AI's Impact on Content Marketing

Today I'm interviewing John Gillum with originality ai and he has an incredibly interesting story and how he got started in what have you, but let me tell you a little bit about him. He is the, uh, founder and CEO of originality.ai, and they provide a complete tool set that helps website owners, content marketers, writers and publishers hit publish with integrity in the world of generative ai, as well as specializing in AI [00:01:00] detection technology.

He is also the co-founder of Motion Invest, where they help you buy and sell content sites quickly. How cool is that? Welcome, John. Really great to have you. Thanks guy. Thanks for having me. Yeah. So John, before we get started, so what is your backstory on getting involved in AI and marketing? Yeah, so, uh, started, you know, going way back to to school mechanical engineer, when the workforce wanted to move back to, to the hometown, my hometown, um, built up some online businesses.

Those online businesses all revolved around publishing content on the web. Um, built up some extra capacity within that. Um. Portfolio of work where we, so basically were turned ourselves into a content marketing agency, had sold that content marketing agency, and through that process had seen this wave of generative AI coming.

Yeah. You know, it's interesting now, I haven't played with, uh, Jasper. We've been doing some stuff on our own. And, um, one of the things though, you know, if you're using AI to programmatically generate hyper personalized content for tens and hundreds of thousands of email recipients or what have you, is how do you know that what it's generating is correct?

Fact check, grammar, spelling, readability, and then, uh, editorial guideline compliance is, is coming. So every check that an editor does, we make that process, uh, easier. Yeah. But how do you, uh, make sure that the, the content is right? I mean, I, I granted that what you're doing is incredibly important in making sure you know that all of that's there.

But how do you make sure then, then that the, uh, the actual content is right. Yeah, so the, if there's the factual, so we have a fact checker. Um, so that may, that checks if the facts are correct. The, the question of sort of like, is this content going to perform well in Google? Is, is what a lot of our customers are, are after, and we have a content optimization tool.

Awesome. And that's, uh, such a critical piece of the, of content generation and what marketing a lot of marketers do when they're mar in the marketing space. So, fantastic. So now you have an interesting story about how you guys, uh, launched. Uh, related to when, uh, chat GPT kind of, uh, went public, tell us a little bit about that.

Um, but quickly the ability to tell whether or not content was AI generated or not became a societally important question. And, and we got, yeah, extremely lucky on [00:05:00] timing, hindsight, 2020. Wish we'd pushed the pace a lot harder and been in the market for, for a while. But, uh, yeah, it was, uh, yeah, certainly, certainly some exciting timing.

Yeah, exactly. I, uh, you know, it's, it's one of those things, how fast do you push the envelope and, and, uh, you know, there's never any good time to launch other than sooner than what you'd like to. Yeah, exactly. Yeah. Exactly. Now, uh, so if you were gonna launch today, what would you do differently? You talk a little bit about, uh, you know, the transparency and transparency gap.

And so we try and drive transparency through that entire workflow so people understand how much AI has been used. Where it's acceptable to be used, um, and, and then make sure all the other sort of checks are in place. What would we do differently? N now i, I think it'd be a lot of the same stuff where it's ultimately what we, what we want is everyone to be understanding of, of how AI has been used.

I think what's, I think still missing on the transparency piece is an understanding from all people involved in that workflow of what. Is isn't acceptable, uh, for, for the use of ai. We see a lot of marketing companies and, uh, website owners that are not well informed on how much AI content is being published on their site.

So tell us a little bit about that and how originality, uh, can handle that. Yeah, so exactly whether, whether or not you're sort of pearl. Yep. We're happy to have this content written by ai. We, we accept the risk that. That publishing AI content brings. And, but we're gonna benefit from this efficiency and we're gonna publish this AI generated content, but nobody wants to publish factually in incorrect information.

And, and so the way our fact-checker works is through a combination of, of RAG and some other, some other sort of layered strategies. Um, it goes out to try and truly sort of identify what the, what the known understanding of sort of identifying a fact and what the known understanding of that fact is in the world.

That it had been one AI generated, and so it could have heightened the sort of the quality control checks that would've been needed. And then second, the fact checker identified that those 10 books were actually not, uh, real books. And so sort of for ten second effort and 14 cent cost, um, the amount of embarrassment that those major, major newspapers experience could have been totally avoided.

To help you versus just writing it yourself. And, uh, it's kind of, then where is that, you know, how do you determine where that good enough is? Where it's, you know, good enough to publish where, as opposed to, oh my God, there's a real red flag. Like in your example with the, uh, Chicago, uh, sun Times. Yeah. And, and that's where the human, the human in the loop, I think is, becomes so critical.

I mean, I think the dream, the dream of all of us is like when we see AI as marketers, we're like, wait, where's the magic button that I can press and the whole thing is done for me? Um, and I think the reality is that doesn't exist. So what is good enough is ultimately up to the human in the loop to understand their audience.

Yeah. So let me ask you a question. You know, a lot of, uh, uh, because you're, it looks like you also do plagiarism detection and stuff like that. Um, so if I just say to, let's say chat GPT or Gemini or whatever, Claude, I say, you know, write me this content content and, and do it in a plagiarism free manner.

Does that actually make it plagiarism free, or is there still a possibility that it has plagiarized something? AI will generally produce content that is plagiarism free. Um, AI is synthesizing its existing knowledge. It will rarely, rarely produce content that is a, that is plagiarized, meaning a direct quote from existing content unless you asked it to provide a quote.

I was always, uh, curious as to whether it was. You know, copying and I don't know how many words you need to be to be plagiarism free. So if I take five words or is it 10 words or is it three words, you know, interesting that, uh, that you say that, that actually makes me feel a lot better about what I've been using from ai.

Yeah. No, it'll be, it'll be, so under, under 5% is sort of the like, call it gold standard of, that's clearly plagiarism. Plagiarism free, and, and AI will constantly produce content that has. A below 5% score in in plagiarism. Oh, fantastic. Well, thanks for that. So, uh, now there's this challenge of, uh, invisible characters.

Informative, authoritative, high quality content. And they also have a bias to being sort of as trustworthy as, as as possible and be perceived as sort of, um, an authority in the content that, that, that l produce. Uh. Invisible characters are characters that are used at sort of when editors are doing like books and, and professional editors not used when you're just using your, your own keyboard.

And so what's the, the problem? Well, so what the theory initially, when people discovered that these, um, invisible characters were being entered by by LLMs was that that was their attempt at fingerprinting. That is, that is not the case. It's not open. AI is not trying to watermark its content using these invisible characters.

But these invisible characters are being sort of injected into, uh, content right now without the authors really being aware of what that is. And it can cause a couple challenges. One, it's an easy tell, uh, visually that that content was likely generated by ai. No one writes with with m dashes unless it's been copied out of outta chat, bt.

Uh, or how many hidden characters have sort of worked their way into your, into your text and then lets you remove them, uh, strategically. And that's just a, a free tool that we've developed, but to try and sort of educate people about, mm, the text that is being generated by, by these lms. Yeah. Interesting.

And, uh, yeah, I hadn't heard about that. And, uh, that's, uh, really a, uh, you hear about these hidden pixels or these invisible pixels, but that's a totally different, uh, purpose, uh, as opposed to this, which it sounds like it's a, an after effect of the generation process that, that OpenAI is doing. Very interesting.

And, uh, certainly, uh, AI generated content can, can help there. But one of the challenges that, that I could see is that. If everybody's using these tools and everybody's generating a ton of content, I mean, that's kind of what AI is doing. Then all of a sudden the. SEO and how Google is now going to rank you and index you is their, their algorithm now has to change it would seem like.

And you're competing against LMS that can produce words, you know, a near a near infinite number of words that near zero, it seems like zero cost. It, it's a fascinating challenge. Um, so the, I'd say the, the sort of. Concept that we talk about at, at originality, where we get, you know, million and a half visitors a month, organic traffic, uh, and have a, have a large scale sort of content marketing strategy being deployed, um, is to think beyond words.

Um, and that's how we, how we think about it internally from a, from a content marketing standpoint. And what we know is that Google is absolutely punishing AI generated spam. And so if you're, if you're mass publishing AI generated content, Google is gonna find you and punish you. And then it's a question of like, we know that's bad.

We know Google will punish that. And then it's a question of how far. How far in the other direction do you, are you going to swing? Is, is it, you know, human generated original research with a tool that's like, kind of on the other extreme of, of like, okay, this is, this is obviously not AI generated spam.

And an AI overview can't. Or an AI mode can't solve that problem. And so we think a lot about unique research and, and tooling. Yeah. Interesting. So, uh, and, and you know, and I think over the next couple years, and maybe it's just the next couple of months is, uh, content is gonna proliferate like crazy. I think that's gonna be a, a, a critical thing for how.

We as sellers or e-commerce or whatever, wanna be found on the, uh, by Google. And yet there's now just gonna be so much more noise out there, even though it could be perfectly valid content, there's just gonna be so much of it. So, yeah. Very interesting. Now, one thing that, that's critical, so it, we work with a lot of regulated industries and, uh, they are, I don't know what you'd call it, but they're so hesitant to even touch AI and be able to use it.

Now, of course, the marketers are sitting there using their phones and they're then typing something in over here, and then they're, I don't know how they're copying and pasting it over, but they're basically doing the AI on their phone. And then they, they, they generate some good stuff and type it in on their, on their, on their laptop, the company laptop.

And, uh, so tell us, you know, the compliance is part of that and, uh. And so tell us, is there like a checklist or something like that that's, uh, really critical that regulated industries and even unregulated industries, uh, can use to mitigate the risk that they might have in publishing some, I don't know, stray content, I guess?

Individual video. Again, not that, not that like mass text can't also move society, but not, not in as sort of a, an acute way as, as sort of image video, audio. And so I think from a regulatory standpoint, that's where the focus is gonna land on watermarking text is gonna be, I think, very driven by the company's policies.

We wanna make sure it's all factually correct. Complies with their editorial guidelines. Um, we want to be that, that layer in that sort of compliance process and helping companies make sure that they're, the content that they're producing meets their, their requirements. Yeah. Yeah. Well that sucked too, is uh, uh, I mean, we suffer through it.

Uh, 'cause our clients force it on us and yeah. And, uh, I mean, it's critical. No question about it. It's important. You gotta have it. Uh, there's so many bad actors out there that, uh, and it's so easy to make a mistake, uh, that you want to at least do everything you can to mitigate that risk and especially.

That makes a, that makes a whole lot of sense. So, uh, yeah. Fantastic. I can see a very bright future for what you're doing. Yep. Okay. Thank you. Yeah, absolutely. So, uh, one last question. Um, and, uh, that is, uh, what advice would you give a, uh, new upcoming, uh, new marketer? Yeah, I think, uh, embrace ai, embrace your own personality, and, and I think especially when it comes to thinking about content marketing, think beyond words.

But yeah, embrace, embrace or the, the dichotomy of, of AI and your own. Uh, uniqueness and then think about how to add value beyond words. Yeah. Fantastic. I think that makes a lot of sense. And, um, uh, 'cause I, I think the, there's gonna be a lot of challenge for new marketers coming into the market and if they don't somehow stand out, uh, then, you know, as the kind of the hiring manager.

And that was, that was the reason why it was or sort of call it presence online was the, the most sort of. Loud, complete, clear personality coming through, consistent personality coming through in, in all of her sort of professional social media. And, and so that was, that was sort of the difference maker for hiring her, was that Yeah.

Very, a very clear personality coming through, uh, that that could be, uh, you know, hopefully harnessed for, for the sort of advancement of our, of our goals within our own marketing company. Yeah. Yeah, absolutely. So, uh, well, uh, John, thank you so much. Uh, this has been great and, uh, really appreciate it. Uh, so where can folks, uh, find you and reach out to you and learn more about you and your company?

AI marketing machine. And of course, don't forget to sign up for more episodes, but more importantly, rate this one with five stars and, uh, and then, uh, hopefully, uh, tell your friends about it as well. John, thank you so much. It was great to have you on today. Awesome. Thanks guys. Thank you.