Eric Siu and Neil Patel break down how creators use ChatGPT, DALL-E, and Canva together to produce infographics fast and cheap. The catch: AI nails the data and initial graphics, but the visual flow falls apart, so you still need a human to manually fix the design and narrative flow.
All right, so today we are going to talk about the crazy new way of doing infographics at scale. So Neil, have you seen this yet or no?
I have not seen it yet, but I know that you can end up creating infographics with ChatGPT and DALL-E. I have messed around with it, but it sounds like it's much better now than it was in the past.
Oh my God. So in the last episode, you talked about not killing the lead and you just killed the lead and now our retention graph's just going to go right down. So anyway, So ChatGPT and DALL-E, I saw this recently. There's this YouTube channel that I follow sometimes called Income Stream Surfers, and he talks a lot about AI as it relates with SEO. And so he's got some pretty good stuff. He's grown pretty quickly in the last couple months. So good job to him. I think he's based in Italy. And so the whole thing here is that he has worked in the fashion industry. And so what he does is you layer on ChatGPT plus a Canva integration. So Canva is the It's kind of like a Photoshop. It's not really a Photoshop competitor, but it helps you design stuff easily. So banners, ads, that type of stuff, right? And so you layer in ChatGPT with Canva, and then you add in DALL-E, which is— DALL-E is the image generator for OpenAI. So OpenAI owns ChatGPT, they own DALL-E. And so you layer these things together. There's plugins that you can have for ChatGPT. And he just says, make me an infographic. and help me make an infographic with the latest styles or traditional men's styles. And basically this infographic created, it's really cartoony, right? And it's like, oh, here's a lapel and then you need to have cufflinks over here. And then this is a style that you can have. There's 6 or 9 different styles and it looks really damn good. This is like an infographic you'd probably pay a couple thousand dollars for back in the day. And then he's like, well, you know what, ChatGPT? Hey, I don't really like the cartoony style, Make it more like, make it more like you know human, right? Like serious, like more real life. And so it changes the whole infographic, makes it more serious. And then the the the models almost become like mannequin-like, but they also look like humans as well. And it's just like dude, like you can do this. Like you and I, we're not designers, but to be able to go out there and and do this for like pennies on the dollar now, I feel like every blog post is just gonna have these infographics. And like that that is like. The very like beginning of this.
So, dude, so we've tested out creating a lot of infographics through ChatGPT. Here's the issue that we faced.
How did they look? How do they look?
It's not about the quality of the image that was actually the issue. The issue was the flow. So the flow from how you go from one point to the next point and you just continually transition, that's what was continually messed up when we were creating—
So what was wrong? What, what did you guys do when the flow was messed up?
We had to have a human go in and modify it. We were trying to fix it as much as possible with ChatGPT, but it just didn't flow as well. And when I mean flow as well, I'm not talking about one point to the next point. I'm talking about the graphic and the design and how the elements tie in together. That was a big issue that we were facing.
Yeah. So, and were you guys able to edit the infographic at all?
We did, but the editing didn't work as well as we wanted to. But yes, it, In essence, what we ended up doing is imagine you have 5 main points in an infographic. Let's talk about like how colors affect purchases. So you may have, you may start off with like different colors have different meanings, red, blue, orange, and red means anger, you know, green means happiness. I'm making stuff up right now. I don't know the actual meanings of colors. And then you'd talk about what colors resonate with different industries. Right? And that would be your next point. And then your next point on that is when you test out different colors, what did they have on impact on conversions? Let's say if you're doing call to actions or whatever, again, I'm making this all up. The issue we had is when you try to modify the image to make the graphics flow well with each other, it was really hard to modify them. But what we would do is maybe the first image worked out really well and the third one worked out really well, but the second one was eh. We would just delete the second one and redo it manually and then tie them in together. So it would save time, but it was really hard to modify it in a way where you can just make it flow. So the easier thing for us was you just delete the sections of the infographic that weren't working well, and then you just have a human recreate that section of it and tie it all in together.
And when's the last time you guys did this?
Last time we did this was probably A month and a half ago, 2 months max.
Yeah.
We do it every few months, but we're starting to find that a lot of times it's quicker to just have a human do it. Where AI has been really useful for the infographics is the data collection and the research, and then you just drawing it out. So we still use AI to help us come up with the data points for the infographic. And what we mean we still use AI for the data points We do that on a weekly basis because we're creating infographics on a weekly basis for our customers. We're just finding the design element because we're so picky on the design quality to not be that great using AI yet.
So I think the key takeaway for everyone listening to this or watching this right now is, so he did this a month and a half ago, right? Like this stuff rotates so quickly with the velocity and now ChatGPT is multimodal, right? Like you can take a picture of infographic that you sketch out and then have them like work off of that. And so it, It has changed quite a bit. Like, you know, we're talking— if we're talking 45 days ago, we're talking years when it comes to this AI stuff, right? And so your mileage may vary, obviously. You know, it's just because we ran a test today, it might not have worked, or maybe it worked differently, doesn't mean it's not going to be different tomorrow, right? I would say it's significantly different now. When I was messing around with it, it is pretty much there. And keep in mind, ChatGPT still hasn't, as of this recording, hasn't been out for a year yet, right? So this stuff's just going to get better and better over time. And I think, you know, a lot of this content creation lift is going to be taken off. But I do agree. I think for the most part, there still needs to be a human in a loop right now. But, you know, that will be less and less as time goes on.
Sadly.
Yeah.
Sadly, yes. I do think it'll be a time where you won't need a human in the loop. I think the infographic will still come out choppy, but then you can just, through prompts, you can fix it versus having a human All right, so that is it for today.
Check it out. Please don't forget to rate, review, and subscribe. It helps us grow, and we'll see you tomorrow. Bye.
