Episode #7: AI Content Generation

I'll talk about it on the pod. Fuck it. We'll do it live.

• Get another hot water. - Hot water. It's the thirst mutilator. (upbeat music) And welcome back to the Big Cheese AI podcast. I am Andre Harakas, the world's 24th best moderator, joined by Sean Hise and Jacob Wise, two tech leaders in Indianapolis. And last but not least, Brandon Corbin, one of the brightest AI minds in the Midwest. Today, we're going to be doing AI content generation and its impact on the industry in general. There's been a lot of things that came out this week, especially with the advent of, even from a couple of weeks back with ChatGPT Turbo 4 and DALI and the upgrades it had. Now we're watching memes come to life. And there's a lot of things happening in the content world that are wild. Another thing that's wild is we officially have a super famous fake girl. And that's going to be where we kick off the episode, because Jacob has done a ton of research on this girl. - He deep dived. He deep dove.
• He's done a deep dive on her Instagram page. So Jacob, what's happening?
• Oh, yeah. So I think the origin is a Spanish modeling company. And I don't know, they were just pissed off about having to pay too much for models and they're too hard to deal with. So they just made one, an AI generated one, and she blew up. She is making them a lot of money, doing a lot of endorsements, and she's very good looking. We'll have a link probably in the show notes that I can share. But yeah, it's pretty interesting because she's not real, right?
• I haven't seen this. - She has like 150,000 followers on Instagram. - It's bringing in about 10 grand a month right now in ads. And if you go through the comments, there are just a bunch of sims out there doing like, "Oh, you're the prettiest thing I've ever seen. Can I get your phone number?" Oh, no. No question. I guarantee it, right? Because that's what's going to be happening. She is just the precursor of what our future is. Where before you had to look like Brad Pitt to be the super monster famous person. Now, they're not going to exist. It's not going to be real. I mean, the future is fake. No question. - With AI, there is a new layer of catfish that is emerging. And it's going to be rocking people's worlds. I don't think that we're ready for that. To have not just one girl who's famous on Instagram for being beautiful, but 1,000, 10,000 girls that you can interact with and have conversations with. And obviously, we're all familiar with different platforms that you can interact in things like that.
• Right. - That's going to blow up.
• There was another one too that was like, "Hi, Grandma. This is Julie," or whatever it was. And I forget who put that video out. But the premise was, is like, "I'm kidnapped and I need $100,000."
• That call has made to my grandpa before he passed away. He called me and he was like, "Why are you in Vegas in jail, blah, blah, blah." - Really?
• That whole scam works. - Well, but did they...
• It's heaven for a while. - So, they probably texted him, right?
• No, they called him. - They called him.
• That whole scam has been around for a long time. - But did they use your voice?
• Well, no, I don't think so. But they don't know. They're old. They go old.
• They're old. - They can't recognize voices.
• But the point that Brandon is trying to make, I think-
• It could be a FaceTime at this point, right? - Snickerdolls.
• Where it's just like, "Oh, hey, I'm trapped in a Mexico prison. You need to send $500,000."
• Yeah, and I think they've been talking about deepfakes for a while, and you're worried about a famous person going on and saying something crazy, but now the local version of that is the deepfake of you. I think I used an app, I forget what it was called, but I created an AI-generated version of myself last week, and it was five minutes of recording. It came out, and it says, "You're a cautious techie guy." I'm like, "Oh, you son of a bitch. How dare you?" But it sounded pretty much like me, right? On the other end of a call, I bet it could fool my family. - They only did voice, though, right?
• Yeah. - There was no photo stuff.
• That one didn't do visual, no, but yeah, it was mostly focused on voice and personality, mimicking me there. But like you said, I'm sure there's video versions that you can do today.
• Well, we're still trying to find those because they all suck. We went through and we tried to generate some 3D art versions of ourselves, and we tried character.ai or whatever some of the other ones were. And I don't know about you guys, but I never had any luck.
• Oh, you have made the best one by far of anyone, though.
• Well, which one? - Mine still looks nothing like it.
• But yours, you're like, you were describing yours to chatGPT or to DALI.
• I did actually both. - Oh.
• So I described mine to- - DALI through chatGPT.
• Through chatGPT. - Right.
• In mid-journey. - Which I tried that.
• But the mid-journey one, mid-journey has a feature where you can upload an image. Jacob and I sat there for three hours one night and tried it. You can upload an image and it will generate, and then give it a prompt. So you basically, you upload the image and then you attach a prompt to that image. And mid-journey will then generate four versions of that prompt. And it'll use the data that it would attain from the image to try to make it look. But it's not able, maybe it's the prompt, whatever, it's not able to generate it with the likeness. So you're really telling it, "Please make this look as much like me as possible, but cartoon."
• So when I was doing stable diffusion, you had the ability to where you could go and basically generate, I took a bunch of photos of my daughter and then did kind of a, I forget what the terminology specifically was, but basically created a lore, I think it was a lore, L-O-R-A, which then let me basically go in and put in anything in my prompt, as long as I would tag it as Maddie, then it would end up looking kind of like her. So I was in, I mean, that was now, I mean, that had to be a year and a half, if not two years ago. And still though, to this day, I went through one and I uploaded like 20 photos of my ugly mug and it looked nothing like me. It was like a dude with a beard and glasses and that was about it. - Now I've seen, so there's like a lot of AI personalities on YouTube and elsewhere. I've seen, it is, I think it's possible. It's not mainstream, but either that or some people are just getting an artist to do it for them.
• Well, so my LinkedIn photo, and it's the one that you've used, was AI generated at this point.
• Yeah, that one was pretty good. - Right, that one, but it was the same exact thing where I uploaded like 30 different photos. - Wait, that's AI generated?
• Yeah, if you zoom in on it, you'll see my iris of my eye is like melting. Because my wife's like, "Here's the thing though, is I hate it. I think it looks great, but you would never wear that suit." And I'm like, "Yeah, you're right. I probably never would wear that." Because again, I only wear black. But yeah, so that one actually was one of the best ones. And I still, I've tried to go through my email, trying to figure out like, "Where did I do this? Because I want to do it again." But I still want us to have the ones that are kind of like the Pixar inspired styles. And man, it's been tough. - So really, we're talking about content generation. And right now, what you're seeing is an explosive growth in non-text-based content.
• And it's completely streamlining and automating the content creation industry as a whole right now.
• Right. - Yeah. I mean, it's one of the things you're seeing. I think the biggest thing you're seeing is people are using it to generate content on the web.
• Right. - Have you guys been seeing anybody who's actually tagging that articles were AI generated? - Yes.
• Right? - Yeah.
• Well, I remember there's a publication. Well, first of all, Sports Illustrated came out yesterday. Did you see that? - Yeah.
• And they were like, "Holy shit, our articles are being generated with AI." And their journalists didn't know. - Yeah. Well, it went as far as to like, they generated fake journalists to say that they were, that's who wrote it. And they didn't disclose it. So I, to your point, I think what we're seeing is like, there's a brave new world. And I don't think generating AI generated content is necessarily bad, but there's this conversation being had like, you need to let people know that it was AI generated or is, I don't know.
• Yeah. - But then I think this story tells, maybe yes, maybe. - So that's, I think that's an interesting topic of conversation here. Is disclosure of AI content, should that be just a thing? - I mean, I think if you're maintaining journalistic integrity, right? You're supposed to source your sources. You're supposed to have an, you're supposed to do those things. I'm sure that those industry standards have to be there. So I would say yes. - Right.
• Do I think it matters? I'm not sure. - Yeah.
• I think it's- - I mean, I think that is the question. Does it matter? - Does it matter?
• I don't know. I mean, I think that it depends if it's, there's people that write in non-factual information on trusted publications all the time. - Newsmax?
• Yeah. - I mean-
• Just kidding, Newsmax. - I think that one of the things about AI is it's, is the other thing is the bias. It's like, where is it scraping its data from and where is it harvesting its data from? - Right.
• And how is it perpetuating correct information or misinformation?
• Yeah. - I think to make matters less complicated, I think there's a definite use case for aggregation, summarization.
• Yeah. - Who needs to go write the AP summary of the Phoenix Suns versus Boston Celtics game last night?
• Right. - Right?
• Yeah. - I'm not sure that that needs to be done by a human anymore. - Yeah.
• I get a summary every week of how I did my fantasy league.
• Right. - That's not written by a human.
• Right. - That's written by AI because it's just taking data that's coming from a system that's based of fact, right? And it's summarizing that. But I think where it starts to, if you don't want to get in a huge rabbit hole about journalism, but yeah, I mean, there's a huge risk to journalism and people are going to be-
• But that's journalism, right? So like, what about this Athena chick on Instagram?
• Right? - I mean, you're threatening, I think it's, in my opinion, it might be a threat to all of those industry, marketing, journalism, entertainment. I mean, you have models on Instagram who have a hundred thousand followers. There's people that worked for years to get that level of following and worked on their physique and everything that's included being a model, which is not easy to get famous and start generating revenue. Now you can just create it automatically. - Yeah.
• You have people in marketing like me today. I was interacting with coaches on Twitter and Aaron and I created a GPT that basically we could insert the tweets from coaches and it created responses for my company. - Right.
• Well, we can just basically turn that on now. - Is content the most important or most impacted current state aspect of AI in the business world? Is content the most impacted aspect right now? - It's gotta be, right?
• Absolutely. - Yeah.
• I mean, because- - I think absolutely.
• We've got text, we've got images and now, and video, which is coming. So stable diffusion video stuff is slowly starting to happen and they're starting to get it to where it can like, each frame can be kind of consistent and whatnot. So yeah, it would seem like that would be impact.
• And so if that's truly the case, you can focus on it diluting existing really good content sources out there. I think there's ways around that. If I'm a truly connoisseur of information, I know I go to the right author or the right book or the right website or, you know what I mean? I can, I'm smart enough to be able to sift through that kind of stuff. Where I kind of go is more the positive side where it's like augmenting what you're doing to make it better. Like I see videos, there's awesome things you can do now inside PowerPoint and Canva and all these tools to take your crappy presentation and in a few minutes use AI to make it awesome. To make your data and your information talk. - Well, that's kind of a theme that we've always talked about is how to get people to spend more time doing creative things, right? Like lowering the barrier to entry to do the monotonous text generation, image generation work, stuff that doesn't add a ton of value but takes a lot of time and then get it to like, okay, curate the data or put it together in a creative way that is easier for someone to digest or consume. So that's exciting to me, but there's a whole host of, you know, worms or can of worms that that opens up a little bit. So one thing I was interested about was early, early on, Jad GPT comes out and everyone's like, oh, we're going to use it to write blog posts. And then everyone was concerned that this was going to be a Google black hat thing. And cautious people pumped the brakes and said, not yet. We're not going to use it yet. But Google straight up came out and said, we don't care how your content is generated. If it's AI generated, as long as it's useful, we aren't going to penalize you for that. The only thing they care about is, is it accurate and useful to users and all of the same metrics that they use now to rank content. So.
• Is it accurate? It's an interesting, how are they determining if it's accurate or not?
• Well, I think one of the, I think, I think maybe the more impactful thing Jacob said is are users getting value out of it? And they know that.
• Okay. Yeah.
• Think about it. You know, with GA4, they know the engagement level on a page, right? They know if people are reading it, right? And if people are coming back to it, if they're, they can take those, the engagement statistics and they can use that to rank the quality of the content.
• Right.
• And that might circle around the truth factor.
• Yeah. And it honestly, it probably accuracy is, wasn't, isn't true. It's more about do people find it engaging and useful? And, you know, you could make an argument that people tend to, you know, things that are useful or things that are accurate, but sometimes it's not.
• Jacob and I spent a long time with Crafted trying to figure out a way to convince our clients to build headless websites. Okay. And so a headless is a concept that is really, really hard for some people to understand because it's more technical, but it really achieves a lot of things that you need, which is around technical SEO, engagement, performance, cross, cross platform integration, all these things. If you, so we, we had to educate, right? And so Jacob and I took all our ideas and we really, really put it into a really organized system, but then we utilized AI to help us kind of finish it because we're busy as shit. And we don't have time to finish this entire massive content project.
• Yeah.
• So we, you, you, we created these pillar pages on our website, this pillar concept Jacob created, and really we just started just chipping away. And eventually we ended up creating this headless guide, which is on our website and it's pretty useful. It's, it can be improved, but at the end of the day, like for us that we don't spend a gobs and gobs amount of time doing content marketing, I would argue that it was useful and helpful for us.
• Yes, it was. And I think, you know, we wouldn't have done it without chat GPT. It still took a decent amount of time, but I mean, it's just one of those things that it lowered the barrier to entry and we were able to communicate our thoughts in a more concise and effective way. So yeah, I mean, I love it. I love the idea that these tools can, can help us to accelerate those things. And if you're busy, you don't want to spend a lot of time on, on sort of the monotonous stuff. Like you reach for this stuff. So.
• Yeah. I think it's a huge cost saver too. I mean, I was on the negative side before, and now I'm back on the positive side. It's a massive cost saver. Like you were mentioning, like, yeah, you can create an entire presentation instead of having to send it off to a professional presentation maker or some sort of designer. Now it's just beautiful. And that allows you to go out and present and with your startup founder, you can go raise capital better and you have beautiful touch points. It's like, it is, that's great.
• Last night, my girlfriend was creating a PowerPoint and she was looking for a Labrador with a doctor's code on it, right?
• She just looking for a Labrador with a doctor's code?
• It was vital for her presentation. And, and she went on the internet.
• Was it a black lab or a white lab?
• It was, it was a, it was a, it was a golden retriever actually. Sorry. And she found, she found some great images on, on the internet. Right. But if she hadn't, I was like, you could just go to Dolly three and say, give me this. Right. So I think it's helping to shortcut some of these, these approaches to generating content. It was like, I don't need to go search 20 different places for it. I can just generate exactly what I'm trying to look for.
• That's I think the most fun. And we've, of all the things we've been messing with at Big Cheese, we love the image generation stuff.
• It's so fun.
• Because you can truly take your imagination and create a world out of it so quickly. I mean, if you go to our podcast at bigcheese.ai website, we generate a mouse image for every episode. And some of the images that Jacob's created are, they're just, I mean, I just love them. And they're, they're like creative to the point where you, you give it, you give it this prompt and then the AI kind of just like, kind of puts its little spin on it.
• It does. Yeah. It adds its own like nice little touches to it.
• Well, I think my favorite part about that is I've always wanted to be able to draw and I've tried a lot to draw. I suck at it. I just, whatever my brain does is in my, to my, you know, hand, I can't do it. I just can't do it. I have, I have ideas and I can't get them on the paper. Right. But these tools help the creative side come out a little bit more and I can actually have a playground and visualize what I'm trying to do. So that's fun. I like that.
• And that's what's going to happen. So you talk about images and text. I mean, that's the basic fundamentals, right? People are building open world games and videos, videos, and you're going to see movies coming out that are entirely generated in an artificial world. And you're going to see some stuff coming out of that. You're talking about Easter eggs. Right. You're going to see. And the interesting thing is it's using all human knowledge to create this stuff. Right. So all this stuff is inspired from the real world, but it's, it's going to, so that the outcomes of it are just going to be things that people have never seen and they're going to be drawn to it. - No, I think it's like, I just saw this on, on X's past week. I thought I sent it in the Slack chat. Apparently I didn't actually click the send. It was just in my log. But it's a platform called Pika. And if you guys watch this video, it's one of the most magical. - How do you spell Pika for our listeners?
• P I K A. If you, if you look up Pika, you need, all you do is you type in a prompt and we'll, it'll create an entirely custom video. Like it's beautiful. What it looks like. If you type in creative walking junk robot, that walking junk robot looks like a walking junk robot. I mean, like, I don't know how anybody would look. - They just raised $55 million.
• It seriously is the most incredible video, like chat GPT or AI video creator that I've ever seen. And like, that's, they're going to create movies on that platform.
• Oh, by the way, female CEO, female CTO. Let's go.
• Love it. - That's cool. Hey, what was that thing we were on YouTube the other day playing around with and it warned us about AI generated content or whatever. Do you remember that? Or it was saying,
• Oh, it was going to start tagging videos that were generated by AI.
• Completely. And we were, I think we were, we were more thinking like they're going to, this kind of stuff. It's like completely, - Versus like Opus clips.
• Yeah. Cause just for the listeners out there, we use AI to augment some of our videos to add, you know, like, - We use AI for
• A lot of things. - Everything we possibly can.
• But it's real video of us. That's just being, you know, edited slightly. Whereas like, I think YouTube's - Except for Jacob. He's not real.
• I'm not real. - He is an AI generated character.
• Follow me on Instagram. - So here's a funny idea. This is actually something that Aaron brought up. What if as a part of some of our guests on our podcast, we had, we interviewed AI. - That would be fun.
• Yeah, that'd be cool. - Like we had fake guests.
• So we interviewed like, I don't know, some like deceased awesome person. I mean, you know, Napoleon. - Yeah, we could totally, we're going to do it. We're going to do it. - Because that would be huge.
• Yeah, we're going to do it. - That'd be cool.
• So, and we could absolutely just leverage chat.gpt's voice.
• Can we, can I, can I, can I recommend the guest?
• What if we interviewed Napoleon? - Bobby Knight.
• Oh yeah. - Let's interview Bobby Knight.
• We need to get all the prompt in there. All right.
• And make sure there's no chairs around. - That would be, yeah.
• I just got that. - And definitely it's going to be an R-rated episode. - That would be a very cool.
• Just Google Bobby Knight in a sand trap before you go home and watch the video.
• Right. - It's the worst thing I've ever seen in my life. - We've been, so both, so my wife is learning Italian. My son has been doing Spanish for a long time and he stopped paying for some app because he just wasn't using it enough. And I'm like, well, what's the experience like? He's like, well, I was just able to have a conversation with it. I'm like, well, shit, I think we could totally do this with chat.gpt. So we just launched the app and said, hey, I want you to act like a Spanish tutor and give me a challenge one at a time. Let me reply. Well, we know tweaked it a bit. And so, yeah, and we did the same thing for my wife for Italian. And they were just going back and forth, having full conversations with these people, being like, how would you say blah, blah, blah, blah. - And it's just a text response?
• Well, and so, but you're using, you launch it in the audio conversation. And so then it's like, how would you say this in Spanish? And then he says it in Spanish and it's like, oh, that's either right or wrong. And again, now that's, I mean, it's not free because we're paying for, you know, plus to have that functionality. - I think Duolingo came out recently and was talking about how much they're leveraging. - Oh, they've got to.
• Yeah, well, but to your point, I mean, it's how accessible is that? You can just do that in chat.
• Yeah, so, and that's kind of that piece is that a lot of these-
• Is chat.gpt the ultimate, if it can be done in chat.gpt, it will be done in chat.gpt?
• Probably. - Yes.
• Yeah. Someone surely has created a gpt for it.
• Yeah, but we're going to be the first podcast to interview ancient heroes.
• I wonder if anybody's done that. - It could be drunk history, right? Do you remember, did anyone ever watch that show?
• Oh yeah. - Yeah, but really we're drunk and we're talking to historical figures, right?
• I think so. I've been reading books on, I just have this delusion that I can, yeah, that's a-
• I'm reading book. - I read one single book.
• First 30 pages of a book. - I read a book.
• Which is all you need usually. But it's all about writing movies and screenplays. And with these context windows that are so huge now, do you think, because the books I've been reading about what makes a good story arc and plot and subplot and what, basically what makes a good story, which is plot subplot, right? Do you think that there could be a gpt that could take in an entire book and analyze the structure of the plot and analyze if it was good or bad? - I would assume that you could with the turbo because I mean, that's 300, what's it? 300 pages, I think is kind of what they're saying for the four turbo model that as long as it's less than 300 pages, it shouldn't be able to take it. - I mean, think about if you were, think about people that read screenplays, actors. - Oh yeah.
• Like people all in Hollywood, right? - Right.
• Think about if you could literally just review every screenplay that's in your queue in 10 minutes and find out which ones are the good ones and which ones are the bad ones.
• So this is kind of full circle 'cause like, I don't know, 1600s, they were trying, the first like generative AI-ish stuff that they were trying to do was like creating the perfect story, right? And it was like a choose your own adventure, but one word at a time kind of thing. And the whole process was like, we know, well, I'm gonna butcher this, but like, they kind of knew like the structure of a plot, right? And like how that would work. And then they built these like big, just manual word lookup things. So it's like, they've been trying to do this kind of stuff, but now you're just on the other side, right? Like, I was like, I wanna consume that, but with AI.
• Well, there's a lot of, I think there's a lot of generation, but I like to know about how to analyze if something is good, something is bad, and it's auto-correct.
• Yeah, that's an interesting question because good and bad is so subjective to the individual that can you...
• Netflix will be like, will this be good for this sub-segment of people? 'Cause that's what they already do is like, they just try to...
• Well, think about it, if Netflix has that data, right? And they know, and they could basically turn these stories into math, they could easily create plots and screenplays from AI, and then feed that into an AI and literally produce the movie. And it could be a movie that people would want to watch.
• That's the Black Mirror TV show that we were talking about last week, right? So I think it's episode one, I think it's season five of Black Mirror is what's out now. And episode one is the premise that basically, and it's Netflix, I mean, they don't call it Netflix, but it's got the logo that looks kind of like Netflix, the N, you know, the boom, boom, and the colors and all that stuff. And that they're dynamically generating the TV show in near real time based on this one person who they're like, yep, that's who's gonna be our actress. And they're using our phone data and all this stuff. And just because she signed up for the service, she agreed to basically let them use her phone and her data to basically be able to use a story.
• You know what would be a show that everyone would watch every day? Would be a dramatized, how the hell do you say?
• Dramatized show about you.
• Dramatized show about their day. So like, you know, someone shows up to work and there's a punch in the clock and do it. But like, it's taking in that data, right? And then they go home and watch like a fantasized version where when their boss gets in their face, they're like, I don't even know what they're doing. It's gonna be crazy.
• Right, okay. So, okay. So I've kind of, I have this kind of. So I've been doing where I now track my, I track, so I, yeah. I can say, hey, Siri, track. And it'll be like, what do you want to track? I'll be like sleep. And it'll be like, what value? And I'll be 7.5. And would you like to leave a note? I'm like, yeah, I had a rough night of sleep. Right? And so I track all my stuff kind of throughout the day and I can just go to my large language model and basically be able to like, write a story about my day. And it does. And it's like, you woke up at 7.50 and you had a good night of sleep at 7 blah, blah, blah. And then you did some pushups and then you had some this, da, da, da, da. And it's just like, this is actually kind of cool. Like, I mean, it's telling me a story of my day.
• That's coming. So think about what Apple's gonna do with their AI about.
• In the journal. I was gonna say, I think Apple wins again in this situation. There's probably a world, if you really think about it, take a step back. There's probably a world in the future where like, we're all talking about content and like creating on social media where the movies you're watching, the experience you have on a day-to-day basis in the digital world, whether it's on your phone or it's like overcast on VR is going to be completely unique to you and what you want to see. A little kind of correlation is Aaron told me a story about he was in the airport and there was like this new thing in the airports where like on the screens, if you look up and like, it knows, you scan a thing and you stand in front of the screen, everyone else is like, there's like different dots along this path and like this big screen. And if you scan your ticket, it'll show you like where you're supposed to go in the airport, but everyone's looking at the exact same screen, but seeing something else. And that's like, exactly. So it's using like the angle. Based on where you are, you're seeing something else on the screen and someone else is also seeing something else. That's what it's going to be. Have you ever seen that movie? Minority Report? You've never seen Minority Report? I have not. Okay, tonight you literally need to watch Minority Report. That's one of the greatest, most creative, not greatest movies. It's one of the most creative movies. Yeah, it's an excellent movie. And I will say, as much as you want to talk about Tom Cruise, Colin Farrell, stud, and then maybe the FBI guy or whatever it is, stud. He was ready for that one. Yeah, do we know why they printed them onto the wooden balls though? No. But there's a scene that, where one of the pre-cogs, so it's basically this concept that you have these people who have precognition of what's going to happen, right? But there's one where she's like, "Run!" And I watched this movie with my daughter when she was like, I don't know, like eight or something like that. And to this day, if I say run like that, she just gets chills down. She's like, "I can't." That just scared the shit out of her. So basically Tom Cruise runs the pre-cog crime unit and they catch criminals before they commit acts. And basically there hasn't been a murder in LA or whatever in 20 years. And the first scene in the movie is this guy's about to commit a crime of passion that's super intense. And they come in and they bust his ass right before he kills his wife. It's crazy. It's such a good movie. I've got to pee really quick right back. Such a good movie. Yeah, I'm down for it. So I have a question for you, Brandon. If I'm a person and I want to write an entire book using ChatGPT, how would I go about doing that? So the first thing is I think you need to ditch chat.openai.com and go to the Playground. And so the Playgrounds, I think it's like a beta.openai.com/playground. And what the Playground gets you is you have the option of picking which models you're working with. You're able to adjust your top K, your top P, you're able to adjust how many characters are output. Where when you're in just the chat interface, you don't have any of that luxury. And so the first thing that you need to do is you need to ask it to generate you the outline for your book. So it's going to generate your outline. And then you're going to take that outline and you're actually-- so the other thing that the Playground has is it has a system setting that's over on the left side where you can basically just say, I want you to act like a pirate and only answer things like you're a pirate. And so then anytime you ask it a question, it's going to be like, "Yarr, matey." And it's going to-- whatever, it's going to output it. But what you do is then you take that outline and you drop that into the system prompt. And now you say, I want you to generate section 1.1. Section 1.2. Is it going to be able to maintain context the entire time? And it can do it, yes, because that system prompt's always going to be present there. So now what I have found, because I was trying to generate entire podcasts. That was one thing I was playing around with where I was like, I want to just give it a topic. I'm going to generate a huge podcast. And then I'm going to have some audio. And I forget which one I was working with. To be AI to put in our podcast. Yeah, to then generate the voice. So we're being replaced. Yeah, so then we ultimately get replaced. Damn, yeah, we did last week. We did seven episodes. Fuck. But it worked. It was actually-- fuck, what did I call it? The United States of History or something like that. And it was basically I picked a location. I would go do a bunch-- I would have it pull a bunch of research about interesting things that happened in Indianapolis. And then it would write a whole thing. But again, it would. It would start repeating itself. It would start pulling a lot of the same conversation that you'd have in section one would be in part of section two. So that's when I moved it to-- and I started just doing it manually in the playground. It ended up working pretty well. And now that you've got these huge context windows, that I think it stays with it better. But yeah, so the playground is absolutely where you want to be. And you want to do things like I want you to act like a blog writer who has 30 years experience. And I want you to output this. And so you can really give it a lot of-- you can prime it a lot. And you can tell-- and you can even give it examples. Like if it's not outputting things that you want, I want it to be in this kind of format. Or I want this type of thing. Or I don't want this kind of thing. And all of those things go basically into your system prompt. And then you just go over and you start having conversations. No matter what, there's not an answer right now where you can just go and be like, write me a full book about the beauty of the-- I don't know, whatever. I think the future of work is you're not generating the content. You're curating what's coming out of the editor. Exactly. You're copying. You're pasting it. You're editing it. You're mixing it together. There's so many movies that there's just this nice flow, like Space Odyssey. That's more of just asking questions and interacting. But there's a lot of movies-- I think there's even Star Wars episodes where they're working on a vehicle or they're fixing a car. And they're getting information and feedback from the robot. But the robot's doing all the work. But they're the ones that are making the decisions. Yes. You know what I mean? And that's really where I think a lot of this is going is you're, once again, augmenting or taking the lift off of the actual output. And then you're taking that output and then you're refining it and making it better. And just using that to probably, at the end of the day, still come out with your creative input. Right. Yeah. And how you deploy it is also-- that's very important. Totally. You can have amazing content, whether that be articles or videos or pictures or what have you. Right. But if you can't get the eyeballs-- But if you can't get the outcomes, depending on the channel, then it's all good and well. It doesn't matter. Well, here's a question. How big of a disadvantage are college students right now that are being asked not to use ChatGPT? That's-- and it's bullshit. It's absolute-- I'm sorry. It's total bullshit that they're like, you can't use ChatGPT. I'm like, to my son, I'm like, use it. Abuse it. Go in there. Because this isn't going away. This isn't like-- And I actually say that. My niece is a senior at Ball State, and she's in marketing. And her marketing teacher tells him to use ChatGPT all the time. See that-- OK. So that's-- But that's a-- they're thinking-- he's probably from the business background. And he's thinking execution. Yeah. Right. Where you might be in the legal or the medical or the-- you know what I mean? Or even in the writing. Right? Where they're like, you need to absolutely have all this. And this needs to come from you or else you're cheating. Yeah. But I think at the end of the day, they're going to be at a disadvantage. I mean, I think about my 8-year-old and my 10-year-old. I mean, by the time they get to college, they'll have been using AI for years. Yes. Yeah. And they're going to be using them to learn and create things. Yep. Right. And there's going to be people that come out into the workforce that didn't have to learn that. You know, most people do learn it on the job. But let's be honest. But I think that there's probably a thought or a risk of not using AI to create right now. Or at least learning how. Because you just blathered off. And you use it all the time. All these different ways and means to do those things. And so I think it's probably right now, it's like, go start with ChatGPT. Yep. Right? And Dolly, go get your-- pay $20 a month for the extra. Yep. And start using that in your daily workflow. Or a lot of your creatives. I mean, you're going to have a very large advantage over those who aren't using it. Yeah. Now, my son apparently did-- might have gotten called out on one article that he turned in that might have been heavily inspired by ChatGPT. It might have been actually one of his computer science classes. Because he's like, first of all, I hate programming. And I'm just like, oh. But that's fine. Whatever. I get it. He likes Photoshop better. I'm like, all right. Cool. I can appreciate you wanting to be a designer more than a programmer. And he's like, so I'm just having to write ChatGPT. Just output the HTML that I'm supposed to be generating for this class. And I'm like, well, that's what I do. I mean, if I have to go and have it write-- I don't know how many times I need to ask ChatGPT to write me a function that does x, y, and z. And I'm just like, all right. This is fine. Because it does what I need it to do. And so for me, I'm like, there's nothing wrong with that. But yeah, apparently he got in trouble for using it. Not to harp on this too much, but I've said it so many times. Having a fundamental understanding of what you're doing and using AI is the way to go. Yeah. Using AI, and you're using it a lot, and you're only 20 years old. Into a decade from now, you're 30 years old. You're naturally going to just learn how some of this stuff works. Yeah. So I don't know. I mean, I agree. The whole fact-- any teacher who's like, you can't use ChatGPT, because the fact is this is going to be integrated. This is going to be a new tool that everybody's going to be using. So let them use it. Let them figure out how to get better at prompt engineering. Let them figure out how to communicate with these AIs. And to all the people out there that have been asked to do so much at work that have to work 60 hours a week-- Right. IE, I'm telling my wife all the time to do her Sunday messages and all this summarization of all this and cohesion of all this information that is being sent to her. And then she has to blast it in a concise manner to her people. Use ChatGPT. Use tools like that. If you're out there and you're overworked because you have to produce a bunch of stuff, respond to a bunch of emails, you're the number one candidate. Because you can reduce your workload from 60 hours to 35 hours to 30 hours and spend those 10 hours doing something that you actually wanted to do. And it gives you a better output. I'll give founders out there who-- I mean, you guys are all really hacky anyways. But I'll give you a little cheat code. So when it comes to investor updates, oh boy, every 30 days you've got to whip something up and figure something out. So when it comes to investor updates, where you're going out to talk to an investor, and you're going to have the pitch. And then they're going to send you back, typically, if they're interested, a series of questions. And you're going to type out those questions. And you're going to think, this is the best thing ever. But you don't know what you don't know. So you can put that in a ChatGPT. You can have it help you write the answers to the questions. But the next layer is then going into the chat and saying, hey, if you were an investor, what are the kind of questions and concerns you would have based off of these answers? And now give me those. And then based off those questions and concerns, you say, OK, rewrite my email. And you can do that a couple of times where you basically have a perfectly impenetrable email where it's perfectly worded. You're kind of alleviating risk, potential risk you might have created because you said something wild in an email you might have-- shouldn't have said. I've used that before, a little bit of a cheat code. Or to have you tell you what's wrong with it and then have it make it better. And then you do that a couple of times. And then you're really going to be a better-- Yeah. And I think you're better. You're going to be a better founder for going through that process. You should probably also do that, too, where it's like, why was this better? Yep. And you can absolutely be like, I want you to correct it, but I also want you to tell me where I was wrong. And it'll go through. And it's just going to tell you. So the ones I wanted to bring up on the show was copy.ai. Oh, yeah. Yeah. So I started playing around with copy.ai. Copy.ai has a lot of really interesting tools in there to help you do long format where you can copy this to the editor. You can highlight things. You can have it rewrite. But copy.ai, in terms of generating long form content, seems like one of the better tools to check out. It's just copy.ai. Have you used Jasper? It's a similar tool. It is. Jasper's around pre-AI. Yeah. Jasper's been around for a long time. Yeah. I've used copy as well. And those tools are amazing for those longer form generations. Well, guys, we've talked about AI content generation for writers, for marketers, for blog writers, for you insert creative, and we've talked about it. There's a bright and beautiful world ahead of us, especially for Jacob with the famous Instagram model. This is the Big Cheese AI Podcast. I'm Andre Herakles, joined by Sean Heise, Jacob Wise, and Brandon Corbin. We cannot wait to see you guys next week. As a reminder, on episode 10, we will be having guests join the podcast. And based off this, we might be interviewing someone from the past. So check that out. It'll be the first podcast ever to interview someone who's not here. And also, if you have any questions for us in the Big Cheese AI team, whether that's for technical things or just things you want us to cover in the podcast, email Big Cheese. Podcast@BigCheese.ai. Thanks, guys. Have a great one.