Episode Transcript
[00:00:00] Speaker A: Foreign.
[00:00:05] Speaker A: Welcome to ChatGPT Curious, a podcast for people who are, well, curious about ChatGPT. I'm, um, your host, Dr. Shantae Cofield, also known as the Maestro, and I created this show to explore what ChatGPT actually is. Really, though, are the files in the computer, how to use it, and what it might mean for how we think, work, create, and move through life. Whether you're skeptical, intrigued, or already experimenting, you're in the right place. All that I ask is that you stay curious. All right, let's get into it.
[00:00:38] Speaker A: Hello, hello, hello, my curious people, and welcome to episode 22. That is Lex's favorite number of chat GPT curious. I am your grateful host, the Maestro, and today we are talking about the fact that JGPT will never write like you. And that's a good thing. Perhaps some of you let out a little sigh of relief when I said that, uh, in sitting down to. To outline this episode, I went back and forth in the title.
The. The one, the knee jerk one, the Shoot from the hip one, is obviously how to get Chat GPT to write like you.
But. And my goal was like, okay, I'll. I'll make that the title, and then I'll do a bait and switch, because y' all already know me and be like, it'll never write like you. And that's a good thing. And here's how you can train it, because I do. We're. I am going to offer up some. Some tactical things, but I sat with it and I was like, nah, that ain't it. Let's just hit him with the truth from the jump. And then I can still go into some of that technical, tactical stuff as it relates to training Chat GPT to sound somewhat like you, because I do believe there are some legit, you know, ethical use cases for that, uh, which is, namely, when you ask it to rewrite something, you'd want it to regenerate something or generate something that you'd actually say. So I have no qualms with people being like, I want it to sound more like me. Yeah, I get it. And so we're gonna talk about that. But, uh, this episode was inspired by the Dope Workshop webinar that I just ran. Chat GPT for online business owners. Apologies there for that little whistle that just happened. I have a gap, and sometimes I get excited and then the air comes out and it whistles. But the workshop was dope. It was amazing for me. If you were there. Thank you.
Uh, if my commercial last week about it scared you so I'm sorry. Uh, Amy, I'm sorry. I know that little scratch that was in there, I thought I had it down, the volume down enough, and I didn't. So apologies. Um, but if you were at that workshop at the webinar. Thank you. Or you watch the replay, you're a better person than me. Thank you. Uh, clearly I really enjoy this stuff and it is absolutely the direction that I'm moving my business in. And I am just exceptionally, extremely grateful for those of you who are early adopters and you're here with me and you're supporting me and you're clapping for me and you're showing up for me and you're trusting me. So thank you. Same, same for all of you listening to this. Thank you.
Uh, but for the workshop and for all the workshops and webinars that I do, I sent out a Google form questionnaire beforehand asking folks what they want to learn.
And the number one question was about how to get ChatGPT to sound like them. And I get it. Right, right? My people are good. You people listen to this. You're good. So I never, I didn't interpret that ask as how do I fully outsource my voice and thinking so I never have to do any work again, but can maintain the trust with my audience and run a successful business. Like, I didn't interpret it like that, but I do know that there are folks out there that be asking that.
And I don't love that question, but here's my two pennies. CHAT GPT will never write like you. And the number one reason for that, you're alive and it isn't. I will forever continue to be amazed by the math that is required for ChatGPT. Like, it is insane.
And at some point I will take a deep dive into the overlap between like the computer math that's required for this and that LLMs do and then. And the, the calculations that are going on in our brains because there's overlap, there's similarities, right? Our brains are even more amazing than these LLMs and, and gpus and tpus. Like, our brains are amazing, right? But my point, my original point, LLMs are not alive. They are not sentient. Right? And you know, a little side note, the requirements for sentience and for self awareness, that really does intrigue me. But again, a deep dive for another day. But ChatGPT will never write like you because your voice is like a fingerprint, right? It's like a fingerprint. And that it is uniquely yours.
This, this voice you have, it's shaped by your lived experience, by your brain, your knowledge, your intelligence, your world view, and what ChatGPT calls the internal physics of how you move thoughts. Yes, I had a chat with chat about this and it was really good. And I was like, that is a really interesting phrase that you just said to me.
Uh, and so we did a little dive into that. But the internal physics of how you move thoughts. And I was like, all right, cool. But that's unique to you. Computer didn't have that. Right? So, yes, number one, your voice is like a fingerprint. And then it's uniquely yours. And it is also, I would liken it to a river because it moves and it changes, right? It loops. Water again. You are alive. So all of those things that I just listed before, they're dynamic, they're changing, they're growing, they're evolving, they're shifting from moment to moment. Right? You have all experienced this before. You write a caption for Instagram and, uh, you write it on Monday, but you don't share it because it's late or whatever. And then you come back to that caption on Thursday and you're like, I'm gonna post it. Or maybe you come back on Tuesday and you know, I'm gonna post it now. And then you're like, actually, I hate this same person, but things have changed, right? Uh, and LLM can't capture that. And I hesitate to say the word yet. Right? And LLM can't capture that, period. I don't want to say the word yet because I'm not daring anyone to achieve this. I don't want this. We don't need this. We don't need it to be sentient. We don't need self aware AI. We do not need AGI. We don't need Skynet. We're good right here.
[00:06:18] Speaker A: I would, however, like to offer up a few tactical things to help ChatGPT sound more like you as it is. Right. For the LLMs as they are, I would love to offer you find listeners, uh, a few tactical things to help it sound more like you're a little, little less robotic. Because I'd be lying if I said that there wasn't, you know, some utility in, in, in doing so.
Because, yes, we're gonna use CHAT to assist us. Not to replace us, but an assist that has your vibe and your values is way fucking better than the alternative. We know that whether it's a human or a robot, an assistant that already is in line, it's gonna sound like you, at least closer.
That's nice.
Right. It's easy. And for, for anyone, I don't get any pushback because this is a podcast. Nobody's arguing with me. But if your brain goes to like, well, it's nice to have differing opinions or whatever, you can do that very easily with ChatGPT. That's the problem right now is that it's like, it doesn't sound like me. They're sounding like that over there.
So the goal here is, and I do think there is utility is getting it to sound more like you, because it's nice to have an assist that has your vibe and your value. So to start off with, and I shared this in the the webinar, ChatGPT is good at three what I call core, core COR functions. Clarifying, organizing and refining.
If the thing that you want it to do is outside of those three domains, it's probably going to be pretty shitty.
It's not probably. It's definitely going to be pretty shitty. That's good to go make a picture. And you're like, that's really bad, right? Writing something from scratch, not good at it. Generating content from scratch, it's not good at it. Not good at all.
But if we consider those three core functions, we can maybe hopefully see a way that we could work with ChatGPT to help it sound more like us, right within those three domains of clarifying, organizing and refining. So the two things that I would say, I'm not gonna make you guess like, oh, what it can do, number one, use ChatGPT a lot. The more you use it, the more data information that it has about you and how you write. And it can then clarify, organize, refine all of that information you've given it.
The second thing is ask it to create what is called a voice anchor. Now, admittedly, I had no idea what a voice anchor was until I chatted with chat about why it can't ever capture my voice. And then it, doing its chat thing, asked if I wanted to create a voice anchor so that it could at least get closer. And I was like, what the fuck is that? And it told me, a voice anchor is a short set of rules, patterns and preferences that helps an AI tool right in your actual voice, instead of some generic or AI sounding version of you, it captures things like how your thoughts naturally move, how you open and close ideas, the tone you default to, what you never want the AI to do, and the internal logic that makes your writing feel like you. It's not a style guide, it's not a list of favorite words. It's a baseline map of how you think on the page.
Once someone builds one, ChatGPT can write with far more accuracy, consistency and alignment to their real voice without feeling cheesy, robotic or off.
[00:09:37] Speaker A: Now, before anyone again, uh, no one comes at me, but I. That's just what came out. If you're thinking, oh, maybe I feel kind of bad. I don't want. I don't want to, like, train it on me.
I hear you and I'm just going to share how I feel. I don't give a.
And I say this because to me, it very much feels like once your is out there, it's out there, whether it's my writing on my blog or the writing that I put on Instagram or anything that's out in the, in the Internet. And yeah, it would be great if, like, that wasn't the case, but it's the case. And so I am not investing my energy trying to figure out how to like, stop that.
Like, I can have it toggled off in the settings of like, no, don't use my, my data for training. But like, these gonna do what they want to do.
And to me, my fight back pushback is not, you know, does not use it. It's get really good, understand what this is, become proficient, gain other skills, and then I can fight back in that in a different way. Uh, so if you're like, I don't want to train it on my voice, that's fine. Don't use it, by all means. My guess is that you don't care that much because you're listening to this podcast. But for those of you that have those feels totally valid. And I don't judge you. I'm not trying to convince you of anything otherwise. Me, I don't give a fuck. I thought it was actually really cool to create this, this voice anchor mainly because, uh, I just wanted to know what it would say. Like, we know this. Humans love learning about themselves. Everybody's taking these fucking tests. And she had him being like, tell me what house I am. I. I don't know. I am not a Harry Potter person. I'm a Lord of Rings person. Not that they have to. You have to be one or the other. But I like high fantasy. And Harry, uh, Potter is not high fantasy. It's not, it's not, it's not my bag. But if you like it, cool. J.K. rowling. But if you like the, the books and stuff, good for you. But people like learning about themselves.
To that end, I was like, I would actually love to have some language around how I write. I like how I write. I consider myself a good writer, but I don't know why. I don't really know what I do. I know when it feels right. I know when this, when. And it was really cool because one of the things that spit out was that I end things when the feeling has ended. Not like I don't remember the words that it used. And I was like, yes, is that, like, that is how I write. And so to capture that, I was like, yeah, that's actually really cool. And to label that and to name that, I thought that was really cool. So I, I went, uh, through the voice anchor thing with voice anchor creation, with chat. Um, and it was cool. Like, no, I'm not saying that I take everything that Chad says as Bible, but I definitely agreed with what it said and it was cool to have language. So if nothing else, perhaps that's a fun drill for you and to get some language around how you write. So again, the more that you use chat, the more that it knows you. And I'm saying this because before it even prompted me about creating a voice anchor, it described how I write. But based on all this that I've uploaded over the past three years, I stay putting my stuff in there, right? I asked it, I don't ask you to usually change things. I actually ask it for, um, what is the word I'm looking for? I ask, I look for comprehension. And I have it give me a summary of the things that I write. So this way I know if the main point that I was hoping to get across actually got across. That's the number one way that I use chat. And, uh, then I also, I use it for bolding, because I hate bolding. And I will let Chad do that. Um, but yes, I have all this stuff uploaded and it knows how I write and it bases it, it based it off that. And I asked, I'm like, why? What do you. Like basically, what proof do you have? And it was like from this email, from this thing. And I was like, oh, okay, you're not lying.
So you know the description that it, that it gave me to, uh, asking that question and why it never sounds like me was like, it was like, because you do xyz.
It was literally because you abcdefg. It wasn't just xyz. And I thought that was really cool. And maybe you will think that is cool too. So to create that voice anchor, here is the suggested prompt from chat. Very simple. Help me create a voice anchor so you can Learn my voice, ask me for three pieces of my recent writing, then analyze them and create the voice anchor.
I actually didn't. I, uh, put that in the prompt and it told me to put that in the prompt. But what was interesting is that, uh, when it asked me if I wanted to create it, it asked me to upload them. So the reason I'm saying this is because I always want to give you folks, like, the shortest prompt and, like, it's not that serious. And I was like, hey, can I just tell them help me create a voice anchor so you can learn my voice? And it was like, yes, but it, but Chat may not ask you for the pieces of content. And I was like, but you did. Uh, so, you know, just sharing my experience with using it and that. Yes, I did upload three pieces of my recent writing.
It asked me as well. It's like, what do you love about your writing? I didn't feel like answering that, so I didn't answer that.
Um, but you upload the things and it'll do its Chat GPT thing and it'll spit out an in depth answer. And you, my curious friend, can do whatever you want with that. I'm not going to tell you what you should do with it. I just think it's cool. You do whatever you want to do with that. All right, I'm gonna wrap this section up real quick. We're gonna go through how I use Chat GPT this week, and then we'll, we'll do the official wrap up. So if you're new here and you don't know, each week I include a section where I say how I use Chat GPT that week.
This week, it's actually about how I use Chat gbt, uh, a few weeks ago, but I forgot to share it.
Uh, and that is I used it to play catchphrase and charades.
Cool. So when Lex and I went back to my sister's house for Thanksgiving, we wanted a game to play that my nephews could play. They were like 7 and 8.
And so, so we settled on catchphrase and charades. We were going back and forth with, with heads up as well. But, like, heads up is really not as good for, like, a group.
So I used my phone and I had chat generate the phrase, the word by prompting it that we were playing. I said, we're playing charades. Well, first we played catchphrase. I'm playing catchphrase. It knows what that is, right? It was training the whole fucking Internet. So it knows what charades are. It knows what catchphrase is. And I said, we're playing that and that I wanted it to give me a new word whenever. You know, spit out a new word whenever we type the word next. And that was it. Like, how fucking cool. Of note, you can also type in if a kid is going, and it will generate easier words and phrases. So, like, whenever I pass it to my. My nephew, they're two different ages, so, like, one reads a little better than the other one. And I'd be like, young kid. And, like, really young kid going now. And then I would type in next, and I would let him hit the little, like, arrow, and then the word would come up, and he could read the word and act it out. And it was, like, really, really cool.
Just as an aside, playing charades with kids is really cool. And playing. Actually playing catchphrase was actually probably a little bit cooler. I think we're both cool, honestly. It's just cool to see their brains work and, you know, they're competitive. And if you guess the word too fast, they get frustrated because they don't realize, like, it's a good thing that you guess the word fast.
Um, but. And I mean, I'm like, that means you gave a great clue. And one of them was. I don't even remember what Jameson said. As soon as you said extinct, we were like, dinosaur. And he was like, I didn't even finish saying it. And it was like, well, it's extinct. And then he was like.
Threw out some number. He was like, Is it some 98 of mammals are extinct? So. And I was like, listen, man, I don't know that we.
All of us just know that the dinosaurs are extinct. So that's the word. And then the. We finished it. We ended up playing the next morning. So he wanted to play. Kept asking, can we play again? Play again? And I was like, yo, man, I'm like, kind of tired. I'm gonna chill. And I actually had work to do. Um, but I was like, if you go to bed, do what your dad says. We'll play in the morning. Before Lex and I had to drive back to the airport. And so we played again in the morning, and we were like, okay, last word, Jameson, you go.
And, uh, the word was. The clue that he gave was jumping in a pattern. And the word. What was the word? The word was hopscotch. And I was. I was like, hopscotch. And I got it. And he was like, yes. And I was like, that's, like, really cool, dude. Like that to watch your brain be able to Come up with this and give that clue. Like, that was a really cool. A cool thing. And so, you know, all that to say, uh, y' all know I love AI, I love LLMs, I love generative AI, but there are things that it can do, like this, that break, you know, that's part of bringing people together. That's using it for in person time as well. So I just wanted to share that. I thought it was a really cool use case. And as always, I'd love to hear from you how you're using it.
Feel free to, uh, don't email me. I really actually never want emails. But you can text me. 310-737-2345.
You could shoot me a DM. The. The account for this. This podcast is ChatGPT curious podcast. But I'm never really over there.
Uh, so best bet is to hit me up at the movement. Maestro, I would love to hear how you're using it. Kate had hit me up.
Um, Amy hits me up. And I do love hearing what you're doing with it. So shoot me a message.
I appreciate you, but let me wrap this up. That is all that I have for you today. I got a little excited about charades and catchphrase, but hopefully you found this episode helpful. If you did consider leaving a little review, a little rating. We're up to 21 ratings.
21. Let's fucking go. Y' all are the best.
Don't forget, I also have a companion newsletter, the Curious Companion, that drops every Thursday. That is basically the podcast episode in text format. So if you prefer to read or you just want a written record, join the newsletter fam. You can head to chatgpt curious.com forward/newsletter or check out the link in the show notes. As always, my friends, endlessly, endlessly. One, um, more time. Endlessly appreciative for every single one of you until we chat again next Thursday, stay curious.
