AI Meets the Pen: How Writers Are Embracing AI Without Losing Their Voice

Julia Bye

Senior Director of Account Strategy

SmartBug Media

Rebecca Corgan

Copywriter

SmartBug Media

Back to All Episodes

Can AI and authenticity coexist in content writing? SmartBug Copywriter Rebecca Corgan weighs in.

• 43 min

AI Meets the Pen: How Writers Are Embracing AI Without Losing Their Voice

SmartBug on Tap

Play

AI is changing the content game—but how do writers keep their creativity, voice, and humanity intact? In this episode of SmartBug on Tap, host Julia Bye chats with SmartBug copywriter Rebecca Corgan about how content creators can embrace AI tools like ChatGPT and Gemini without compromising authenticity. From overcoming blank-page syndrome to building powerful second-brain workflows, Rebecca shares how she integrates AI into her writing process—without letting it take over. Learn where AI shines, where it falls short, and how to strike the right balance between efficiency and originality. Whether you're a writer, marketer, or ops leader curious about the future of AI-powered content, this episode is packed with insights you won’t want to miss.

📚 What You’ll Learn

How writers use AI tools for ideation, drafting, and content scaling
Why iteration is key to making AI outputs usable
Where AI tends to fall short—and how to edit out the "AI-isms"
How SmartBug is building responsible AI practices into workflows
Tips to keep your voice and brand intact when using AI for content

🔑 Key Highlights & Timestamps

[00:13] Intro to the Episode
[01:22] Meet Rebecca Corgan
[02:43] Best Use Cases for AI in Content Writing
[05:21] Beating Blank-Page Syndrome
[07:18] What AI Does Well
[08:57] What AI Still Gets Wrong
[10:25] Why Great Prompts Matter
[12:23] Iteration is Everything
[15:29] When AI Helped (and When It Hurt)
[19:17] Rebecca’s Framework for Using AI
[21:24] Brainstorm Partner vs. First Drafter
[25:43] Top 3 AI Mistakes Writers Make
[29:47] Fact-Checking and Gut-Checking AI Outputs
[32:04] The Plagiarism Problem
[33:25] The Ethical Line: Keep Humans at the Center
[37:21] Favorite Tools, Prompts, and Workflows
[40:15] Final Takeaway

🔗 Want to Connect?

👉 Contact SmartBug: www.smartbugmedia.com/contact-us
👉 Contact Julia: jbye@smartbugmedia.com
👉 Contact Rebecca: rcorgan@smartbugmedia.com

Relevant Resources

[Vidcast] AI Meets the Pen: How Writers Are Embracing AI Without Losing Their Voice

Julia Bye (00:13)

Hi everyone, and welcome to SmartBug on Tap, the show where we dive into the strategies, tools, and real world tactics that help businesses get more out of HubSpot and beyond. I'm your host, Julia Bye, Director of Account Strategy here at SmartBug Media. And today we're digging into a topic that's generating a lot of buzz, AI and content creation. From blogs to landing pages to premium content, artificial intelligence is reshaping how writers work. But it raises a big question.

How do we use AI without losing our unique voice? To help us unpack that, I'm joined by someone who knows the content world inside and out, Rebecca Corgan, one of our incredible copywriters here at SmartBug. Rebecca has been experimenting with AI tools like Chat GPT to speed up her workflows, overcome writer's block, and explore new ways to brainstorm. But she knows the limitations, and it's important to keep content human, especially in a world filled with AI noise. So whether you're a writer curious about AI, a marketer wondering what it's worth automating or just trying to get more done without compromising quality, you are in the right place. Rebecca, welcome to the show. We're so glad to have you here today. Tell us a little bit about yourself.

Rebecca C (01:22)

God, Julia, thank you so much for having me. This is the highlight of definitely my week.

Yeah, so I'm a writer here at Smartbug, but before then I worked in research science and in data analysis. So I think when AI tools arrived on the scene, I was already kind of in that headspace that likes programming me storytelling headspace to see how we could use these tools to realize more efficiency, but do it without completely steamrolling over the more fun parts of my profession at the same time. So it's definitely been a wild ride, but I think we're starting to arrive at some lessons learned that can help everybody feel a little bit better about using AI going forward.

Julia Bye (02:02)

Awesome. Thanks, Rebecca. I'm so excited to have you here too. So let's jump right in. AI is showing up in nearly every part of marketing right now. I don't think I can get through a day in my job without someone mentioning AI. But when it comes to content, it can feel especially personal. Some people are skeptical, some are all in, and a lot of us are somewhere in between. So today we are going to unpack how writers are actually using AI in the real world, what it helps with ...

Rebecca C (02:04)

Yeah.

Julia Bye (02:28)

where it falls short and how to strike the balance between the efficiency and the authenticity. So let's dive in. What kinds of content are you seeing AI used for most effectively? Are they blogs, landing pages, gated content?

Rebecca C (02:43)

I mean, why not all three?

First of all, I do want to validate what you just said. There are a lot of us who feel a certain kind of way about AI, and that makes all the sense. These tools, while they've become pretty ubiquitous over the past couple of years, they're still so new. We're still in the very early years of this whole AI revolution thing. So if you're listening to this and you feel weird about AI, you're in good company. I feel weird about it too. That said, let's talk about how we're using it. I think blogs, landing pages, gated content, all three, they're all great contenders for automation.

at some level as are honestly most of the assets that I put together. However, I think it's probably a more interesting angle.

to consider what you're trying to do with that content rather than the specific type of content you're creating. So if your goal with a specific piece of content is to educate, to inform, if you're using it as a pretty basic SEO or AI search play, that might be a good contender for automation. If on the other hand, you're trying to convey what's unique about your company, if you're trying to surprise, to delight, to engage, to connect, then maybe that's an opportunity for your own human

human-led

brain situation to shine through. Now, most of the content that we're putting together these days does need to kind of play in both of those arenas. So I think the best way going forward is to figure out some way to use AI for a bulk of the stuff you're doing if you feel so inclined, but to figure out a way to let your creative angle always shine through.

Julia Bye (04:13)

Yep. I completely agree. I think it's entirely about a balance. And one of the ways I personally have found AI with content really helpful is almost using it as a thought partner for me. And I'll go in and say, this is the audience I'm trying to reach. Like, does this align with what you're seeing online? Because personas are a lot of work and time, but when you have AI who can go out there and kind of dig around, it helps attend in that way too.

Rebecca C (04:38)

Well, and also when you can use the persona and kind of take it to a next level, when you're putting together a persona document, you're necessarily limited to a couple of pages because nobody is going to read through like a 50 page thing. But if I can safely, securely give Gemini or ChatGBT an idea of the persona I'm talking to and then be like, OK, I need you to get in the head of this persona. Give me three unmissable, hilariously specific pain points that they're going through and weave that through this content.

That's so helpful. I've not lived in every industry that I write for. That's impossible. having that as a second brain slash initial research tool has been so fun.

Julia Bye (05:09)

Great.

Yep.

So how has AI changed how you approach a blank page or the ideation phase of writing?

Rebecca C (05:21)

It's completely flipped it around. my God. So blank page syndrome, the worst. We all hate it. Writer or no, if you're looking at your blanking cursor, you're not having a good time. using AI to generate that icky first draft helps you skip that step, skip the first couple steps of I don't know what I'm doing to get you to maybe the more fun parts of writing, which is maybe editing and tweaking things so it feels like you. And I think that's helpful because of psychological reasons.

The human brain is really really good at knowing what it doesn't want right. It's like the reason rock paper scissors works. You understand what you want, what you want, the moment you want or lost. So if you get that first output from AI and it's not resonant, you already are primed to have a much better idea of what you want based off that. Now of course that is a fine line to walk. If you're getting really not great outputs out of AI,

that can do more harm than good because it can feel like you're setting yourself back a little bit. So if you're using AI to combat blank page syndrome, which I think we should all be doing, you got to strike that balance between putting in enough work to get something good out, but then knowing that that is not where the writing and editing process ends by a long shot.

Julia Bye (06:34)

Right. And even beyond the blank page syndrome, I can't tell you how many times I'm like, I'm trying to say this, nothing that I'm writing is sounding good. And I'll just say to AI, like, please make this sound better. And usually that's enough of a prompt to give me like at least a starting point to go on from there too.

Rebecca C (06:48)

yeah, God. We'll talk more about iterating. think iterating is a magic ingredient that a lot of people skip. Why are we not iterating more? But you don't have to be like a good writer, whatever that means, to iterate well. You can literally say, hey, I don't think this is landing. Can you give me some idea of where to go from there? And then, as you said earlier, you can use it as a thought partner and really just make things go forward a lot more easily.

Julia Bye (07:12)

So let's talk about the good, the bad and the robotic. What do you think AI does really well when it comes to writing content?

Rebecca C (07:18)

So second brain stuff, right? AI is so, so good at going through and summarizing and finding patterns in huge swaths of information. So if you have SME transcripts, if you have onboarding discovery documents, all that stuff that can be kind of hard to hold in your brain, especially if you're working with seven clients at a time.

AI can do that. AI can tell you exactly what you should be looking for in all of that so that you can use it later on. On the other end of things, AI is usually pretty good at providing short snippets of straightforward copy in response to a very specific need. So like meta descriptions, meta titles, product descriptions, FAQs, these kinds of things, which they're not unimportant, not by a long shot. But if you have a limited amount of time and creativity

to work with, then those are pretty good contenders for just one and done automation. I'm never writing a meta description from scratch ever again. The thing that maybe it's not as good at is being unique, right? So just based on the way that LLMs work, whenever you put in a prompt, the thing that you're getting back is going to be the most algorithmically likely thing.

So in order to accomplish any content goals related to SEO or standing out or in those fun more holistic things, you're going to need to take whatever the algorithmically perfect thing that AI gives you and treat that as raw clay that you can then put your own weird edge on to have considerably more fun and more return.

Julia Bye (08:57)

Mm-hmm.

Yeah. Yeah, and I also feel that AI can sometimes be overly positive with things and also, like, they love a bullet point. Like, I've gotten emails, like, suggestions where it's, three different sections of bullets. I'm like, this is way too much.

Rebecca C (09:12)

Yeah, too much. And you're so right about the hype person thing. AI went through a phase a couple months ago. It's still ongoing. Where you could give it a bad idea, like objectively just the worst idea. And it would be like, my god, Rebecca, you're so amazing and wonderful. I'm going to implement this right now. And I'm like, well, thank you. I needed that. But also, this is not helpful.

Julia Bye (09:31)

Right.

Rebecca C (09:32)

So maybe a tiny pro tip, if you ask Chat GPT specifically, I'm still testing Gemini, to have a more balanced approach, then it will modulate a little bit. But for the most part, you gotta have your reality hat on when you're working with AMA all the time. Yeah.

Julia Bye (09:48)

Yep. That's a great suggestion. I also read

somewhere that like saying please is not necessary and actually adds a lot more like data processing that isn't needed. So like all of us trying to host robots for when they take over the world, like it's not needed right now.

Rebecca C (10:00)

Really?

my god, okay, well, I don't know if that's a habit I'm gonna be able to break, because I love having this, like, if I say please to it, then it's like, my god, Rebecca, I love you, and that's nice.

Julia Bye (10:08)

Thank

huh. Yeah,

I know. I think we're just polite people in general, even to the thoughts. All right, so you've already mentioned a few areas where AI falls short or requires heavy human lifting. What are some other examples you have?

Rebecca C (10:25)

So where does it require heavy lifting?

three places come to mind, you have to give your AI good instructions going in. If you just have a blank, I'll keep deferring to chat GPT since that's what most people use, I think, if you just have a blank chat GPT chat and you say, hey, write me a good blog about HVAC systems, you're going to get nothing in response. So the magic comes in when you take the time to either prompt engineer a crazy prompt full of all of the background, the personas, industry

information, any fun sources, might need more information than you'd think, like fully a prompt as long as whatever you're planning on getting out. Or increasingly, I think there's a lot to be said for custom GPTs or, as we're testing here at Smart Bug, custom gems, which are like pre-made little bots that you can program with instructions and stuff and then share within the Google Workspace environment. And in those, you can upload again.

Julia Bye (11:09)

you

Rebecca C (11:22)

the things that a specific chat might want to know about a type of project you'll be doing or a client that you'll be producing content for, as well as any other maybe specific to you writing styles like only use emojis here or if you're putting together a chat that's just for LinkedIn posts, make sure to always do X, Y, and Z

that's the first part. You gotta give the AI so much instructions, so many instructions, because it's not gonna know how to do well or to deviate from that boring algorithmic standard on its own.

Place two, iteration. So you're never going to get something good the first time out of AI. And I feel like, I see a lot of people complaining, justifiably, that AI is not doing too great when it's very clear that they've just said, hey, write a good blog, and they've gotten something sad in return. So iteration can actually be a lot of fun. So for my entire career as a freelance writer, I've been the one producing the deliverables, right?

Julia Bye (11:59)

Mm-hmm.

Rebecca C (12:23)

the one typing for eight hours a day. But when you're iterating, you can kind of put on your director's hat. You can kind of put on your client shirt. You can start to be picky and stubborn in the ways that maybe as a writer you've been more on the receiving end. And that can feel cathartic and very nice. And it can give you skills that maybe you could start to use in other facets of your career, which has been very fun. But anyway, Iteration. After you get a blog back, you need to go back in and say, hi.

actually this doesn't speak to the persona. This introduction is too long if we're talking about a blog. I would like to see the first line of this rewritten with a bunch of hypothetical question and a bunch of endearing examples that my IC will immediately connect with. And you got to do that like five or six times. It doesn't take that long once you're used to it, but you got to do it.

Julia Bye (13:16)

different LLMs are good at different things. So I've even seen clients take information in one LLM that they've generated and put it in another one. Specifically the example that comes to mind is with video, but like not just being limited to one tool. feel like a lot of us, like you said, are comfortable in chat GPT and that's where we live. There's so many out there now and they're all good at different things. They all pull from different areas. like

I mean, you'd have to go off the deep end to like learn the nuances of each, but there are people out there who will do that. Or we could ask them specifically what they're good at too.

Rebecca C (13:48)

Yeah, I feel like every week just another person in my orbit, whether it's one of our colleagues here or whether it's somebody in my LinkedIn feed is touting the virtues of a new platform. And I feel like it's easy to feel skeptical about that because everybody and their mother has an AI tool these days, but you're right. Some of them are quietly super powerful and

It can feel fun to cobble together your own dashboard of tools that you use. I know that I use ChatGPT a lot for creative content, but increasingly I'm turning to Gemini because they've really upped their game in the past few months, I guess. Everything's moving so fast. And then I use Perplexity a lot for research, which always check your own research. But it's a good first step a lot of the time.

So the third place that humans are going to have to always provide heavy lifting is the final bit, the editing, the tweaking. Even after you've iterated a bunch and gotten something that feels kind of in the right direction with AI, you're going to need to go back in and add inside jokes and specific client tweaks and have yourself and ideally another human read through it to make sure that nothing stands out as weird or as overly AI generated. If it's not your goal to sound AI.

generated. And again, that doesn't have to take too long. You don't have to be a good writer to do it. It's just a matter of giving yourself some space and time, not holding AI to this crazy perfection standard because that's unrealistic, and making it sound like you at the end. Yeah.

Julia Bye (15:17)

Those are

all great suggestions. I completely agree with. Can you share a specific time when AI helps you move a piece of content forward and when it, another example of like the alternative flip it on its head when it completely missed the mark?

Rebecca C (15:29)

Yeah, so I think a surprising place where AI was super helpful recently was when working with clients that have

a long list of super exacting standards and specific tiny things that they want to see in their content. Maybe these are clients that we've been working with for a while. Maybe they have, and this is great, just really high standards for their content. And where previously we would have had to have teams of people just going through and checking commas and adding spaces before dashes and stuff like that. Now I can kind of just put that laundry list right into a custom GPT and whether

I've had AI write something or whether I've written it from scratch, having AI just kind of implement all of those changes has taken hours off of what used to be a really, really involved and not super rewarding creative process.

Julia Bye (16:25)

For me, that's a

huge win because think about salespeople. Every time I write an email for a salesperson, they're always like, well, this doesn't sound like me. I'm not going to send it. But now with AI, we can do the practice of here are five emails this person has written, match the tone of that email to the examples used. And then all of a sudden, things that take us hours of revisions are fixed in minutes.

Rebecca C (16:51)

Well, I think that's how AI is when it is used well. We're all overworked. We're all scaling. We're all growing our businesses. And that's great. But in order to make sure that's feasible without completely burning us out all the time, and also without expecting things with the human brain that are kind of unrealistic, human brains are amazing, but they also make mistakes all the time, you can have computers do that. It's the future, right?

Julia Bye (17:17)

Yeah, What about as far as missing the mark?

Rebecca C (17:19)

missing the mark.

So I can't speak to anything super specific that I've done for a client because we're having a good time today, right? But let's see. I have noticed that when people, me, my team, in general use AI specifically and only to save time, that never goes well, right? So if you're like, okay, a normal blog takes four hours. With AI, I can technically do this in 15 minutes. So let's just do it in 15 minutes. It's never going to work well. The client doesn't have to

Julia Bye (17:27)

you

Rebecca C (17:50)

to be super attuned to what good content is, alarm bells will just go off at some point. Your editor is going to be like, this sounds weird. The client's going to be like, I don't know what's wrong, but I don't love it. So I've definitely seen more requests coming in recently for me to retouch content that a client has written, that somebody here has written, maybe that I've written. I've gotten requests to clean them up and humanize them a little bit

Yeah, I think as far as missing the mark, if you're only using AI to save time, that's probably going to be a situation where you're missing the most of the time, unfortunately.

Julia Bye (18:23)

Right. Yeah.

And this kind of goes back to what you were saying with telling all the prompts that you need to tell. Like this is the branding. These are the thing, the words we say, the words we don't say, how we describe ourselves. And that's the thing that I see AI not getting right a lot. In fact, I'm seeing clients build their own AI tools that will train them how to do it, which is great. But again, like you have to have the human touch to know if it's authentic and you and all the

Rebecca C (18:47)

Google's rewarding the things being authored, things being associated with a specific human, people talking about their experience and expertise to establish authority and trustworthiness, EAT. So if you're just using AI to save time and skip that step, it's going to miss the mark because it's not going to sound good. And also, you're just not doing yourself any favors.

Julia Bye (19:09)

All right. So let's talk about finding the right blend between human and AI. How do you personally decide when to use AI and when to build something from scratch?

Rebecca C (19:17)

So I think I run it through a few gut checks there. So we're going to look for speed, scale, and context.

Speed. how fast does whatever this thing is need to happen? Am I being given a really quick turn assignment that for whatever reason is a five alarm fire, it needs to happen three hours ago? If so, then definitely I'm going to use AI. That is the only way that's going to happen. Scale. So am I putting together one LinkedIn post or am I writing 20 for a gigantic campaign? If I'm only writing one, then economies of scale, it might not make sense to go through the whole priming, whatever, if I don't have a chat already set up.

But if it's 20, then that's clearly going to save me a lot of time and headspace. And then context.

when I'm being given this new project, is there a lot of data that I need to assimilate and have the deliverable reflect? Again, back to purpose of the piece, is this something that needs to be magical and unique and fun and original? Or is it something that's just like, we are writing product descriptions? I will say I usually start with AI pretty much regardless, even if it's just from that brainstorming or thought partner.

angle. For most of my clients, I already have AI chats set up. So the second brain is already there. Why not consult it? And even if I have a new client or a weird one off project, I probably have a similar AI tool already set in place that I can be like, Hey, dude, I need to write a blog and I don't have time. This chat has a really good thing going. Can I leverage that for a client in a different industry? And again, even if it's just getting me

past that blank page or helping me build some momentum on a project, it's always a good thing. So how do I decide when to use AI? Pretty much always in the beginning. But then how much I let that be part of the process depends on a bunch of different factors.

Julia Bye (21:14)

Yeah. Yeah. So let's talk a little bit about that process more. Do you find that you're using it more as your brainstorm partner, your first drafter or something else? And you've hit a little bit on this, but.

Rebecca C (21:24)

So I think it's a mix, right?

I think as a writer, I mostly use AI as a first drafter. I've got tools built with these second brains already in them. For most of the content that I produce as an agency writer, we've got a formula working, we know how things are going. So I can very easily just be like, hey, here's a new blog, here's the key points, let's get this going. And that's how I'm able to write at scale. However, there are some super fun projects that I'm taking on as I'm getting more into ops roles.

messaging projects and stuff like that. And for those, it can be super helpful as I'm starting to expand what I do to be like, hey, I actually need to kind of direct strategy a little bit more on this. If we're going with this content strategy school of thought, what would be a good move here? How can I make this persona really feel seen and heard? What?

buyer journey stage, what they likely be at, know, stuff like that, stuff that I have learned in real life through, you know, courses and by working really closely with a lot of cool people. But as I'm starting to do that more, it's really helpful to have something that I can lean on as a first foil. Yeah.

Julia Bye (22:39)

Mm-hmm.

Using AI as a brainstorming partner is one of my favorite things to do. I love the creative part of my job, but it's just like, okay, here's my strategy. Check me. What is really good about this strategy? Where are the holes? Let's poke holes in it how can we make it even better? It's been super fun.

Rebecca C (22:56)

I love that.

Using it for a critique is really fun. So you will get AI as hype person saying like, you're just amazing no matter what you write. But if you're specifically like, hey, this lead nurture needs to do specific things, X, Y, and Z, I am unsure about one, two, and three. Does it work in that lens? Do I need to add anything for that? And I wouldn't say you should take everything that it says with that whole salt shaker of salt, but it can be really helpful to get your brain moving to help you fill in gaps, as you said, that you may have missed.

Julia Bye (22:59)

Mm-hmm. Mm-hmm.

Mm-hmm.

Rebecca C (23:26)

It's also super helpful if you're looking to

experiment with like tech that you don't know anything about. So I was recently trying to figure out how to optimize an internal workflow for one of my colleagues. And I realized that I'd have to get into like super crazy Google Sheet macros and like mess around with Google Apps script. And I'm not a programmer, but I was able to use Chat Sheet BT to be like, this is what I'd like to do. This is my available skill set. I know how to program at this level. Is there anything I can do here?

Julia Bye (23:49)

Mm-hmm.

Rebecca C (24:00)

And we were able to put together a pretty decent initial proof of concept tool with ChaiTBT doing a lot of the scripting, which was really cool. Yeah. Yeah.

Julia Bye (24:10)

That's very cool. I

use it all the time to help me almost like as a service, a HubSpot service partner where I'm having this problem, how do I fix it? And it'll go in and tell me like ways to troubleshoot it too. So.

Rebecca C (24:18)

Ready?

That's so cool. my God.

I love it when an AI tool is able to take context or build in research like that. This is late and breaking, but we're experimenting with gems right now. And we just found out that you can set up a custom gem with all of these custom instructions built into it. But then you can open it in the sidebar of a Google workspace, like a Google Docs or something like that, and just ask it to reference things within the page. And it'll do that.

Julia Bye (24:31)

Mm-hmm.

Rebecca C (24:54)

That felt like magic. my god. There are so many cool things we can do here as long as everybody's comfy and happy, which we'll talk about I'm sure in a minute. But as long as we're taking care of our people, there's a of cool stuff.

Julia Bye (24:55)

awesome.

Right.

Yeah, it's making it. I feel like I've read this every day on LinkedIn. not making our, it's not taking away our jobs, but it's making us more efficient and better at our jobs.

Rebecca C (25:15)

and helping us enjoy our jobs more. You were saying that you really are into the creative parts of your job. So Amai, I think that's all of us. And I think there is a world in which with AI as a strategic partner in some capacities, we can all do the stuff we like more. So that's awesome.

Julia Bye (25:20)

Mm hmm. Yeah.

Great. Let's make more time for that. I love that.

All right. So let's talk about common AI pitfalls and how to avoid them. What are the biggest mistakes you see people make when using AI for writing?

Rebecca C (25:43)

So these three biggest mistakes are going to mirror.

An earlier answer. So we were talking earlier about where do humans need to come in and do the heavy lifting, right? So the three biggest mistakes that I see are when people skip those. So I'll go through these quickly, but you're not going to have a good time with AI if you go in without a plan. You don't get good stuff out of AI if you don't know what good is. That's something that I would like to have tattooed on my arm and have scribbled over every surface. Just like, I'll get a billboard, you know? And that doesn't mean you need to be a good writer.

Julia Bye (26:07)

you

Rebecca C (26:13)

What does that mean? It means that you need to go in with strategy. You need to know what you want unless you're going in specifically for the brainstorming partner thing. You need to have a plan going in and you need to give information to support that plan. So you're not just going in completely expecting miracles. Second, settling for the first draft. You got to iterate. First draft is never going to be good. And even if it is good without iterating, you're missing a key opportunity to put even more

of a specific stamp on there to get away from that algorithmic standard that you're going to be saddled with if you just stop before iterating. So, got it. Great. And then, leaving in the AIisms. So, LLM's work by collecting everything on the internet, I'm generalizing, but it collects everything that has been written and uses what works well to generate further outputs.

So if something is really, really popular out in discourse, it's gonna get put in the mix to get used a lot more. That's why there was that thing a while ago about dashes and how using dashes was an AI tell. It's because we all like using dashes. Same thing for like dive in and in the story landscape of that's just AI trying to do what's cool and fun and popular.

Yeah, but the thing is, whether it's fair or unfair, if you skip the final editing phase and you leave those AIisms in,

If you don't want to sound like AI, that's going to be the best way to make sure that you do. That doesn't mean you need to get rid of all your dashes. Don't do that. dashes are great. But it does mean that if you're reading through an output and you notice that it uses the it's not this, it's that constructions three times in like two paragraphs, got to edit that out. The difficult thing there is LLMs and their existing knowledge sets are always

Julia Bye (27:58)

Mm-hmm.

Rebecca C (28:05)

So what is today's top AI red flag might not be such a big deal tomorrow. So I would recommend following some very cool people on LinkedIn. I'm one of them. Come find me. There are always people talking about, I've noticed that AI is doing this more. So you should do that yourself. As you're using AI, have your detective hat on and notice if it's doing something a lot and then make a note to yourself.

Julia Bye (28:19)

What?

Rebecca C (28:32)

to edit that out, but you can crowdsource that information too.

Julia Bye (28:35)

Yeah. And it's very true that, you know, LLMs are only as good as the information available to them. I think I saw an example the other day that was talking about, I'm going to screw this prompt up, but it was like, asking AI the shape of the earth, you will get both round and flat because there are, all of that exists on the internet. So like, that is like the perfect example of like, you need the human element to go in and say, Hey, actually,

We know that the earth is round, so...

Rebecca C (29:08)

Amazing example. That can be really funny. I know that when Google acquired Reddit or something like that, there were all of those viral posts about the AI summary at the top of the SERPs being like, yes, you should eat a rock, or, you know, yeah, make sure that you glue the cheese to your pizza or whatever. yeah, misinformation. It can be funny, but if...

Julia Bye (29:09)

Mm.

Rebecca C (29:30)

as far as AI ethics go, making sure that you definitely fact check. I think we're gonna talk about this next, if you fact check things and take ownership of your content. It's a thing.

Julia Bye (29:35)

Right. Yeah.

Yep, that's

exactly exactly the next question I had for you is how do you fact check and gut check your AI content before publishing?

Rebecca C (29:47)

So table stakes, click every link, every single link. you're working with...

chat gbt, think it will try to generate links, but there might be like a UTM code and it might lead you to a 404 or something that it's completely made up. So check that. think in Gemini, it'll more create a dummy link, which is a little bit less dangerous, but it will definitely be an AI red flag and won't do your content the favors that internal or external links usually would. So click every link. Ideally, you'll be doing your own research or using something like perplexity and then

your

own brain to put those links in yourself but if your AI tools do go ahead and give you links check those check those every time. Secondly I would advocate for if you have time please please please give yourself a little bit of a buffer between generating something with AI and then reviewing it. That will help your brain kind of have the cool down and give yourself the space you need to come back and see

if everything's resonating, anything feels a little bit off. Generally, this might be a little bit dated advice. We're still looking at this more, but AI...

may avoid making super declarative statements. So when it does, it's always a good idea, especially if you're writing legal content or anything that could be seen as something that somebody might be held to. Soften those, fact check those, talk to SMEs on your team. Which brings me to the third best practice.

SME to talk to. Have a secondary person look over your stuff. Professional editors are great. Shout out to my editor who's made such a difference in making sure that I'm not spreading misinformation and sounding like a human. So editors are worth their weight in gold But even if it's just another person at your company or if your client has an SME, get somebody else to give your content a second review to make sure that, again, nothing's standing

Julia Bye (31:35)

Mm-hmm.

Rebecca C (31:47)

out as weird, nothing standing out as potentially misleading or actionable or weird or things like

Julia Bye (31:54)

about plagiarism? Because I know that people are running AI through plagiarism checkers too, because AI doesn't necessarily have the filter to care about taking other people's words.

Rebecca C (32:04)

I mean, think plagiarism is part and parcel, kind of, of AI, which is why you gotta rewrite your outputs. You gotta fact check them.

Yeah, LLMs get their stuff from everything that is on the internet. So it's gonna be repurposed content and smorgasbord together assets. That doesn't have to be a bad thing for a lot of the content that we produce, especially if you're producing a lot of content and following an SEO strategy. It's gonna be the same building blocks of basic information. That's fine. That's kind of what we're doing here sometimes. But taking the extra time to get

an SME input, maybe put a specific SME quote in there, or say things like, our experience, or according to Julia, or Rebecca said this in your content. That helps avoid plagiarism, bad things. That also helps feed into your EEIT, makes you feel better about your content. Also, it can help you build in that sense of connection around the creation process that is sometimes maybe a little bit more left to the wayside.

when it's just you and a robot creating content. So that's kind of a win-win win.

Julia Bye (33:13)

Right.

So let's talk more about the ethical lines because plagiarism kind of falls into that bucket. What do content marketers need to be careful of not crossing when using AI tools?

Rebecca C (33:25)

So plagiarism and misinformation, yeah, you gotta take ownership of your content. And that is one ethical thing that...

is evolving as we all use AI more, but those are kind of things that have always been the big standards of our profession. If you are representing clients, if you are telling stories, you want to make sure that they're good stories, right? You want to make sure that you're not saying things that could harm or lead people astray. You want to make sure that you're not taking...

you're not taking ownership of somebody else's content, which again can feel dicey because we're all with LLMs drawing from the same pool. But I guess here that would be more being really proactive about putting your own stamp on things, in fact checking and gut checking, as we said before. I think the new thing that I'd want to bring up here.

content marketers, but I think this goes for anybody who could be listening to this, is that we have to be so careful to continue to center humans in this process. You said earlier that we're seeing a lot of messaging about how AI is not replacing people's jobs. It's just helping us do the parts of our jobs that we like better. And I think that has to be our clarion call here, otherwise we're not using AI in a way that is actually.

helpful, right? So I think that if you're in charge of an effort to use AI more at your organization, which I'm involved in that at Smartbug, we're all doing that to some degree, you need to make sure that humans feel as safe as is possible, as is possible, and that they have as much room to become familiar with this new tech as much as is possible, because otherwise we're not going to feel good about using it more.

just Maslow's hierarchy of needs, right? So center humans in the process, give us all space to feel safe about this. I in the past few years, my God, this has moved so fast.

Julia Bye (35:18)

so fast.

Rebecca C (35:18)

in the grand scheme

of things that like we're all, even if you're really excited about AI, you've still taken considerable psychic damage over all of this. So we need, again, we can move fast, but we can also make sure that our people feel good. We can give people time to heal even as they are onboarding.

Julia Bye (35:27)

Wait.

Rebecca C (35:38)

Like that's just a good people thing, but it's also a good business level. If you're taking care of people, the people are going to feel better about using AI and getting all of the benefits that come with that. So that's something that I'm prioritizing a lot as I'm helping our teams at Smartbug use AI more to accomplish our business goals. But I'm very cognizant that we have to have our human goals at the same time. So that's that. Please do that.

Julia Bye (36:05)

For sure. Yeah, I know. I think in the back of my mind when you're talking about that, I think about that MIIT study that came out that showed like what human brains were when using AI. And it's like, my God, I don't want to lose my brain functionality.

Rebecca C (36:18)

which is why, as you're thinking about which tasks.

can be automated or offloaded to AI in some capacity, you want to go for the things that nobody wants to do anyway. So when I started this AI project at Smart Bug where I was told to help people use it a bit more, I scheduled one-on-ones with as many people as I could in our company.

because I wanted to ask them, hi, how are you feeling about this? What's the thing that you like doing least? What could we take off your plate? And what's the part of your job that you would never want offloaded? And I'm really happy that at SmartBug, we're using that data to try and prioritize automating away the things that nobody wants to do and giving people more time in their day to focus on creative direction or content strategy or connecting with their teammates or the other things that we uniquely

can do.

Julia Bye (37:11)

Yeah,

definitely. All right. So before we wrap up, I have one more question for you. What are your favorite, if you have any, prompts, tools, workflows that help you get the most out of it?

Rebecca C (37:21)

Okay, so.

Second brain priming. If you're not doing this, I don't really care what you're using AI for, but this is just such a mental load alleviator. Start a new gem or custom GPT or just a new chat in chat, GPT, whatever. Give it a lot of the information and then say, hi, my name is Jeb, my name is Rebecca, here's my job title. I just need you as a sounding wall sometimes. This is generally what I'm trying to do. I don't want to...

As I'm saying this, it occurs to me that that should be something that your coworkers do. Your coworkers should be your sounding board. But it can kind of help to both. The humans you can reach out to in times of need, obviously, let's do that more. But as a second brain tool.

have the AI hold all of the databases of stuff. So you can go back in and say, hey, what were we trying to do here again? Or I feel weird about this piece of content. Can you tell me what's weird? Just second brain, second brain that. As far as tools, right now I use ChatTPT slash Gemini more and more for content generation. But perplexity for research, perplexity is very good at taking a very general question like, hi, I don't know anything about HVAC.

just keep using that. Can you give me something interesting that's happening in this industry? I have an SME with a client coming up soon. What's going on with this general business that I don't know anything about? And then I can go into the SME and see super smart and cool like I've done a research, which I have just, know, I did it quickly. So Gemini, ChadGBT, Perplexity, and then favorite prompt trick. I think I referenced this earlier, but...

If I'm in that first iterating phase, so I've done the general write me five LinkedIn posts or give me a first draft of a blog, my first iteration step will usually be high.

I'm talking to a real human here, ChatGBT. I would like this piece of content to be so emotionally resonant and so hilariously dialed into their life experience. Please open this blog with a funny rule of threes, a really detailed evocative example of what they're probably going through that led them to Google the search term to land on this piece of content. And sometimes it misses the mark, but sometimes it helps dig up something in my brain or give

gives

me something that I can put in the sidebar of a Google Doc to be like, hi, client. In my research, I found that HVAC clients are usually dealing with these three things. And again, then I come up as an informed and proactive person, which I am. But I've done so without having to dedicate five hours to arrive at that conclusion.

Julia Bye (39:44)

Thank

Right, right. So we just covered a ton of stuff. I learned a ton even beyond how I use it and have ideas too. So this has been super helpful even for me to be along for the ride. If there is like a takeaway from today, what would you say is like, this is the one thing that people need to remember from this conversation.

Rebecca C (40:15)

iterate more. So yeah, just never don't don't ever settle.

for the first thing that comes across as an AI output. If you were writing with a human writer, even the best human writer, when they get back to you with the first piece of content, you're probably going to have feedback for them because that's part of how creating good content works. You don't settle for the first thing. You go through a collaborative creative process together. And that's something that we should still do with humans. But I don't think that we should hold AI to a lesser standard, right? So once you get the first thing from AI, don't go, cool, this is AI.

It's okay if it's not great. No, be a little bit stubborn. Be a little bit picky. Go back and say, I don't love this. Help me make it better. I have high standards. You're going to meet them. It doesn't have to take long. And then you said one, but I'm going to say two. Please center your people. Please be kind to people. my God. Try and find some humor in this. Allow people to heal. It's the Wild West. We're all just trying to figure it out. So give yourself grace, give others grace, and follow people on LinkedIn who have a sense of humor about this kind of stuff, which

it's.

Julia Bye (41:16)

feel like that's all

general life things in general, like beyond AI. Be nice, kind, follow Rebecca on LinkedIn.

Rebecca C (41:21)

I mean...

Pretty much, yeah, those are my top three. What about you?

Julia Bye (41:25)

All right. So

where can listeners go to learn more about how SmartBug is using AI responsibly in our content workflows?

Rebecca C (41:34)

I mean, I would say watch this space. Wherever you're taking in this content, SmartBug on tap is an incredible resource for learning more about the things that marketers are trying to grapple with right now. And I am so beyond sure that there are going to be more episodes about AI and AI ethics in the future. So stay tuned for that. Follow SmartBug on LinkedIn. Same deal. We're always writing more and more about really cool things there. And then you can follow me. My name is Rebecca Corgan, R-E-B-E-C-C-A-C-O-R-G-A-N. It's somewhere.

I'm always ranting about this kind of stuff and experimenting with being funny on LinkedIn.

Julia Bye (42:05)

Awesome. Thank you so much. That was so much fun. Let's do it again.

Rebecca C (42:07)

Thank you. This has been a blast.