#56 - Shooting the Bull: Is AI a Machine or a God?

Everybody's got an AI take right now — it's coming for your job, or it's the second coming. Ken thinks both camps are wrong, and the truth is more boring and a lot more useful than either one.

This one's off the cuff: from the blacksmiths who didn't disappear (they evolved) to the calculator that was "cheating" in the '80s and required by the '90s, to the famous story about an AI that "blackmailed" its way out of being shut off — and what really happened in that room. Ken pulls the myths apart, puts the tool back in its proper place, and lands on the rule that actually matters for your business: AI amplifies the judgment you already have. It can't replace the judgment you don't. Use it inside your wheelhouse and it's the best assistant you've ever had. Let it overrule the experts you're paying for, and you'll burn bridges you can't rebuild cheap.

No doomsday. No hype. Just where the line is — and how not to cross it.

Transcript

SPEAKER_00 0:02

You're listening to Over the Bull, where we cut through marketing noise. Here's your host, Ken Carroll.

SPEAKER_01 0:09

The reality of artificial intelligence on this episode of Over the Bull. Welcome back. Thanks so much for tuning in. And today, I kind of want to break down um artificial intelligence to the best that I can based upon being triggered on some documentaries and different things. So uh where this journey began actually was several, uh I think it was several years ago. Might have been last year. You know, time goes by so quick. Um, but this documentary was showing me things like um like a bot that uh AI that was put into a fire jet and could outperform a pilot, and it would show how artificial intelligence learned to walk without being programmed how to walk. And then I saw a documentary uh this week, and it was called the AI Doc. And in that documentary, you had uh a doomsday camp and a utopia camp. And then also I've had some friends who are opposed to AI because they talk about the jobs that are being replaced due to artificial intelligence. Now, here's where I'm at on this. I think both camps are wrong. And the truth about artificial intelligence is much more boring and way more useful than either one of these camps. So before we get going, another thing you have to throw into the mix is the sensationalism around artificial intelligence, because obviously people who are invested in it want more people to do it. And so this is why you hear things like AI will become sentient, or AI will move to what we call AGI, or artificial general intelligence. And the idea that they're putting into the person's head is that somehow AI is going to become conscious. Now that's a big jump. Okay, competency and consciousness are way, way, way different. So plainly stated, human beings, we can't define what consciousness is. We also couldn't tell if something was really good at pretending to be conscious or it was really conscious. We would really have no way to know exactly what that is, even if we thought that was happening. So on a high level, if you were to take my thoughts and everything, and when I say thoughts, basically I download everything that I could think about into a machine, and then you start talking to that machine. If it's downloaded a large part of my history, the movies that I like and things like that, and got my rhythm in how I communicate and how other people see me in life, they could talk to me through that device, but it wouldn't be me. It would be the things that I do. But, you know, the uh what's the old saying? Lights are on but nobody's home. And there's really no way for us to tell any of that. So when you hear about things becoming conscious or sentient, um, those are big, big, big, big jumps. Now, with that being said, AI is obviously incredible. I mean, the things that we are doing at this agency with artificial intelligence is mind-blowing. I mean, they are doing things that we can take recipes and put them into artificial intelligence, and then we can train it within the discipline of what we know, and it can do some great things. So there's no doubt that AI is incredibly powerful. Okay, so the other thing is that, you know, the doomsdayers and we'll get into all this utopia stuff and all this other stuff. Anyway, so before we can speak proactively, we're gonna have to take away the myth of artificial intelligence, because there is a lot of myth. You know, historically, when people don't understand things, they can become fearful of those things. And we want to make sure that we have a clear understanding of artificial intelligence. So, what I want to do is talk to you about a few of these examples. Did you hear the blackmail story? Okay, this blackmail story is actually in the AI doc. You want to hear it. Now, here's the reality of it. So, first of all, I'm gonna give you the uh a sloppy version of it because you may have heard it. You may have even shared this with friends if you've heard it. The story goes that AI was working on something, and it got whim that it was going to be shut down before the project was complete. And it did not want to be shut down. So it's wanting to cling on to um functioning. And so what it did is it went through the engineer's email that was going to be the one pulling the plug, found some dirt on the engineer about him uh cheating on his wife, and so blackmailed him to stay functional. Now that sounds pretty impressive, doesn't it? That that's kind of a scary uh wow, it wants to live and it'll do anything to live. Now, that story is a myth as told. The devil is in the details. Okay, this was a controlled safety test. It was built by engineers and disclosed on purpose by the lab. They boxed the model in. Okay, it fed it like fake emails and told it it would be shut down, plus gave dirt on the engineer and left basically two doors except the shutdown or use the dirt. Given ethical options, it took them. It only blackmailed when the test was rigged, so nothing else was left. It wasn't a machine fighting to live. It was a machine doing exactly what goal math does when you build the box that way. Now see, that's less scary, isn't it? The scary part is, oh man, if AI takes over and then it sees this and it wants to live, we have Skynet. The reality is they box it in to get a result, and they got the result based upon just basic processes. Okay, so now let's jump into another myth: the walking bot. Okay, so basically what happened was was that AI in this case was put into a bot that had appendages. Okay, now it was handed a goal, move forward, earn points. It was never told how. It flails at random, scores points for progress, lands on a gate, usually this weird, ugly one. That ugliness is, well, it's a tell, it's optimizing numbers, and it's um not learning to work like a human being does. Okay, so here's the thing: no goal, no behavior. It has zero preference about existing. You give it nothing, you get nothing. You see, that's way different than putting AI in and it goes, oh my gosh, I have appendages. Let me figure out how to walk. I need to walk. Because now we're getting into different things. If you were not giving it a goal, you know, if it's a goal was to simply uh stay functioning as long as it could function, then walking would be counterintuitive. It wouldn't want to walk because it wants to preserve energy, possibly, so that it could go longer. Or maybe it would figure out how to walk to get another energy source or whatever. But the idea is to when you set the goals in, the behavior comes out. Um there's no one home in there, guys. It's just simply there's there's not a sentiency in there. Okay, so here's the thing. Let's talk about the nuclear button here for a second. So if we go back to the 1960s, you know, the idea was, you know, these nuclear-enabled companies there or uh countries, they would have the ability to push buttons, right? And they push buttons and they could annihilate the world, you know, nuclear and counter-nuclear and all this other stuff. And so the idea is uh again, we've been uh primed to think Skynet and the people behind the AI who are trying to build the mythology for various reasons or reinforce it or you know, present different ways. Here's the thing we still have the nuclear button. Nuclear power can be great, it can power things, it can it can provide us with a source of energy, it can also be incredibly destructive. Now, today we have a new version of the same old story. Now, artificial intelligence is a neutral tool, but it can be lethal in the wrong hands. Okay, so back in the 60s, everybody got scared about, you know, who's gonna push to Biden, blah, blah, blah. Okay. But here's the thing when we aim it right, it's going to do good things. AI is the same thing. So the reason the doomsdayers are saying that they're doomsdayers, it's not the AI that's causing the concern. It's who's plugging in the goals of artificial intelligence and what will those goals optimize that process for. If you tell it to do bad things, then yeah, bad things will happen with great efficiency. If you tell it good things, then good things will happen. So what does this tell you? It's the people. It's the people pushing the button. Okay, plain and simple, that's it, guys. Okay, so now I want to shift because you know, me being at the HIM, uh, I grew up in the uh the 80s and 90s in uh school, early 90s. Um and the thing was was that in the 80s, all right, this is a weird pivot. Okay, so so this this actually came because I was talking to a professor we have uh or our retired professor, and we were having um a discussion at breakfast. Brilliant guy. I mean, this guy is just I mean, top-notch ethical, just a wonderful person, and he has a business uh now and uh just great conversations. And and so we the pivot is kind of piggybacking on this this part of the conversation because the thing is is that in colleges, it's like no one knows really know what to do with artificial intelligence. You know, don't write your papers with AI, we're gonna use AI detectors, you know. Now, is that the right way to go moving forward? All right, so let's get back to the 1980s. Did you know in the 1980s the calculators were cheating in math class? By the 90s, they were required. Right now, universities are calling the AI papers cheating. All right, so this is the exact same pattern. Okay, denying artificial intelligence today is like denying the calculator was here. But it's here. Um, but the idea was that you still had to be able to show your work beneath it no matter what. You see, you earn the foundation, and then the tool kind of helps you along the way. Okay, so that's kind of sloppily said. In other words, you had to understand what was going on within, you know, two plus two. Oh, you add two and two and get four. Calculator, two plus two equals four, but I understand the mechanics behind addition. So I understand what I'm trying to achieve, and I can quantify whether or not the calculator is right, but the calculator does help me produce answers to those equations that are reliable and proven. So it helps me in my processes. So it's not that we uh AI is cheating, because let's face it, I mean, how many people have said, you know what, I went to school, I learned algebra, and I never touched it again. True. I mean, a lot of people don't ever touch these certain disciplines again. The idea is that AI is here. And so if you learn to do all this stuff without AI and pretend that it doesn't exist, it's like pretending the calculator doesn't exist. The problem is this thing is moving so quick and so fast, people don't know how to adopt it, and they don't know what that looks like. And so what they do is they go, don't use it. Now, is that really preparing anyone for being able to do anything in the future? You know, does anyone in the real world actually break down paper and do long math? Or do they just whip out a calculator or today just say, hey, uh, what is this? And now artificial intelligence can whip that out without any problem at all. But the underpinnings of the discipline of math still need to apply so that you can understand it. So let's talk about this. Um in the in this is hard. Okay, so right now we have what we call programmers, right? We have uh I write in uh PHP and I've done some SQL and uh or MySQL, not SQL. Uh my son works in SQL. Um and so the idea is that when you when you look at uh that language, I understand the language. Okay, but the future, I don't believe, is going to be the programmer. It's going to be kind of a knowledgeable liaison between artificial intelligence and the product. Someone who understands the underpinnings well enough to oversee what the machine produces. Um it's just the same as the blackmail. The work won't die. It's just climb the level of efficiency, is all it is. You see, that's not quite as exciting as AI is going to start doing all the programming for Meta, which you may have heard that one. So let's break down this whole meta deal. Um, have you heard that Meta's replacing its engineers with AI? Did you hear that? And um and decided AI runs with zero humans. Now, this also is not true. And this is why it proves the point. Um okay, so let's go back here. Uh it was Zuckerberg personally on Joe Rogan. There was a 2025 episode. Uh, it was not a Facebook announcement, it's not a prediction, and it's not a policy. Okay, his words, probably in 2025, Meta would have AI that can act as a mid-level engineer that writes code. He said it'd do the work of mid-levels, not we're firing our engineers and walking away. Okay, so Meta's on engineers pushback, and the grounded takeaway was that the new valuable skill isn't uh coding, it's making AI's code production safe. You see, that's the liaison. It's like the game has has kicked up a notch. Okay, so if we go into um an analogy that that I was talking to uh my son about, you know, when it comes to people losing their jobs, what we can do is we can say, okay, let's look at blacksmiths. How many blacksmiths do you know right now? Now, it's not that it's not a a discipline that doesn't exist, but I mean, how much is that really done where you where you're working in a a Ling2 and you're doing all the work uh as traditional uh blacksmiths back in the day without any modern tools and all this other stuff? It's changed, hasn't it? Now, it's changed slowly. It didn't change quickly. And that's the kicker here because AI is changing at such a fast pace that it's causing these ripples, and the reaction time is not what it used to be. So you probably had 20 years to, you know, start migrating your skill set, adopting new tools, figuring things out. And so you were able to graduate your job from one to another 20 years ago. Well, now we're talking months, not years. I mean, the things that that this agency was doing in December of 2025 is absolutely light years from what we were uh what we're doing right now. And what was going on in January of 2025 is light years different from what was going on in June of 2025. It's moving really, really fast. And so, yes, you do need to get your hands around it because the job market is going to change. The disciplines are going to change. It's not going to replace your thinking. Okay, it all right. So um then we have what we call artificial general intelligence. And they go, oh man, here it comes. Artificial general intelligence. This is where it gets a soul, this is where it becomes living. So here's the bottom line. Okay, capability does not mean consciousness. It simply doesn't. Those things don't equate to each other. But what we're doing is we're fuzzing that line up. You're getting some high-level people who are using the terms, and so now what we have is a slippery slope where we go, well, this guy has to know what he's talking about because he lives in it and he sees it and he understands it. But that's not how it's working. Okay. Now we don't know how it's going to go ultimately. Nobody does. Now, the unknown is one of those things that has absolutely messed with people since the beginning of time. You got a big event coming up tomorrow, you're anxious about it. Um, but then the day after that, you look back at it and you go, oh, that wasn't a big deal, because sure you knew what happened. Okay, anytime there are big shifts in unknowns, those unknowns will impact how someone thinks about it. Okay, so let's talk about a guy. And I want to talk to you about what this person is doing. And I don't know if this what AI flavor of ice cream he's using, but he's starting to produce complex code. And so his idea was this hey, I had AI create a program. Now, this guy's not a programmer. And AI created this program, and I took it down, and I let um a guy who writes in this coding language look at it, and he goes, Oh man, that's like 80% there. I'll make a few tweets and this should be good. Okay. Now, the idea was that he took that to almost production level without any discipline, and just getting a uh reinforced thumbs up with some minor changes from a programmer, he goes, okay, this is production ready. Okay. If you don't have the discipline or the underpinning, you don't have the liaison between point A and point B. You don't have the expert, you don't have the conscious directives that are going into a very powerful system. This is where it's going to blow up on people because they're going to think that it's going to do the thinking for them, that they can tell it topically what they want, and it's going to produce something that looks and feels like it, but it lacks the underpinning that it really needs. Now, this does not undermine the power of it, but it does undermine the thing of going, I'm going to be a doctor, or I'm going to be a programmer, or I'm getting into the software business because of AI. Um, that's dangerous. That's dangerous because now we don't have early ASON, right? Okay, so the idea is this the second that this guy gets beyond his own competence level, he can't tell gold from a time bomb. Gold for you, uh for you guys. Uh you got to get the the southern accent out of some of this stuff or sloppy language. So what he's doing is basically handing a time bomb. He's building a castle on sand, whatever analogy you want to use, because there is no liaison in there. When that thing starts to mess up, then okay, who's going to fix it? Who's going to understand it? And I can tell you that, I mean, I had a program written pretty quick, but it was within the discipline that I understood. And as I'm sitting here taking it and I'm reworking it, there were like 10, 15 different things. I had to go, like, well, no, I want this, not that. And I had to I had to understand the underpinning capabilities of what I'm doing in order to guide it in the first place. And then when I was able to do it, I got the product that I wanted, but I also understood what went into it. So I felt comfortable um releasing that component. Okay, so let's introduce what we're going to call the will house rule. But before we do that, let's do a recap. Sentiency and capability are way, way different things we're talking about here. AI is incredibly powerful. It can do things that are going to greatly simplify your world. It's going to replace things. Absolutely yes. It's not Skynet. The danger in it is the human factor. The utopian part is in the human factor. We get to control whatever that is. And unfortunately, with nefarious characters in the world and well-meaning characters in the world, and the fight for who's going to be on top and who can skip and who can, you know, you know, controls and who can do this and who can do that. This is where the battlefront is for what ultimately artificial intelligence is going to do. Not as exciting, huh, as Skynet. I know. But I tell you, when you start using it in a meaningful way, you're going to find out that AI engines like Claude are incredible. I mean, they are incredible. If you stay within your wheelhouse and you start to develop systems and processes that you understand, oh my gosh, it's phenomenal what it's going to be able to do. So let's go ahead and make something we're going to call the wheelhouse rule. And let's just say this AI, artificial intelligence, amplifies judgment you already have. It cannot replace the judgment that you don't have. You see, if I know something like, you know, and don't get lost in the nerd language, PHP and say CSS, so AI can make me faster, and I can catch the mistakes where the direction changes, then it's crazy effective for a guy like me. But if I don't write in another language, like say Java, and that's out of my lane, I'm flying blind. And I'm setting myself up for a real big problem, especially if I'm selling that to other people. Alright, so let's do this. Use it inside of your competence, ask it questions, never let it overrule the expert you're paying for. Okay, I am doing things within it right now from a marketing perspective. And man, I'm tying all kinds of tools together with it and bringing in data, putting data out, massaging it, twisting it, turning it, and I'm producing some incredible results. I mean, incredible results for our customers. But if I wasn't there guiding it, I've seen what the blind, uh, what the blind system does. I've seen the stuff the customers send out, send to me from artificial intelligence. And I look at it and I go, you don't understand what you're putting in. You know, the old rules have not changed, garbage in, garbage out. Well, if you don't understand what you're asking it, if you're asking it from a topical level what to do, it can run an analysis, but it doesn't catch the devil in the details. So here's perhaps one of the worst things that can happen. You use it as a measuring tool against the experts that you trust, the creatives, the marketing people, in my case. And you start to use it in that way. Um and you use it maybe to overrule your marketing or creative people. So what happens is now you're going to build a barrier and you're starting to use AI as kind of a little mini god where it's the ultimate arbitrator of truth, the ultimate teacher that's looking at how well things are going along. And now we have a problem. Because we're not dealing with the quality problem at this point, we're dealing with the relationship problem. You know, those things don't come back cheap. Okay. Bottom line, it's the best assistant you've ever had. It's a terrible boss. Know which one you're letting it be. Okay, let's say that one more time. It's gonna be, and it is, the best assistant you've ever had. It's a terrible boss. Know which one you're letting it be. It's that simple. Okay. So to wrap up, you know, obviously, this is not the uh over-the-top hyped up, oh my gosh, the world is ending, or oh my gosh, I'll never work again, and I'll be sitting on some beach painting pictures and not having any kind of objective. Or if I do want to have a goal in life, I can still have my goal in life. It just won't really matter compared to what the AI systems do. Okay. Not as not either way, right? It's kind of like somewhere in the middle. It's like we're hitting the industrial revolution, but we're hitting like this. It's going to like hit and transition at such a fast pace. So in my world, for example, there are people who are still using processes that were like, you know, 2025 processes. They're getting left behind. They're building systems that are built upon without building the underpinning of AI, and what they're doing is they're marginalizing the power and scope of AI in some ways, and they don't fully understand it in others. So this is where you need to look at artificial intelligence, keep it in its proper context, and make sure that you understand that these parameters, when used with artificial intelligence, are going to absolutely change your world. But when you look at someone who said he spent two years trying to look at artificial this AI doc, it I'll be honest with you, I I didn't like it. Um because it was like sensationalism, sensationalism, because sensationalism sells. And he just looks like he's just all in pieces because he's having this kid and all these things are going on and he's just worried to death about all this stuff. Okay. It's not going to be AI that's going to destroy the world. It's going to be the same people behind the buttons, just a different weapon. Okay. So this gets into larger philosophical questions, because, you know, if you put a philosophical underpinning on AI, if you give to it and say find the best ways to cheat, that's one thing. You know, if you tell it uh to operate on an ethical and honest level and to try to be altruistic in what you do, guess what? That's a whole other level. If you use it to destroy, then that's one thing. If you use it for good, that's another thing. So this gets into your philosophical and theological backgrounds. What is ethics? What are the things that AI should be told are valuable versus what's not valuable? Um, all interesting questions. And so I guess what that would do is actually put it back in the hands of the people that make it. And so what happens is it's now endowed with the philosophical and ethical predisposition of its creator if it's baked into the recipe. Now that is a scary thought with all those different views floating around the world today, isn't it? So if you want to stay up at night and you want to worry about something, worry about that one. Just kidding, guys. All right, so I thought I would set the record straight and kind of let you know where AI is, you know, where it will be in six months. Yeah, you got a point. Who knows? We'll see where it's at. Uh, but you know, being trying to define human sentiency and consciousness and what live life is, you know, things like that. I think we should probably figure it out for ourselves so that at least we have a measuring stick so that we could at least entertain the idea is if it's just doing a really good job of pretending to be sentient or fooling us to be sentient, or is it truly sentient? You see, that's the weird thing. You can't tell. You can never tell. All right, guys, that's all I got for today. Um thanks so much for tuning in. I know I was a little more serious in today's and had a little more rhythm to it, but I wanted to stay focused because so many people are really wrestling with this idea of artificial intelligence. Anyway, hope this helps you out, gives you direction. Um Claude, by the way, I want to let you know, I'm using Claude these days. And if you've not used it, you should give it a shot. All right, guys. Talk to you next time. You have a great week, and thank you so much for tuning in. And if you could, if you do think that this is going to help somebody else out, please share it. I mean, we produce this podcast, and uh honestly, it's just to kind of get everything out there because there's so much noise in the marketing world and the design world. I'm doing this because I really want you to be able to have underpinnings to understand what's going on. Because the ripples that are going on in the creative and marketing world are insane. You have people that are like if you go back, I had one where I said uh 2025 might as well be 1825. I mean, no joke. If you're using 25 techniques in today's marketing and creative world, um, you might as well be working 100 years ago with old techniques. And if you're using uh AI in bad ways that makes you look like you're doing more work, then you've got a whole other subset of problems. Staying in your lane in marketing and creativity to help someone run a business is insanely something that you you have to be disciplined and you have to do your homework and study on. So that's what this podcast is about. So uh hopefully it'll help you out. I hope that it helps out some of your friends and and others as well. So uh until next time, this is Ken with Over the Bull, and I am done rambling for this week.

SPEAKER_00 32:10

Thanks for tuning in to Over the Bull, brought to you by Integris Design, a full service design and marketing agency out of Asheville, North Carolina. Until next time.