AI Generated Content

Transcript

0:01

Welcome to the latest installment of Brand of Brothers. I'm Doug. And I'm Johnny. Today we're talking about AI content. All right, let's get to it.

Johnny Diggz: 0:17

All right, so today we're talking about AI generated content. Oh, are we? Yes. It's, uh, it's quite, uh, the hot topic these days.

Doug Berger: 0:25

Have you heard of ai? Um, I, I assume you're talking about Albert Instein. No, it's

Johnny Diggz: 0:32

Albert

Doug Berger: 0:32

Instein.

Johnny Diggz: 0:33

Uh, so chat, GPT. And now there's Claude. There's, uh, you know, Google Gemini there gr um. Uh, deep mind. There's, there's so many of these I didn't even

Doug Berger: 0:48

mention Perplexity.

Johnny Diggz: 0:49

Oh, perplexity. Well, perplexity, we'll get into perplexity. I, I kind of put that in a slightly different category, but probably not necessary anymore. Um, but, uh, so all of these platforms can. Generate tons of content. Um, and that's, that's kind of what they do. Sure. You put, you ask a question and it will spit back, uh, an answer. Sometimes that answer is correct

Doug Berger: 1:18

and oftentimes

Johnny Diggz: 1:19

it's not so much. Not so much, not so much so. Uh, which creates, um, a whole new realm of dilemmas, uh, potential pitfalls, uh, potential benefits. And, uh, so I thought that maybe we should discuss where, what, what challenges that you've seen, where you're using it in your workflow, and what you think, uh, what you, what, what you're recommending to clients. And how they do things. So where do you wanna start? Um,

Doug Berger: 1:48

so I, I'd like to discuss a few things. So first, the benefits, um, then the challenges. Okay. Um, and then the ethics behind all of this. And it doesn't have to be in this order, right? Sure, sure. Uh, so, so benefits, um, actually perspective, so. What is perplexity? It's called perplexity. So in, in this, in this context, what, what I wanna pay attention to is why we're using tools like chat, GPT, and it is a tool, right? It is not supposed to be a replacement for talent. Um, it should be so, so in my world, in the world of branding. Um, a, a typical tool that we talk about is Adobe Creative Cloud and Adobe Creative Cloud includes, uh, Photoshop, illustrator, InDesign. These are all tools and can you get existing materials and utilize those materials for a logo? You can, but should you Right. On a, on an ethics perspective, no, you shouldn't. Right. A, a client isn't coming to you to use a template. They're coming to you to create something, unless, of course, that's the agreement that you've come to where you explain to them that you're using this tool and for the sake of expediency. They're paying you to utilize this tool and a series of preexisting media. So that takes us to things like chat, GPT, right? Where where does chat GPT get its content? Uh, how does it generate its content? Um, I think this is really more for you to speak to than for me. Um, but my understanding is that over the past several years. Open AI specifically has been feeding content to their large language models and then training it, training it with content and, and effectively it's using other people's IP and, and regurgitating it on a contextual basis.

Johnny Diggz: 4:04

There, there are multiple lawsuits, uh, active lawsuits, the New York Times, and uh, I believe, uh. Is it Sandra Bullock, one of the major actresses? Uh. That, uh, no, no, no. So not Sandra Bullock? No. Sarah. Sarah. Sarah Silverman, actually. Oh, wow. Uh, because one of her books, uh, was obviously ingested and, um, and she could get huge swaths of her book. Uh. Regenerated by Chad GPT, which would be copyright infringement. And so, um, I, the, the courts have not caught up to any of this, so it's still kind of a wild, wild west out there. Right. Um, which is why it's, it's important to, to, to note this, especially as a firm who is generating content on behalf of clients.

Doug Berger: 4:55

Right. So. There are are numerous a approaches to how content is, is generated and presented. Um, I, I personally feel like being as transparent as possible is, is critically necessary. So when we talk about utilizing chat, GPT. To our customers. With our customers, we want to make sure that they understand that we're using it as a tool, not so much as a a content generation resource, because that technically. It's replacing a person and on the ethics scale that, that we've put together, um, that gets flagged. That's not in the green zone all of a sudden because when you're replacing someone that is, that's destructive, um, but. Being transparent that's in the green zone, making sure that the customer knows that what is being delivered to them. It depending on how you are utilizing chat GPT, that's safe. That's fair.

Johnny Diggz: 6:01

Now when you say generating content, you're not talking about like the, the, the. The image generation stuff. You, I'm not, yeah, you're talking about the like, generating content for social media posts or,

Doug Berger: 6:14

or even articles. Right? Articles. Okay. It, so there, there's a time and a place, right? So if you are, for example, wanting to employ a search engine optimization and a IO, uh, method, AI optimization, uh, thank you for the TLA. I saw the look on your face. Um, so you'll

Johnny Diggz: 6:34

throw out the, those TLAs and, uh Yep.

Doug Berger: 6:37

And those KPIs and all of those others. Um, so anyhow, and, uh, where was I? Which by the

Johnny Diggz: 6:43

way, it's not a A IO or, uh, LLMO, or, uh, I've seen, uh, g uh, GAI as well. Yeah, sure. So they're all kind of the same concept of this, this, uh, uh, optimizing content for. Being consumed by these, these, uh, engines,

Doug Berger: 7:05

right? And, and so, and it's important that a, as a brand, um, especially as a, a commercial brand, that you can be found both on Google. Regardless of Gemini and that you can be found on chat, GPT because chat, GPT is quickly surpassing, uh, other search engines. People are going to their, uh, LLMs that are in their pocket because they can get faster and, and sometimes more reliable information. However, it's important to implement the 80 20 rule, right? 80% of the content that you're getting from chat, GPT is pretty good. It's not, not necessarily great, but it's pretty good. Yeah. But 20% of it is, it's worse than garbage. Right? Right. I mean, we're talking about primordial ooze and, and so it, it not, not

Johnny Diggz: 7:57

only just, uh, just like one of the good, the things that chat GPT and and these others do so well is confidently lie. It will come hundred percent.

Doug Berger: 8:11

And, and they will glad hand you, they will tell you what you want, what it thinks you want to hear, right? So you can argue with it, argue with it, argue with it, and eventually it will come around to your way of thinking, even though you are completely wrong. Um, but the other, the flip side happens too, right? There have been times where I've asked it for information, and the information that's given back is just patently false. And so having. Uh, uh, not necessarily an expert level, uh, awareness of, of a subject matter, but at least having some degree of, of awareness and education is kind of important because you might put out content that's supposed to be representative of a subject matter expert, and it turns out that it. It's making them look like an idiot. It's making them look like a buffoon

Johnny Diggz: 9:04

because there, there can be, um, telltale signs that you're generating AI content, um, specifically in formatting and stuff like that. Oh yeah.

Doug Berger: 9:14

Because it uses Chicago style format, right? Instead of AP style. And so, you know, all of a sudden look, the telltale sign with chat GPT, if you wanna know if content was generated by chat GPT, not only. Are, is there a rampant use of M dashes? I love

Johnny Diggz: 9:31

the M dash,

Doug Berger: 9:31

but there are no spaces surrounding the M dash either. So not only are you gonna get the M dash, you're gonna get no spaces around them. And so if you see that, that's a red flag, that that means run, that means that someone generated the 80%, but they didn't do the last 20% of the assignment.

Johnny Diggz: 9:48

Well, and that last 20%, I mean this, this, this concept of human in the loop. Like that, that someone, uh, that you're not just generating content and then delivering that to the audience, right? Yeah. You are, you are using the content as maybe a starting point as some foundational clay that an, that an expert is molding into something that's usable, much like, like a desktop publishing software. Does, it's not going to lay out your design for you. I love

Doug Berger: 10:20

how you brought that back. That's nice. So it as, as it relates to, to generative content, my biggest issues are, are, are voice consistency. Right? The, there's a voice to a brand that you're supposed to put forward.

Johnny Diggz: 10:38

And in your tone that you established that in your BrainScape.

Doug Berger: 10:41

Right? And, and, and not only that, um, there's continuity, right? So when you are writing something and you are authoring it, originally, you're not. Just editing something from chat, GPT, for example, and, and not to vilify chat GPT, but if, if, if you are just editing something, then that level of authenticity and consistency isn't necessarily there. But when you write, we all have in an inherent style of writing, and that begins to go away when you're implementing the use of generative ai.

Johnny Diggz: 11:21

Um, let's talk about, uh, SEO for a second. We touched on it already, but, um, I found that, uh, are you seeing any kind of like penalties or can, can, are the search engines seeing when stuff is ai? I mean, if, if you can see that it's AI generated. Are the search engines, are they ranking, are they changing their rankings based on AI content? Are you seeing anything like

Doug Berger: 11:48

that? So it it, it's really hard to say. My gut instinct is yes. Right. There are definitely algorithms that are evaluating content on a regular basis, how it's evaluating the content and, and whether or not it's, it's looking at it. Um, for example, from, from Google. I, I don't know, they might favorably look at, at, at crap that's generated by Gemini and, and you know, I wouldn't put it past them to look for some sort of, of digital fingerprinting on content generated by chat GPT, for example. But I use these tools to evaluate whether or not content we generate, uh, appears to have come from, uh, from ai. And I do that for two reasons. One is to make sure that what we're writing doesn't sound too robotic. Um, that's really the more important part, that there's a, that it feels human. That's the

Johnny Diggz: 12:47

human in the loop, sort of.

Doug Berger: 12:48

Yeah. Yeah. And, and, and the, the other part is exactly what you're talking about. Is there a potential negative? And the irony in all of this is there are times. Often where I author something and I put it into one of these evaluation models and it's like, oh, that's definitely ai. And it was 100% me. Um, frankly, I prefer to take content that I originate and put it into AI and ask it what it thinks of what I did and where there are opportunities for improvement. I, I find that to be the better utility.

Johnny Diggz: 13:23

I think that that's really where. Ai, uh, at least the tools where they stand today are, are the, the really strongest point is as a collaboration tool. As, as a, uh, here's, here's what I'm thinking. Help me flush out this idea. Um, it's like having another person. That, that that can, you can do, yeah. It's a sounding board. It's a sound, it's literally like a sounding board and can, can help elevate your content, take you in directions that you wouldn't have necessarily thought of solo. But, um, but it's still not just, it's not just the tool generating, it, it for you on your behalf. You're actually collaborating with it as a tool.

Doug Berger: 14:08

Yeah. And, and, and frankly, a, as AI evolves. Its utility will evolve, its purpose will evolve. The challenges will also become larger and, uh, and, and, and different, um, but the potential becomes greater. The, the key part here is that we don't forget the human element. We don't forget the purpose of why we're creating this content and that we're not just creating content for the sake of creating content.

Johnny Diggz: 14:40

That sounds like a good, good. Places to, uh, to top. I, I know we have, uh, we want to talk about this, this concept of ethics and ai, but I think we should save that for, for a different, uh, because I, I think we can go a lot deeper than we, we have this time in talking about the ethical dilemmas and the potential pitfalls there, because there's a lot of them and icl, including what you touched on, the, the copyright issue. Um, and, but one thing I wanted to talk about was, you know, I, I've noticed, you know, in, in my personal life, friends of mine that, that start businesses or, or whatnot, that, you know, they, they'll go to chat GPT or one of the ai, uh, image generation tools and make their logo. Oh gosh. And I'm starting to see this a, a lot and, and you know, I, on, on the one side. It, it's, it's a really nice, easy way to generate a quick logo. Mm-hmm. If you're just trying, if you're, you know, we've had

Doug Berger: 15:41

clients do it for us too, where they're like, Hey, I was ideating using, uh, Wally. Right. Per,

Johnny Diggz: 15:47

yeah. Perfect. And we just said ideation and is, is a perfect tool way to use it. So, um, how, but like, I just had a, a, a friend of mine, this is not a client, but, uh, that, that built a logo. She loves the logo. That she built. It is a complex, um, logo that is a single bit mapped, you know, uh, PNG file or whatever chat GPT generated. And, you know, she went to put it on a t-shirt. I. And guess what happened? Oops.

Doug Berger: 16:21

Yeah. It, it has its ups and downs. Yeah. Um, from my perspective, listen, the people that are using AI to generate a logo, um, they are the same people that might use a, uh, an offshore freelancer because they don't want to invest in their brand and they don't understand the value and the benefit. Of working with a professional in that context and, uh, and, and it's fine, right? They will eventually discover there are are shortfalls, right? That, that come with it. Uh, first and foremost, you can't own it, right? You can't. You can trademark it ish, right? Um, it, it's a murky area because if you actually get the trademark, it might be rescinded if it's found to infringe on someone's intellectual property. Um, and, and then there, there are other, other aspects like what you're talking about because it, it's a rasterized image instead of a vector image. And for those who are unfamiliar a raster image, it is the resolution in size and scale. At which it's rendered, whereas a vector image can be scaled up or down, uh, at, at, at a whim.

Johnny Diggz: 17:32

Yeah. I think I, you know, and I, we, we may get to the point where. Uh, an AI tool can generate a layered illustrator file for us. But yeah, that doesn't exist today yet.

Doug Berger: 17:44

Yeah, give it five minutes. It, it actually by the time we hit stop on the record button, it's probably a thing.

Johnny Diggz: 17:50

Well, I think that's a good place to hit stop in the record button, so that sounds great. We'll see you guys next time.

Doug Berger: 17:56

Thank you for tuning in to Brand of Brothers. Big thank you to our presenting sponsor, Remixed, the branding agency, along with production assistance from Johnny Diggz, Simon Jacobsohn, and me, Doug Berger. We can't forget music by PRO. Speaking of not forgetting, remember to do that like and subscribe thing and find us at BrandShowLive. com and follow us on the socials at BrandShowLive.