like, Well, okay, maybe, but like, one, what’s it going to hurt to make a better relationship with your students, of course, but also, maybe that’s how we start to deal with this, instead of chasing people down for cheating or plagiarizing or whatever else it is, then we can actually show that we care about them and their success and their learning and the process. Does. When I say that to you, I’m just wondering if that speaks to you at all? Or if you’re like, yeah, maybe maybe not. Because I’m just interested in like, is that something I should be trying to tell teachers out there? Or is it kind of a middle ground, and maybe not so important to what the final decisions are, what you’re doing in with AI.

Austin Wagner 22:49
But I think at this point, it is part of the learning process and the growing process with AI that we’re going to have to cross these hurdles. And these kind of controversial opinions. I mean, this is just the beginning of you know, what I think is going to happen with Chad TBT. And just AI in general, as it potentially replaces jobs replaces human labor. And at this point, we’re just on this kind of plagiarism talk where it is, it will eventually I think, be a minimal issue. But it’s still an issue that we’re going to have to address in this current stage as we go on to figure out what we can, what kind of boundaries we can accept from AI or not mean being actually removed the ability for it to write essays, I believe, due to fear of the plagiarism, that chat to be experienced all that controversy surrounding it. So I but I do think, eventually, hopefully, we’ll reach a place where the AI is seen more as a brush as a tool to enhance the writing of the person. And not so much just a plagiarism tool where they’re absolving themselves of learning entirely, was something that can aid people and bring people who might not have had the same resources or background learning or knowledge to a standardized level, so that we can hopefully all kind of be at this higher level of learning education.

Tim Van Norman 23:59
As as you’re looking at that, and that makes sense. As you’re looking at that. Let’s transition a little bit towards getting quality results out of the AI because, as I mentioned, when I tried to do a couple of things I got what the AI seemed to indicate was accurate, but was completely wrong. Just like one sentence here or there, that type of thing, but enough that if I handed it in, I shouldn’t be marked wrong, as completely inaccurate on that. So what do you do? How do you increase that? What do you do to get better results out of Chet GPT.

Austin Wagner 24:49
So I think one of the biggest kind of hurdles to overcome is understanding the limitations on chat TBT and just that there is this inherent inaccuracy because of where it was trained on is trained on Reddit forum In which, while they do have accurate information times, they also have a lot of inaccurate information. And so I do go into tend to be cheating, Knowing that ahead of time, and I’m looking for these inaccuracies. If I find them, you know, I tell chat to get to them like, hey, you know, you said this person was in the store what they really weren’t when you told me this, and I can find it directly, I asked her, like, Hey, why did you say this? What made you think this, and you’ll Yes, my confusion, and it’ll correct what it said. But I think it’s just about knowing ahead of time that you’re gonna, unfortunately, going to experience these inaccuracies in the early stages of AI, before, we’re able to get to the place where they can have much more consistent databases, and the sets that know ahead of time, or can fact check themselves, and produce much micro data that people are going to have to learn about are going to have to go through this kind of rigorous learning experience, almost of trial and error, you know, what works, what doesn’t work, you know, what the AI understands, and what it doesn’t understand. But the key thing is, is knowing what you want from the AI, and knowing that you have to refine it, and you’re gonna have to deal with the errors. And it’s not perfect, but it is a tool that can help you create things that otherwise would take much longer.

Brent Warner 26:07
Yeah. Now, Austin, in your case, one of the parts of the conversation that we’ve we’ve had some times and I’ve been having with some professors is that the professor’s have expertise in what they’re asking chat GPT to do, right. And so whatever, you know, Professor or whatever, you know, student to it’s not, it’s not that I’m not only meeting professors, but anyone could have a level of expertise in that thing. And then they can fact check the quality of the AI output, right? One of my concerns is that, you know, people will just go into it, trust it, and say, hey, it’s gonna make good quality stuff, right? And then, but they don’t have the background knowledge to go in and confirm that information. And so right now, you in particular, like you’re, you’ve come through a traditional education, and now this, this technology is ubiquitous, right, or fairly ubiquitous. And so you have the benefit of having learned for many years going, going through the school system, building your own knowledge. And then now being able to analyze this, a concern that I have is future generations who have done everything through something like AI, and they’ve never bounced their ideas to confirm the quality of it, or, or everything that they’ve ever learned is through something like these AI systems, which is inside of its own echo chamber, right, by default, in certain ways. And so. So I guess, a one, is that a concern that you hold? And then to, as you know, and maybe I’m wrong on that? And if I’m wrong, then what are some of the misconceptions that people have about using AI? And about how to interact with these things?

Austin Wagner 27:56
Yeah, I mean, that’s a really big point about this AI is how do we factor in how we know the information that it’s giving is correct. I mean, essentially, essentially, it was trained on Reddit, which is an echo chamber of data and itself, these thoughts are collected into just reverberated around by people who share similar ideals. And so you get these inherent biases in the training data that are going to propagate in the AI itself. So when it gets it becomes a question of whether you can accumulate the standardized databases for the AI to train itself on that are factual, if it uses, either like Google Scholar, all those research documents or the fact check itself and confirm these before it presents this to the user. Or if we are just going to have these kinds of wild AI that have, you know, incorrect or correct data. But at the same time, we do have a lot of experience just in people, generally using Google as a search engine now and shift sifting through all that false information that’s already out there. Even on Tiktok, and Instagram, I mean, it does fall to the burden of the user to sift through that false or false information or true information. But I believe, honestly, with the pandemic, and even the last election cycle, all the false information that’s kind of flooded into our site, it might be some sort of preparatory way for people to identify and hopefully draw a need for a way to learn how to identify false information. That is going to be a big hurdle. I think that AI is gonna have to overcome even being was sharing their results and their citations. I think it’s one step forward, but not enough. Yeah.

Brent Warner 29:30
Yeah, absolutely. And that’s where it gets tricky. And of course not to put the, the burden of the answer on you. It’s just kind of as we’re thinking about the, you know, like this, this becomes so tricky. And Tim, you know, we’ve had these conversations to where it’s like, there’s so many ways to move this conversation, so many places to take it right and so, so it is fascinating, but, but Okay, coming back to the prompting and how to talk to GPT what are some of the ways that At that people maybe start and they, they kind of make mistakes and then, and then limit themselves to their understanding of what it can do, right? Because I’ve seen people go in for 10 minutes or 15 minutes. And they’re like, Okay, I gotta it’s kind of it’s, it’s cute, or something like that. And it’s like, isn’t that cute? This is a game changer. So So what are you seeing as kind of the wrong first steps or the or the problems that people have when they kind of start with it?

Austin Wagner 30:29
Yeah, I think that is a big issue. People just don’t know where to start with it. Because it’s such a wild, you know, unrestricted AI, they can ask it anything. And I did think, open, I did a good job of presenting these prompts for people to use. And generally, when I tell them about AI, I’m like, Hey, asked about something that interests you about something personal to you a hobby, something outside of like work or business, something that they can actually get a dialogue going back and forth on and talk to the AI and understand this conversational tone that allows for and the understanding of just regular human commands, they can just tell it exactly what you want it speak to, like you’re speaking to a person. And if it’s something that they’re interested in to that will, I think also maybe make them more interested in pursuing learning about the chatty GPT and then branching off into other topics. But yeah, definitely, it’s hard to get through how do you give someone this tool that has access to everything and anything and get them interested without overwhelming them with all this information?

Tim Van Norman 31:27
So let’s, let’s take me through a scenario, I sign up for a chat GPT account. I’m interested in technology, a technology in the classroom. Okay, which is a very broad subject. How would you suggest that I start talking to the Chet GPT, to chat GPT or any other AI? What type of prompts? Would you naturally see that going through?

Austin Wagner 31:59
I mean, just right now, like, if I was talking to you about that same topic, it’s kind of the same questions that would come to my mind. Um, you mentioned technology in the classroom, I guess, my first question would be, what is it that we want to do about that technology? Do I want to know the different types? Am I looking to analyze it? And so from there, I would kind of form a more specific question. But you can even just put in tech technology in the classroom into 2018. And it will usually respond with you a wide array of kind of different topics that it expects you to find interest in. But the thing you do need to do is find specific direction that you want to follow with it after a little bit. Once you start with that broad kind of general terms, and you specify it over time, as you kind of branch into the AI.

Tim Van Norman 32:45
So I, I was working with somebody last weekend, and they were talking about I think it was Chad GPT. And they had it write a write a song about using chat GPT. And it actually was a pretty interesting tune that it came up with, you know, no, no music to it. But the lyrics were, were pretty good. They were interesting and stuff like that. And so yeah, I was impressed by by how it could write something basically about itself in that way. And it was very readable. And in using it, everything I’ve seen come out of it has been readable. It’s not necessarily the language I would use. But I also haven’t trained anything to talk the way I do. So it’s definitely been impressive that way. And I’m curious as to how other people, I’m hoping that people will respond to this as to how they’re using it, to try things out. What are they prompting for? What are the cause GPT also then gives you prompts back, but you can click on and go into something else and had a direction. I love what you were talking about what that thank you that’s that’s a great way to to develop that. So what do you not like about the DEA at this point? I’ll begin this for Chad

Austin Wagner 34:20
TBT I think some of the things I would kind of like more if they changed were like the reiteration because sometimes it does repeat kind of the sentences. It’ll be like, you know, if you’re seeking legal advice, you know, you need a lawyer. If I was coming to you, I probably would, you know, like I could afford a lawyer or something right? I’m using tragic a deeper reason. And it kind of does reiterate all these topics over and over. It’s very repetitive, it does kind of get lost in itself almost if it has too much text. We will start referencing itself. It kind of tries to force these conversations based on what was inputted earlier, which doesn’t really understand necessarily how the conversation changed to a different topic. So what I learned initially isn’t relevant now but it still has that kind of I logged as, like relevant. And so you do have to like start a new conversation each time starting to chat dialogue in order to clean the slate and get what you want to get and not have any type of exterior information or extraneous information injected into it. But I think that is hopefully something that we’ll be able to overcome, you know, as they refine this as they are able to expand the AI working system and how to be able to differentiate between what’s relevant and what’s not relevant to this conversation outside of just, you know, one chat dialogue, one box.

Brent Warner 35:28
So Tim, kind of what you were talking about was, it was striking me as an interesting point, because you’re saying, Well, you know, it’s like the style of it, or whatever else isn’t going on. But Austin, you and I have talked a little bit about that to where you were trying to train it to write in your voice, right, like to have your stylistic writing or to maybe, you know, your, your written mannerisms? For example. How do you train it to do that? Was it can you actually like, give us a walkthrough on like, what you would prompt it with how you would how you would have it, learn what your style is, and then, and then whether you think it’s successful in producing stuff based on on that input?

Austin Wagner 36:14
Yeah, I mean, I like because when you enter a prompt in chapter two, chat, TBT, it gives you this result. But after a while of reading it and experiencing tragedy at the AI generated text, you can see these patterns, and it’s very under, after a while, you can recognize that they’re AI generated. So what I will do is when I write an essay, or when I try to like write a excerpt, or film or brief repair, or whatever I’m writing, I will copy and paste a section of my own writing that I’ve developed, I will tell it to look for nuances in my writing, like, I want you to look for the sentence structure, the different words, I use the adjectives the way I phrase and vary my sentence length. And I give a trait that I would expect a human piece of writing to have, so that it can mimic it, but I give it specific things to look for, instead of just giving it a flat copy of paragraph text. Because you know, we’ll know what to do with it necessarily, you might just copy and paste it, you might just start putting it into your essay. And generally also tell it I’ll be like, Hey, don’t use you know, this excerpt, don’t use a snippet of text. This is just for you to learn from, not for you to use in whatever you’re writing, not for you to use in the future things.

Brent Warner 37:20
Okay. Okay, that makes sense.

Austin Wagner 37:24
So after kind of using it for a while experienced, what it does is kind of errors that it makes and accumulates, where it’ll try and copy and paste something you’d write, add it in. And I do have to go, I will go like paragraph at a time, like a single paragraph or single, you’d have to fact check it and revise it pretty frequently. In order to steer it in the right direction. I think it’s kind of the best way to envision it.

Brent Warner 37:47
Okay. Okay, so then. So then you put it in, and you say, like, recognize how often I use adverbs, for example, or, you know, what’s the frequency of my, you know, pay attention to the frequency of my feedback on this or whatever else it is, right. And so once it starts paying attention to those things, then it can plug it into its own responses. So you’re getting to

Austin Wagner 38:12
me, here’s, here’s one that I sent to Chad GBT, exactly, it says, Here’s a snippet of my own writing. So you can understand the style that I like to use, primarily the words flow sentences, and the immersion of main points in the text. So the elements are not so blatantly presented to the reader. Note that I don’t want you to use excerpts or clips of this reading in the essay, but just look at the style and writing and how it sounds overall, so that you can better write my essay, here it is, and then I would copy and paste my own extra writing.

Brent Warner 38:40
Mm hmm. Okay, I got it. So. So you’re really, it’s, I’ve played with this a little bit, right? Where you just kind of tell it to say to understand something, and I’ve prompted it with the end of say, respond with yes, if you understand and then it just gives you a yes. Right. So like, I’ve done that kind of a setup where it’s like, you know, I said something like, you’re, you’re a grammar expert, you you know everything there is to know about you know, subject verb agreements and punctuation. When you when you see mistakes, you you you what did I write, when you see mistakes, you you point them out in a gentle way, and then make sure that you give a clear explanation as to why it’s a mistake. Type respond with yes, if you understand yes. And then I said, Okay, I know you’re in there so that that idea is inside of there. And so then I started with the actual prompting of the things that I wanted to do. So I said, here’s a paragraph that I that I wrote, please give me feedback on it right and then it then it starts processing it in the in the mindset is, I want to I want to avoid the humanization at some at some levels, but But in that in that processing now it’s going, Okay, this is who I am, I’m role playing as this grammar expert. And here’s how I’m going to be responding. And so that background set up is something that I think a lot of people using it don’t really understand that you can do. So you can kind of plug in, you know, almost like plugging in a specialized computer chip into a computer. And it’s like, oh, now it’s boosting it for this specific purpose to help me get the work done. I, this is what I’ve been playing with so often. And that’s kind of lining up with what you’re talking about as well, I think. Yeah. So but you have also done several other ones, several other uses. And Tim, this, this might be interesting to you to the coding conversation. So so often, can you tell us a little bit about you mentioned before that you said, Hey, you play with a little bit of coding, but can you kind of get into a little bit into the weeds about what you’re doing with coding or what you’re learning through the coding inside of GPT?

Austin Wagner 41:03
Yeah, yes, I was looking at developing an app and Android app. But in order to do that, I needed to learn about like the previous languages that computers use, just so I can kind of learn the ins and outs of programming. And so I would just pick a language like Visual Basic, I would pull up the command terminal, which I would use to code in, and then the tutorial. So I could just get started. And then Apollo chat GPT as well. So I have these three windows. And so I follow the tutorial, I start coding. And then whenever I kind of reach an obstacle, or I get stuck, or when I start branching into my own kind of coding, I want to diverge from the tutorial, I start pasting in chat to BTS are asking me what I can do. And then from there, I just go back and forth, usually with chatty btw, we’re all tell it what I want to do, it gives me an example. Or it’ll tell me how to do with these code. And I’ll either copy and paste it into my terminal, run it, encounter an error, and I’ll let it know what errors I have or what parts I have questions on. But I think one of the best parts is that if you have a question on line 32, you can ask it be like, Hey, I don’t understand. Line 32? Why is there a colon? Or why am I getting this error? Why am I getting this message when I tried to run it, and it will explain it to you and provide in depth, reassurance and understanding. And you can just keep asking these questions and get tired of your responses. It’s not human, right? It’s just an AI robot, like continue asking these, you’d have to be careful because some of the code that it’ll give you back sometimes is not completely accurate, though, because it is still in the working phase and like the learning phases. So I did make an app with Python using the chat GBT to write the code and go back and forth, correct to, you know, specify what I wanted, in order to pull text out of PDF documents. And it took a while, few hours of working back and forth. But I was able to make come up with this code that works for my intended purposes, and does everything that I wanted it to. And I did learn along the way how to use the code, how to interact with it, and what each section means. Variables, all this coding, the language of the, for that coding language.

Brent Warner 42:56
I like that to be just because that’s active learning, right? That’s you going in and saying, Hey, I have a goal, I have a project here, right. And this is the thing that I’m wanting to do. But at the same time, now I’m learning Python, or I’m learning, you know, like the coding that maybe you wouldn’t have taken the time just to sit down and by yourself or some pre generated reason for it. And so you can kind of jump right into the activities that you’re interested in with it, and then use that as a support system as well. That’s, that’s super cool.

Austin Wagner 43:27
Yeah, it seems like if you can think of it you can basically do with tragic at night, you can learn, you know, I even started a whole like company through jet TBT. Just using applying for like the business license. During the LLC, the trademark, setting up the PIO box and all the things just through chat DBT. We’re all otherwise someone might need like might need a lawyer, you might need all these other resources. And just a time it saves the accessibility to people. It’s just something unrivaled. Absolutely.

Tim Van Norman 43:57
So you’re you’ve started to get into my next question. I think this might be one of our last ones as we start to wrap up. But so you looked at setting up a business. What other cool ideas or are you finding are you thinking about with this technology.

Austin Wagner 44:15
And there are so many of ideally only reaches, because we’re now there’s three branches of AI where you have visual image recognition, you have a language learning and you have like a sound based one. And right now they’re all separated. But hopefully in the future, maybe within like five years, they’ll be able to come together. And Bing is already working on this kind of image recognition, video recognition and language learning, mixing them together. But ideally, you could have a personal assistant on your phone that does everything you need to you can manage your assignments can tell you when your next assignment is due just by letting it know you can go check Canvas for you. You can go check a website for you if you want to apply for housing or food stamps, someone who doesn’t have the resources can tell it what they want. And it can go back and forth and tell them exactly where to do it and run this. So essentially it was make accessible things more easy for people, they can follow their dreams, hopefully, without having to have this larger barrier of entry of expertise that they would otherwise have to have.

Brent Warner 45:12
Nice. I love it. There’s so much to keep thinking about here. We are unfortunately going to run run out of time here for today’s show. But but Austin really appreciate like all the all the ideas, incredibly thoughtful on how you’re looking at using it. And, and I think that that honesty about it too, is a big part that I want the teachers listening to, to kind of walk away with is like, here’s an example of a student who is actually using this stuff and trying to figure it out and see what’s going on. And often it sounds to me like you’re saying to the teachers, hey, come talk to me. Right, like, you have a last word for teachers.

Austin Wagner 45:53
Yeah, but anyway, I think just having a two way communication is the best thing that we need, because it’s going to be used to students are going to use it, people are going to use it. And just knowing what they’re using it for and how to enhance the learning and how to just go back and forth and create a better atmosphere incorporating AI into learning. I think it was opened the door for so many students and teachers to just go forward to get their education.

Brent Warner 46:17
Excellent. That’s great. So we’re gonna jump out. Thanks so much, Austin.

Tim Van Norman 46:27
Thank you for listening today. In this episode, we talked about a student perspective on ChatGPT with Austin Wagner. For more information about the show, please visit our website at the higher ed tech podcast.com. There you’ll find our podcasts and links to the information we’ve covered.

Brent Warner 46:42
As always, we do want your feedback. So please go to the higher ed tech podcast.com and let us know your thoughts. And if you have ideas for future shows, there’s a link over there where you can give us your topic ideas.

Tim Van Norman 46:53
For everyone at IVC. That’s listening. If you need help with technology questions, please contact IVC technical support. If you have questions about technology in your classroom, please stop by the IVC Training Center at a 322 or contact me Tim Van Norman AT tvannorman@ivc.edu.

Brent Warner 47:11
And if you want to reach out to me about the show, you can find me on LinkedIn at @BrentGWarner.

Tim Van Norman 47:17
I’m Tim VanNorman,

Brent Warner 47:19
and I’m Brent Warner and we hope this episode has helped you on the road from possibility to actuality. Take care everybody