
On the Brink with Andi Simon · 2026-06-26 · 38 min
This episode tackles the gap between AI hype and responsible implementation, featuring two practitioners with front-row seats to both success and failure. Shelley Tench (Shelton, a digital marketing agency) explains how generative AI democratizes content creation for small businesses - enabling affordable blog writing and custom GPT strategy building - but warns that amateur users who skip human oversight create disasters (she cites a heavy machinery company that accidentally generated carnival imagery for their job site photos). Heather Owen (Genuine Stewards) drills into the technical truth: generative AI is probabilistic, not deterministic, meaning it completes patterns rather than retrieving facts, so hallucinations aren't bugs but features. She contrasts Deloitte's expensive failure (AI-riddled government reports costing hundreds of thousands to refund) with IKEA's success (reskilling 8,500 customer service workers and unlocking $1.2 billion in new interior design revenue). The episode's real alarm centers on security and compliance: financial services panelists confidently claiming they've stopped hallucinating and encouraging firing marketing teams represent dangerous misinformation, especially given that unencrypted public-facing AI tools expose proprietary data to thousands of daily bot attacks. The hosts argue the issue isn't whether to use AI, but whether leaders understand what they're actually wielding before deploying it at scale.
Generative AI (launched publicly with ChatGPT in late 2022) works through probabilistic systems running statistical probability equations on massive datasets to predict the most likely next words, unlike deterministic systems (Amazon recommendations, Netflix feeds) where one input reliably produces one output. It completes patterns rather than retrieves facts, which is why hallucinations occur - it's doing its designed job, not making mistakes.
No - public-facing AI tools are targets for thousands of daily bot attacks by cybercriminals, and inputting personally identifiable information, client data, or confidential information exposes that data to vulnerability and potential breach. Only enterprise-level AI tools with dedicated security staff and encryption provide real protection.
AI can automate specific tasks (blog writing, copyediting, initial customer inquiries) but companies that simply lay off workers without reskilling them face failures; IKEA's success came from reskilling 8,500 customer service workers in AI-assisted interior design, unlocking $1.2 billion in new revenue, while Deloitte's cost-cutting approach led to inaccurate government reports and million-dollar refunds.
Misunderstanding of the technology - generative AI by design prioritizes pattern completion over accuracy, so hallucinations are inevitable and constant; claims that the problem is solved represent dangerous misinformation that leads to unverified outputs being deployed in high-stakes contexts like government reports and marketing materials.
Understand the specific tool's architecture and limitations, develop clear prompts built around brand voice and compliance needs, require human verification of all outputs before publication or use, avoid plugging in sensitive data without enterprise-level security, and consider reskilling employees to work alongside AI rather than replacing them outright.
Computed from the transcript - who did the talking, and the words that came up most.
Artificial intelligence is changing business faster than most leaders realize. But are we adopting it wisely? In this episode of On the Brink , Andi Simon welcomes marketing strategist Shelly Tench and AI educator Heather Owen for a practical conversation about the opportunities - and risks - of generative AI. They explore how AI can help small businesses compete, improve marketing, and increase productivity, while explaining why human oversight remains essential. If you've wondered whether AI will replace people, how much you can trust ChatGPT, or how to avoid costly mistakes, this episode offers thoughtful guidance without the hype. Learn how to embrace AI as a powerful tool while keeping human judgment, creativity, and connection at the center of your business. About the Host Dr. Andi Simon is a corporate anthropologist and founder of Simon Associates Management Consultants. Through her consulting, speaking, writing, and On the Brink podcast, she helps leaders and organizations see opportunities others miss and navigate change with confidence. SimonAssociates.net Author of Rethink Retirement , Women Mean Business , and On the Brink ️
Transcribed and scored by The B2B Podcast Index.
Speaker A: Shipping, billing, admin, payroll, marketing. You're managing all the things, so why waste time sending important documents the old fashioned way? Mail and ship when you want, how you want. With stamps.com print postage on demand 24, 7 and schedule pickups from your office or home. Save up to 90% with automated rate shopping. That's why over 1 million small businesses trust stamps.com go to stamps.com and use code podcast to try stamps.com risk free for 60 days.
Speaker B: Welcome to on, um, the Brink with Andy Simon. It's always a pleasure to greet you today, like all of our podcast days, because I'm so happy to bring to you wonderful people who are going to help you do like an anthropologist, see, feel and think in new ways to adapt to changing times. We call this on the brink because I want to get you off the brink. And that means you have to begin to think about things differently. The tendency for people, a little anthropology is that you take what you know, that story in your head and try to fit the new stuff into it as if it's simply a replacement part in some fashion. Remember, the stories guide our daily life and we believe it to be true. But the only truth is there's no truth. And so today we have two wonderful women who are going to help understand what's going on in the world of AI which is embracing everything, touching everything that you're touching and, and you're not quite sure what do I do with this and how do I make it more efficient for me and is there something here I should worry about? And this all came about after a meeting where a panel of people were talking about using AI for legal issues, for managing their business, for doing marketing. And we said, oh, lots of, uh, m expectations that are going to lead to disappointment. So who do I have today? Let me start here. I have Shelley Tench. Shelly is an entrepreneur, a speaker and a marketing strategist who does amazing work. She's a founder and CEO of uh, Shelley 10, a digital marketing and brand awareness agency dedicated to elevating brands through authentic connection and strategic visibility. She is truly committed to people over profits. She has a strong sense of purpose. It isn't just about the bottom line, it's about how we get things done. And she spent more than a decade helping entrepreneurs, nonprofits and business leaders grow their impact through international, uh, marketing. She has lots more to her story, but I'm going to let her tell you a little bit more about her journey. With her is Heather Owen. Heather is the founder of Genuine Stewards, where she Helps entrepreneurs and business owners adopt AI thoughtfully, not just quickly. Now, the word thoughtful is a very important one because it's still filtering through their thoughts. But now we're going to really pay attention to it. When AI tools started becoming accessible to everyday business owners, Heather went all in. She taught herself rigorously, tested everything, and realized very quickly most people are rushing into adoption. Didn't have any idea what they were doing, she wrote, anyone helping them build a real foundation. That's the gap where she does her best work. Now, I must tell you, we all agree that AI is extraordinarily powerful. It amplifies creativity, opens new revenue streams, can really help small teams do things that used to require much larger ones. Makes some people completely eliminated. My copywriter, for example, called one day and said, I think I've been eliminated. And I said, yes, thank you. That was the last time we worked together because I didn't need her anymore. And it's not because I didn't like her work. It's because there are lots of ways of getting a really good copy written today. Not always perfect, but better. So, my friends, let's start. Shelly, let's talk about your own journey. Then I'll have Heather's, and then we'll get to the what's happening? How are we seeing and feeling this? And then why is this so important for us? Shelly, please, may I start with you?
Speaker C: I grew up. The name of my company is actually Shelton. It's Shelly Titch put together.
Speaker B: Ah. Um, my apologies, Shelton.
Speaker C: It's okay.
Speaker B: Thank you.
Speaker C: It's a throwback to my dad, because my dad is the one that named me and my dad is the one that taught me my work ethic. And so I named my business after my. In honor of my dad. I grew up in Chattanooga, Tennessee, in a religious cult, actually. And I was not supposed to go to college. I was supposed to marry a preacher and be a. A pastor's wife. And anybody that knows me now thinks that's funny because. No. And I got married young, had kids, got divorced, and then at 40, decided to go back to Colle College. And that was a journey by itself. I ended up going to community college first, and then a smaller college in Georgia. Uh, and that's where I discovered that all of the things that they had told me as a kid weren't true. Like you weren't good at math. That wasn't true. You can't accomplish these things. That's absolutely not true. And I ended up transferring to NC State. And while I was a student at NC State. In my mid-40s, I. I got asked by NC State to study abroad at Oxford. So I went to Oxford in the summer of 2009. And while I was over there living my absolute best life, I discovered social media being used in a PR capacity. Now, social media was a buzzword in the United States at that time, much like AI is a buzzword now. And they were very much using it in a more evolved manner than we were. So I came back and said, hey, we should start doing social media. And my professors looked at me like I was crazy. They were like, we don't. Because they were still teaching traditional pr, which is what my degree is in. And so I threw myself headlong into learning, uh, very much kind of like what Heather and I have done with AI. There was somebody to teach, so I threw myself into it. I ended up starting a company doing social media for small business owners in the Raleigh Durham area. And it has evolved from just creating Facebook posts to now. We come in, create strategies to raise your brand awareness. We use online and offline strategies. I always go in and ask my clients, what is the thing you would love to accomplish but you're scared to say out loud? And they sell me. Then I'm like, all right, let's figure out a plan to that. Because even if you don't accomplish that, you'll accomplish way more than you think you will. Right? I don't, I just can't. I don't believe in thinking small. So that in a nutshell is kind of where we are right now. I live in Washington state and I'm relocated out here after spending 13 years in Raleigh because, you know, Covid happened and my kids are here and my family's here and so. But I travel all over and so, yeah, I have clients all over the nation.
Speaker B: And so Shelley is our dreamer very much. I love it. I think big and let's make it happen. I love it. And Heather, how about your story?
Speaker A: So I'm going to stick a little bit more just to the professional story for myself and that's that I've. I've been in the digital marketing space since 2016 and there's been a lot of evolutions of my work in that space in that time, but I ended up working on a lot of high profile teams on the back end. I am very much a systems thinker and I am very much a big picture person, but I also immediately see all of the details that have to happen and fall into place to achieve that big picture. So I have always really Excelled in back end operations and don't always really like to be the front facing person. It's not as much fun. I, uh, don't love the content treadmill, that kind of thing. So in that back end work with some of these higher profile people in the digital marketing space, AI, uh, of course started to come up over and over and I started to see it being used and being taught in certain ways and I knew I just couldn't avoid it any longer. And so I'm the type of person, if I, I want to learn a thing, I want to learn it deeply and really well and then turn around and teach it to other people. It's just kind of the way I'm wired. And so that's what I did with AI. And, and uh, as you said at the beginning, you know, AI is such an amplifier of so many good things, but unfortunately it's also an amplifier of risk and exposure if we're not using it in the right way. And so I looked around at what was being taught in these popular spaces and I was horrified because I knew people didn't fully understand the tool that they were wielding. And we've already started to kind of see the fallout from that happening and rolling out in public. And so, you know, there's this saying in the army that slow is smooth and smooth is fast. And so that's my message for anybody who is facing AI integration. Slow is smooth and smooth is fast. It's very true. We want to understand that tool before we deploy it.
Speaker B: So let's stay. Shelly, I'm going to stay with Heather for a moment. What is this thing called AI that we're talking about? Because I would be helpful if we sort of got more clarity about what exactly you're thinking about.
Speaker A: Absolutely, absolutely. So when I say AI in this context, I'm actually talking about generative AI. You know, AI has been around for a long time in giving us Amazon recommendations, Netflix Watch recommendations, that kind of thing, feeding us the right ads we want to see. So it's not that it's new, but generative AI is much newer. We didn't have generative AI until about 2022, at least not public facing use tools. And so when ChatGPT launched in late 2022, that, that was the first of the public facing AI tools. So generative AI is. I'm going to get a little technical, but I'm going to try to keep it, I'm going to try to keep it friendly. We've always worked in technology up to this point with what we call deterministic systems. Right. You put in one thing, you get out and output, and you can be fairly sure and certain that it's going to be accurate.
Speaker B: Right.
Speaker A: Uh, very controlled parameters. Generative AI is different because what it's doing, even though when you're in a conversation with ChatGPT or Quad or any other large language model, that's what those are called, it seems kind of like it understands you, like it knows you, like it's giving you insight. But what it's really doing is it's completing a. Because these are actually probabilistic systems. And every time we go in and type a prompt into that generative AI tool, it's running a very long statistical probability equation to bring us back the words that are most likely to fit what we have asked it for based on human language patterns. Because it's just working with these huge data sets that our brains couldn't hold by themselves, but they can hold and parse that data and come back with those patterns. So when AI makes a mistake, or has a hallucination, as we call it, it's because it's doing its job, the job it was designed to do. It's there to complete a pattern. And when you understand that foundational aspect of generative AI, it helps you understand what's happening so much better, and it begins to help you work with it better, to leverage it for better results.
Speaker B: So let me swing over to Shelley with that in mind. Now that we have this new stuff, new tool, right? What are all the wonderful things it's doing and all the challenges that it's posing? How is it becoming a concern for you as well as an opportunity, share with the listener or the viewer. What do you see?
Speaker C: So the beauty of AI is that it tends to level the playing field quite a bit for small business owners. They can go in, they can create things that they don't have to pay huge teams necessarily to create, especially people that don't have huge marketing budgets. Right? We can create huge campaigns now by using specific tools, which you can't do if you don't have any training, is go in, plug in random ideas to AI without having a human that knows what they're looking at, look at it and then come back. So we go in and we'll, for instance, plug in our clients, brand, voice, their values, their missions, whatever it is that they're trying to get across. Right? And we have specific goals in mind when we plug it in. When we build these prompts, we also build custom GPTs for each client that has all these tools. And so we go in and we build these, and then we ask it for a strategy. We still tweak it. It still gives us the wrong information. It does it because it's a tool, right? And so we go in, but it's a great. It gives you a great icon. It like a skeleton to work from. It comes out with a strategy that you can then build on. Now you can go deeper, you can take that strategy, you can break it apart, you can plug each piece in. And I've seen it create amazing things and come out with amazing ideas. The other part of it is like, we generate a lot of our blog because we do a lot of blog writing for our clients. And in the past, it has been a very expensive endeavor. You have to write it, you have to hire somebody to do research, you have to hire a copywriter. Blogs can be very expensive, right? That's not the case anymore. Now we build these prompts where it generates blogs because most of them are general knowledge anyway, right? And it comes in and we're able to offer that. Just build it right into our package. And people love it. And we're able to get it already SEO optimized or GEO optimized already. I had a bookkeeper here locally that just her blogs have moved m her up to number two on the Google search. Just her blogs. So that wouldn't have been possible before this tool because of the money history that you have to pay all these other people. The bad part about it is that now people are convinced they can do it themselves. They don't understand all of those strategies that I just talked about. They plug it in, they'll pull something out, they'll put it online. Case in point, I was on a plane talking to a man who, uh, works for a very large company that is global, that produces heavy machinery, I'll put it that way. And we were talking about AI and he said, look at this. And he showed me a picture that his global company had used AI to produce. Apparently nobody looked at it. It was supposed to be a job site of them using this particular heavy equipment. And instead it looked like a carnival and there was like a food truck and clams. And they had put it on their website and we're using it in their marketing materials. And I said, how in the world did that. How is that? Did nobody look at it? He said, that's the problem. They plug it in, they give it a quick glance. They don't look at it in detail, right? And then they post it or they put it on their marketing material. That's the problem with feels like an easy fix and it's just a tool and there's a big difference there.
Speaker B: Heather, I'll come back to you. I'm watching your face responding to Shelley's stories and her commentary. And what are you seeing and how are people using, abusing and trying to delude themselves into what this is all going to do for them.
Speaker A: So I would actually like to, to answer that question, I'd like to sort of share a tale of two companies. These are two very large, very public companies. This has all been in the news. So back in the summer of 2025, so not quite a year ago, Deloitte, which is a huge global consulting firm, had, ah, been commissioned to do some reports for both the Australian government and the Canadian government. Now Deloitte, of course, is using enterprise level AI tools. That means they have extra security, they have customization, they have a lot more things in place than a small business, a small corporation or you or I would in our AI use. Okay, so, so I know that that instills a bit of a feeling of confidence, but it's very clear that they did not understand that need for continued human oversight that Shelley mentioned. Because the reports to the Australian and Canadian government were found the month after they were turned in to be riddled with AI hallucinations. And it cost Deloitte several hundreds of thousands of dollars that they had to refund to those two arms of the government because they had not done their due diligence, did not fully understand the tool, and assumed that it would be entirely accurate. They didn't verify the research. Now, if we contrast that with Ikea, I love what IKEA did. They were contemplating laying off 8,500 customer service workers because of AI. They thought they could replace them. But before they made the move, they sat down and really thought through it and they thought, you know, that's a lot of knowledge capital to lose at one time. What else might we do? How might we use AI to spark a new revenue avenue? What they did is they reskilled all 8,500 of those workers. They taught them, um, AI within the context that they would need to use it. And they unlocked a $1.2 billion stream of revenue for IKEA interior design. I feel like that tale of two companies really kind of illustrates what we're looking at across the board. You've got a segment that's like, hey, we're going to replace all these people with AI. We're going to Put the tool in place and not need the people anymore, and we're just going to forge ahead. We can reduce costs. Reduce cost, and you can do that. But it's not really working out very well for some of the companies who've tried. Again, that Deloitte example, they didn't even lay anybody off. They just didn't employ human oversight. So when you start to do the layoffs, you. You're definitely removing that human layer. That's very necessary. Then you've got Ikea and the other mindset who says, hey, let's amplify our creativity. Let's produce new revenue. Let's see how we can take our existing people and their skills and turn that all the way up to unlock avenues of, uh, business we haven't thought of.
Speaker B: I love the creative process or the box process. I'm in the box. This is how I do it done. Or, uh, how do we get creative about opportunities? I will go research the IKEA story because I want to know how they turned it into a million, um, dollars of revenue. There's something that needs to be explored and expanded as case studies because, you know, the abstract isn't useful. You know, data out of context has no meaning. But you put it into meaning like that, and you go, how'd they do that? I want to do that. You know, where can I get some real benefit of it? Shelly, when you were listening to the panel that you and I were sharing, you were concerned about the misinformation that was being communicated. And I didn't want to ignore the dilemma of, uh, sitting on the sidelines saying, God, gosh, you're making it all bad. Can you share some of the things you were listening to and what you wish they had understood?
Speaker C: We were at a conference. We were listening to a panel on AI, and there were five people on the panel, and only three of them own marketing companies. And one of them, I will not name names, but she is in the financial industry. And she said, I plug everything into AI and it spits out what I need. And you heard people that knew better gas, because we were like, you can't plug intellectual property into AI. Um, you certainly can't plug things into AI that have compliance regulations unless you have a tool that has been created to handle that. And Heather can address that more in a minute because we've talked about this, because. And so we all, uh. And I knew she didn't have that. And so she literally looked down at the crowd, laughed, and said, go fire your entire marketing department, because you don't need them anymore. As she sat on the panel with three very successful marketing firms. And. And the facilitator actually has a company that does cybersecurity and AI compliance. And I remember just being aghast that people that run million dollar companies had no concept of what the tool was capable of on the good or the bad. Because I said to this person later, if you've plugged your client's information in there, you just made it public knowledge.
Speaker B: That's right.
Speaker C: And I believe you have broken some laws, because I believe in your industry, that's against the law plugging it in. I know there's. I'm not, I'm not up on all the compliance issues. And the other thing was that they were saying, you know, you can put anything in there, you can trust it. It has stopped hallucinating. I'll never forget them saying that it doesn't hallucinate anymore. Me and the, uh, doors, another person in the crowd just look at each other with our mouths just hanging open. Because it. A lack of intelligence on display.
Speaker B: Yes, but the certainty was compelling and
Speaker C: they weren't ready to say something with such confidence. And I went to the person that ran the conference and said, those people could be liable now. They could be liable. They gave information as a trusted source that is completely false. Completely false.
Speaker B: Yep.
Speaker C: And why was she even on an AI panel? She owns a financial company. That's crazy. Absolutely crazy. So I, uh, want to ask you
Speaker B: what the response was, but I do know that you were so upset that we really did want to bring this to the public in a way, to set some parameters. Not the five rules about why you should or shouldn't do what you're doing, but you're basically outlining, this is high risk and it could give you a high return. But if you don't understand what you're actually doing, you know, Chat's my buddy, but he's not, you know, and. And unless you're a, uh, critic or at least aware of what's going on, you're going to be burned. And, and I am anxious to see as these begin to materialize, what, what actually happens. Heather, you're smiling. Something to share about Shelley's. The two of us were sitting there going, oh, really? Please.
Speaker A: Well, as I explained earlier, hallucinations are not a flaw of AI. It's literally generative AI doing its job. The way the technology that underpins it, the system architecture is designed to work. It's designed to complete patterns. We have to know that it prioritizes Pattern completion over accuracy. That's how it was built. We have to do our human verification. And, you know, again, I think that the Deloitte story illustrates that whether you're, you know, a financial planner or a global enterprise, we have some mass misunderstanding because I think, uh, most of us are going into this with the mindset of these deterministic computer systems that we've worked with for most of our lives, not understanding that we're now dealing with something that is probabilistic. It's a completely different animal. And now I am no cybersecurity expert, so I want to make that very clear. But I share Shelley's concern regarding the security because as I explained earlier, if you're not an enterprise level company like Deloitte, where you have an entire staff of security experts who are working with AI who are performing audit log audits and all the things that are above my pay grade, right, then you're using what we basically call public facing AI or public AI. And the problem is that there are thousands upon thousands of bots being run by the cybercriminals daily, targeting users like you and me, because we are small business. So if you're putting personally identifiable information into that, and then let's say you've set up agents or you've connected your AI tool with other tools, you are super vulnerable. And now you've exposed your clients to that same vulnerability, just if their information exists in your email. You know, m. You think you're just keeping your email inbox straight by having an agent perform sweeps daily, but those bots that are out there, scouring, looking for those vulnerabilities can come and inject a prompt that pulls all the confidential information out of your email. So we just have to understand the dangers. There's so much powerful, good, amazing, uh, opportunity created by AI, but it has also exponentially multiplied the risk to the same degree. And we just have to be aware, we just have to be responsible. We have to learn about this powerful weapon, basically that we're holding.
Speaker C: Heather and I are best friends at that conference. As soon as I left the panel, I stepped outside and called her and I could barely breathe, I was so stunned. And I called her and said, am I overreacting? This is crazy, right? Am I overreacting? And shared with her what happened and we had a long conversation about it. And then the woman that facilitated the panel came up to me afterwards and she also. The state of shock was so profound. And then you saw us shortly after that. So, yeah, it's like holding it's like it can be a tool or it can be a weapon.
Speaker B: Well, but we don't know what it is. And it's not going to shoot you, but it is going to kill you. And it becomes important to have some do's and don'ts. Have you found some, you know, the three or four or five things one could do or not do when using this as a, um, vehicle, a tool for answering a problem or a question, having anything to help our listeners become more aware of. Because we're not doing a training today, we're doing an alert, right?
Speaker A: So I would say the first thing is just be very, very thoughtful about what information you're giving it and also, you know, go into your, go into your settings in whichever large language model you're using and turn off their ability to use your chats for training data. It's just an extra little layer of protection. Uh, you know, never give it any personal personally identifying information. We call it, definitely don't feed it proprietary information of a client. Maybe any kind of frameworks that aren't copyrighted yet or trademarked yet, you know, that could give you an exposure there. And one of the biggest things that I think people don't realize, especially with a lot of image generation and video generation, is that they need to be keeping what we call provenance logs. And that means that you need to have a record of the prompts that you used, the date that you created this thing, the tool that you created it in. Because without that you have no basis for copyright. You have to prove significant human creative direction in that process. And it's not just about being able to get copyright, it's also being able to protect against copyright infringement. And a lot of that is just very in flux right now because of the way some of these models were trained and some ongoing lawsuits. And we really don't know how all that's going to turn out. So it's, I guess, document, right? That's kind of been the cardinal rule of business our whole lives, right? Document everything and hold on to that documentation and just take a lot of thought and care before you enter information into it.
Speaker B: Interesting.
Speaker C: I would say we build custom GPTs for people. And so what we plug in that we have found super useful is we go in and find out their brand, voice, we find out their mission, we find out their statement, we find out and we plug all that, we build that into it, which helps us create some really good stuff. I still tell people to this day, I really need you to give me original Photography. Because AI generated visuals do take algorithm hits sometimes, a lot of times. And like, especially with Google, especially some of the platforms, they will straight out just, you know, rinse, like it. And I'm like, why don't you get some good visuals in there? And then, like, the photography still video is still king, right? Create your own. And then. Because the other thing is that people still want that human touch. People can now are starting to be able to recognize what's AI generated and what is genuinely created. Right. So it's been great for when we plug all that in, producing some really quality, good, you know, content and ideas and stuff. But like she said, we also keep every prompt that we build for people so we can go back and say, this is how we got to where we are now. Which also helps us, uh, train the AI because we can see in the prop, okay, we need to build this better here, we needed to add that information there. Or like that whole thing. You do have to keep detail records, you really do, of how you got to where you are. So. And it's still going to hiccup. It's. I work with a company in Raleigh that has been in Raleigh for 30 years, and I swear to you, every time it produces something, it says it's in Texas. And I don't, I don't know how to. I don't know how to make it. Every time I go, it's in North Carolina. Oh, you're right. I know. I have arguments with it all the time.
Speaker B: Well, you have the same conversations with your, uh, chat buddy as I do. And it is. Everyone is learning on the job and trying to stay out of harm's way at the same time. And I don't think we'll cross a threshold which says, you've arrived, it is now safe, it is now protected. Because the whole nature of it is to take data and make patterns out of it. And if the data isn't good, there are no good patterns coming out of it. And, uh, it's making stuff up and so are you. And they're not the same. Claude is different than Copilot, is different than Gemini is different than Chat.
Speaker C: So.
Speaker B: And then you say to yourself, well, how can each of them have a different way of creating the pattern? Heather, go ahead.
Speaker A: I can actually answer that question. So it's because the different large language model platforms had different training approaches in the beginning, right? They had different priorities, different developer priorities. And Claude, or Anthropic, they have what they call a constitutional approach to training. And so they had safety and rigor in mind. First, far more than any of the other LLMs in terms of developer priorities, they have different context management, different waiting, a longer context window. So there are a lot of other technical components and how they've built their version on top of the transformer model architecture that go into that. The Gemini system is within Google's ecosystem, so it excels at anything where you're talking about like conversion strategy and local SEO and geo, and it does a lot more than that. But anything within that Google ecosystem it excels at, you know, Claude is much better for long form content because of the way they manage that context, the way the waiting is. Chat is great for creativity because it's not going to interrupt your brainstorming so much. It's going to kind of flow with you. It gives you good scaffolding. They definitely are not built equally and they cannot escape their original scaffolding no matter how hard they try. Because once you start, it's kind of embedded. You can try to alter it, try to change it. We've watched chat do that as it's tried to become more responsible and stop certain what we want to call moral failures in its conversations. But it's found it very difficult to do that and to do it successfully. And they have come a long way. But, uh, again, we're not training. This is just, uh, an awareness thing.
Speaker B: Well, it is experientially. You can ask each of them basically the same prompts and see how different the responses are. And then as a human, you have a problem of figuring out which one is the right one for you to use for what purpose. And it makes it pretty interesting about how your buddies in each of them can see the world through very different lenses. And they are intentionally designed to do that. So it's, it is an interesting time. So remember, let's. We can probably begin to wrap ourselves up. The, um, whole point of this podcast is to take people off the brink, to get them off the brink and to help them see, feel, and think in new ways so they can do better. So, last couple of thoughts you'd like to share. Maybe I start with you, Heather, and then I'll wrap with Shelley a couple of things.
Speaker A: So I would say the last thing is there's such a huge push right now to just adopt agentic AI processes, and that is autonomous decision making that you are turning over to AI. So I just want to throw up a big, huge, giant red flag no matter how much it's being pushed. Please, please, please, uh, do that very cautiously, if at all. But don't be scared of AI. Just get educated about AI. Just get some understanding. You don't have to know how to build the engine, but you do kind of need to know how it works under the hood so that you don't put the wrong kind of gas in it and get stranded on the side of the road when you're trying to get to your destination.
Speaker B: Thank you. That's a great metaphor because that's just how people feel. What do I do now? And am I going to delete it or what am I doing with it? And Shelley, your last thoughts, ma'? Am?
Speaker C: I would say that I remember when social media started and people were hesitant about getting on social media because it was that social media thing, right? They waited, and then by then the people that were on the train were already ahead. And I told people, social media is an eight ball. You can be in front of it or you can be behind it, but once you get behind it, you're never getting in front of it. Now's the time to adapt AI, however, it's a tool. Don't forget it's a tool. And what really, really, really helps small business owners is human connection, Right? So I think the businesses will thrive that use AI to enhance their human connection and not replace it. And that, uh, don't be afraid of it. And if you are afraid of it or you don't know, there are people out there like us that do know, ask us questions, ask us for help. You know, I think it comes across as a very easy thing to do. And it's not just like marketing is not a super easy thing to do, even though it looks that way. Go find somebody that's an expert in it, let them help you, ask them to train you, but don't be scared of it. I know a lot of people are terrified of it right now, but you're not going to escape it. You're not going to escape it.
Speaker B: So, yeah, uh, I'm laughing. We went out for dinner with an old friend of ours who was an accountant, and he leaned over the table and he said to us, so do you guys ever use ChatGPT? And I looked at Donald and my husband uses it all the time. And I've been using it for a number of years. And I said, okay, Donald, is this a trick question? He said, well, none of my clients use it. I said, well, that's a segment of people who are your clients. But we've been using it since, I don't know. As soon as I discovered it, I did A proposal that I got, uh, an ethnographic research project in 2022, uh, on it. So, you know, now the question is, how do we use it? And, oh, by the way, I like Claude for my long stuff and Chad for my short stuff. And Copilot took lots of transcriptions and helped make sense out of them. And each of them is different. How do you use it? And he said, oh, you're ahead of us. And I laughed and I said, well, just, you know, keep moving along. It's an app. Very smart guy. But he was very, very concerned that his clients and he was keeping them in the dark.
Speaker C: He's an account.
Speaker B: Yeah.
Speaker C: It's funny to me that people use ChatGPT instead of Google. Now. That part is really funny. And that's where geo optimization comes in. Because if you want your stuff to be found when people are Googling chat using ChatGPT, I hear people sell it all the time. I use ChatGPT instead of Google, and it just boggles my mind a little bit. It's kind of funny. I'm like, for what?
Speaker B: Well, for whatever chat will find for you. But I actually asked buddy, how do I make sure that I'm optimized for generative AI? And gave me a whole list of what to do. Uh, and you can go to Google and do that as well. My job is to stay on top of all the search. You know, you're looking for a corporate anthropologist, cultural anthropologist. You want someone to help you change. I want to be on the top of your search, whatever you're searching at, so it becomes an interesting time for us. Am I right?
Speaker C: Yeah. Yep.
Speaker B: All right, my friends, we've had two wonderful women here today to talk about this new world that we're in, and it is time for us to sort of say goodbye. I didn't mention that Heather has, uh, a YouTube channel called Clarity First AI and, uh, it's there. Shelley, do you have something that they could refer to as. Well, because I didn't see it on your bio.
Speaker C: I would say just connect with me on LinkedIn. Yeah. I would say that you can find it in my name. Yeah.
Speaker B: Or Google search for Shelton, Right?
Speaker C: Yes. Or Google search for Shelton, which is S H E L T E N. Um, there's no O in the name. Yeah. Where you can search that.
Speaker B: So good for all of you who do come. I truly appreciate it. It's great to share with you folks who are going to help you see, feel, and think in new ways. Our books are all on Amazon and they do love your reviews, and I appreciate it as well. And now that we're rethinking retirement, I will tell you There are 75 million boomers who are rethinking life. And I've written a couple of articles for my substack post and for LinkedIn about how, uh, all the new technology is going to impact on us getting older. And I don't even want to talk about on the retirees, because they don't want to be called retirees. They don't even know what they want to be called. So there's a new world that's emerging, but, you know, technology is going to be extremely important in many phases of it. So I'm going to say goodbye. Please have a great day. Don't waste a day. It's every day is a gift, and I'm enjoying sharing with you. Bye. Bye, now. Bye, ladies. Take care.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.