Why You Should Never Fully Trust What AI Tells You (And What to Do Instead)

Pop your AI baseball cap on backwards if you want, but if you were going— I know you probably use AI in your personal life, in all of the things, right? You're Chief IA Officer.

Yeah.

What if you're a— if somebody's a franchise business owner today, what are the 3 AI technologies that they should be using or could be using every single day?

It'll depend on who you are, but there are 3 things that I think that I use that would be great.

Personal preference. This is always a personal— you got your Claude, you know, you got your open Claude people, you got your Grocs out there, you got your, I built everything in Chat and I'm gonna keep doing it. Where do you go?

All right. So, 3 things that I always do. If you're a franchisee, you know, you're probably taking, you know, you probably have financial statements and you're probably looking at your performance. You have data in Excel, right? Collecting data in Excel. I don't know of any franchisee that does not use Excel. There's Claude in Excel. Claude in Excel is an AI that is built directly into Excel. And you can just— you don't have to know anything about macros or modeling or, you know, how to do right predictive algorithms. If you can describe what you're looking from the data, it'll find that— it'll do that for you. And I use Claude in Excel all the time. It's very easy to use. There are tons of videos. So I think if you're in a franchisee, you know, if you own a, you know, a business, you should use Claude in Excel because AI will give you insights on the data and on things that you should be doing, that you could, that you should be looking at, that you would have not thought of earlier. That's, I think, one tool that you should use because always use AI that is embedded into products that you already have that you can benefit from. The second thing is that I think when you're running a business, it's always very important to know what's out there in the market, what's going on, and like how what you are doing compares with others. And so again, I use Claude, but you can use Gemini, you can use Grok. So when you are facing situations or problems or, you know, like you have questions about something, use AI as an advisor for you. But pick the AI of your choice because the AI that you pick learns from who you are and what you are doing. So you have to describe to the AI who you are, what you're trying to do, and then use the AI sort of as an advisor. I think that's a really important thing to do, which is, you know, pick one of your choice and then use it as an advisor to run ideas off of them. So the second one, first one was, you know, Use AI and tools that you have. Claude in Excel is a game changer. Second one is use AI regularly as an advisor, ask questions. And then the third thing that I'll tell you is sort of like might seem contradictory to the second one, which is just because an AI responds and tells you something, it doesn't mean that it is correct and that you should believe it. So one of the agents that I built was built recently is an AI that sort of like that has a discussion between them. So what I'll do is I'll pose the same question to, you know, you could pick Grok, but I say OpenAI, Gemini, and Claude, right? If they all agree with it, then you're fine. Like, it's very likely that, you know, that the question that you've asked, you know, that is the response to it. But sometimes they will disagree. Like, you know, you say, hey, how should I do this thing? And they will disagree. When they disagree, what you have to do is you have to— you can— what you can do is you can post the disagreement from one AI to the other AI. And what's amazing about this is that eventually all three will come to an agreement. And what this shows about AI, when people talk about AI, is that all AI tools may start with a different perspective. And I could get into technicality why that thing happens. But as you're feeding in information, even from other AIs, is that they will come to a consensus because they have this ability to learn and like absorb what you're providing it, and it'll come to consensus. So again, the third thing is that, you know, use AI, leverage AI to your advantage. And even if AI gives you, you know, in this case, you know, you have an AI consensus, but remember, you are an expert. Don't— you don't have to believe exactly what it says. If you disagree with it, push back and get the data. So like, in a, you know, in a simple way, like, you can make AI a part of your everyday life, right? So you can use AI as an advisor. You can have AI run scenarios and, you know, thought process about it. And AI is also very sophisticated because it can do your business modeling and you can do your forecasting and, you know, PL analysis and so on and so forth. And like doing it within the AI is valuable because it learns from you and adapts to it. And over time you can gain more and more value out of it.

Just to point people to a resource on that advisory point. I had somebody named Geoff Woods spelled with a G, G-E-O-F-F Woods, who wrote the book AI-Driven Leader and has a mastermind to that effect. But he shared in the first 15 minutes of the Unemployable podcast that he was on. I don't know what episode number it is, but just go to Unemployable wherever you get your podcast, put in Geoff Woods on the Unemployable podcast with Geoff Dudan. The first 15 minutes of that, if you're a nascent AI user, if you don't have— if you're not building things already, or even if you are, this CRIT methodology that he does, which, which stands for context, role, interview, and task. And he taught us in the first 15 minutes of that podcast how to use in any AI model this methodology to basically create an advisor. And then he went on to create He has an advisory board, his future self. I think he's got Steve Jobs on his advisory board. So he basically has an entire cadre of personalities that he's engaged with and it creates just great outcomes. So he shared a couple things too about how to run your annual planning using these tools where everybody's using the CRIP methodology all at once. And it was a really fascinating podcast. It's one of the most shared that we have. I sent it out to all Homefront Brands employees and all Homefront Brands franchisees, and I challenged them to use the methodology to come back to me as to how they're solving business problems. And the responses that I got were amazing. So now that's interesting though, but we moved forward in AI as an organization as a result of that one thing that we rolled out. Now when I interview our employees, they will say, You need to tell me what's next, what to use, where to use, and how to use it. So I'm gonna kind of get to a, I'm gonna do an interesting question. I don't wanna go too deep on it here, Geoffrey, but at what point do employees, developers like you might have, coaches that we might have, at what point do they say, if I continue to develop in this AI, it's clear to me that I am building my replacement. And at some point they stop, you know, using the technology to do the work out of survival mode or whatever it is. And I think I actually, you know, as I thought about that last night in anticipation of our meeting here today, of having our conversation, and I thought, you know, as I think we just had, we had the CEO of Microsoft on talking about, you know, what percentage of white-collar jobs are going to be gone over the next 18 months. I think it's going to be a more and more common thing where people are looking at it and they're start— they see the writing on the wall where what they— 80% of what they were doing, they're now putting into the AI, and they're seeing their roles become maybe less relevant. They get concerned about that. Are you seeing that in your development teams? How do you deal with that? Or is everybody just like, this is going to make our life better, easier, it's improve our utility of our product. Tell me what you're seeing there.

Yeah, so I won't sugarcoat it. I'll just say it as it is, right? So the first thing is if you're doing a job or role where the task is repetitive, you know, I do the same thing all the time, or I'm writing, you know, like a junior developer, I'm going to write, you know, like simple code, right? That's simple things. Those things are probably going to be replaced by AI because somebody above you or, you know, somebody more senior is going to be able to replicate that and is not going to need a junior developer. I mean, like one example was, I think, like Elon Musk. There's a story where, you know, he had this executive assistant and she wanted to go on vacation. And then, you know, she came back and then, you know, she wanted a raise. And he said, go on vacation and come back. And then he said, hey, I don't need you anymore because I'm able to do this on my own, right, with the tools. Right. So in some cases, depending on what you're doing, I think you could be replaced. But what I'm also seeing is that if you are creative and you have ideas, it was being held back by development or held back because there was this friction point of like, how do I go from what I have an idea for to what do I'm gonna create? That's not going to be replaced by AI. Because AI, remember, when we say generative AI, it doesn't come up with like new ideas. It comes up with suggestions and things like that. But you still, as a person, still have to do that execution. The other thing is like in franchising, we talked about engagement and other things. The human relationship element is really important. Like I don't go to a doctor because, you know, I couldn't like upload the file and kind of like self-diagnose myself. I go to the doctor because I sort of like trust that person and I recognize what they're trying to do. So I think when technology initially comes in, we focus so much on what the technology can do, and then we sort of extrapolate that, hey, it'll finally be able to do everything. But the fact is that, you know, it will not be able to do everything, and then we will see the limitation. Remember, uh, like, like full self-driving. Like, you know, we'll always keep getting closer to it, but it'll be very, but the last mile problem is a real problem. So I think like, if you are a person in any job or anything that you do, if it's creative, if it's design-oriented, if it's something where it's relationship-based, where, you know, you have to gauge the empathy of somebody else and like, and interact with them, at an emotional level, which is what makes us, you know, makes society what it is, cannot be replaced by AI. Because AI can pretend to have emotions, but at the end of the day, even when you type in something to AI, it converts it to zeros and ones, right? It converts it into a token. And then, and what a lot of people don't realize is that when an AI is responding, it actually doesn't know what the next word is going to be. It's based on a probability theory, right? It uses probabilities and then says constructs things together. So it sounds really smart and really interesting. So I think if you are, you know, if you do repetitive tasks, obviously that's a workflow, it'll be, you know, automated. But if you, but if you are, you know, dealing with people, creative, solving problems that are innovative, and those, I think you will still need people, and AI will make them better. So the speed of innovation will improve because of AI. AI.

Yeah, it's interesting that there was such a movement towards the analytical, the technical education, coding, tech, people making money in tech and all that. And now because that's more easily automated, it's kind of the rise of the creative and the person who can ask, who can see the problem clearly, who can ask the right questions. And at some point, like, I don't know how a calculator works. I just know that it does.

And you use it. It didn't make you any less valuable than before. This is the same thing with like, you know, when printers came along. I mean, like, it didn't mean that it reduced people working in offices. You still, you know, you still use printers. I think that's the thing with AI. I actually wrote an article recently which says, you know, AI makes people into builders. And that's what it is, right? It's moving the managerial class into builder class. So if you're just managing people, that's the other area I think that AI will probably get rid of. Unless— because even if you're managing, you still have to be using AI and building ideas and concepts that— so it'll create a society probably more of builders and less of managers and less of, you know, people doing the same tasks over and over again, at least on the white-collar realm.

And I'm hopeful that franchising is an area that's gonna be more insulated. And the reason for that is one of my basic premises when I talk to people about franchising is that franchising works best when a product or a service needs to be delivered locally by a human. So by and large, that's what franchising is. If somebody, I've had people come to me with franchise concepts that were based on, a technology or based on something that was gonna be shipped. And I'm like, that's not Amazon-proof. I said, that's not, that you can't, that's not gonna be, need to be done forever. It can be done by Amazon. Or if it's a technology, you just do it centrally. So there's some businesses, there's some organizations that just lend themselves to be managed centrally. You don't need a franchise system because at the end of the day, the local execution of the introduction of the product, the delivery of the service, the fulfillment, fulfillment of it, the personal touch, the analysis. We're going to paint this, we're not going to paint this, we're going to move this furniture, we're not going to move that. All of the things up until this point in time, those are perfect franchise models. So now we're sitting here and it's always worth putting the date out there. So we're sitting here in February of 2026 and there's mid-short videos all over the internet about Chinese robots that are dancing all over the place. And I think Elon Musk just decommissioned a couple of his Tesla models to allocate those 14 million square foot factories to Optimus production of Optimus robots. So, and he expects that Optimus robots are going to be in consumer production by the end of 2027. So, you know, that being said, a humanoid, you know, who I You know, most of the conversations that are going on right now are around the AI replacing our thought work, our white-collar work. Okay. But shortly after, you know, after that, if you're paying attention, then what's going to happen to the blue-collar work? What's going to happen to the service delivery? Are we really going to have a crew of robots jumping out of a self-driving car with painter's overalls on coming out and painting our house? I'm not sure. I don't know anymore.

Yeah. So I think this is another interesting question, right? When I talked about the last mile problem, where if it's repetitive or it's mostly repetitive, I think AI can do it, or whether human interaction doesn't matter. AI can do it. So I think, you know, drones, good example of AI. Anything with military, like, you know, you don't need to put people to fight wars for you. You can have, you know, an army of drones, or you can have, like, you know, these robots go and do that for you. I think, like, you know, like fighter pilots, I don't know if you need fighter pilots at some point. AI can actually, you know, probably, you know, you can use that because there's no human, you're not engaging a human being on the other end. Right. But, and even like, and even for constructions or any of these like other manual tasks, like where, you know, if I pick a brick here, put a brick there, put big, like, you know, like car assembly line type stuff, assembly line stuff. Yes, you can do that. Because we've been trying to build robots that can do complex tasks and think about it, like, you know, like rearrange your home and all. Right. Very hard to do. Like, yeah, there's one thing to say, I have a vacuum cleaner that goes and, you know, vacuums out. Stuff, but it's not mission critical, right? If it misses a piece of dirt, nobody really cares about it. All right. But what about a robot doing surgery, like a heart surgery on you, right? Like some people might, you know, be okay with it at a really low cost, but most people are probably gonna still wanna have a surgeon associated with it. But I'll make another, what's called argument, that if you AI-fy or robotify human interactions, it can actually hurt your brand. So think about it. And I always think about, say, Starbucks, right? So what made Starbucks special was that you could walk into a Starbucks and there was a human being and they would know your name, write your name, you'd sit there, you know, you could, you know, it was a meeting place for people. They looked at the data and said, you know, how do we make Starbucks more efficient? And they looked at all of this data and analytics and said, you know what, we're going to build this app and you're going to order order in advance and you know, you're going to get all that stuff. We're going to do drive-throughs and all of this other stuff. Remember, like they went through it. As a result, it lost its character of what it was. Starbucks no longer focused on the human interaction. It was more about efficiency. Once it became more about efficiency, it became just like another fast food place that you go to. And for efficiency, and if I'm trying to pay a premium, why would I just go to Starbucks? So that's another example of like, even though it was nothing to do with AI, but what it was is that when you move from human to mechanical, you lose something in the process. For some businesses, it might be a positive, and it's okay because that's what you're trying to do, right? But for some businesses, you know, the, the reason you have customers is because of that human touch, right? I mean, order a pair of jeans from Amazon, but I don't think I'll order my daughter's wedding gown from Amazon, right? You know, like, that's the thing, Right?

So might order the husband from Amazon, or he's from Amazon.

My wife works at Amazon AWS.

Yeah.

So anyway, so that's the way I think about, right? You know, I think AI is a great technology, is very capable, but we always have to keep a balance of where it is. So I think on the development side, like, you know, I do a lot of work on like, you know, things I can do now I couldn't even think about in 2024. And I know that things we can do in a few more years can even not even think about what, you know, uh, you know, today. But still, there are like, there are, there are so many unknowns. And, and, you know, how societies work, right? So maybe AI will change society completely, but, you know, often we overestimate the impact of technology. Like, we sometimes underestimate as well, sometimes we overestimate what technologies can do because we get blown away by the initial and we think, hey, we're going to have flying cars. Remember in the '60s, people thought we were going to have flying cars by now.

Yeah. And look, this technology is different because, and there's several becausees there, right? But to your point, I mean, the vast overwhelming majority of people by population anyway aren't really paying attention because we believe that our future is going to be some small deviation of the past. You know, if you talk to— if you, if you read anything that Ray Dalio talks about, it's like people, you, if you have to look far enough back, 100 years, to understand the cycles. And once you understand the cycles, then you're going to understand, are they going to come faster, are they going to come slower? But there's always a cycle, right? So This is a technology that is different because it has the ability to evolve on itself. So think about this. You've got, you know, the ability— today, robots in automotive factories are basically single modality tools. They're screwed to the floor. They're an arm. They pick up a piece, they put it in, the next robot comes in with the screw gun, screws it down, and the car just moves along. And there's 1,000 of these single modality robots that are not mobile and they're not really learning anything. They're executing on a set of code to do what they're doing. Sure, do they have sensors that tell them if something's off and how to get the screw right in the hole? And sure they do, but that's certainly not, intelligence. Now you take AI and you put it into a mobile multimodal humanoid robot, and now not only does it have the network that it's connected to that it's getting smarter, it has some proxy of visual. So it's not, it's, it's not only getting, getting smarter from the network, it's also getting smarter from its environment. Because it's seeing and it's observing and it's taking that into account. And something that it learns is getting learned by the other robot. So, and then at some point, when things are learning so fast, it almost, while it may be programming, it looks different to us because we don't have the ability to understand it as quickly. So that's, I think that, you know, so the change is it's, yes, it's another technology. Look, you know, we went from whales to whale oil to steel to tech to, you know, all of the things, right? Or you wrote, there was a great book I read years ago. I think it was called The Big Switch. And it went from the evolution of the fact, you know, to get power, you would have to put a factory on a stream. And then when the Industrial Revolution came, you had to build a power plant in each factory. But when they built a— when they built the network of electricity, the infrastructure, then you could have a centralized power plant and then you could power everybody. And then that led to a big thing, right? And it was the same thing with the internet when the internet happened. So you have this cycle pattern of basically networks These are different types of networks that are technology-based that basically create a new normal, and then everybody has to adopt and all the jobs need to change, and then it runs like that for 30 or 40, 50 years, and then the next thing comes. The difference in this is just the freaking rate of change. And it's at a rate of change that we are going to struggle to adopt. 'Cause you could create a new iPhone model every day now, but what good would that be because you can't manufacture it fast enough and we can't adopt it and we don't need more tools on our iPhone. It does plenty.

You can build new apps. I think, I'm not disagreeing with any of this. I think AI is—

Well, you probably, if you knew my IQ, you would.

So no, I'm not disagreeing with the fact that AI is, is a huge disruptor, more than anything else we've seen in the past. In the same way that I think the internet was a huge disruptor before anything else before. If you think about now, without the internet, AI would not have been possible, right? It's built on it. Like we stand on the shoulders of giants, as they say, and technology works the same way. And I think AI is— there's a lot more to come with AI. But what I was referring to mainly is When people think of AI today, they're thinking of large language models. There's a lot more AI that's happening beyond large language models. And I think what we will see in the next few years, this is just my instinct, is that we will start seeing the limitations of a large language model-based AI. But there is other AIs being delivered, and I think we'll see the next generation of AI. And, you know, and AI will, you know, right now the bottleneck really is on the computation side. They're building these large data centers centers and, you know, it's processing because, you know, AI, if you give it enough time with enough data, enough parameters, it can basically answer any question on anything based on any prospect. But there's the limitations on processing capabilities. So, you know, so that is going to change. And I think, you know, we are going to get, you know, obviously I like Ray Dalio as well. The rate of change is going to get faster and we don't know what's going to happen in the next few years, but we have to be ready for it. Yeah.