How to Leverage AI in Marketing: What We’ve Learned | Part 2

As AI technology evolves, its role in B2B marketing becomes increasingly impactful, yet understanding how to use it effectively can be a challenge. In this episode, we continue to take a deep dive into our team’s journey with AI, sharing insights on the real-world applications and hurdles encountered along the way. From streamlining tedious tasks to refining SEO, we discuss what’s working well and areas where AI still falls short to support marketing efforts. 

(00:01:25) The learning curve of AI tools
(00:02:45) Experimenting with prompt refinement
(00:06:03) Using AI for customer research
(00:07:49) Enhancing buyer persona insights
(00:10:08) SEO optimization with AI
(00:12:07) Automating repetitive tasks
(00:13:26) Challenges with image generation
(00:17:04) Limitations in website reviews
(00:18:43) Struggles with financial data extraction
(00:20:04) Fact-checking AI-generated statistics
(00:22:00) Privacy concerns with sensitive data
(00:24:28) Building contact lists and data limitations
(00:26:07) Effective prompting techniques
(00:28:00) Understanding AI’s “hallucinations”
(00:31:43) Using AI to complement human work, not replace it

The Demand Gen Fix is hosted by Deanna Shimota, CEO of GrowthMode Marketing. Listen to our team of marketing experts and featured guests drop knowledge on how HR tech companies can maximize the success of their marketing for today’s B2B prospects. Learn more at www.growthmodemarketing.com.

Transcript

Jenni Geiser 00:00:01 Hey, everybody, it's Jenni GrowthMode Marketing. You're listening to The Demand Gen Fix podcast, where our team of GrowthModers and our guests discuss the ins and outs of demand generation, and why we believe it's the key to long-term sustainable growth, especially in the HR tech industry. 

Deanna Shimota 00:00:20 Welcome back to our multi-part podcast series where we are talking about AI in marketing. AI is rapidly transforming B2B marketing by automating repetitive tasks, enhancing creativity, and offering insights from large data sets. If you're not experimenting with AI yet, you're missing out on a competitive edge. So Greg and I are back. This is the third in our four part series. In episode one, we explored how to learn about AI and the resources that are out there, and the little encouragement around how you can start experimenting with the basic AI tool in episode two. Which was actually episode 70 of the Demand Gen Fix podcast. We covered some ways that we've been integrating AI into our work and experimenting with it here at GrowthMode Marketing, and today we're going to dive into more examples of what's been working for us, as well as digging into some areas where we're not quite getting the results we're hoping for.

Deanna Shimota 00:01:25  And we'll do one more episode after this on AI, where we will look at how important the AI and human collaboration is, because there's a lot to think about as you start to integrate AI into your marketing efforts.

Greg Padley 00:01:41  Yeah, we've been mainly experimenting with the paid version of ChatGPT for, but obviously there's plenty of other tools out there from Google, Gemini, Claude, all those. And then there's also more purpose built models like Grammarly for writing assistants and Jasper for marketing teams. But like I said, we've been mostly working in check GPT at this point. But it's worth mentioning whichever tool you're using that AI adoption is still evolving in B2B marketing, and a lot of marketers are still in experimental phase. The learning curve can be a little bit overwhelming, right? Because there's so many different things you can try and different things to think about. And sometimes people even just getting started is overwhelming because there's so many options, right? It's like you don't even know where to begin. Sometimes it seems like it's faster to do things the old fashioned way, or try testing things out, and you spend 45 minutes trying to test something, and then it comes out and then you're like, I could have just done that myself.

Greg Padley 00:02:38  But the more you do and the more you play around with it, the better you get at it. And the, you know, more efficient it gets overturned.

Deanna Shimota 00:02:45  Yeah.

Deanna Shimota 00:02:45  You bring up a good point. It does feel sometimes that it is easier to do it the old fashioned way. I've certainly had some things where I tinkered around so long and I was like, oh heck, I should've just done this. But the more you play with it, the more comfortable you get with it, the more you start to understand what it can do well and what it maybe doesn't do so well. And I think sometimes the tinkering around helps you refine it, where in the future you know how to do it. So it really will maybe not save time the first couple of times you try to do something with it, but over time, as you learn like what prompts work well and what it can do, I think that's where the real efficiencies can be gained. So I know last episode, Greg, you and I, we were talking about some of the things we've experimented with that have turned out to actually be pretty valuable when using AI in marketing.

Deanna Shimota 00:03:40  Let's talk about a few more things. But to add on top of that, today, let's also talk about some of the things that didn't work so well for us, because we've got plenty of those stories too, right?

Greg Padley 00:03:53  Yeah, yeah. I mean, it's funny because when you think about it, the AI isn't working well. Or is it? They're we're not asking it to do the right things or in the right way. So yeah, something always here. And also it's changing so fast, right. Like just because something doesn't work today doesn't mean it's not going to work next week. Whether you learn how to do it where the model got better or they just keep coming out with new versions, all the terms.

Deanna Shimota 00:04:17  Yeah.

Deanna Shimota 00:04:18  Very true. And I think a lot of times when you don't get good results out of it, there is something to be said about doing prompts and knowing how much information and how detailed and how to even state things on it. And we can talk about that a little later in the episode.

Deanna Shimota 00:04:36  But just because it doesn't work the first time doesn't mean you should give up on experimenting and trying with it. Because my first experiences with ChatGPT over a year ago, when it first became the buzz around town, when everyone was like, oh my God, this was a game changer. I was so frustrated with it. I kept playing with it and I kept getting crap out of it. And I'm like, I don't see why everyone's so excited about this. Like it's not working. And now I'm singing a different tune because we have learned a lot as an organization, and we have figured out like some things that it works well for and some things that it doesn't. And I still get frustrated with the I again might be I'm not asking it the right things, and sometimes there's just limitations to what it actually can do for you. But it can be a great tool for a marketing team when if you step back and you think about what are some of the things that we do that feel inefficient, I think that's a great spot to start to say, okay, what can we potentially use AI to help us streamline and become more efficient? So it's not replacing marketers by any means.

Deanna Shimota 00:05:49  You still need to have that creative brain. You need to have that people oversight. But there's a lot it can help do to make it faster, to do some of the tasks. That is, marketers, we commonly do.

Greg Padley 00:06:03  We we are. One of the things that worked out really well, I think, was when we were doing some customer interviews and we had transcripts from all these interviews. And in the past, you used to sit and go through everyone and highlight things or cut and paste things or whatever, but we could put those into a and get some insights out of it, and it would find them for us, instead of us having to do all of the heavy lifting and then obviously have to go back and confirm that there was not a mistake made in the process, but it pulled some things out that just made it a lot quicker to look through those transcripts.

Deanna Shimota 00:06:37  So, for example, let's say you're doing buyer personas or you're going out into the market and interviewing potential prospects to get information for your ideal customer profile, those type of activities.

Deanna Shimota 00:06:51  It's really great because if you recorded those conversation, like you said, Greg, you have that transcript. As long as you ask ChatGPT or whatever tool that you're using the right questions, it can summarize that information for you and it can pull out those key themes. And it really is a huge time saver because we all the time are working with clients where we're doing those type of interviews and to go in and manually have to read the entire transcript and highlight everything and pull it out and put it in a spreadsheet. Takes hours at times, especially if you have a lot of those interviews. So that's been a great tool for us going out and having those interviews with whomever it is that you're talking to to get the summarized version. Now, does that mean you never read a transcript again? Probably not, because you want to make sure that you're not missing things that maybe the eye is, Ideas, but it certainly speeds the process up.

Greg Padley 00:07:49  Yeah, you can prompt it to get certain attributes that you want out of it.

Greg Padley 00:07:54  You could even get sentiment. There's ways of prompting that get to get people's sentiments, looks for certain words that would lead you to believe that somebody was happy or they were dissatisfied or things like that. You know, it can really depending on how deep you want to get into prompting, you can really get some interesting articulation out of those terms.

Deanna Shimota 00:08:11  Yeah. And there's for anybody that's serious about using AI to help with marketing, it probably doesn't hurt to do some kind of training on prompts to understand, because there's things I didn't realize until I went through a training where the person giving the training was talking about, you have to basically input your customer profile like, here's your ideal profile. Then you have to explain what you're an expert in. So for example, like I am a market researcher who's really good at doing this. Like you're giving all these characteristics to the eye as direction. And when you do that, it really helps them to your point grade. When you talk about you're trying to capture sentiment.

Deanna Shimota 00:08:56  If you just put a prompt in and say, summarize this interview for me with the key points, what you're going to get back is a lot different than if you have all this very specific direction around. This is what I'm looking for in this, and this is what I want to pull out. And this is how I want to sort the information. And when you get to that much finer detail, I find what you get back is going to be much more in line with what you need. Then when you have the vague prompts that you put in there.

Greg Padley 00:09:29  Absolutely. It's like we said in our life, become the last podcast where you have an intern and you have the teacher intern, Greg, what to do and what to look for or else. And after we get that stuff, you walk from your intern.

Deanna Shimota 00:09:42  Right?

Deanna Shimota 00:09:43  But if you spend time doing prompts and you find that they work really well, like document those prompts, because then the next time you do it, it just makes it so much easier, because then you've already got the whole prompt mapped out for you, and you can just copy and paste that in, upload any other information you need to coordinate with it, and it'll deliver results for you.

Greg Padley 00:10:08  We've also used ChatGPT to help us with some SEO. As far as making sure that we have the right keywords, like we have a keyword list and then content is written, and we use that to compare the keyword list to the content to make sure that we captured it all. Instead of manually looking through a document and shipping it next to the spreadsheet of keywords, and just saves a lot of time to make just to confirm that things that are correct. Really? Right. Proofreading in a way. Right?

Deanna Shimota 00:10:41  Yes. It's like a gut check for you. Did we work in the proper keywords and phrases from an SEO standpoint? And do we have our H1 and H2 and help us write a meta description? I wouldn't recommend that you use ChatGPT, or maybe some of these other generative content tools to research your keyword. And I guess I say that with an asterisk because you can use it to research keywords, but I wouldn't do that as your soul way to go and identify it, because it can come back with broad stuff.

Deanna Shimota 00:11:14  It's not necessarily that kind of tool where if you use like SEMrush instead, you're going to have much more detailed information. So you just have to think about, okay, this tool is here to help ensure that we're working SEO into it. It's not necessarily the best tool I guess, is what I'm saying. Right? Right. To identify and feel confident about the keywords that you're working in to your content.

Greg Padley 00:11:41  There are a lot of things that we just say to you. Time. Really? Yeah. You know, creating lists, creating tables, it can give it data and say, make this into a bulleted list or put this in a table, I want this column. One is named list. Column two is name that. And instead of sitting there and cutting and pasting all that stuff you can, he just haven't done for you in 10s, which is really nice.

Deanna Shimota 00:12:07  Yeah, absolutely. I also think you can do it with pretty simplistic things that are just time consuming and not fun to do.

Deanna Shimota 00:12:15  If you have images or PDFs that have content in them, I makes it simple to pull the text from mouse so you don't have to go recreate the content if you don't have a word version of a document, right? It can pull that in. You can take that content. You can then drop it into the word document instead. I also think like cleaning up duplicates and spreadsheets. So I've used ChatGPT personally to say, okay, I've got this list A and I've got list B, I need to compare them to see how many duplicates there are and clean that up. Now, one could argue you technically could do that in Excel. I find that kind of cumbersome at times to be able to do that and make sure I'm not accidentally deleting the wrong things with ChatGPT, I've been able to put those two things in there and been like, okay, identify the duplicates here, and it just pulls that information out for you and helps you clean up. And then you can ask it for a spreadsheet that doesn't have any of the duplicates, and it will do that for you if you have the right prompts.

Deanna Shimota 00:13:25  Yeah.

Greg Padley 00:13:26  Yeah, I guess speaking of having the right prompts, some of the things that didn't work well for us, which like you said before, maybe it's our prompts that are no good, or maybe it's just that we're not using the right tool for the job. One of the things that you know, I've played around with is using Dall-E and or ChatGPT to create some images, and it always seemed to get so close to what I want, but I never quite get there. There's always something if I say, you know, I did a picture of a robot, right? And I wanted it to have a red heart on its chest. And then I had a great robot that I was happy with, and I was like, put a red heart on this chest. But then the heart was too big. So I was like, make the heart a little bit. And then it would change the whole robot. Yeah, not like the robot was fine. Why did you mess with the robot? Obviously, that's probably not the best tool for creating graphics, but that was what I was playing with.

Greg Padley 00:14:17  But it was like, man, why is this so annoying? And it all. It likes to always put words in graphics too, and but it always spells everything wrong when it puts words into graphics. Somehow the systems don't jive together and and it always puts typos in there and and you can't get it to fix the typos because then it changes the whole picture. But it still puts, in other words, that have typos. This is crazy. And then even sometimes, like I'll say, don't put any words or text or copy in the image and it still puts words in the image. I'm like, why does that keep happening? Like I said, it's not totally not really the best tool for that job, but it's one of those things that doesn't quite work the way you would want it to.

Deanna Shimota 00:14:59  Yeah, I've had the same experience. I have played around with the Dall-E integration with ChatGPT as well. And then here's the scene I'm envisioning. And let's say I'm trying to create a photo for a blog post.

Deanna Shimota 00:15:13  Instead of going and searching through all these, I'm like, I'll just create an original one here. And same thing. You get one image and you're like, okay, just refine this and this and it's completely different and not even like somewhat similar. Like it's so different that you're like, okay, go back to the other one. Nope. And then I've had times where I uploaded an actual image and I'm like, create an illustration that looks like this. And it's we we don't do that. It's just outright it's probably a copyright thing, right? Even though it was a personal photo that I uploaded, I think you can maybe get like blog images or basic imagery type of thing out of it. It's not something you'd want your graphic designer to use. There are certainly other AI tools out there that are better for creating images. One thing we haven't tried this, but we heard the suggestion from others in using the photo functionality or the graphic functionality and ChatGPT is you could say, create some images, give it all the information about what you're trying to convey in the image, and it can give you like concept, like a bunch of different concepts or treatments to the imagery.

Deanna Shimota 00:16:29  So if you were creating a mood board, for example, that you wanted to show your team. Like, here's the different type of styles that we could do. You could potentially do that with ChatGPT. But again, I feel like there are better tools out there from an image and design perspective that use AI. That would do a better job than ChatGPT does at this time, right?

Greg Padley 00:16:54  And I guess each thing's going to have its right. Each one is going to have its own strengths and weaknesses. So there's definitely other things out there for creating.

Deanna Shimota 00:17:03  Right?

Deanna Shimota 00:17:04  Another thing that we've tried to do. So we do these market differentiation audits that we offer to prospects as a free opportunity to compare, like how your messaging is out in the market, where we look at 3 to 4 websites. And we wanted to streamline that because this is something we're doing for free for prospects out there. So we couldn't go and spend 12 hours on each of these audits. We thought, let's see how ChatGPT does with comparing this, right.

Deanna Shimota 00:17:36  So we would input the four websites. We would say pull out the key themes. We give them all this information to do it. And I would say ChatGPT wasn't great for doing this, at least not the last time we tried to do it. And we've tried it more than once to do it because it struggles with evaluating the user experience and navigating these live websites, and it lacks the context to be able to do it well for you. So we've always had to do those manually.

Greg Padley 00:18:06  Yeah, I wonder too, there must be some like how you think about it. The website has all these different types of it. We'll have a video, it'll have a static image, it'll have a animation. So I don't think it's able to get all of the information that's in the website either. It's got to be missing. It is missing something. But I think maybe that's part of its relevance because it's trying to get all this information from these different types of files. It's not just giving them one PDF or one image, and it's trying to pull information from there.

Greg Padley 00:18:37  I wonder if that's part of why that never really has been successful for us.

Deanna Shimota 00:18:42  Right.

Deanna Shimota 00:18:43  Another thing we've tried doing is pulling financial data from PDFs. So let's say somebody sends you a PDF, and it started out as an Excel spreadsheet and you're like, oh, I need the original list to play around with it, right? We've tried to pull that information like the exact information from it, and ChatGPT has struggled with that. I think when you have numerical data from complex table formats, and again, maybe we just didn't know the right prompt, didn't figure that out to get it there, but what it would pull bits of pieces of the information and like when I was playing with it, I would try a bunch of different prompts and kept trying to get it to do it, and I just could not get it to go there. So is great as ChatGPT is at pulling information from PDFs. If it's copy, I think it works really well. If it's like number heavy Excel spreadsheet type of data that you're trying to pull in.

Deanna Shimota 00:19:45  That was a different story for us here.

Greg Padley 00:19:49  And obviously the like everything, right. You always have to fact check it. But if you've got numbers and you really need accurate financials or accurate statistics or something like that, you really have to double check that because those sometimes fabricate information.

Deanna Shimota 00:20:04  Yeah it does. And statistics are a whole nother thing. Like we'll be doing research for a client. And as part of our unique point of view exercise, a lot of times we're trying to find data and research out there that supports the position that they're talking about in their unique point of view, saying like air, like 95% of air. Leaders are frustrated with x, y, z. We have used ChatGPT to try to identify those metrics and statistics, but I've actually had it where I've gone back and said, please tell me the source where you pulled this information from. And ChatGPT has actually responded back and said, this is an example. I made this up. Oh boy. If someone doesn't realize that you shouldn't be pulling facts as your source of truth from ChatGPT, you're running with it.

Deanna Shimota 00:20:58  It's just making stuff up at times. And you don't expect that, right? Like you just think it's pulling everything from the World Wide Web. It's just a database of all those things. Surely this is accurate and surely it can tell me the source. It doesn't. So I'm like very leery now. I will not use ChatGPT to try to find metrics and statistics. And if I do, I go and try to find the actual source online before I would ever use it?

Greg Padley 00:21:27  Yeah, absolutely. When you Google search things right, sometimes you get told you get to a fact, but then you look at the source of on that website and you're right, I don't know. I'm not really too confident in the source. So it's similar to just going on the web. And if you think about it, all of the information that ChatGPT has was all public information off of the web anyhow. So it could be could be some website from who knows where it doesn't that that it refers to. So you really got to double check if you're going to use stats or facts like that.

Deanna Shimota 00:21:58  But even it could be users inputting information and maybe they're making up stuff. Or maybe they did their own internal minuscule research. Right. And they talked to ten people, but they're inputting stuff when they're using ChatGPT unless they have it marked as private. All the things that they're inputting into that anybody's inputting into ChatGPT or some of these other AI systems becomes part of the public domain. And even I just saw it on LinkedIn last week. Now, anything you put on LinkedIn, unless your profile checks that it can't be part of their public database, they're using it for their own AI tools on there. And I can guarantee you for the majority of people, that is going to be the case, because now your profile, like it's automatically marked to you can share it unless you go in and change it. Right. And most people don't know that's happening. It's the same with ChatGPT and and these other AI tools. They're usually set by default to share anything you put into the system, unless you go and change that.

Greg Padley 00:23:05  Yeah, I saw that with the LinkedIn. Cheers. And it's funny because it it's not just LinkedIn that also says LinkedIn and our affiliates. So that means what was there affiliate for you. Everything you write and everything you've closed unless you click that one little button that says don't share my information. Who is their affiliate? They're getting paid. Somebody's paying them to use. Then get me somewhere. Pay me.

Deanna Shimota 00:23:30  Yes. Pay me. You're taking my knowledge and my brain dumps, right? Yep. One of the other things that we've tried to do that hasn't worked well for us was trying to build a contact list from websites, so I couldn't reliably pull contact information from websites, especially when the data was inconsistently formatted. And so an example that's specific for us. There's what, 400 exhibitors at the Air Technology Conference, for example, and they post that on their website. So as an HR tech marketing firm, we want to know the companies that are out there in the tech space. It's very manual to go in and compare it against our existing list and find the companies that weren't on our list and not our rater Moderator before that.

Deanna Shimota 00:24:18  So we asked ChatGPT to help us with that. And I know that's one of the things I throw at you. Okay. Figure this out. And we never figured out how to make it work. Right. But we spent a lot of time trying to get it to there.

Greg Padley 00:24:33  I don't know how to fix that. I wish I could because it's frustrating. Right? It's like you have this puzzle. How can I undo this puzzle? But then how much time can I spend on doing this puzzle?

Deanna Shimota 00:24:42  Right? And that's the exactly the type of thing. Like when you step back and say, what are things that are inefficient in our day to day work as marketers? That's one of the first things I'd say is, oh my gosh, I wanted to build this contact list of information that's on the web. Whether you're going to one site like we did with the technology conference, or you're trying to scrape information from multiple sites to build your list, because we haven't found a way to do it.

Deanna Shimota 00:25:08  And it feels like that should be an easy one. Right?

Greg Padley 00:25:11  Right. That's I think that's why it's so frustrating that this should work. And why doesn't this work for me.

Deanna Shimota 00:25:17  Yeah, they.

Greg Padley 00:25:18  Will figure it out.

Deanna Shimota 00:25:19  Yeah, someday. And it's going to be such a time saver for us and for our clients if we can figure out how to do it.

Deanna Shimota 00:25:26  Well.

Deanna Shimota 00:25:26  Yeah, absolutely.

Deanna Shimota 00:25:28  So I think there we've talked about some of these things as we're talking about it, but there are definitely some things to always keep in mind when you are using AI tool. The first one being from smarter. Quite frankly, how you frame your request is crucial. And when it comes to prop design, clarity and specificity are key. So instead of saying give me a summary, try something like summarize the key findings of this report in three bullet points, and then go deeper from there. Because the more specific you prompt, the better the output.

Greg Padley 00:26:07  Yeah, absolutely. And there's a lot. In our first podcast on this week, we talked about some resources of where you can learn more things and who to follow on LinkedIn and newsletters that that go into more detail of effective prompting.

Greg Padley 00:26:21  So there's definitely tons of information that you could find out how to do that and how to get better on it.

Deanna Shimota 00:26:27  Yeah, and there are people out there who have built GPT that are publicly available that actually already have like prompts built in. For example, if you're trying to create a profile of your ideal customer, there are prompts out there that have enough questions and things built in to pull the information out to help you get there. Now what I say that should be your strategy from a marketing standpoint, to just go to one of those GPT and enter the information. And there you have your ideal customer profile. No. But if you know who your ideal customer profile is and you have it documented this, it's like the thing I'm thinking of, the example, it was actually a prompt creator, meaning it asks you all the questions. And then it came back with a succinct paragraph and it was like, this is what you should use in all your prompts when you're creating content. Moving forward to explain who your audience is to the AI.

Deanna Shimota 00:27:28  So there's tools like that out there that can help you. so you don't have to come up with prompts from scratch for everything.

Greg Padley 00:27:39  In speeding them. Everything. Don't forget to fact check everything, right? I know we talked about that already. Hey, it's great for giving you responses, but sometimes they're not really true. So you always gotta verify that that's true and that it's written the way you want it. You don't just cut and paste something without reading it, and make sure that it's really the way you want it to be said.

Deanna Shimota 00:28:00  Yeah, always fact check everything. Couldn't have said it better myself, and I think I gotta leave them to the next thing that you need to keep in mind, which is called AI Hallucination. So sometimes I provide made up answers like we talked about before, or even nonsensical information. So you always want to verify the responses. sometimes it's really obvious that it's an AI hallucination. Like, I've had some experiences where I felt like my prompts were clear, they were specific.

Deanna Shimota 00:28:35  I was trying to get very specific information from ChatGPT cold out of like, transcripts that I had, and it would come back and what it was talking about. I'm like, where is the story coming from? This is not anything we talked about in the transcript that I put in there. And I kept trying to tweak it, and I kept trying to tweak it, and it just the AI hallucination. I don't know that ChatGPT was on mushrooms that day or something, because it just could not get it right. So I had to trash that conversation and start from fresh. And then I did it and it worked. So it's a really weird thing, the I hallucination. Like I said, sometimes it's pretty obvious that it's doing it. It's not out of you all the time though, and that's why you should always verify responses. And that's why if you're trying to pull like statistics and research and things like that other, that is that critical information that you have to fact check everything.

Greg Padley 00:29:35  And if it's private critical information, make sure that you have your privacy set up.

Greg Padley 00:29:40  You don't want to share your any of your sensitive information to the AI. So make sure that you go into your settings and make sure that you're not sharing any of that sensitive or proprietary data.

Deanna Shimota 00:29:53  Yes. So for example, let's say if you're just entering prompts and pulling information from ChatGPT, that way you probably don't have to worry about the privacy setting. But if you are, for example, building a GPT, which for those that don't know what a GPT is and ChatGPT, you're basically building a database of information and prompt that you can use, and you can choose to make that public or private. Right. And let's say you're uploading all of your company's marketing content, for example, because you want to create a database of that content so that you can create article lines and social media posts and things like that moving forward from your existing content library. You probably don't want that to be public, because if it is public, everything that you uploaded on there now becomes part of the general ChatGPT database.

Deanna Shimota 00:30:51  And what does that mean when somebody else is talking to ChatGPT and they're trying to come up with really good marketing language for their home page of their website. Some of your smart Thinking.

Deanna Shimota 00:31:06  May end up.

Deanna Shimota 00:31:07  In the recommendations that come to the next person, right? Like the language you use and things like that. And that's where there's a risk. And also there's a lot of concerns from an AI standpoint for copyright and who legally owns content and things like that. So you just need to be mindful of that. And I think it's always just safer to turn it to private. But you can only do that on ChatGPT if you have a paid version, I believe.

Deanna Shimota 00:31:36  Yeah, I think.

Greg Padley 00:31:37  That's the way it is with most of them too. It's like you have to pay to play.

Deanna Shimota 00:31:42  Yeah.

Deanna Shimota 00:31:43  So to sum it up, as we wrap up this episode, AI is a tool to complement, not replace your expertise. And we suggest you start by incorporating AI until 1 or 2 tasks that you're comfortable with, such as content, research or proofreading, and build it out from there.

Deanna Shimota 00:32:03  I think you'll find the more that you use it, the more value you'll discover. So hopefully this episode and the previous two have been valuable. We have one more episode that we are going to hover AI in marketing in, where we're going to take a deeper look at human AI collaboration and the importance of human oversight and all of the things that you do. So tune in again next time to the The Demand Gen Fix. And if you have any questions about AI and marketing and how we're experimenting and learning about it, feel free to reach out to our team at GrowthMode Marketing.

Jenni Geiser 00:32:40  Thanks for joining us on The Demand Gen Fix, a podcast for HR tech marketers brought to you by GrowthMode Marketing. I sure hope you enjoyed it! Don't forget to subscribe for more perspectives on demand generation and B2B marketing strategies. Plus, give us a like. Tell your friends. We'll see you next time.