Pod Insights Voices of Search // A Search Engine Optimization (SEO) & Content Marketing Podcast Putting AI To Work For Specific Tasks About this episode Britney Muller, Founder at Britney Muller Inc., delves into how artificial intelligence can be effectively utilized for specific tasks. Show NotesConnect With: Britney Muller: LinkedInThe Voices of Search Podcast: Email // LinkedIn // TwitterBenjamin Shapiro: Website // LinkedIn // Twitter See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.
Transcript 0:00 Speaker 00 The Voices of Search Podcast is a proud member of the I Hear Everything Podcast Network. Looking to launch or scale your podcast? I Hear Everything delivers podcast production, growth, and monetization solutions that transform your words into profit. Ready to give your brand a voice? Then visit ihereeverything.com. Welcome to the Voices of Search Podcast, a member of the I Hear Everything Podcast Network. Ready to expedite your company's organic growth efforts? Sit back, relax, and get ready for your daily dose of search engine optimization wisdom. Here's today's host of the Voices of Search Podcast, Tyson Stockton.
0:43 Speaker 01 Hey, what's going on? My name's Tyson from previsible.io and joining me today is Brittany Mueller, who is the Data Scientist Advocate at Brittany Muller, Inc., which specializes in providing innovative digital marketing solutions achieved through a combination of cutting-edge technology and creative strategies. Today, Brittany and I are going to be discussing putting AI to work for specific tasks. So with that, here's my conversation with Brittany, Data Science Advocate at Brittany Muller, Inc. Brittany, welcome to the podcast.
1:14 Speaker 02 Woo-hoo. Thanks for having me. This was awesome.
1:17 Speaker 01 I was, I was looking forward to having you on here. I think as we were talking before the show, I have a shared interest in the education space. So I think it's really cool what you've been working on from kind of AI education and trailing and kind of that entire front. So thanks for joining us.
1:36 Speaker 02 Yeah, thanks for having me.
1:38 Speaker 01 And I mean, we're doing kind of today's conversation, tomorrow's as well, but today we wanted to jump into more of like different applications of AI. And I think most people are a lot of the narrative tends to drift towards content, generative AI. I have a suspicion that you're kind of thinking in a maybe a wider lens. Like, how would you set the stage for the listeners of like, what should our perspective be? Like, what's the breadth of options that we can really be working towards instead of just content?
2:18 Speaker 02 Yeah, I love this question because I think so often, like especially right now with just all of kind of the AI hype and things exploding around AI. We are inundated with AI messaging that it's going to make our work faster, more efficient and productive. All of these vague productivity terminology is typically used to describe AI today. And what's interesting is it's a lot like programming in terms of application.
2:56 Speaker 02 You have to get really, really task specific to have it add value to your day to day life. And so oftentimes, you know, people kind of overlook the task that might take them 20 minutes a day. But if you figure out a way to streamline and automate that using something like AI if applicable, you know, over the course of a year, you're saving over 86 hours of work, right? Over two full weeks of your time. So I think it's important to kind of create a fun playground to experiment with really, really task specific applications, right?
3:39 Speaker 02 And and to take a step back and identify what are the common tasks that you do each and every day? What does that look like? Right? What are you and your team working on? What sort of insights would be super, super valuable for, you know, maybe your project or this quarter's goal? How could you start to kind of uncover different insights that AI can support you with? So in my opinion, AI applications are everything from helping to clean up and organize your inbox to categorizing massive amounts of customer insights and reviews and conversations on platforms like Reddit.
4:24 Speaker 02 The ability to surface real time kind of Temperature checks on how people are perceiving your tool, your services, your brand, et cetera, are more attainable and valuable right now because it's more accessible than ever. You're able to use that and wield it in ways that we've never really had access to before. And then it's all the way to the research support side. There's um, my friend had helped found this company Monkey Learn, where they did sentiment analysis at scale. And to show the power of sentiment insights, they decided to analyze over a million hotel reviews from around the world
5:08 Speaker 02 within like the last however length of time using Google Hotel Reviews. And the insights that they uncovered were incredible, you know? Like, the biggest disappointment in Paris was croissants. Bangkok hotels have a cockroach problem. Um, Wi-Fi is a common issue in like Northern Europe. There were all of these interesting things that, you know, from a concierge perspective or someone doing marketing or SEO for these hotels, you can be like a quick differentiator in some of these areas and be able to communicate fixes a lot quicker using this real-time insights.
5:50 Speaker 02 And so it's really about getting specific on what it is you want to use AI for. And a big part of that, too, is understanding what it's qualified to do and what's not qualified to do. And so that's something that, you know, through a course that I teach, it's so fun to sort of break apart why AI isn't good at, you know, data-driven research in and of itself, right? It's not reliable. Whereas on the other hand, it's really great at summarizing massive amounts of text or querying against a big document, you can quickly find insights. And so there's different applications that, to be honest,
6:32 Speaker 02 I think you also only learn and get a true handle for through experiencing it. And so it's fun, I would challenge people listening to to play with the technology and just to sort of have fun using it in different ways and verify the outputs.
6:50 Speaker 01 On those last points, so it's like through some of the examples of the positive, like it fits with that like synthesizing, summarizing, condensing down large volumes of text. You made the comment of like, hey, from a data analysis, there's some kind of limitations or like risk to it. Can you kind of elaborate into that? Because I think I've heard from kind of like the general public too. It's like the assumption that AI is going to be amazing at processing or digesting massive amounts of information, regardless of its text or other forms of data.
7:29 Speaker 01 So can you kind of elaborate on the areas that may be like the yellow or red flag of, hey, you're going to want to express caution in using it in those ways?
7:40 Speaker 02 Yeah, definitely. I think it's also maybe helpful to sort of lay the foundation of how this technology works like at a real high level to understand where these errors come from, right? And so essentially, these large language models are processing essentially all of the internet. And what they're doing when you enter in a prompt is they are predicting sort of what comes next. They will generate the most average output of everything they've ever consumed on the topic of your prompt to provide value to you. But it's important to remember, it really is like a probability distribution behind the curtain.
8:25 Speaker 02 And so if it's seen, you know, Voices of Search podcast mentioned tens of thousands of times next to like world-class SEO, if someone prompts, you know, what's a world-class SEO podcast, Voices of Search would show up, right? It's a numbers game, it's a popularity contest. But because of that, we really miss outliers and we also miss real-time updates and information because these models take about a year to train. And so unless they are reliable. Connected to something like Google or Bing, which sometimes they are, but even those search results, right, are not always the most dependable.
9:10 Speaker 02 You sometimes won't get the most up to date information. And so it's like a great example is it's such a terrible calculator, right? It's not, it ha- doesn't have calculator capabilities yet. They will eventually stack that on to these L L Ms and make it look like it has that functionality. But at its core, it's a language model. It's not a numbers model. It literally just predicts the most probable next set of words. And so if you put in like some random math problem, it's never ever seen before. It's going to really struggle to come up with an answer. Um, similarly, if you, you know, give it an A C T question or something it's seen on a lot of
9:51 Speaker 02 practice, PDFs and websites and whatever, it'll just rattle off the answer because it's seen it so many times. And in that case, we refer to that as like a stochastic parrot. It's just parroting back information. But oftentimes that's not the most reliable, right? And it also can go a bit off the rails in terms of making things up. And so we see cases of that going wrong when people try to use it as like the de facto research tool. I remember early on, a bunch of S E Os were getting angry at me because I was saying, you know, don't use chat G P T for research, like do not use it for research. And all these people were, were adding me and screenshotting their keyword research with chat
10:35 Speaker 02 G P T. And I would just tell them, run it again, like run that same prompt again. And I never heard from any of those people because I know they ran it again. And the, the average monthly searches were probably drastically different, right? It's predictive. It's a predictive engine. And so it's so important to understand like, yeah, a predictive engine is great at explaining a complex topic. It's read a lot about. It's great at translating text into a different language because it's seen lots of that. It's not so great. Right, uh, surfacing real-time information or like academic resources.
11:16 Speaker 02 There's so many other tools that are useful for that. It's and it's also it's not a truth engine. Right. It has never been trained on what is true and what is fiction. And that's why we saw some of like the funny airbud examples like early on. People were asking, you know, has a dog ever played professional basketball? And it would say like, yes, yes, they have, because it it doesn't know, right? Like it is, it is one of the most unsophisticated, but like appearing so brilliant systems we've ever had access to. It's fa- it's absolutely fascinating. But yeah, I I and hopefully that sheds some light on, you know, why it's not qualified to do many things.
12:01 Speaker 02 And on the other hand, it is wonderful and an incredible tool um for other things. And I, the last thing I'll add to this is there's been tons of research on this topic of like how to improve outputs from language models. There's been all these academic papers, all of this work in the research community and at NERIPs and hackathons. The number one thing that has surfaced among all of this research and all of this effort to improve prompts is examples. Your output of LLMs will drastically improve and compound in the improvement.
12:42 Speaker 02 If you're able to provide real-time examples, examples of sort of what you want it to do, right? Here's um how this would look if X, Y, Z, here's this, you know, give it specific examples. It's also referred to as few-shot learning. Or you'll hear zero-shot learning and they're just these like fancy words for not providing any examples and providing a couple. It's so stupid. There's so many gate key keyword terms that the AI community uses is so unnecessary, but but yeah, just you use examples and like spoon-feed that information to an LLM so it can better serve you what you want essentially.
13:24 Speaker 01 I wanna, I wanna dive more into those kinda like best practices and how to serve like the examples in. But like, going back to like the scope, I feel like the predictive piece. I mean, that came through loud and clear. Like, so I think that's like one that I almost feel like people uh would like to look past.
13:47 Speaker 02 Yeah.
13:48 Speaker 01 Because it's like you want to think it has this like uh truthfulness or like authoritative. But it's like, I think recognizing that that is a limitation.
13:59 Speaker 02 Yeah.
13:59 Speaker 01 Which is for all rights, like the populace, like what is the most popular still has its own value. But knowing that that is the strength. I mean, it also is kind of interesting because I feel like it just really feeds into like, I don't know, public media, like
14:15 Speaker 02 Yes.
14:15 Speaker 01 interest, whatever, because it's like hitting on like the most common, so the most popular, so you're gonna get the most interest. But kinda like that factual outlier, like that I feel like is really valuable. Because if you take those principles of knowing like, these are strengths, these are fringe or kinda weaknesses. It could guide you into the task that you're applying it towards.
14:36 Speaker 02 Yeah.
14:37 Speaker 01 Coming back over though, so examples, giving those pieces for it, like what recommendations would you have for like the structure to give the examples? Like, are you thinking more in the terms of like, this is done, like this is what I'm looking you to achieve? Or is it kind of in a different standpoint?
14:59 Speaker 02 I mean, it's task dependent, right? So depending on what it is that you're trying to do, it's gonna look a little bit different. Um, I throw my students into this actionable AI exercise, uh, class one. And it's so, I just wanna go through the tabs with you because I think these are so specific to what people can use it for. So like, there's a, and with this exercise, we use G P T for sheets. And so it's pulling in, you know, either chat G P T, you can use Claude, you can use a bunch of different language models right within Google. Sheets to do these specific tasks. So the first tab is sentiment analysis and I already have pulled in all of these IMBD reviews
15:44 Speaker 02 for Roadhouse. And so essentially you would ask, you know, you would write the formula to observe the review within that review cell and then output the sentiment in like the next column, right? And so walking through that, there's also summarization of recent tech news articles that are quite lengthy. Um, there's categorization where I have all these product descriptions and you have to categorize categorize the products and these are at scale, right? There's quite a lot of them and it's fun to see to see individuals play around with this a little bit, play around with the prompt and sometimes less is more, right? It is really, really task specific.
16:29 Speaker 02 Sometimes you don't need to feed it a bunch of examples about product descriptions. It's already really good at that because it's seen so many, it's seen so many categorizations of products on e-commerce websites all over the internet. So it doesn't need a whole lot of guidance. And then there's other examples like this one, I always feel a little bit mean. It's called the research tab where I have a bunch of random elementary schools and I encourage students to enter in your elementary school in a column. And then see if you can get chat GPT to correctly predict or correctly output your elementary school mascot. It has never gotten it right.
17:12 Speaker 02 Like it's literally never worked and it's just a funny example of like, again, it's observed so many schools with this kind of name and it has no idea. It's just gonna confidently regurgitate something that it believes could be a mascot. And that's everything from, you know, customer service responses, which I think are really exciting because that takes such a mental, you know, it's mentally taxing for different um, support staff to handle various things. And if you can help at least get an outline. That's the other part of this is there should always be a human at the center. Right? You should never set and forget these workflows. They should just facilitate a framework or a first draft of sorts.
17:58 Speaker 02 Um, and that's everything from language translation to Excel formulas, Google Sheet formulas. I never want to look up another complex formula again for as long as I live. I just want to go to Claude and say, hey, you know, this is my, and I copy and paste a sample of the data set I'm working with. Hey, I want to transform this, right? I wanted to take this and help me remove, you know, the first part of the email and just get the, the websites of all of these contacts. That's sort of like, it's so quick and accurate at providing you with support like that. Um, and yeah, I could go on and on. There's just, there's tons of examples like that.
18:39 Speaker 02 And then it's also really fun to see people play around with Notebook LM. Everyone's heard about REG models, right? Retrieval Augmented Generation. It's a brilliant framework to provide more consistent and reliable outputs depending on like, let's say your HR handbook. Let's say you work at a huge tech company and you have a, you know, an HR internal uh, employee resource guide that's a hundred and fifty pages long. And you just want to know if there's like parking at this facility. And so you don't have to read through the whole thing anymore, right? You can have that uploaded to something like Notebook LM or AWS has a different,
19:24 Speaker 02 a couple of services that provide this. Google Cloud makes it really easy. You can upload these documents and then have like your own chat interface to query against those things. Notebook LM cracks me up because not only can you do that sort of thing, but you can upload entire transcriptions of classes or educational videos and generate your very own custom private podcast where two podcast hosts communicate about the content.
19:54 Speaker 02 And the information provided within whatever you uploaded. And it's so hilarious because they worked so hard on this feature. And the defining moment that made this whole thing explode and be successful was a secondary model that added human mouth noises. And like ums and ahs to make it sound real. Uh, no one wants to listen to a robot crap, you know, like we've all done that sucks. We don't wanna do that. And so it just, it adds like these creepy human noises and pauses and laughter. And it is wild. It's so interesting.
20:36 Speaker 01 Yeah. Meanwhile, uh, to Lisa, who's editing this is second guessing like which things to remove. That I'm making on that. Um, I, I mean, and I was like kinda smiling like during that because recently I went through our HR handbook, which are tedious, horribly annoying documents. Because it's like they're just, you try to like bring human elements into it. But then you're just confronted with legal language that is difficult.
21:08 Speaker 02 Yeah.
21:09 Speaker 01 To say, like to say nicely. Um, and I was literally just playing around with that of like, hey, maybe we could just put this like in our little like hub resources. And then that way if someone's like, I wanna know about like, uh, I don't know, like maternity policy. Like they could just chat it in instead of having to read through a bunch of dense, annoying legal language.
21:31 Speaker 02 Yeah.
21:31 Speaker 01 When it's like, really, they're just knowing how many weeks or how many months am I gonna have off here?
21:36 Speaker 02 Yes, exactly, exactly. And it's so funny too, because I'm such a fan of just like exploring the new Google AI products anyways. Because you see all this crossover, like in Notebook LM, if you did something like that, it cites the source. It's, it would cite how many, it would reference or say however many weeks maternity leave is. And then it would provide a link in a reference to where it got that from. And we see that in AI overviews. It's like you see all these crossovers or experimentation over here that maybe will, will eventually show up over here. And so, yeah, it's very, very interesting.
22:14 Speaker 01 Which is, I feel like there's two directions that I want to head in the conversation. There's like, one, I think like some underlining elements that like we keep kind of like touching towards, and then a little bit is kind of like more like the forward thinking piece. But I feel like being that there is such like, I mean arguably speaking, like infinite applications to this, it seems like the common elements that you've been hitting on is like the strengths of the actual systems and hey, like it's predictive, it's language, like there are these elements or characteristics that it's going to be just more suited towards.
22:58 Speaker 01 And then there's a little bit too of like, what is the data set, like what is the training set that's being applied to it? Because there's the generic, but then we're talking about like notebook elements where you're like kind of confining it down to an area. Is there anything else that you would say is like more of these underlining principles that could guide someone to then maybe think a little bit more creative on applications that are going to be well suited for these systems?
23:28 Speaker 02 Yeah, it's funny because I think oftentimes people newest to this technology have the best ideas. And what I mean by that too is like, we always talk about this bottleneck in AI. And a decade ago, we talked about data and training data sets and computational resources being bottlenecks. We no longer have that problem, right? The bottleneck today and moving forward is ideas. It's people with real world experience and ideas to apply this technology in new and interesting ways. And the easiest way to start to just frame some of this up, I actually have a framework, I should just
24:13 Speaker 02 publish that somewhere on my site that we can maybe share with your listeners because it's pretty straightforward. But the big takeaway is like, write out, if you have a lot of List of task ideas, evaluate them and evaluate them by risk. Right? Would something really bad happen if you got this wrong? Maybe leave that to human, human work, right? That shouldn't be maybe automated just yet, right? So first we evaluate risk. Secondly, we, I have students kind of reorder them by impact. So which one of these will provide the most value to your work, to your life, if you had support or assistance with this task?
24:57 Speaker 02 And then I have them write out the task as if someone was in another room and they had to pass a sheet of paper to a person in another room with really clear, specific steps, detailing exactly what each part of the, the workflow is. And what's interesting about breaking it up like this is you'll often discover that the AI part of it is just a teeny tiny role somewhere down that line of items. And that the first, you know, several items is usually around the, the data acquisition. So maybe you wanna identify leads on LinkedIn
25:41 Speaker 02 that are interacting with these kinds of posts. That part of the process is a heuristic model. That's a rule-based system. That is use these keywords to identify trending or popular posts on LinkedIn. And then use a tool like Phantom Buster and Zapier to basically crawl, you know, those interactions and those names of people that have commented or liked, whatever. And then we would use AI to maybe, you know, draft up some message or based on what they said or how they interacted. And so it's, it's interesting, like I see it in so many applications. And at the, at the end of the workflow, it looks like a magic trick
26:24 Speaker 02 with a teeny tiny bit of AI, but AI was just a small, small part. And then there, there should also always be like, we refer to this as a human in the loop. Right? It's again, it's never just like a set and forget it. The the people that are doing that will fail. It's inauthentic. It's gonna make embarrassing mistakes for you. That's not how it should be used. There there should be someone in the loop that is authenticating, verifying, making things more tailored, making things more personalized, more authentic, in a way that makes sense for what you're trying to do, whether that's outreach or evaluating potential clients or whatever. You can do all sorts of things, but
27:07 Speaker 02 it's oftentimes just a small piece, but breaking down tasks, getting really, really specific. And oftentimes I will see like the first go at like breaking down a task. There's like one or two items that need to be split out several more times, right? So that's typically how I would encourage people to go about that.
27:25 Speaker 01 I feel like to that last point is super valuable because basically with the current state of the technology, it has this strong ability in like that specific use case where we're not, like you're not advocating of, hey, apply AI to an entire workflow.
27:45 Speaker 02 No.
27:46 Speaker 01 It's a very specific touch point, like linchpin, so to speak, of the process, where it's not this, I don't know, apocalyptic, AI is taking all our jobs, but it's like, it seems like it still keeps coming back to like one of the most human elements, which is creativity. And it's like whatever limitation we have in AI is stems from what is our limitation in human creativity.
28:15 Speaker 02 Yes. Bingo, bingo. And this is why AI agents are bullshit. I call bullshit. I just do. Like if you're touting like AI agents or using that, it's this interesting show of hand that like you're not clear on your thought process. You're not clear on your value adds. It's such a vague, undescriptive word that is so incredibly general and just sort of the hot new trending marketing term. It drives me absolutely insane. And and you also don't want agents to complete that cycle. Do you know what? Like there's so many, I have so many issues with AI agents. It's insane. Yeah.
29:01 Speaker 01 Which, I mean, when I'm laughing, because it's like, yeah, a hundred percent, but it's like, still, I feel like the majority is just, you throw AI in anything, like, marketing, investor, whatever, and it's like, great. But it feels like, and I think you're alluding to too, is like, agents is going to become that new buzzword that is like, maybe it's more, more present in your day to day than like the general public, but I have also started to see that same kind of trend starting.
29:33 Speaker 02 It's crazy.
29:34 Speaker 01 Because obviously, it's, it's good and bad thanks to that. It's like, it's great in the sense of it's increasing general public awareness, demand, interest, but it very much feels like kind of the, I don't know, greenwashing of like the early 2000s, like this is sustainable.
29:51 Speaker 02 Yeah, totally.
29:53 Speaker 01 Environmentally friendly, et cetera.
29:54 Speaker 02 Yeah, exactly.
29:56 Speaker 01 Maybe to kind of land the plane, so to speak, on the conversation, I'm assuming it's not going to be agents, but looking ahead, what excites you the most? What interests you the most? Like, what's the more emerging area that you're like, that is where I really want to pay attention?
30:18 Speaker 02 Yeah, and maybe it's important too to clarify, like, don't get me wrong, AI is a powerful technology and I have it doing incredible automated things for me. I just don't refer to it as agents, because it actually, the different workflows I've applied it to have specific names. They have specific, you know, task-specific efforts, I guess. And so, yeah, I just think we need to be, especially with the field of AI, the number one issue in the world of AI, as we talk about AI, is the lack of clear definitions. What is intelligence? Right? How will we know when we've reached this?
30:58 Speaker 02 And so something I challenge, like, new people to the space, marketers, try to be really thoughtful about how you're speaking about different things and be clear on what. It is that you mean what you're intending because it it could come across very differently. And again, this agentic AI, it just, it doesn't communicate anything of value. Um, and so I would just encourage people to get clear about that. So I'm curious, as I explain the, you know, how to identify AI tasks and break them up and figure out, you know, in what areas can AI support different things?
31:41 Speaker 02 A, did that make sense to you? But also, B, what are, I'm dying to know, like, what are the things you're using AI for on a day to day basis?
31:50 Speaker 01 Okay, so first, a hundred percent makes sense. Like, I think it's also just a a healthy refresher to to like ground people because it's like that predictive element in language is so critical in how you would look to apply in it. Because I think if you're not super, I don't know, in the weeds or like thinking about these systems, you're gonna get very kind of distracted by the, this language sounds so great, therefore it must be able to do everything.
32:23 Speaker 02 Yes.
32:23 Speaker 01 And I think like that's the piece that could potentially be dangerous. I mean, who knows? We'll see. But I think like that's the easily, like the slippery slope that it's very easy to fall into. So hearing those pieces, for me, it's more of a reinforcing, like reestablishing the strengths of the existing systems to then kind of, I guess, give me more thought or creativity of where I can apply it.
32:52 Speaker 02 Yeah.
32:53 Speaker 01 For where I'm applying it, I think it's a little bit, it's probably quite a bit different than most of our listeners because there's like, like the day to day of SEO execution, I'm not really involved in anymore. So it's like, I'm not working hands on with many of our clients. Like, there's a small handful. And I'm spending more of my time on like the recruiting and training, whatever kind of backend office of the business.
33:22 Speaker 01 And I think for recruiting, it's definitely helped in a sense of communication, also in the sense of like, looking at people that have applied and kind of getting a better sense of like, the overall impact of it. We have applied it in some very specific use cases within an SEO job study that we're in the middle of conducting. So we collected basically every SEO job that was posted in 2024, and then scraped the job description of it, and then using LLMs to kind of pull out some, I don't know, common skills and sentiment. And it's like,
34:06 Speaker 01 most SEO job descriptions aren't going to say like stakeholder management.
34:10 Speaker 02 Yeah.
34:11 Speaker 01 But they're going to say something that refers to that. And so in the like, the clustering of skills to make something more, I guess, helpful for an SEO that wants to progress their career, I think it's been really helpful. So I think like, those would be probably more of the common top of mind areas.
34:29 Speaker 02 Yeah.
34:30 Speaker 01 But I think it's very different than what the common day to day of like, you know, an SEO manager or someone that's more of executing on SEO work versus me kind of like talking and working around SEO work.
34:44 Speaker 02 Yeah, for sure. And do you program at all, or have you ever explored getting into some of that or not really?
34:51 Speaker 01 Embarrassingly, very light. Like in some regards, I think I'm more aspiring, but have not put in the hours to be very confident or proficient in it. So it's like, I've scratched the surface and that certainly is of interest to me because it's like, great, now the barrier is coming down a lot.
35:11 Speaker 02 Exactly. That's awesome.
35:13 Speaker 01 But definitely not to the level that I feel like I should be at or I would like to be at.
35:19 Speaker 02 I feel like we're so hard on ourselves, especially if we're technical SEOs or of like any technical SEO background. We're like, oh shit, we should know JavaScript or this or Python better. And honestly, like, that's not part of our job description, you know? Like, where are you supposed to be spending the time and resources to do that? It's, it's difficult for most people to acquire some of these skills, but I will say you're exactly right. It's never been easier to get into it, right? Like Claude's Sonnet 3.5 model, and now it came out with 3.7, is far above and beyond the most helpful code assistant available on the market today.
36:03 Speaker 02 Its ability, you could go to it right now and have it build a Pomodoro timer in a theme that you like or care about, and it can just do it, you know? And it can help you learn how to go through some of those things, or how to teach yourself some things. And it's just, it's, it's so fun. It's, it's made programming so much more fun, especially with cursor. It's like a video game, because you're communicating to it in English. You know, like, hey, take this tool I just, you know, spun up and make it look like a retro video game, right? Make it look like this. And it does it beautifully.
36:44 Speaker 02 And so people are, I just heard of this today from a mentor of mine, but people are referring to vibe coding, like the vibes. And they'll just like, they just play with it. They just have fun. And they're so bizarre. Like, what world are we living in?
37:00 Speaker 01 Well, okay, so, so with that, then it comes back down, I guess, going back for previously, like, too hard on ourselves. It's like, yeah, I think that's, I guess, comes with the territory of perfectionism and that kind of
37:11 Speaker 02 Yes.
37:12 Speaker 01 part. But as far as like the last piece on how, how far it's come and into that like area of focus, it like feels then like we're going back into the conversation around like leveraging and leaning into human creativity. And so
37:30 Speaker 01 As maybe, I don't know, I guess this is a kind of heavy last question, but like, so with that, we know the hardest thing or one of the areas is the piece of human creativity. What advice or recommendations would you have to people, not necessarily of like, encouraging them that they should use these systems or tools, but more of how, how can they flex and lean on that human creativity and kind of like strengthen that, which would then create more opportunities of how to use this technology?
38:03 Speaker 02 Yeah, I think today and moving forward, we will all continue to think more about what makes us uniquely human and what unique aspects do we bring to the table on a regular basis? And we have a lot to appreciate about our human experience and a lot to offer. Uh, you know, AI has no friends. It has no real life experience. It has no strong, unique opinions about things. And we see this actually in search results because Google has tried and is having to find a way to supplement what AI completely lacks.
38:48 Speaker 02 And that is that real world experience that we all share and have and connect with each other on. And because AI doesn't have any ground truth, it has no, no grounding in the real world. Now we're seeing the rise of all the forum sites of Reddit and Quora and different because it supplements exactly what AI will never provide, right? And it's those unique shared experiences, it's those stories, it's anecdotes, it's quotes. And so I think especially us as SEOs, like I really highlight the aspect of double down on those things in your work, right?
39:29 Speaker 02 Double down on authentic authorship and unique storytelling and unique writing that isn't AI slop, right?
39:40 Speaker 02 And I think about the marketers and those SEOs that take risks will stand apart. Right. I think about Duolingo all the time, like AI can't write that. Right. The social media account that's doing really funny things or um. And I also, so I have this kind of hot take that, you know, real authentic photographs are, you know, are going to be more popular over time with like all this AI crap. And same with writing, like, we know when we see real, unique, powerful writing, right? Like, I think of Ed Zitron that has been writing the pieces about Google.
40:21 Speaker 02 Or um, you know, the blind five year old dot com, right? Like, he writes incredible things and you can hear their voices so clearly and distinctly. And that's what I think we we're we will crave because more and more people will gravitate towards automating things and AI slop. I swear to God, if I see delve in one more email subject, I'm going to lose, like, I might just create a filter to throw all of them in the trash. Like, if it has the word delve, throw it in the trash. Because I'm, you know, we, I think, over time, we're going to just want more and more authentic communication and marketing. And I think old school marketing is going to make its way back around.
41:04 Speaker 02 Like, I'm, I'm hoping to see like old school, like guerrilla marketing come back, like more unique things that AI cannot do.
41:13 Speaker 01 I could not agree more. Like, it feels like in this weird way, it's like a pendulum swinging back to like older school tactics and I mean, I guess not tactics because it's not being done in the same way, but it's like old school principles.
41:30 Speaker 02 Yeah.
41:31 Speaker 01 Where it's like, I feel like more and more things are coming back around to like this digital PR and kind of branding elements and less into, I don't know, the the cheaper tricks maybe that we've gotten away with for a period of time.
41:48 Speaker 02 Totally.
41:49 Speaker 01 But it's kind of fun, too, in the sense that you have that, like, slight deja vu of like, I've seen this before, but then it's like that additional opportunity to like layer in. But I think that's like the big, I don't know, there's, there's like a human catch in there of like the interest towards authenticity.
42:10 Speaker 02 Yeah.
42:10 Speaker 01 And regardless if it's like a hundred percent true or not, I feel like there's always been that appetite and you see it through, I don't know, whether it's like the rise of like social media or reality TV or like these things that are trying to show a open window or a genuine view.
42:30 Speaker 02 Yeah.
42:30 Speaker 01 It's like people gravitate towards that. And so if that's the area that we're limiting here, that's essentially going to be the area that then we have to address or lean into.
42:41 Speaker 02 Exactly. And it's a little bit of a dance, right? Because we should be doubling down on that aspect of our authenticity. And, you know, flexing our creative muscles and coming up with interesting things for customers and social and whatever. And on the other hand, we unfortunately, you know, this genie is not going back in the bottle. And so we have to play the game in a way that increases our odds of showing up in AI outputs. And that is why I continue to say across all these platforms, you know, mentions are the new backlinks. It's a popularity contest, right? So back to that.
43:23 Speaker 02 If voices of search podcasts occurs in all of these places next to these keywords, like, and if someone prompts using those words, your probability of showing up increases tenfold. Um, and we need to think about that as marketers, right? How can we, you know, encourage others to share our messaging, right? How could, it's just, it's like same old school link building stuff, but really through like creative brand mention. It's so funny, like I think a lot of, you know, Garrett French, like OG link builders, and they've always had creative solutions to this. And going back to some of their incredible resources and
44:05 Speaker 02 using that for, you know, brand mentions, consistent brand mentions with specific references to what it is that you want to show up or care about, et cetera, that will improve your odds.
44:18 Speaker 01 I'm I'm on the same page. Like, I think we're probably not gonna call it the same, but it's like we're swinging back towards backlinking and we're gonna call it digital PR.
44:29 Speaker 02 Yeah, digital PR, a hundred percent.
44:33 Speaker 01 Like, conceptually, it's the same thing. It's like now we're just, we're probably gonna call it something different, like, cause it's like backlinking has its like stigma and whatever impression in the industry.
44:45 Speaker 02 Yeah.
44:46 Speaker 01 But it's like, it really feels like that's where we're heading back towards. And maybe, like, not necessarily in a bad way, but just in a like, yeah, that's gonna be, like, it's gonna be less of optimizations only on your website. Like, that's gonna be like your table stakes. And then it's how much can you increase your mentions, link, just social channels, doesn't matter, like, your collective online presence to show up more and more in the systems.
45:13 Speaker 02 Yeah, exactly. Do you think, uh, mention juice will take off?
45:19 Speaker 01 Um, hopefully in that, in that specific naming, no.
45:24 Speaker 02 Okay, good, yeah.
45:25 Speaker 01 Just cause if I had to talk about that, in the same sense, like, it would be difficult.
45:33 Speaker 01 But like, I guess, conceptually, like, in some ways, yes. Like, I think, yeah, it's like, there's gonna be a volume play, there's gonna be like an authority play to it. Like, I'm leaning, I'm trying to like lean towards, I want people to just, like, let's just use digital PR. Like, that's, it's simple, it's basic.
45:55 Speaker 02 Yeah.
45:55 Speaker 01 It's encompassing. But yeah, that's, hopefully it's not mention juice.
46:00 Speaker 02 Yeah, fingers crossed.
46:02 Speaker 01 Well, um, I've kept you, uh, far beyond our scheduled time, so greatly appreciate it. But with that, that's gonna wrap up this episode of the Voice of Search Podcast. Thanks to Brittany Mueller from Brittany Muller, Inc. for joining us. And in part two, which we're actually not gonna publish tomorrow, but we're gonna give a little more time to give it a full kind of attention and topic to it. But Brittany, are you gonna? Jump back on and we're going to talk about ethical AI considerations. And if you can't wait until then, you'd like to learn more about Brittany, you can find a link to her LinkedIn profile in the show notes, or go on over and check out her company's website at data sci 101.com.
46:45 Speaker 01 Also, Brittany has been nice enough to gather a customized discount code for her actual AI course for our listeners, which will also be in the show notes. So be sure to check that out and take a look at her course.
47:01 Speaker 03 Okay, thanks to Tyson Stockton, our guest host. If you'd like to get in touch with Tyson, you can find a link to his LinkedIn profile in our show notes. You can contact him on Twitter where his handle is Tyson underscore Stockton. Or if your team is interested in SEO consulting or organizational education, you can always head to their company's website, which is previsible.io. That's P R E V I S I B L E dot I O. Just one more link in our show notes I'd like to tell you about. If you didn't have a chance to take notes while you were listening to this podcast, head over to voices of search.com where we have summaries of all of our episodes and contact information for our guests. You can also subscribe to our weekly newsletter. And you can even send us your topic suggestions or your marketing questions, which we'll answer live on our show. Of course, you can always reach out on social media.
47:45 Speaker 03 Our handle is voices of search on LinkedIn, Twitter, Instagram, Facebook, or you can contact me directly. My handle is Ben J. Shap, B E N J S H A P. And if you haven't subscribed yet and you want a daily stream of SEO and content marketing insights in your podcast feed, we're going to publish an episode every day during the work week. So hit that subscribe button in your podcast app and we'll be back in your feed tomorrow morning. All right, that's it for today. But until next time, remember the answers are always in the data.
Previous SEO Tactics That I see Driving Growth & Results Next Human-Driven, AI-assisted SEO Page updated November 15, 2026 · transcript and summary are machine-generated