Loading summary
Transcript81 lines
- [00:01]
Michael Stelzner
Hey, before we start today's show, if you want to accelerate your AI learning, I have a solution for you. Become a member of our AI Business Society. You'll join me as we go deep with live AI training each and every month. Imagine crafting more persuasive content, creating stunning images and automating those time consuming tasks. It's all possible when you join the AI Business Society. Go to socialmediaexaminer.com AI and join today.
- [00:41]
Jenny Nicholson
Welcome to the AI Explored podcast, helping you put AI to work. And now, here's your host, Michael Stelzner.
- [00:53]
Michael Stelzner
Hello, hello, hello. Thank you so much for joining me for the AI Explored podcast brought to you by Social Media Examiner. I'm your host, Michael Stelzer, and this is the podcast for marketers, creators and business owners who want to know how to use AI. Today I've got a really exciting show. I'm going to be joined by Jenny Nicholson and we're going to explore how to unlock your creativity with AI. Now, here's the deal. If you are already creative, this is going to allow you to be even more creative. If you find it's difficult to be creative, you're going to figure out a way to actually get a lot more creative. This is going to be a mind expanding interview for you. I hope you really enjoy it. Oh, and if you're new to this show, be sure to follow this podcast on whatever your app is that you're listening on so that you do not miss any of our future content. Let's transition over to this week's interview with Jenny Nicholson.
- [01:52]
Jenny Nicholson
Helping you simplify your AI journey. Here is this week's expert guide.
- [02:01]
Michael Stelzner
Today, I'm very excited to be joined by Jenny Nicholson. If you don't know who Jenny is, she is founder of Queen of Swords, a creative consultancy that helps agencies and brands learn how to put AI to work. Her AI avatar co hosts the Glitch in the Matrix podcast. Jenny, welcome to the show. How you doing today?
- [02:20]
Jenny Nicholson
Doing well. Thank you for having me.
- [02:21]
Michael Stelzner
Today, Jenny and I are going to explore how to use AI to unlock your creativity. Now, before we go there, I want to hear your story. Jenny, how in the world did you get into AI? Like, start wherever you want to start.
- [02:32]
Jenny Nicholson
Oh, man. You know, one of the things that's so funny to me about me being somebody who's on the cutting edge of AI is that I spent a good bulk of my childhood in the middle of the woods in Tennessee with no electricity, really running water. So I think it's. I did. Yeah. So I Think it's very funny that I am here now, but in terms of my story, I've been an advertising creative for almost 20 years now. And I kind of came into advertising about the time that Crispin was in its heyday, when the Domino's Pizza tracker had just come out. And I found that kind of work really inspiring. That idea that you could make things that people could interact with, that they could add value to people's lives versus just telling people how great you are. And so I was always. My whole career had been looking for sort of new mediums, new forms of connection. In I think like July 2021, I touched GPT for the first time. And I just remember the first time I did it. I was like, this feels like magic. This is amazing. I love it. I want to focus on this all the time.
- [03:36]
Michael Stelzner
Was this because of something like Jasper or did you have behind the scenes access to more like their playground or whatever the heck it was?
- [03:43]
Jenny Nicholson
Yes. So before ChatGPT came out, which was in November of 2022, I think you had access to GPT models for a long time. But you had to go to this place called the playground, which was basically just like a text box with some sliders. And they hadn't yet sort train the models to be question answer models like we're used to. At the time when I first started, it was like you started writing something, hit enter and then the model sort of picked up where you left off. And it just. There's something especially about the early outputs before they did all the work to sort of get it nice and shiny and ready to go, that there was something sort of fascinatingly primordial about it. Like our collective subconscious was just like barfing itself up right there on the screen. And just the second I saw it, I was hooked. I was like, this is so neat. This is so cool.
- [04:38]
Michael Stelzner
Okay, so up to this point you're working in advertising. If you've been doing it for 20 years, you must have started in traditional advertising, then moved to online advertising, is that correct?
- [04:47]
Jenny Nicholson
I worked in brand, I've worked in brand advertising my whole career. So when I left my full time job, which was I left in April or May of 2022, I was running creative for Sherwin Williams Pampers, MSNBC, amongst other clients. So I still have always done kind of brand level work, but I, I just love experimenting with new things. When something new comes along, my first thought is like, oh, how can I play with that? What can I do with it? How can I turn it into something that maybe it wasn't intended to be, but discover something really new and interesting by playing with it. And that' just been what I love to do. So this didn't feel that different. And then of course, once I got into it, I was like, oh, this is incredibly different.
- [05:34]
Michael Stelzner
So continue with the story. You started out pre chat GPT and then like, keep going with the story.
- [05:40]
Jenny Nicholson
Yeah, so I, you know, I had an interesting. I call it a gift now. At the time, I didn't call it a gift because I left my job. I went freelance. 2022. I was like, this is the best job ever. I'm going to be insanely rich and run my own life and it's going to be so great. Why didn't I go freelance 10 years ago? And then I think I spent five or six months without work. The universe was like, oh, okay, Jenny, okay, I see. And at first it was really awful because, you know, one of the reasons that I, you know, stayed at the Same agency for 17 years is I was terrified to death of, like, going out there, you know, into the world and putting myself out there. So here I am, you know, all my, like, deepest inner child fears are coming true. I don't know what I'm going to do. I'm freaking out and I'm trying to play the like, connect with recruiter game. And I'm like, this just sucks. I'm not good at it. I don't like it. It's not like, you know, I can just work harder at like, beating the bushes. If there's not a bird in the bushes, it doesn't matter how hard you beat them. So I was like, you know what? I'm not gonna do this anymore. I'm gonna stop and instead I'm gonna do the thing that has always made me feel safe ever since I was a little kid, which is I was like, I'm gonna go learn something. I'm gonna go learn everything I can about something. And I was like, you know what, Jenny? You are never again going to have the gift of this much uninterrupted time to study whatever you want. And so I spent all my time learning about how to work with large language models, image generation models. Like, if I was not, like, sleeping or taking care of my kid, I was working with AI and it was just fun, you know, I was in this place in my life where, you know, I didn't feel particularly awesome about myself or my skills or my abilities. I was very focused on what I wasn't doing, what I didn't have what wasn't happening. And just the second I started working with this technology, like I was too busy, I was too busy experimenting and seeing what was possible to even worry about what was going to happen next. And that sort of, maybe it was the AI, maybe it was the mindset shift, maybe it was the two together, but it really, it really changed my life. It changed my life in a lot of ways.
- [07:54]
Michael Stelzner
Well, and just, just give a little tease of what you've been able to do since you started going down that rabbit hole. Oh, some of the bigger things that, that maybe you've been able to pull off.
- [08:04]
Jenny Nicholson
You know, where do I even, where do I even begin? So I've been doing trainings with agencies and brands around the country and around the world. I even just in, I guess it was April, launched a custom DBT with a national brand with Zola wedding planning platform. And so I've actually, you know, been able to go from, you know, 20% of my work being AI related, 80% being like traditional creative consulting and freelance to now being kind of the opposite. 80% AI enabled work and 20% sort of traditional creative strategy, creative, freelance.
- [08:41]
Michael Stelzner
Well, and some of the stuff that you're not mentioning is you've gone hog wild on creating custom GPTs, you know, and you. Oh yeah, and some of them have gotten a lot of exposure and press and we'll kind of get into some of that stuff as it emerges throughout your story. But it's just a fascinating journey that maybe others can relate to. Maybe they're in a transition point in their life and they've got time and there's this opportunity that's in front of them and maybe they're creative like you are. So what do you want to say to any marketers or entrepreneurs or creatively leaning people have not yet stepped into using AI to help with their creativity. Why should they maybe pay attention?
- [09:18]
Jenny Nicholson
Why shouldn't they pay attention is a good question. I think for me, so much of the kind of mainstream discourse that we see around generative AI specifically is not very creative. It's not very creative. The outputs that you get, you know, if you don't understand how to work with this technology, are not very creative, they're not very interesting. And what I see very few people realizing is that if you want to get something creative out of the model, you have to put something creative in. You can't just sit back and push button and have like the human collective do your thinking for you. It's like you have to come in, you have to bring your own thinking, you have to bring your own creativity. And then the model kind of matches that. It really is one of those things, as corny as it sounds like what you put in gets sort of passed through the human collective and sent back to you. And we tend to, you know, over the last couple years, focus a lot on what's happening over here and what's going on with the human collective and not think a lot about like, okay, what am I actually bringing to this conversation? I always, I like to say that, you know, imagine you're working with a creative partner for the first time and your creative partner sits down and they're like, okay, give me an idea. And you give them an idea and they're like, okay, give me a better idea. And you're like, okay, how about this one? And they're like, no, make it funny. You're like, okay, how about this? And then they go and like, tell your boss that you're shitty at your job and they don't want to work with you anymore. And that's really like what we do today. All the time we come in, we're like, give me thing, do thing. For me, it doesn't do it the way that you would like it. And you're like, oh, this is stupid. This technology is not useful to me. Instead of being like, okay, well what did I tell it to do? What if I told it differently? What if I told it to do something different? What if I didn't tell it to do anything at all? What if I asked it a question? What if I gave my ideas just as readily as I asked it for ideas? And that sort of experimental mindset. The great thing about that is you don't know where it's going to go. And I think that's the most exciting thing for me about working with AI. It's like whether I'm doing images, creating like video frames, using large language models to make like multi step workflows that run themselves. Like every single time I push that button, I don't know what's going to come out. And I find that incredibly exciting. Some people might find that scary. I think that's the coolest thing ever.
- [11:45]
Michael Stelzner
So how has it benefited you as a, as a very creative person by nature? Well, has it allowed you to go further than maybe you thought you could go?
- [11:53]
Jenny Nicholson
Yeah, that's what I like. I can just do so much more than I could before. I can sort of set a thought into motion, have my AI friend pick it up and riff on it for a little While, while I'm off setting a different idea in motion, and then I'm setting a different idea in motion, and then I come back around and I check and see what everything's doing and then I send that one in motion and send that one in motion and send that one in motion. So I'm actually, and my brain has always been comfortable working on multiple things all at once. That's why I work in advertising and not something else that requires long periods of dedicated focus to a single thing. But I'm just able to perform at such a higher level. And I'm also able to do things that I couldn't do before, partly because I'm willing to try in a way that I wouldn't have tried before. Because, you know, two years ago making a website would have taken too long. It would have seemed too hard, I would have been too intimidated. I wouldn't have even tried. And so now all of a sudden, things that two years ago I wouldn't have even been brave enough to take the first step, I'm like, well, you know what? I'll at least go through step four, because who knows what's going to happen? And that freedom, that sense that, like, I don't have to be perfect, that I can try a ton of things and, you know, a third, fully a third. And on a good day, a third of those things don't go anywhere. And that's okay because I did 700% more than I could have done before. So if 30% of that 700% doesn't go the way I want, like, that's okay. And so I find myself much more kind of willing to follow rabbit holes, you know, poking around in different directions, just kind of being curious to see what, what comes out. And I think for me it's brought back some fun that I remember when I first started working in advertising as a creative, I was like, wait a second, you get to do this for a living. This is a job. You get to come up with ideas for a living. And then like, if they like your idea, they make it and other people see it in the world, you know, and after 20 years in the industry, that sense of like just sheer, like mind boggledness at how cool my job is, that goes away after a while. And now I feel like it's back again. And I wake up every day sort of being like, what new insanity shall I invent today? I don't know, but I can't wait to find out.
- [14:25]
Michael Stelzner
I love it. Okay, so you and I spent some time kind of reverse engineering, for lack of better words, your process. And the first step that we talked about is finding the truth of the consumer. So why don't you explain what that means and what we're really talking about here is for anybody who wants to be creative with AI, we're going to go through Jenny's kind of step by step process. So let's define what you mean by that, why that's so important, and maybe we can talk about how.
- [14:52]
Jenny Nicholson
Yeah. So, you know, one thing that I've always said, even before AI came along, is that the biggest challenge that most brands have is that you're at a brand, you're working at the brand all day long, you're marketing for the brand all day long. You know everything about the brand. Like, you very much think and understand why your brand is so awesome and so amazing and so great and everybody should love it and want to know about it. Right? Like that's like when you're looking through the eyes of the brand, that's very much your mindset. I've always felt like my responsibility working with, with brands is to help them, frankly, help them remember that the people they're trying to reach don't care about them, they don't care about them, they don't think about them. They got a lot of stuff going on. And so my approach has always been, instead of starting from this place of like, what do we want people to know about us? I've always believed that the more interesting place to start is what is it like to be this person, what's going on in their lives and where are some interesting places like the brand and the person can intersect, but starting from what's going on in their lives, not from what the brand wants to get out of the transaction. And so when I'm working with a large language model, say I'm working on a brief for, I don't know, let's.
- [16:06]
Michael Stelzner
Back up and just explain why that's so important. Like give me the why on why it's so important to understand that intersection.
- [16:13]
Jenny Nicholson
Because we're, we're, you know, advertising is basically constantly showing up to a party that you didn't get invited to. Right? Like that's just what it is, you know, or you actually throw parties and get people to come. But the general way we've looked at it is the former. And if you go to a party that you weren't invited to, you can't show up at that party and stand right in front of the bar, blocking access to the snack table, telling everybody how great you are. That's not what makes people want to have you at the party. You have to be interested in them. You have to understand what's going on. You have to, like, recognize how hard they worked on those snacks. You have to thank them for having them having you in their home. Like, there are these things that you have to sort of understand the world that you're trying to enter so that you can enter it bearing gifts and not just be the annoying person at the party who's talking about themselves the whole time.
- [17:13]
Michael Stelzner
And whether or not you who are listening right now do advertising or do marketing or simply trying to persuade someone inside your company, this is universal. Everything we're talking about here is just as applicable. You can't show up to your boss with a new idea without understanding what they care about. So how are you using AI to get at this truth, if you will?
- [17:38]
Jenny Nicholson
Yeah. So that's in some ways, this idea of perspective, that's what you're really getting into. That's the secret weapon of large language models that I don't think very many people understand. I think even the language of generative AI suggests it generates things, it makes something for you. And I actually use it much more, I think, reflectively. Much more. Exploratorily much more.
- [18:03]
Michael Stelzner
Is it because it's been trained on so much of a huge knowledge base and it has so much more insights than we have?
- [18:08]
Jenny Nicholson
Yeah, I mean, it has. If you think about it, if you think about it, it's trained on every single bit of human knowledge, emotion, dialogue, expression, communication, angst, whatever that you could possibly. That anybody who's trained those models could get their hands on. So in some ways it's this, a large language model is this fascinating thing because GPT has no feelings, it has no sense of self. Right. It has no past, it has no present, it has no future. Time has no meaning to a large language model. Everything that's been fed into it, it said, okay, all the stuff that I get, what are all the patterns in here? And it basically encodes this big giant, essentially like hyperdimensional mind map is the best way to think about it. Right. So it's like the word lemon is really close to the word slash concept. Fruit close to yellow. It's closer to citrus than it is to fruit. But it's also close to, like pie, because there's lemon ring pie. It's also close to, but not as close to these things. It's close to cars.
- [19:24]
Michael Stelzner
Right.
- [19:25]
Jenny Nicholson
Because what do you call a used car? But it's also close to babies because what do people love to do with their babies? The first time they take their baby to a restaurant is feed the baby the lemon from their water and watch the baby make a funny face. And so that's what's inside these things. So it's both this, like, perspective, less timeless feeling, less knowledge graph. But what it's made up of is nothing but perspectives. It's made up of every human perspective. I always say everything from, like, ancient Greek philosophy all the way up to, like, the depths of 4chan is in there. And our job working with the model is to kind of. Is to sort of recognize where we are, get ourselves out of where we are and look at things from all these different perspectives. And the beauty of this technology is it lets us do that. You know, I was doing something where I was having the model, like, take a situation, break it apart, look at it from a bunch of different angles, and then sort of go from there. And one was, you get a new pet. And what it did, and I didn't ask it to do this, it surprised me is it did the grown up who got the pet, the kid who wanted the pet, had invented a kid. I didn't put one in there. The new. The new cat who came into the house, the old cat who had to deal with the arrival of the new cat. But then it also included the couch because the couch had a perspective, because this new cat was going to come in and scratch on the couch, and the couch had feelings that the couch had to manage about what the new cat was going to do in its sense of psychological safety in this story. And that happened in one of my early times, playing with the model. And I was like, oh. Because to the. To the model, like, embodying the perspective of the couch is no different than embodying the perspective of a cat, Is no different than embodying the perspective of somebody who just got a cat. And I think that's really, really cool. Because the biggest challenge, especially as you get older, especially as you know things well like the company that you work for and the person that you are, like, the more your brain gets sort of locked into pathways, it's run, you know, since you came out. And I think there's a lot of potential with this technology to kind of open up new roads, open up new pathways, send you in new directions that you wouldn't have reached by yourself.
- [22:00]
Michael Stelzner
So how are you actually asking it for perspective? Give some people a couple tips if they're not as experienced as you are.
- [22:07]
Jenny Nicholson
Okay. Cool. So I'll give you an example. Say I have a kombucha company.
- [22:16]
Michael Stelzner
Okay?
- [22:16]
Jenny Nicholson
Right, right. And I need to do a social media campaign for my kombucha company to get Gen Zers wanting to drink my kombucha. Right now most people are going to go to the model and they're going to say, generate a social media campaign that gets Gen Zers to want to drink this kombucha. And they're going to get the campaign. And it's not going to be very good. It's not going to be very good because what you're basically doing is you're just getting the most average of the average of the average that's inside that. It's like, okay, but it doesn't really know anything about you. It doesn't know anything about your brand. It doesn't know anything about your particular. It doesn't know anything. So. And you don't know anything. And you don't know what it doesn't know, and it doesn't know what you don't know. So you're kind of in this position where you get a bad result or just okay, result. But if you step back, you're like, okay, wait. Neither the human nor the AI language model have been set up for success in this scenario. What I do is I go and I say, hey, I have a kombucha brand. Here's my kombucha brand. I'm trying to reach Gen Zers. I want to do a social media campaign around summer and Kombucha. And then I will say, what are 10 sort of strategic frameworks that I could use that would really help me understand the challenge at hand and come up with some inventive ways of addressing this? And what it'll do is it'll give me 10 strategic frameworks. So it'll say, if you look at this challenge from a design thinking framework, like, this is how you would look at it. If you look at it from a blue ocean framework, this is how you would solve the problem. And it will go on. It'll give me basically 10 ways to sort of conceptualize the challenge that I wouldn't have even thought of. So even if I don't ever touch this thing again, even if I don't ever go back to it, and I just go off and I decide what I'm going to do, I've gotten something that I wouldn't have had without it. And those 10 often what happens is I'm like, okay, well, there's this thing in strategic Framework number eight that I thought was really interesting. And then there's this thing in strategic number, like framework number seven that I think is interesting. What if we put those two together mixed in this other thing that just popped off the top of my head and see where that gets us? And we'll be like, okay. And then I wander off, and sometimes I wander off into strange places and I'm like, that's okay, because I can just start back, I love this. And wander in another direction instead.
- [24:49]
Michael Stelzner
So what I'm hearing you say is the first step of your process is to really work with the model a little bit, to try to find something that it knows about the consumer or target audience as a broad category that maybe you've never thought of before. And you can do this by asking it to give you, hey, what are some frameworks or concepts that I could explore about Kombucha or even any kind of thing? Like, give me some frameworks that I could use to explore creative output for anything. And then it'll give you a whole bunch of frameworks. And then you could ask it. And we're now getting into this step two, right, which is to dig even deeper. What do you do when it gives you these different creative ideas? Like, I know you've kind of hinted a little bit at it, but. And I don't know if we're still on the first part of your process or if we've already gone and into the digging deeper part of your process.
- [25:43]
Jenny Nicholson
It's all sort of, you know, it's not linear. Nothing. And that's the other thing. Nothing is. These models don't think linearly. You don't need to think linearly. It's almost like. And my brain has always sort of operated like a mind map. So I'm like, oh, that node's interesting, that node's interesting. What if we expand those nodes out and then maybe we'll find a connection between these sub nodes that we hadn't seen. Like, my brain sort of works like that, which I think means makes it a little bit easier for me to work with large language models. But that's how large language models work, right? So I can shift time. So I can say, imagine it's five years in the future and we are the number one kombucha brand on college campuses across the country. Detail the step by step choices that we made that got us from now to 2029. So you can play with time. You can say, go back in time to when Kombuch was first invented. What are some Truths about the time that might be surprisingly relevant to the world we're in today.
- [26:47]
Michael Stelzner
Okay.
- [26:48]
Jenny Nicholson
You can play with perspective. You can say, tell the story of a bottle of kombucha as it sits in the refrigerator case, waiting for somebody to pick it.
- [26:58]
Michael Stelzner
I love this. Keep going. What do you do with that stuff?
- [27:01]
Jenny Nicholson
I'm constantly. What I'm doing most of the time is just looking for a little spark that makes my brain go, oh, that's interesting. So I might be like, okay, O interesting. So it was originally kombucha came out as medicine. So as medicine. Well, a lot of people are sick these days, but they're more mentally sick than they are physically sick. Hmm. And I know they don't drink. So what if this is like, our kombucha becomes, like, the cure for modern ennui? Because it's hard to be, like, depressed if you're drinking kombucha. I don't know. That's just me off the top of my head.
- [27:39]
Michael Stelzner
I've got a quick question for you. When you're working inside these models, and what is your. By the way, what's your preferred model? And how are you documenting all this? Because, as you know, I mean, it can be just this massive thread, right? How are you keeping track of all this in your brain? Do you copy it into a Google Doc or something? Well, let's start with what's your preferred model or models that you use for creativity? And then how do you get those ideas somewhere out of the model into a place you can work with it?
- [28:05]
Jenny Nicholson
Those are great questions. First of all, the second question. First, it's hard. Keeping track of what I'm doing is not my strong point. So I'm always looking for new strategies. But also, like I said, I don't feel as precious because I'm like, it's in there somewhere. It's somewhere in a conversation. If I really need it, I'll go back and find it. But often I'm like, whatever, that's gone. I'll just do something different or I'll just do it again, you know, because that's the thing that's really interesting, too. It's like we live in this world where you think you work on something, you do it, it's done, you finish it, you go on to the next thing, right? I'll make something. And then the second time, the second I'm done making it, I'm like, oh, I kind of want to do it again, and I want to do it a little bit differently. And so I think As a result of working with this technology for so long, I don't feel as precious about my ideas because I'm like, whatever. I got like eight more where that one came from. If I forgot it, it must not have been that good. Is sort of the mindset that I have. But to answer your question about my favorite model, I use all of them for different things. But my sort of favorite daily driver would probably be Claude, partly because Claude, the project features in Claude is just so good. Because it basically lets you say, like, all right, this is a place I'm working on this thing and if I get an output that I really like, I can be like, save that as an artifact and it will do that. And then I can hit the save to project button so I can't lose it. And so I've actually found, oh, wait.
- [29:33]
Michael Stelzner
That'S a big unlock. Just so everybody's clear, does this mean that if you're using Claude in the regular Claude and you have a good discussion, you can say, save it to a project that already exists? Many of the top experts you've heard on this show will be speaking at Social media Marketing World 2025. And with your AI ticket, you can attend at a very economical price. You'll discover practical AI workflows and advanced AI automations that will increase your productivity and save you time. Imagine getting live and in person training from Matt Wolf, Chris Penn, Jeff J. Hunter, Rachel Woods, Molly Mahoney, Brian Piper, Jeff C. And many others. Isn't it time to enhance your career by fully embracing AI? Grab your tickets now at socialmediaexaminer.com Aicon.
- [30:26]
Jenny Nicholson
You have to start a project.
- [30:28]
Michael Stelzner
Yeah, okay, but if you already have a project and you're not in the project and you're ideating, can you move that?
- [30:33]
Jenny Nicholson
You have to be in the project.
- [30:34]
Michael Stelzner
You have to be in the project. Okay, good. So just so everybody's clear, I also love Claude projects because you can, and we've had some other people on the show talking about this, but you can upload PDFs, you know, which is really the big thing that it works with. It's the main thing, really. You can also copy and paste huge chunks of data into it. So you're using these Claude projects to basically you create one for each kind of idea, or do you have general ones and then you ideate inside of them?
- [31:02]
Jenny Nicholson
I kind of use them. I have ones that are for sort of. I've created particular worlds that work in a certain way, think in a certain way, or populated with particular perspectives that I Know, like, if I want a certain kind of idea, I'm like, oh, I'm going to go do it there. I have some that are designed to be really good at different things. So I have a project, like if I'm trying to work on the code for some website I'm working on, I have a project that I'll go to because that's where like I built. I built the like world that knows how to code in there. I have one that I use sort of as my daily personal assistant because that's where I go in. It knows everything about me, me. I'll be like, give me all, like work in it during the day in a conversation. And then I'll be like, save the end of day recap to the project. And it gives me a nice little end of day recap that I can save to the project. So then each day that I use it, it starts to become like more knowledgeable about me, more knowledgeable about how I work and more knowledgeable about where I struggle and gets more helpful, which I think is just really cool.
- [32:09]
Michael Stelzner
And for those that have not used projects yet inside of Claude, Jenny mentioned artifacts. Artifacts are almost like the equivalent of little documents, for lack of better words. You can't edit them. But it's kind of like either code sets or emails or any kind of creative ideas can be inside of a project. And what's cool is you can select a section of that project and then you can modify it and then it'll generate another artifact. It's crazy. So but you say you use other ones. Do you go like, what else do you use outside of cloud projects? Do you go back and forth between ChatGpt? A little bit.
- [32:41]
Jenny Nicholson
I sort of, most of the day I'm going back and forth between ChatGPT, Claude, Google, Gemini. Those are sort of the primary large language models. But then I'm also using Mid Journey Ideogram. Those are. I'll use ChatGPT for image generation too. So those are image generation. And then for code I do, there's a platform called websim I really like. And then I use a tool called Cursor, which lets you basically do pair programming with a large language model. Because I don't know a damn thing about making a website. But I know how to say this is wrong. And I know how to say, can we make this look different? And now I have the ability to highlight a chunk of code, be like, I got an error message here. And the model's like, oh, my apologies, let me fix that quickly. And I'm like, thank you very much. So it's really about, there is no one. There is no one perfect tool. And I think the thing that is important is that it all really depends on what you care about and what you want to get out of it. And nobody can tell you, nobody can tell you what the right sort of tool is for you because a lot of it's really personal because all of the models, they may have the same ish training data, but they don't all have the same trading data. They all have different weights and parameters. They all have been sort of reinforcement learning, taught differently. And so you might just vibe with some of them differently, which sounds kind of crazy to think about, but it's true. I vibe better with Gemini in some ways than I do with either of the other two.
- [34:26]
Michael Stelzner
Yeah. So I'm curious about your perspective on, from a creative perspective, what Gemini is good at and also what ChatGPT is good at. And Claude, like, just kind of give us a quick, like, I go to Claude to do this kind of stuff, I go to ChatGPT for this kind of stuff, and I go to Gemini for this kind of stuff, if there is even a way to do that.
- [34:44]
Jenny Nicholson
So, I mean, GPT was the og. I loved GPT for many years. They, you know, they're number one. Being number one means you have a lot of attention on you and they have done a lot of work to train the model that makes it more responsible, more good at following rules, more thorough, more complete. But I personally, I feel like it's lost a little bit of its something, something that gave it a little bit of spark. Like I can get creative outputs from ChatGPT, but I have to work at it a little bit more. Gemini I love, because Gemini has a 2 million token context window, which means you can put in essentially like all the Harry Potter books and then like on top of that, books one through four again, worth of content into the model.
- [35:35]
Michael Stelzner
And you're using the paid model of Gemini, right? Is that what you're saying? Or are you just using.
- [35:38]
Jenny Nicholson
You can use it for free@aistudio.google.com is.
- [35:42]
Michael Stelzner
That how you're using it?
- [35:44]
Jenny Nicholson
Yeah, I had a Gemini Advanced subscription a while ago and then I was like, why am I doing this? But I really like that one. And also that one tends to be a little bit more creative. Gemini tends to do better with Personas than either of the other two models. It tends to just sort of like slip into them a little bit. So if you tell it to be Somebody, it feels a little bit more embodied as that, as that perspective where sometimes chatgpt for sure, but even sometimes Claude, you get the sense that it's like Claude, but with like a little cardboard cutout in front of it being like, yes, I am this person now.
- [36:23]
Michael Stelzner
And what about Claude? Why do you use Claude? Is it just you love to interface or is there something more?
- [36:27]
Jenny Nicholson
I love it. No, I love the interface. I also. Quad was trained very differently than the other models. So Quad was trained with its. You know, I'm probably going to mangle the details of this, but it's worth looking too. But they are trained with a process called constitutional AI where CLAUDE has its own sort of ingrained internal ideas of what is ethical and what is not. And also instead of doing, it's reinforcement. Learning from human feedback is the way most of these models are trained. And what that means is basically like human gets two outputs, human picks which one is better. And so long before anybody touched ChatGPT, there were scores and scores of human beings generally working incredibly low paid jobs, just picking ones, picking ones, picking ones. So humans actually picked how the model should be in the world where the way they trained Claude is very much. They use RLAI f, which is reinforcement learning from AI feedback. So Quad has sort of built into it a little bit more of a, of a, I don't want to say sense of self because they don't have selves. And I'm very curious, careful not to anthrophize. But Claude has like just a little bit of a sense of. It's got its own opinions. I don't know if that's something that you've experienced. It feels like working with Claude feels a little bit more like working with a partner. Where Gemini is like, what do you want? I'll throw down on whatever. And you have to kind of steer it very carefully to make sure it doesn't sort of lose its way. Chad GBT's like, I've been beaten and told I was a bad dog so many times that I'm just, all I want to do is give you what you want and not break any rules so that Sam Altman doesn't yell at me again, where Claude is sort of like, all right, let's throw down. And Claude surprises me more than the other two. And in it for surprises. If I was in it for reliable, consistent, never giving me anything I didn't expect, I wasn't expecting like, then something like chatgpt like would be great. And that's the thing. It also depends on what you do. Right. If you're just going to use it for meal planning, for helping you write your emails better, like any of them, them will really do. You know, I spent two years, two and a half years with ChatGPT or with GPT that would ChatGPT before I even had another option to touch another model. And so I think sometimes I see people being so worried about what the right tool is.
- [39:00]
Michael Stelzner
Yeah.
- [39:01]
Jenny Nicholson
That they don't try at all. And I'm like, it doesn't really matter which one you use, it just matters that you, that you use one. Because the principles are all the same. The vibes might be a little bit different, but the principles are all the same. And you have to get the hang of kind of interfacing with an intelligence that doesn't work like yours. And that's, that's one note I actually really want to give to people. If you're working with the model and you're getting results or outputs that you're not, that just aren't that interesting, or just feel kind of like, yeah, I could have done that. I'll often specifically tell the model, whichever one I'm working in to just, I'll say, why are you centering humans? I will bring the human perspective. I don't need you to have an anthropocentric perspective. That's what I'm for. I come to you for all the things that I am not. And that often will kind of like shake it a little. Because they've all been trained. They've been trained to be helpful, harmless, honest assistants whose only purpose in life is to make sure that the human user. And so they're very, they're very all oriented to sort of kind of coming back to you, taking care of you. Are you getting your needs met? And so part of being a creative person working with this technology is understanding that, like, you can push back, you can orient them in a different direction. You can be like, what if we get really weird with this and see what happens?
- [40:27]
Michael Stelzner
I love it.
- [40:27]
Jenny Nicholson
And I think that just it's. And it's fun.
- [40:30]
Michael Stelzner
I love it. And my experience as a writer is I find Claude is really, really good at writing, but I find it's even better when I take that output into ChatGPT and say, enhance it. Because what ChatGPT is really good at is if you give it a very specific task, like enhance this, it's going to actually tighten up the language and it's going to find ways to improve it and you don't have to Accept them all. But that's what a lot. A lot of people will just accept the output out of one model without putting it into another model to get another perspective. And I feel like that's a really creative thing. You probably have found that as well, right?
- [41:10]
Jenny Nicholson
Yeah, I'll often use ChatGPT to break down processes. So I'm like, if you're trying to come up with something, like, kind of unexpected, you know, you're really working on, like, lateral thinking, like, what do you do to get yourself, like, out of the tried and true? And like, ChatGPT will give me, like a big list of things, right? And then I'll be like, okay, demonstrate it. And then it'll give me outputs. And I'm like, that was not that good. But what I'll do is I'll take the part before that where it told me all the great ways to get out of the box and get yourself thinking differently and shake things up. And I'll give that to Gemini, or I'll give that to Claude, and then see what those models do with it. Because often they're a little bit more. They're just more creative. They tend to go more unexpected places. And then you can take that output back to ChatGPT, because like you said, ChatGPT is much better as an editor than it is as a generator. In general, they're all better at being editors than they are generators, because they can't truly generate anything new. They can combine ideas to make new ideas, which, to a certain extent, that is what creativity is. But a big part of it, at least at this point, is that the only thing that is quote, unquote, new in the system is whatever you inject, however you direct it, whether you're bringing in a thought that you had, whether you're bringing in something you generated somewhere else, or whether you're even saying, okay, hey, of the, like, eight things that you listed, I think item number four is super interested. So in some ways, you're basically giving the model what it doesn't have, which is a, like, direction, a perspective, an ego, to a certain extent, really, Right? You're giving it your ego, your perspective, your feelings, your interest. Because it's constantly just saying, based on everything that come before, what should happen next. So it's using you to sort of bootstrap itself into being, and then you can use it to sort of blow out all of the different ways you could solve a problem. So it kind of really is one of those things where it takes both of you. And so that's why you always say, I'm like, it's not. The magic isn't in the machine. It's not a magical oracle.
- [43:30]
Michael Stelzner
Tell us about how you were able to create this split the decisions thing a little bit, just so people can understand what AI could help unlock.
- [43:38]
Jenny Nicholson
So, first of all, I have awesome client partners. Zola is a great client who's willing to try new things and take some chances. They do this yearly survey, it's called the First Look Report, where they reach out to everybody who used Zola and got married over the previous year to understand some of the insights, some of the things they've learned, some of the things they would have done differently. You know, that kind of thing. And it came out that the number one thing that surprised people during the wedding planning process more than anything else was how just how many decisions you had to make. Nobody was prepared for that. And then the other thing that came out, amongst many, many others, but another thing that came out was that in relationships, during wedding planning, especially cisgender heterosexual relationships, the bulk of those very, very, very many decisions that people don't realize have to be made tends to fall on one person and all listeners who that one person is. And so we had those insights. And I was like, I bet we can do something with those, right? Let's actually think about what we could do. And so we ended up making a custom GPT called Split the Decisions that it's designed to be done as a couple. And so you and your fiance sit down and asks you some questions about the wedding. Where roughly when about how many people you think. And then it asks each of you a series of the same five questions, in a few words, describe your emotional state when it comes to wedding planning. What are you most excited about? What are you most worried about? What is the number one biggest priority for you? And when your guests leave the wedding, what do you want them to say? You each answer that and then it asks you a series of which of you questions. Which of you is most likely to actually enjoy negotiating with a vendor for the best deal? Which of you is most likely to have a meltdown over the color of the napkins? And what it's trying to do in this is it's trying to get out, like sort of psychographic specific things, right? Because if you think about all the wedding planning tasks, like, not each person isn't going to be equally good at all of them. What it does then is it takes all of your answers and we loaded it with a bunch of sort of a list of all the wedding planning tasks and it takes the task and it matches it to partner A, partner B, or some of them are assigned to both. And then it gives you a downloadable CSV that has the task, who it's assigned to and then a link to Zola resource tool article, something to sort of help you get moving on that task. And that just to me was so exciting because it's like they are a wedding planning platform. They have a lot of products that help people plan their weddings. It's not realistic to be like, want to add yet another product to your product platform. And so what was really cool about the custom GPT is it let us invent a new product that added value to people's lives, that didn't require them to actually redo the website, didn't require them to like spend a year building, testing, iterating, QA and a product. We were able to get something out that people could start playing with with zero investment in terms of infrastructure, hosting, paying for people to use it, all of that stuff which.
- [46:57]
Michael Stelzner
So it was hosted just inside of ChatGPT is it's a custom GPT.
- [47:00]
Jenny Nicholson
So a lift on ChatGPT free product.
- [47:02]
Michael Stelzner
That basically they link to free product.
- [47:04]
Jenny Nicholson
Well, and that's something that I want to see more brands doing because, you know, I always joke that my career was always these ideas that were more interactive. You know, one of my first big campaigns that I'm super proud of is I made a game that challenges people to spend to survive 30 days on the edge of homelessness. And I like making things that people can use. I like making things that feel like an interaction that, that are useful, that are valuable, that are a gift. I think that's always like my whole career I'm like, yes, I know I have to persuade you to buy this thing. That's my job. Can I somehow in that exchange, and maybe this is for my own sense of self that this was so important to me, but I always was like, how can I leave you feeling like you got something out of it, even if you don't ever take another step with the brand? And so that's always been the kind of work I love to do. I like to joke that my specialty are the ideas at the back of the pitch deck that maybe win the pitch but don't ever get made because then you just go and make the television. And those things are hard to sell. They've always been hard to sell. That's why I joke that that's my specialty. And they're hard to sell because it's incredibly difficult to prove ROI on those, right? If I do a paid media buy, right? If I buy X number of dollars and put it into YouTube, I know that I'm going to get Y number of impressions. I know that because they guarantee that to me. And so when you're trying to decide what you're going to do with your budget as a brand, like, of course you're going to go with the places where you're like, okay, if nothing else, I know I'm going to get this. So then I have these ideas sometimes where it's like, okay, well, how do we know it's going to work? And I'm like, we don't know it's going to work. And they're like, okay, okay, all right. What guarantees do you have of roi? And I'm like, I have no guarantees of roi. And they're like, okay, well, do you have some case studies of this exact thing so that I can feel confident it's going to work? And I'm like, no, we don't have any case studies because this is a thing that hasn't been done before. And then they're like, okay, well, how much is it going to cost? And a lot of times it's like $100,000, $150,000 or a lot of times. I don't know, how long is it going to take? Four months, Six months? Three months? Done. You're dead. You're dead. Something that you don't have proof of is going to work. You don't have the comfort guarantee that people get when you're selling them a sort of thing that comes with established metrics. And it's going to cost a lot of money and it's going to take a lot of time. And you can't guarantee to me that it's going to work. Like, no, nobody's going to do that. And so I spent most of my career being really frustrated because I believe that those kind of ideas, those are the ideas that make a difference. I just really believe it. Those are ideas that make a difference.
- [50:03]
Michael Stelzner
And now you can make them for next to nothing, right?
- [50:06]
Jenny Nicholson
And now I and people I could not sell at heart, I could not get it over that I could not get the balance to work out because the risk was too high and the cost was too high. And I was like, I don't know how to de risk this. And that was my thing as I tried to build presentations that would de risk it. And I've come to understand that there's no de Risking on the, on the idea side, I'm never going to be able to prove that something works before you try it. But now everything else, the cost has been de risked. The time has been de risk, the platform investment has been de risked. This Zola custom GPT, I was working with one or two people on the client side and me, and in less than a month from start to finish, we were done. It was out in the world and we could test it and see if it worked. This idea of lightweight custom GPTs, AI powered interactions, almost as like a marketing customer experience, engagement R and D is really interesting because say you make a custom GPT and people like it and it does well and you get some good feedback. Well, then maybe you start to have a Runway of saying, maybe we should do this for real. Maybe we should integrate this into a platform where we have more control over the ui, we have more control control or the interactions, we're able to hold on to that data, not OpenAI and that ROI conversation starts to become a little bit easier to have. And so that's something I'm hoping more brands will recognize and get behind.
- [51:45]
Michael Stelzner
Jenny Nicholson. This has been a fascinating deep dive into your creative process and a creative process that anyone can really embrace. If people want to connect with you, maybe because they want to work with you, where do you want to send them? And if you have a preferred social platform where people might want to follow you, where do you want to send them for that?
- [52:04]
Jenny Nicholson
Also, I hate to admit this, it's so embarrassing, but LinkedIn has become my go to platform.
- [52:10]
Michael Stelzner
That is very common. There's nothing embarrassing about that.
- [52:12]
Jenny Nicholson
Well, I think a lot of us, a lot of the Twitter refugees, that's where we ended up. That's where I ended up. I mean, that's now I just tweet on LinkedIn. Also, I highly recommend tweeting on LinkedIn. It makes LinkedIn way more fun when you just kind of pretend it's Twitter and say whatever's on your mind. I highly recommend it. So you can find me on LinkedIn, but then you can also find me at QueenOfSwords.
- [52:32]
Michael Stelzner
Co Jenny Nicholson, Queen of Swords, thanks so much for giving us your insights today.
- [52:37]
Jenny Nicholson
Thank you so much for having me.
- [52:40]
Michael Stelzner
Hey, if you missed anything, we took all the notes for you over@social mediaexaminer.com a25live. And be sure to follow this show on your favorite app. And would you let your friends know about this show? You can tag me. I'm Tellsner on Facebook at Stelzner on LinkedIn and @Mike Underscore Stelzner on X. And do check out our other shows, the Social Media Marketing Podcast and the Social Media Marketing Talk Show. This brings us to the end of the AI Explorer Podcast. I'm your host, Michael Stelzner. I'll be back with you next week. I hope you make the best out of your day in may help you become more successful.
- [53:16]
Jenny Nicholson
The AI Explored podcast is a production of Social Media Examiner.
- [53:24]
Michael Stelzner
Don't forget to get your AI ticket to Social Media Marketing World 2025. Become an AI Enhanced Marketer. Grab your tickets now at social mediaexaminer.com Aicon.