AI’s Cool, But It Can’t Replace This: Where To Keep It Real in 2025 (Part 1)

In this exciting first-ever guest-hosted episode of the Simple and Smart SEO Show, Sara Jensen and Shea Karssing from Brighter Messaging step in to share their expertise.

They discuss two pressing topics in the world of SEO and digital marketing: why relying solely on AI for content creation is risky and what their predictions are for visibility in 2025.

Sara and Shea explore the nuances of maintaining a human touch in marketing, the limitations of AI tools, and strategies for creating meaningful, effective content. 

Tune in to discover actionable tips and the latest insights for SEO and digital marketing in 2025.

Key Takeaways:

1. AI as a Tool, Not a Creator: AI can assist with brainstorming and summarization but cannot replace the creativity and personal touch required for content creation.
2. The Risks of AI Dependence: Over-relying on AI for marketing can lead to generic, duplicated, and sometimes inaccurate content, harming your brand's trust and SEO rankings.
3. The Value of Human Connection: Authentic human interaction and unique perspectives build trust and differentiate your brand in an increasingly AI-dominated space.
4. Future of Visibility in 2025: Personalization, authenticity, and professional expertise will remain critical for standing out in a competitive digital environment.

Memorable Quotes:

• "Trust is built human to human. You can’t replace that with AI." – Shea Karssing
• "If your graphics look like everybody else's, they’re not going to stop the scroll." – Sara Jensen
• "AI helps with the grunt work but can't replace your creative process." – Sara Jensen

Listener Action Items:

1. Evaluate Your AI Use: Assess how you're using AI in your marketing. Focus on tools that complement your creative efforts rather than replace them.
2. Invest in Human Creativity: Prioritize hiring professionals for critical content like website copy and brand design.
3. Focus on Authenticity: Share your unique story and value proposition to stand out from AI-generated content.
4. Plan for 2025: Develop a visibility strategy emphasizing human connection, tailored messaging, and high-quality, original content.

Links Mentioned in Episode:

• Connect with Shea Karssing on LinkedIn
• Connect with Sara Jensen on LinkedIn
• Learn more about Brighter Messaging 

Chapters

• 0:00 And then the outcomes that come with that.
• 0:02 Of having a brand that's unique.
• 0:03 Because if your graphics on social media look like everybody else's,
• 0:43 Welcome to the Simple and Smart SEO Show
• 1:28 Introduction to AI in Content Creation
• 1:54 The Role of AI in Interviews and Summarization
• 2:31 AI as a Brainstorming Buddy
• 3:40 The Pitfalls of Relying on AI for Content Creation
• 4:09 Detecting AI-Generated Content
• 5:13 The Importance of Human Touch in Writing
• 9:26 Challenges with AI-Generated Stats and Sources
• 13:34 Gray Area Thinking and AI Limitations
• 16:14 The Human Element in Marketing
• 17:15 Trust and Personal Connection in Marketing
• 18:34 The Role of Professionals in the Age of AI
• 20:12 AI in Graphic Design: The Canva Example
• 24:40 The Risks of Self-Referential AI Content

Transcript

[00:00:00] Sara Jensen: And then the outcomes that come with that. 

[00:00:02] Of having a brand that's unique. Because if your graphics on social media look like everybody else's, they're not going to stop the scroll. 

[00:00:10] Crystal Waddell: Welcome to the simple and smart SEO show podcast, where we talk all things, brand building, SEO, helping you connect with your audience, elevate your visibility and grow your business. 

[00:00:19] I'm your host, Crystal Waddell, here to bridge the gap between SEO strategy and real world business success. By bringing you insights, stories, and conversations from the SEO community and beyond. 

[00:00:30] Whether you're an entrepreneur, marketer, or SEO enthusiast, this is your place to learn, share, and build a brand that stands out. 

[00:00:38] So grab a coffee or your favorite tea, and let's dive into Smarter SEO for your business. 

Welcome to the Simple and Smart SEO Show

[00:00:43] Hi, and welcome to the Simple and Smart SEO Show. I'm Sara Jensen. I'm here with Shea Karssing, and we are with Brighter Messaging. 

[00:00:53] Sara Jensen: A digital marketing agency that helps small businesses manage their online presence, and SEO is a huge piece of that. 

[00:00:59] Obviously, I'm not Crystal. Neither is Shea. She asked us to just do a guest appearance today and talk about two big topics that I think are probably on a lot of your minds out there.

[00:01:11] One being why we don't rely on AI for SEO content creation. And then the second one being what do we recommend or what are our predictions for visibility in 2025. 

[00:01:24] So we're going to go ahead and jump right in and I hope you enjoy. 

Introduction to AI in Content Creation

[00:01:28] Sara Jensen: So to get us started, Shea, tell us a little bit about how do you use AI in content creation at this point?

[00:01:37] Shea Karssing: So I think key to what Crystal pointed out is why we don't rely on AI content creation, not why we don't use AI whatsoever. There are plenty of use cases where AI has made my job as a writer so much easier. So an example that springs to mind is. 

The Role of AI in Interviews and Summarization

[00:01:54] Shea Karssing: Interviews. Often ghost written pieces or thought leadership pieces that require quite in depth interviews with those people.

[00:02:02] And previously I would sit with either a word document in front of me or a notepad. And I'd be scribbling furiously trying to make sure I was catching up. And, then your brain gets stuck on your notes and you forget to ask the next question. 

[00:02:15] We can just shove that all into an AI generated transcript and it's amazing.

[00:02:19] And you can pull direct quotes from there. So I found that really helpful. Also just summarizing those notes. What are the key points from this conversation? And there's some wonderful tools that, that do that for you now. 

AI as a Brainstorming Buddy

[00:02:31] Shea Karssing: And then also I see chat GPT specifically as a large language model. As my brainstorming buddy.

[00:02:38] I work on my own as a freelancer. I don't have any colleagues to bounce ideas off straight away. And chat GPT is there. 

[00:02:45] Look, it's not, we're not exactly having interesting conversations about our weekend or any water cooler gossip or anything like that. But. It is really great if I'm stuck on an idea for a subject line. Or a headline for an article. And what I'll typically do is I'll ask it to give me a 10 different options.

[00:03:03] And it's very rare that I'll actually take verbatim one of the responses or one of the options, but it gives me ideas and then I think, Oh, let me build on that or let me take that and and customize it accordingly. Or I'll combine some of the ideas. So definitely as brainstorming buddy. And just for those real grant worth tasks that we actually don't need to be doing come 2025. 

[00:03:29] Sara Jensen: Yeah, I think one thing that I want to just clarify in what you said, because you described a lot of ways that you're using AI in your content creation process.

The Pitfalls of Relying on AI for Content Creation

[00:03:40] Sara Jensen: You're not relying on AI for content creation itself. So when we talk about relying on AI for content creation, what we mean is telling chat GPT, Hey, write me a blog about this. 

[00:03:50] And then taking that and pasting it up onto your website and going from there. To your point, it's really good for a lot of those sort of grunt work or preparation or even assisting you in your creative process, but it's not replacing your creative process.

[00:04:07] Shea Karssing: Exactly. 

Detecting AI-Generated Content

[00:04:09] Shea Karssing: And I think building on that too, nowadays, it's so obvious now that people have been using chat GPT and other large language models for a while, it becomes more and more obvious when someone has just copy pasted straight from whatever it is. 

[00:04:25] Onto their website, onto their social media posts, I don't know about you, but I feel like now I can sense it a mile away. In the early days not as much but now that.

[00:04:35] They've been around for the last, two years on mass, you can definitely notice. 

[00:04:40] Sara Jensen: Yeah. I see it a lot in LinkedIn comments. I'll post something on LinkedIn and then somebody replies and just like summarizes what I just said. It's thanks. Why? Why did you comment that? It's just, it's not helpful.

[00:04:53] It's not having a conversation. It's not advancing this topic in any way. 

[00:04:57] Shea Karssing: And I think this is something we're going to discuss later, but. Does that not actually take more time? If time saving is the goal, right? 

[00:05:05] Yeah, does that not take you more time than actually just commenting what genuinely pops into your human head?

[00:05:12] I don't know. 

[00:05:12] Sara Jensen: Yeah. 

The Importance of Human Touch in Writing

[00:05:13] Sara Jensen: Yeah, We're both writers and so writing comes easier to us than it does for other people. Which is not to say that it's easy. But it's not as challenging as it might be for other people. 

[00:05:26] And some of the time saving things and things like that, if you're not a writer, it probably seems super appealing to say, let me just have one of these AI tools write all my content for me.

[00:05:37] But, what starts to happen, when people are doing that? 

[00:05:41] Shea Karssing: They're not bringing themselves to any of their engagements. And it's difficult then to discern what makes your response different to anyone else's vanilla generic response that's coming through? 

[00:05:55] In a way, and I've seen people post about this on LinkedIn too, I would rather see someone with maybe hasn't got the best command of English. Or isn't great at grammar.

[00:06:05] It's almost like you would rather see that than some real Spam, AI generated spam. 

[00:06:11] That sure it's polished. There's no glaring errors per se, but it hasn't been written by a human. And I've even heard of people sneaking into their own writing like writers. making mistakes on purpose so that people can see that it's, I wouldn't take it that far, but it's a crazy world when you're actually thinking of sabotaging your own efforts for the sake of not sounding like AI.

[00:06:34] Sara Jensen: Yeah, it really is. I think too, it's about that tone and style or your unique voice. So if AI is creating all of your content, to your point, you're not in it. But also, you sound like everybody else whose content is being created by AI. And the deeper that we get into this AI world that we're apparently in, the better we're all getting at spotting it.

[00:06:59] I know we joke about, phrases that are, just show up over and over again. And that's this is a clear tell that This is just AI content. 

[00:07:09] Shea Karssing: So one that I'm actually particularly sad about is the em dash. So I love the em dash. It is my favorite punctuation and I know this is super dorky writer conversation.

[00:07:22] So please bear with us for a moment, but. Please, ChatGPT, stop using the emdash. It's my favorite and I want to be able to keep using it. I 

[00:07:31] Sara Jensen: wholeheartedly agree with that and I refuse to let go of it. It's such a useful piece of punctuation. That's right. It was ours. We staked our claim. Yeah, that's so true.

[00:07:41] Getting beyond the voice space. piece of it too, I think are concerns about the content itself that's being put out. Is it accurate? Is it plagiarized? Is it a duplicate of what somebody else has already put out there? Which we all know in SEO, duplicate content is a big no as far as trying to get your page to rank for something.

[00:08:03] That'll just get you de indexed. 

[00:08:05] Shea Karssing: Yes. Often on people's websites where I've noticed it anyway. Where the content reads like a summary. And something that you should be saying in that industry. Or it's quite jargon heavy. 

[00:08:19] So I'm trying to think of an example now, but when it's talking about making an impact on business results, for example. It is so generic.

[00:08:27] It's a message, sure, it's not incorrect, and it definitely applies and is accurate perhaps to what that company is trying to achieve or do for their clients or do for their customers.

[00:08:36] But that tells me nothing about why I should choose you. And that's something that AI can't do. 

[00:08:42] It all feels duplicate.

[00:08:43] Sara Jensen: Yeah, we had an example from our agency. We had one of our clients was experimenting with ChatGPT. 

[00:08:50] And he sent me this blog for their website that he had let ChatGPT write and I read through it and was like, okay, this is okay. 

[00:08:58] And then I thought, just for fun, I'm going to run this through, not, I ran it through I think five different plagiarism checkers and they found anywhere from like 40 to 70 percent of the content.

[00:09:10] Was plagiarized. And so I know there's a lot of you know It's not a hard and fast science as far as what is and isn't Plagiarized that these tools are putting out. But that is not something that you want on your website 

[00:09:25] Shea Karssing: Agree. 

Challenges with AI-Generated Stats and Sources

[00:09:26] Sara Jensen: Yeah, another one. I know that I've heard you talk about is stats And, hard numbers for things. If you can explain a little bit more about that.

[00:09:36] Shea Karssing: I remember when I came across someone told me about perplexity, which is meant to be pretty good at pulling stats and sources. I was like, wow, this is amazing. 

[00:09:44] Really hoping again, speaking to that content creation process, that it was going to save me some time because. For most of the blogs I write, I like to see find some recent research and include some stats and back up some claims.

[00:09:58] And I end up spending a fair amount of time digging through the search results. 

[00:10:02] And trying to find studies that are recent enough and look to be Authentic and well researched and scientific, et cetera, et cetera. 

[00:10:11] So I was hoping to find a tool that would be able to do that for me. So when I have, and perhaps it's user error, maybe it's my prompting.

[00:10:17] That's not great. I'm not sure. 

[00:10:19] When I've put a topic in and said, please provide stats and sources to back up X, Y claim I get a list and there's always a lot. 

[00:10:26] There's a lot of data there, which is great. But when you actually click on a lot of the source links, it takes you to more of those sort of stats, aggregate pages. 

[00:10:34] Or whatever the top Google results is for a similar search using those kinds of keywords.

[00:10:40] And then I think, okay, fine. That's given me something to start with. And then you end up in this loop of trying to find where the original source scientific sources or study or survey. Okay. 

[00:10:52] And all you find is a whole bunch of articles that are just referencing one another. And you can never get to where that research actually came from.

[00:11:00] Although I haven't seen hallucinations per se. I also am wary of some of the stats that it spits out. And by the time I've gone and validated the information that's been served, I've wasted more time and I could have just done it myself.

[00:11:16] These tools are always improving. So that is something to remember too. I think we've got to put improving in quotation marks because as we'll chat about, there's also issues of model entropy and collapse. 

[00:11:28] So there's that. But I do believe that there is potential. 

[00:11:31] And I'm hoping that one day there will be a way that I can and please, if anyone has some tips on doing some really great.

[00:11:39] Science backed research using AI, we would love to hear from you. 

[00:11:43] Sara Jensen: Yeah. I think anybody who is just alive in the world today, is at least hopefully somewhat aware of, you have to check your sources when you hear data. 

[00:11:54] Because it is so often twisted or used in a way to mean what it didn't actually say in whatever study came up with it.

[00:12:03] I think there's just this like sense of well, it's AI and it's giving me all this information right here and it must be authoritative. 

[00:12:11] You really have to be cautious about that. Do your research, back up, whatever you're saying. 

[00:12:17] Shea Karssing: And make sure that the way you've prompted. Which again, this is all very new to many of us.

[00:12:23] And perhaps the way that I've prompted it is biased. Or inaccurate. Or whatever it may be. So it might even be a user error. That's reflected in the outputs I get. 

[00:12:33] Sara Jensen: Yeah. Maybe this exists. I don't know. It wouldn't be super practical for what we do. 

[00:12:38] But if you're in a very specialized field like in medicine. And then there's an AI tool that can scan all the published studies in a certain medical journal or something like that, That's where this stuff starts to become really powerful and just amazing for synthesizing so much information.

[00:12:54] But just asking it generalized questions and hoping to get some reliable information back. I think we've seen it's not there right now. 

[00:13:03] And 

[00:13:04] Shea Karssing: like you say, I think, especially with the publicly available tools. So I think. I would presume that a lot of companies are setting up parameters. Private large language models, or some kind of AI tools.

[00:13:16] I'm thinking the legal field, for example, that can, instead of a lawyer having to comb through pages and pages of documents, that they can have that information much more accessible. So I'm sure those, they must, they do exist. But when you're working in the public domain, you have to be really careful.

[00:13:32] Sara Jensen: Yeah, definitely. 

Gray Area Thinking and AI Limitations

[00:13:34] Sara Jensen: Another thing that we had talked about the, things that AI is not good at is this concept of gray area thinking or ambiguity. Say a little more about what that is. 

[00:13:45] Shea Karssing: That's what, in my opinion, part of what makes us human is our ability that to think that, and two things can be true at once.

[00:13:55] Whereas In code or a robot for want of a better word. They just don't have that ability to see that there's this and there's this. 

[00:14:04] And there's a middle line in between that as humans, we know because we can read it using our empathy and our emotions. And the kind of intangibles that you can't program into a computer.

[00:14:17] Yes. A lot of that comes with our own bias too, for sure. 

[00:14:20] Because those thoughts that we have are shaped by our experiences and our perceptions and our own realities. But that's what makes us human. And your gray area is different to my gray area. And that's amazing because we can talk about it. And I just don't see a world yet.

[00:14:38] Where the bots are able to replicate that. How do you feel? 

[00:14:42] Sara Jensen: Yeah, I totally agree. 

[00:14:43] I just saw this news story. I have no idea if this is even true, but it was a news story. 

[00:14:48] I saw that they are testing these AI robots in China to be police, basically. And there are these little things that kind of roll around, and if they're pursuing somebody, it can throw a net at 

[00:15:00] Shea Karssing: them, 

[00:15:00] Sara Jensen: capture the person, and it's just oh my gosh, how terrifying. Because, I think that's a great example of something that looks like criminal behavior, maybe if it was programmed into this thing.

[00:15:13] Maybe it's actually not and it's just normal, whatever. I can't necessarily think of a great example off the top of my head, but it's like a, Ooh, do I really want a robot making a decision to chase me? It could even chase people through the water. It was crazy. Oh, 

[00:15:30] Shea Karssing: wow. Frightening. That sounds like a horror movie.

[00:15:33] It really does. 

[00:15:34] There's so much nuance in the way we live our lives. For the police robot, for example. Let's say you're with your kid at the store and they run out into the street. 

[00:15:45] You're still holding something in your hand.

[00:15:47] Your priority is going to save your kid, not necessarily thinking, Oh, I've just walked out of the store with this bag of cookies in my hand. 

[00:15:54] I don't know if the Chinese police robot will feel quite the same. And will pursue you and your bag of cookies and your child into the traffic.

[00:16:01] Sara Jensen: Yeah. 

[00:16:02] That's a great example. 

[00:16:03] There are limits to these things. 

[00:16:06] And what makes us human, I don't personally believe can ever be completely replicated by technology. 

The Human Element in Marketing

[00:16:14] Sara Jensen: But marketing is about showing people how you're unique. Compared to other companies in your space. 

[00:16:22] Every single client we work with, no matter what we're doing for them. We always start by defining, who is your target audience?

[00:16:31] What is the unique way that you solve their key problem? And how is that different from the other people who do what you do? 

[00:16:39] And, I think that's another piece that if you're literally outsourcing your marketing to AI, you're going to miss.

[00:16:47] As much as you can, feed it all of your content and your value proposition and all of this like AI is pulling from kind of the masses of what's out there.

[00:16:58] And so if you tell it, I'm a custom home builder and here's my website. And here's some things that I've written and here's some other stuff about me. 

[00:17:08] It's still going to go with what it knows. And talk about you like the other custom home builders that have been fed to it. 

Trust and Personal Connection in Marketing

[00:17:15] Shea Karssing: Trust is built. Human to human. A decade ago, maybe a little more. 

[00:17:19] Before the proliferation of the internet, people would consume content. Maybe that watch ads, see ads on TV or something, and that would be a brand and they would pursue a particular narrative and perhaps we believed it because it was all we had.

[00:17:33] Then we got so much information, literally at our fingertips.

[00:17:37] Disinformation, misinformation. 

[00:17:40] And now it's come to a point where we're looking not necessarily at brands to trust. 

[00:17:47] But humans and people. And that's not AI. We see it, for example, on, on LinkedIn as a professional platform where company pages. 

[00:17:56] Sure, you keep it running with a bit of content here and there, but that's not really where you building an audience or getting your engagement.

[00:18:03] It's on those. Personal profiles that are unique to that person and in this very specific voice that seem to gain more traction. 

[00:18:11] So more and more, we need to bring that human element into brands to keep them relevant and keep people trusting and coming back. 

[00:18:20] And most brands find that their best customers come from word of mouth and referral, because that comes with that element of trust straight away.

[00:18:29] Sara Jensen: Yeah. I think that's such a great point. That interpersonal trust between people. 

The Role of Professionals in the Age of AI

[00:18:34] Sara Jensen: And I think that extends into the marketing field. 

[00:18:37] Because as much as there are tempting ways to automate the people out of it. 

[00:18:45] There are still people who have a certain expertise. That has taken years to develop to get their instincts to be able to say that's not going to work.

[00:18:55] here's how to fix this problem you're having. I liken it to the accounting space where we have had these tools that enable you to do your taxes yourself. 

[00:19:07] Like TurboTax. But that hasn't erased the need for CPAs. Maybe there's some people who would have worked with a CPA who don't anymore, but there are still plenty of.

[00:19:20] Reasons why you should not be doing your own taxes. Depending on a case by case basis. 

[00:19:25] It's like you talked about at the beginning. It's a tool. It can be helpful in certain ways, but it's not going to replace an entire profession. 

[00:19:34] Shea Karssing: A chat bot on a website is great to a point and depending what you need. 

[00:19:38] There are times when you actually just need to speak to a human being who can understand the nuance of your situation. 

[00:19:44] And why perhaps, your solution that you need doesn't fit into a tiny little programmed box that takes you down some kind of branching from here to here. Like sometimes you need someone to say, this is a solution that suits you, specifically. 

[00:19:59] Based on what I know that perhaps isn't written in the code, it's based on my experience that I'm able to help you.

[00:20:06] Sara Jensen: Yeah. I think we've all had that experience of getting so frustrated trying to deal with those chat bots. 

AI in Graphic Design: The Canva Example

[00:20:12] Sara Jensen: Another example too, is what has happened with graphic design. 

[00:20:16] With the rise of tools like Canva. 

[00:20:19] I should preface this by saying I have nothing against Canva.

[00:20:22] We use it in our agency. It has its place. It's another great tool. It's not the only thing we use. We also have a real human graphic designer on our team. 

[00:20:33] If you're using Canva to create graphics, you're probably noticing that you can tell, now. When you scroll through social media. 

[00:20:41] Like what was made with Canva. Versus what was made by a real human designer. 

[00:20:46] You just start to see the same thing over and over again. I do it with stock photos, too. 

[00:20:52] Because we spend so much time in building websites and things like that. It's Oh, I've seen that stock photo before. Canva took something that used to be in the realm of professionals. And it made it accessible to everyone to DIY.

[00:21:05] And there are definitely benefits to that, but. You also lose the creativity. And the trained eye of a graphic designer. And then the outcomes that come with that. 

[00:21:18] Of having a brand that's unique. Because if your graphics on social media look like everybody else's, they're not going to stop the scroll. 

[00:21:26] Because people have already seen them before and they're just going to keep going.

[00:21:29] So I think that's a great analogy in something that, Is probably a little further along than AI for content creation of what starts to happen when you're using these tools, when you cut the professionals out of the process. 

[00:21:44] Shea Karssing: And I know even from using it myself in my own business, it's amazing.

[00:21:47] It is game changing. I'm not necessarily in my little one man band going to be paying a graphic designer for social media graphics. I did pay for my logo and my website and the important things. 

[00:21:59] But I can tell I'm not a designer and despite having a template, I'm never 100 percent happy with the output.

[00:22:07] It always looks a little messy or just, it just lacks because, I've taken something that's been partially generated for me. 

[00:22:16] Tweaked it a little, and I must admit it's, Never as good an outcome as if I'd paid a professional human being to do it for me. You also need that eye on these tools that has the experience.

[00:22:28] So again, going back to writing. 

[00:22:30] Yes, we use AI as part of the process. But the outcome is polished because we're able to bring our experience as writers to that. And I think the same applies to any tools. 

[00:22:41] You can get a baseline with AI. You can have a minimal vibe or product with AI at the moment.

[00:22:48] But to really give it that polish and finish it off. 

[00:22:50] You still need that expertise that comes from experience and proficiency. 

[00:22:55] Sara Jensen: Yeah. And I wanted to come back to something that you said just now. 

[00:22:58] You paid somebody to create your logo and your website. 

[00:23:01] And, I think website copy. For SEO. 

[00:23:06] Is just becoming rampant with people putting chat GPT content up on their website.

[00:23:12] And, if you're going to have a website, you want it to do something for your business, right? 

[00:23:18] Invest in that. Invest in the platforms that you own. Where you control it, you want it to deliver results for you. Most businesses want their websites to generate leads.

[00:23:30] And so that's where you get what you pay for. 

[00:23:33] If you're making a continuum of when I'm going to use AI for content creation, I would say your website copy, please do not do that. 

[00:23:41] Shea Karssing: That's a good point.

[00:23:42] And yes, again, like we keep saying, it's not that there's no place for it whatsoever. If you're not a writer, but you've got something that you want to say. 

[00:23:50] And you've got an email to send or whatever it is, and you want it to look fairly good and polished, go right ahead. 

[00:23:55] But yeah, for those real assets to your business, you want to make sure that you're getting them right from the beginning.

[00:24:02] Sara Jensen: We touched on this earlier with the stats. These large language models are sucking in a lot of content. But where, like, where is it getting it from?

[00:24:14] Is it an opinion piece that's being presented as fact? Is it just old information that has been proven wrong? That's not always clear. 

[00:24:23] Shea Karssing: No. And especially something like chat GPT where you're not given sources. You, you could be given reams and reams of texts that you have absolutely no idea where it's been aggregated from. 

[00:24:34] Even the Stats that are, that can be given in an article with no backing to them.

The Risks of Self-Referential AI Content

[00:24:40] Shea Karssing: And I think something else that we need to be aware of as these large language models proliferate and people are using AI generated content on their websites. 

[00:24:50] Is the idea of it becomes self referential. It's just then, AI referencing AI referencing AI, and there's nothing new being added to the conversation whatsoever.

[00:25:00] Sara Jensen: It's really easy to see with visuals .

[00:25:02] I saw this one where somebody told an AI tool, make a picture of this particular statue. So it was like a real statue somewhere in the world.

[00:25:12] And so AI did, and then they took the picture and fed it back into the AI. It was like, use this as a reference point. Make a picture of the statue. And they did it ten times or something like that. 

[00:25:24] And every one got further and further away from what this statue actually looks like to the point that it was completely unrecognizable at the end.

[00:25:32] With content. It's not always as easy to see. But it's it's almost like a massive global game of telephone. 

[00:25:38] By the time the message gets all the way around the circle, it bears no resemblance to what you originally started with.

[00:25:44] This is changing a little bit with search GPT because that will tell you the sources.

[00:25:48] Shea Karssing: Another example that, that just sprang to mind is I get a newsletter from a writer who's talking about AI. 

[00:25:54] And she is white and she does have like really crazy curly hair. 

[00:25:59] And she getting AI to generate an image of herself. And she said, with really wild, vibrant, dark hair. 

[00:26:08] And every single image that was generated was of a black woman and she could not get it to conceive that as a white woman, she might have some kind of different hair. 

[00:26:19] So that was also really interesting that those inherent biases that come with these tools. 

[00:26:27] Sara Jensen: Yeah, it's like a constant regression to the mean, 

[00:26:29] no matter 

[00:26:30] Sara Jensen: what 

[00:26:30] Shea Karssing: it is. I loved 

[00:26:31] Sara Jensen: the example, you sent me one last week, where you asked ChatGPT the latest score on a sports game, and it gave you the score and said the date that the game took place, and it wasn't, it was a game from six months ago, it wasn't the latest, that you had asked it for.

[00:26:47] But if you just ask it that question, and look at the answer. You're not, it like it's a, it's just a beautiful example of something where it's this is not even, it's technically accurate, but it's not what I asked for. . 

[00:27:00] Shea Karssing: And I did ask, just to be sure, I said, what what year is it next year?

[00:27:04] And it, I must say chat, GPT did say 2025. So it. Perhaps have some concept of time, but that was quite an easy question I gave it there too. So yeah, I'll pass on that one, but definitely not on the sports score. So if you are into sports betting or whatever it may be, do your own research, find something more recent do not rely on AI for that just yet.

[00:27:30] MacBook Pro Microphone: That is awesome! Thank you so much, Sarah and Shea. And that was just part one of their conversation. 

[00:27:36] Make sure you come back tomorrow for part two with Sara Jensen and Shea Karssing from Brighter Media. Thank you ladies so much for hosting the simple and smart SEO show.