00:00 - Introduction to the Episode
05:00 - AI's Role in Enhancing Productivity
10:00 - Practical Uses and Limitations of AI
15:00 - Ethical Implications of AI Usage
20:00 - AI's Impact on the Marketing Industry
25:00 - Concluding Thoughts and Advice

[00:00:00] Hello there. It's that time, the week again we are doing Geeks week. Today we have a special guest. Of course, my name is Phelan with Seymour digital media, where we focus on search engine marketing. And my guest today is Ashley. Writer gal. Awesome. Yeah. And today we are talking about how you can use AI in your content marketing, like what are the best use cases for it?

[00:00:23] What's the, what kind of your do's and don'ts in order for you to optimize this as a tool to help you get more [00:00:30] work done and get it done properly and as well get the most out of it because the, these tools are super powerful, but they can also. Lead to some issues where people become over reliant or don't understand where it fits in to your workflow and getting a better understanding on that part.

[00:00:49] Yeah. And I know that's the huge thing is that it can fit into your workflows. It can support you, but it's not meant to replace you. I've been writing for 17 years, various stuff. [00:01:00] And it's only in the past few years, this has been a thing that people are asking about. I'm getting people that aren't even in my niche asking me, is AI going to replace your job, actually?

[00:01:10] And it's not if you use it in the right way, which, we'll have a chat about today. Yeah, exactly. I don't think that's the important part is understanding its context within your business and understanding that it is not going to solve every issue you've ever had. And it's not like a, like a silver bullet that's going to replace a bunch of stuff.

[00:01:29] Like it's meant to [00:01:30] extend your business. What your team's already doing, because you'll still need to have writers. And I know that we've had, we've heard the stories of the air Canada, where, they're on the hook all of a sudden because of a chat bot that they had and they didn't put set the correct prompts.

[00:01:44] They didn't verify things and they didn't have a person in the loop. Which is going to be super important to, I think it's going to be like the motto of what we're talking about today is that it's supposed to automate boring things, think of it like that, rather than [00:02:00] replacing my writer it's just mundane, repetitive tasks that, or, they're overly simple.

[00:02:05] I think that's the key cornerstone that we have when we're, when we prep for this today. And it's all about, not letting AI do it all. There's always going to need to be a human element. Checking it and testing it, like with that air Canada example, I can't remember exactly what they, what the bot promised them, but they were now on the hook, and we don't want anyone else to be on the hook for something that AI might've said about their business, [00:02:30] especially when you're writing about things like maybe healthcare or safety, if AI.

[00:02:36] Write something for you. That's not correct. And remember, you're the expert, you still need to have that eyes on it to make sure that AI is doing it right. There's the right way and the wrong way to do it. Yeah. And I've heard as well that like newspaper companies as well that are trying to replace full writers wind up spending as much time editing it down because it makes a bombastic claims that it just are not backed up by anything that they have.

[00:02:57] And so they have to retract because they're on the hook [00:03:00] as publishers that they don't want to libel anyone or anything like that. And so they understandably will edit those ones out. So I, I think that, yeah, understanding that it's not just like A perfect solution out of the box, but they, you need to also have it as part.

[00:03:15] Yeah. I think of it more of as a tool rather than a person I think it's the mental model that we're coming to. And so what are the, what are some of the uses that you've seen with using it? Yeah I think the biggest thing when it comes to creating any kind of [00:03:30] content, I focus on writing, but this can apply to.

[00:03:33] Videos, social media, anything is using it as a brainstorming tool. Phelan, like you mentioned, it's, we're looking for ways that we can use AI to save time. And when I'm teaching blogging, when I'm teaching writing sorry, one second, when I'm teaching writing, one of the joys of sharing an office, right?

[00:03:52] Yeah. Sorry. When I'm teaching writing, one of the big roadblocks people have is brainstorming, figuring out what to write, and that takes them [00:04:00] forever. But with an AI tool, I can type in one word and get You know, 10, 20, 30, as many as I want blog post titles or topics. And it's right there in two seconds.

[00:04:14] And then you can vet the ones that you like and move forward with that. So I definitely find that brainstorming, getting things going is a much, is a really great way to use AI. I don't know if you've used AI for that kind of brainstorming as well. Yeah. I think Pip [00:04:30] and I, Pip Morseau, sorry. My computer is very upset.

[00:04:34] I have to close some windows down because it is not happy camper right at the moment. Sorry about that. I'm seeing a bit of lag here too. Yeah. Yeah. No, it's my computer trying to do too many things at once. I yeah, okay. But yeah, what did yeah, Pip and I had used it to do stuff like find me five new things about something that I'd never heard about or something like that, where it's ones where you're trying to find, like, maybe edge cases of something or ones that are out [00:05:00] there that are, like, not You know, they're not normal ones that you would find and so there's definitely some Brainstorming is a good use case for it as well it's a really good one for first drafts as well just because I think you talked about writer's block as well because when you're looking at a blank page It's tough to get going Whereas if you use the ai to even just get something as a first draft, you go I don't agree with this paragraph, but then it gives you something to You Bounce off of and then rewrite it better So that you are not just staring at a [00:05:30] blank page where you can actually look at it and be like, oh, yeah Actually, I'd like this the way this is constructed.

[00:05:35] I would change these words You know, you like you have something to work with because I find that I'm better when I don't have you know blank page is hard to work with but even if it's a templated thing or something that I could just I have a much easier time going, I do, I agree with this.

[00:05:51] I like that. Maybe actually my research showed that this was actually going to be more of the case, but at least it's something that you could work off of and refine rather than [00:06:00] just start whole cloth yourself. Yeah, and I've used it two ways. I done, I said, Ask the AI to give me an outline for a blog article, which is really great.

[00:06:08] I can take and choose the outlines that I like, the subheadings, or write the first paragraph of a blog about this. It'll get my ideas going. I would never use that paragraph. I'd always rewrite it. But it really kicks off that, that blank page is very scary. Once you have, that outline and that's how I, even when I do it myself without using AI, I always [00:06:30] start with an outline.

[00:06:31] I say, okay, I've got 500 words, a thousand words. Here are my headings. I have approximately this many words for each heading. And it helps me organize my thoughts. And that's basically what AI is doing for me. And I pick and choose what makes sense for me. Starting a blog post, starting social media.

[00:06:48] I don't know. Phelan, have you guys ever used it for getting like ad copy or SEO keywords? Yeah, actually funny you mentioned that one. That was the most recent one. The Pips done is a really good use case for it [00:07:00] is to basically PIP Grabbed our search console data. So that's the stuff that we're showing up for in Google.

[00:07:06] She fed it into, fed it the CSV, cause on the chat GPT 4, you can feed it data. And then ask the question, find me the top keyword I would have for this page. And this is the title of the page and what it did, what it's, this is a really good use case for chat GPT 4. Chat GPT because there's a distinction with AIs.

[00:07:27] One is that there's one mode that's called [00:07:30] generative. So make me a picture that's of a dog eating a birthday cake. And it's going to generate you that image. So it, but it has to know what a dog is. It has to know what a cake is and know what eating is. And so it has to approximate that. Whereas if you say something like there's , selective or regressive AI, which is the, which is what most of these large language models, right?

[00:07:51] It takes a ton of information and it regresses it down to whenever you have this word follows it. That's what it does. But if you give it data and you say, [00:08:00] find me the keyword that's like I'm ranking for the best. That's one of its things that it will do really well with. And so you could use that information because it's computers are just good at algorithmically Grinding down something and being like, actually, we found this trend in the data.

[00:08:14] That's a strong suit. That's what it's really good for. So we've definitely used it for something like that. Ad copy, I found it to be, again, a good jumping off point. But I find it to be not as good as it thinks it is. Is what I would say. It gets a little repetitive. It gets a little it [00:08:30] uses certain verbs and certain adjectives.

[00:08:32] Like All the time, or at least that's been my experience is that you can feel that there's like a couple of words. We use it for our own and it wanted to give us like a nautical theme. And I didn't really get yeah, the ad copy was all like taking sale and like on your journey.

[00:08:46] And and it just decided that we were going to be a nautically themed company. And I didn't quite get why. . That goes to show, you don't just, Take what AI gives you blindly and put it on your website. Cause if you did, which a lot of business [00:09:00] owners do, they just take what AI gives them with the generative stuff, gives them and puts it on the site.

[00:09:05] And then they find out later, Oh Seymour digital isn't a nautical based company at all. Yeah, you got a fish and you live by the ocean. Yeah, that's no nautical theme to it. Yeah. Yeah. It was Yeah. It was like all the, yeah, the like calls to action were all very nautical. It was just a very, I didn't get where it came from with that one, but yeah, it was it was definitely an interesting one.

[00:09:27] I think also one other thing that kind of [00:09:30] came up when we were talking about this in the prep is that keeping the AI also constrained, like giving it a lot of constraints and like very small segmented parts, it's easier for it to do of only generate me generate me 200 headlines that are exactly 30 characters long.

[00:09:47] If you can get, it does a better job with that. Though I found also it doesn't, it likes to ignore character limits. If I give it to it, I feel like it'd be very specific. It's just make it 30 characters long and then it's 60 and I'm like that's No, 30 [00:10:00] characters and then it was 45 and I was like, okay, like seriously Oh, the AI is already rebelling against us, right?

[00:10:06] Yes, exactly. Oh dear. Which is also a funny thing that's come up as well is that on the forums I don't know if you've experienced it. We've done a couple of them With that the AI, like Chachupichu just doesn't want to do the work because it's been trained to be like a person. People don't always want to, they find excuses.

[00:10:23] And so apparently, like there's a lot of complaints on the forums of yeah, I asked it to generate me 30 images, it generated two and then told me to [00:10:30] finish the rest. There's, it's entertaining what's like the, how would I put it? The weirdness that comes out of AI because of you don't always know how the model is going to speak, give it back to you. And so sometimes it's it acts so much like a person that it's yeah, I don't want to work. You do the work. It's trying to keep us honest, maybe it's a good thing, because we shouldn't be relying on AI to do this.

[00:10:54] It's reminding us that, hey, we're real people here. We should be doing some of this work too. So I guess when you think about it, it's actually a [00:11:00] good thing that is not doing all the work for us. I think that Yeah, I think that a lot of, especially if you read like in the big newspapers and all that stuff, the way they talk about it is it's going to be, a great apocalypse.

[00:11:10] And I'm not of the opinion that it's going to be like coming for all our jobs and like it because, yeah, there's just there's just things like hard limitations on the amount of computing power that would be required to do that. Have all of us compute these things. It just would be like the numbers that they would require.

[00:11:28] Open AI is trying to get 10 [00:11:30] billion, $10 trillion to build out hold infrastructure networks to build out new data centers and all this stuff. And it's just the amount of what they require when they calculate it out, it's just gonna be so crazy that it would not, yeah, I'm starting to see a lot of articles and stuff online shop about the environmental impacts of using AI.

[00:11:49] We just use it thinking, Oh, we can just use it. It's online. It's on the computer. We don't think about, all the servers that you mentioned that they have to have. All the power that takes all of that computing power is going to be [00:12:00] damaging if we rely on it too much. If we use it too much for things that we don't really need to.

[00:12:05] Yeah, exactly. And as well that there's won't go into like rare earth metal stuff, but like those are. Difficult to get out of the ground. That's why they're called rare earth metals and they're difficult to get from places process All the stuff just to make a gpu to run these data centers I know that's gonna it's gonna be a huge one that is Gonna have a lot of long term effects and especially for the expansion that they want to do with [00:12:30] Ai It's going to be a lot of work but I think that, yeah I think, yeah, just being more thoughtful about how to use it in your mix of your business, I think is going to be a good one as well.

[00:12:44] I don't know about it like growing past this point like That's the part where I'm a little like I see it as a little bit plateauing a little bit Because to scale up more it's Gonna be crazy. We've got a question here as well How do we feel about training an [00:13:00] LLM, which is a large language model, which is what chat GPT and all these are basically, it's like a really large model of language and it knows things about it.

[00:13:10] So it's a great question. I don't know about like specific ones like the ODIN AI or Umbrel. I don't know about them specifically, but I think in general, I think that they, they have some really good things going for them. Because if you have a closed network, you're [00:13:30] the. It's one of those, like they, they say it in computer science is like garbage in garbage out.

[00:13:33] So if you give it bad data, you're going to get bad data out of it. So with these ones, I think that you can control what's going in and have it trained more to like your own separate data set, I think that would be, it'd be interesting and I think it could be good for businesses. I know that there's ones of like banks and insurance companies that have been using a lot of this as well.

[00:13:57] Cool. For just doing they're having to. Tons of [00:14:00] computation. So if you can get an algorithm or large language model to compute through a lot of this stuff, that's going to save you a lot of time. And so they only train it on their own, like privately held information versus the internet, which is, that's one of the big ones, right?

[00:14:13] It's like the more people are using this stuff, the more it's out there, the more that the chat to PT gets trained on. It's like a recursive loop of getting fed on its own. Giving it what they call like digital dementia where it's you're just getting fed your own part and it's not working as well.

[00:14:28] Yeah. I've had a client that [00:14:30] came to me recently with that and she had her own model. I can't remember what site she was using. She put in all of her like PDFs and all of her program information and the stuff that. Chat, or it wasn't chat GP, whatever the program was, whatever it was generating for her and content wise was factual, which was great, but it wasn't her.

[00:14:50] Like it included a little bit of her, the way she talks and from the language, but it still wasn't, it didn't include like the personal stories, the the human element of language, but [00:15:00] factually it was more correct than if I put the same prompt into chat GPT or something more generic. In that respect, it's good, but still I didn't take any of the content that gave me.

[00:15:10] I just rewrote it all in my own words, but it was, it was interesting thing to see how that works. Yeah. Yeah. I, it's, I think that's where the interesting part I've heard of some people even training it to sound like, themselves of here, read like a hundred blog posts that I've written.

[00:15:27] And it cuts, it's in their own voice that it [00:15:30] is super useful for them to be able to do that. And so they're saying that they are want to geek out more offline with PIP as well. I think there's an, they have a bunch of awesome comments. Thank you, Laura, for all these. And yeah, I think that this is, I think that this is a good way to do it.

[00:15:48] And I think that, cause that's a big one that's coming up as well is issues of copyright. A lot of what AI helps for businesses is I don't want to say like getting around copyright, but you could [00:16:00] be like who knows where it got all this information from?

[00:16:02] And so that's a big one that is, it's a big looming thing about AIs. Is that like how, attribution, who wrote this, like it and I know that there's been one court case that said if it's generated by AI, it technically doesn't have a copyright if it's like purely generative but then there's like the, how did it get trained?

[00:16:20] Who gave it that information, which is, I think that's a big one that it's not for me to decide, like I'm not a lawyer, I'm not, I don't run open AI, but that's definitely [00:16:30] one that is going to have to get figured out. And I think that's a big one that is, I don't know where that's going to shake out, but I think that's going to also really define the boundaries of it.

[00:16:41] Not only like we talked about physically, like how much it can compute, but also who's legally responsible for what it says and all that part. I think that's one that's going to be, I'm not, I don't envy the lawyers that have to figure that part out because that would, that it's a big headache that I don't want any part of.

[00:16:57] No, and what if somebody goes out there and they [00:17:00] probably have used AI to write a whole book and published it, what impacts does that have? That's why if you, like we talked about it before about using shorter bits of text. If you ask eight, I generate one sentence.

[00:17:14] Chances are, sentence structures can only go so far, so many different variations. There might not be, again, I'm not a lawyer. There may not be any copyright issues with one sentence that sounds like something else that's out there, as opposed to larger pieces of [00:17:30] text. So again, it goes back to Short text and rewriting still, adding your personal touches, personal perspectives to anything that AI generates for you.

[00:17:40] Yeah, I think that I totally agree with that, that it's, having it small and condensed and, using it for the brainstorming or like a short intro paragraph that you just kick things off I, I think that is, a good use. A good use and a good prompt, and it saves you from any issues later on because then you know that you [00:18:00] wrote it.

[00:18:00] You're not like I didn't even know you don't wanna be responsible for something. Or even if it made an, yeah, not an allegation, but maybe made it a, an unsubstantiated. Like assertion and then you're on the clock for being cited for that. And so it's definitely one where I think that will really help a lot of businesses just to not accidentally say something that they didn't mean to or double check.

[00:18:23] Even a way around that would be use AI for research as opposed to content generation. I do that for [00:18:30] clients that I don't know anything about their niche and I need some top resources or links or just, definitions of certain things, use AI for that instead of generating the content that you're going to use in the end result.

[00:18:42] Yeah. This was a interesting point as well, that I think yeah, I think that it's a good idea to treat it. Yeah, how to view the copyright. I think that, yeah, that's a good way to do this. I think that, yeah, there's going to be, I think there's going to be a lot of conversations between people [00:19:00] about figuring this part out.

[00:19:01] I was going to say I had a thought it evaporated from me at the last second. But yeah, I think that it is, Oh, I was going to say what you're saying also worked with, like I'd heard the content manager for mailer light. The email marketing platform that they, the way they were using it is basically they have they have the research and they know their voice and then what they do is they just get basically the AI kind of what we're talking about to do a first draft of Hey, write me an article about [00:19:30] why.

[00:19:30] Cosmetic companies are getting a 5 percent increase in their open rates because of this type of email and so they have the data They know that they're putting that in and then they just the it does the boring part of actually writing like a structure And then they go through and add their voice on top of it.

[00:19:45] And so it's a good workflow of it's not It's like to me. That's a really good use case where you've got your fax, right? You've got voice, you and then the boring part of like actually writing out like A boring article that like, you've got to hit these points and it's [00:20:00] just that's the fun part That's the part.

[00:20:01] I like the writing part I'm a weird breed though Yeah no, it's good Like I think that and if that's what you're looking for like obviously you're using it for your own way That I think it's like looking at it as part of your workflow and how you work with it I think that's going to be the most important part I and I think that the other part that's going to be interesting as well is that google Didn't officially say but they've basically said that like It's good.

[00:20:28] There's so much AI content now [00:20:30] that they can't really distinguish the two. It's just so much. And because it's like a, because it's like a regressive, like an algorithm, they would use a regressive algorithm as well to be like, okay, in our large language model, we found that these words all came together, but it's it's becoming so much that they just can't even detect it.

[00:20:47] So that's going to be, as humans, I feel like, when you're reading something, you can definitely feel when they, Yeah, you could, it's just one of those weird things that humans do well of that uncanny valley where it's just it's close, but it's not [00:21:00] quite like I, why would a person say it this way kind of thing?

[00:21:03] Yeah. And people are smart. People know, they may not know what's wrong with the copy, but they'll read it, whether it's your website, your blog, your social media, and they'll just get that feeling that this isn't right. This isn't on brand. This doesn't sound like. The person behind the business and that, that really stands out, and I think maybe one way to get around not get around that.

[00:21:23] That's not the right word. One way to make it sound less AI is that if you do use portions of AI is to work in those stories [00:21:30] that AI doesn't know about, if you were to do a, write something about today's thing about today's live, we would be throwing in our own ideas and stuff like that.

[00:21:39] Our own stories. Like we've thrown in stories that we've heard in the news. I've thrown in stories from my business. And that will make it sound less robotic, AI still sounds, even if you ask it to be light and cheery, it still sounds robotic a lot of the time, people know, people will see that.

[00:21:54] Yeah. Or like a twee is another emotion that I noticed a lot with it, [00:22:00] where it gives me like a, Oh boy, like responses. And it's just it feels inauthentic, but that's because the machine wouldn't know that is. Would come across as inauthentic. They're just trying to be positive and happy and then but they don't get that It sounds like overly sincere.

[00:22:16] And so it comes across as insincere to people because most people are just using Yeah, or just have more world experience than the AI does and doesn't know how interactions on an emotional level work. And [00:22:30] so that's a limitation that it has. And that's just the brass tacks of it.

[00:22:33] And I think that yeah, I think ultimately like I would say yeah, it's not, I don't think it's coming for your jobs. I don't think it's coming. It's going to be the end of the world. Even looking at the example of the. Amazon grocery store remember the one that you could walk in and walk out?

[00:22:49] And it, they're just saying that it was all AI. And then it turns out like, no, it was thousands of people in India watching the cameras. And then I'm manually like charging people and it was, they, [00:23:00] yeah. Yeah. So like the AI vision of where like businesses think it's going to be and what it's going to do just doesn't line up with kind of the limitations.

[00:23:08] I think that it, yeah, I think that. ultimately it's just one that we're just going to be able to use. It's just like a easy automation tool. I think personally, that's where I see it really, it's future is not like taking over, but basically just fitting into the slot of all of our tech tools and how we use them, I think is a easier way to conceptualize it.

[00:23:29] Yeah, it's [00:23:30] like I said in the beginning, it supports content creation. It supports our businesses, but it's never going to replace us. We were always needed. And yeah, from people like me, I'm getting those questions all the time. So I have to have those responses on why they should hire me and not trust the AI.

[00:23:45] So it's adding a little bit more work to my plate now, but people are, AI has been around for content generation for, a couple of years now, the past year, really. And people know people understand that. And if your budget is a little bit shorter, you're a new business.[00:24:00]

[00:24:00] Maybe you might rely on a little bit more starting out, but as you grow, as you build into your own voice and your own perspectives and experience, you can start to pull away from relying on AI and using more copywriters, using your own voice and content. Yeah, I think that's a good way to look at it is, making your own resources go farther if you've got tight budgets, if you've got, you can look at this as a way to spread it out a little longer, make things work a little better because yeah, you're going to have a lot of, [00:24:30] you may have constraints or something.

[00:24:32] But yeah, I think that, looking at it more pragmatically as like a, a glue that fits into gluing more spots together than it is a replacement for whole departments. I think that's just optimizing business practices. What we're all learned when we learn about business, look for ways that you can optimize.

[00:24:47] I think this is one way we can. It's just, again, with the warning, don't rely on it as, don't fire all your employees and replace them with the AI. It won't last long. Yeah. Yeah. And it just won't, it [00:25:00] won't be as good as we expect. And yeah, there's a lot of comments here from Laura as well.

[00:25:06] I'll just add this one. It gets comical. Yeah. It totally does. Yeah. Yeah. It definitely, yeah. It has some issues that come out to the fore. And yeah, I think yeah, I think we got a good spot with it. I think that we looked, we're, I think both of us are going to be using it.

[00:25:25] I think that. It's just going to be more fit more as like a tool in our arsenal. [00:25:30] Then, yeah, I think that's where I see it's coming up. Compared to like, where, the hype, like of what people think that it's going to be. I just don't see it there. Yeah. I'm in complete agreement.

[00:25:41] No, let's use it to support us, to help us all grow and scale our businesses, but not replace us. It never will. I don't think it will. Yeah, I agree. All right. So I guess we will close it out there. It's been a great talk, Ashley. And again, my name is Phelan with Seymour Digital Media. And I'm Ashley at [00:26:00] WriterGal Marketing.

[00:26:01] Awesome. Bye for now.