Why True AGI is a Century Away and What Comes Next

[00:00:00] John Thompson: Most people — like my sister, your mom, whoever we want to talk about — they’re not in our ecosystem. They don’t understand that it’s not a model. Most people are thinking, “Oh, I’m prompting a model.” Well, you’re not anymore. You’re prompting an orchestrator, which is prompting an ensemble of models behind it. That’s how we’re going to go forward. That’s how we’re actually going to get there. If you look at many of these people — Kurzweil and all these other folks, Bostrom — they’ve all come off the AGI track. Now they’re talking about superintelligence. We’re going to have superintelligence in these domains, or in these small areas, or something like that.

[00:00:36] Scott Kinka: Technology is changing faster than ever, but the most important conversations aren’t about the tools. They’re about the ideas and the people shaping what comes next. On the Bridgecast, I sit down with the leaders driving change, from AI and cybersecurity to the infrastructure powering modern business. Together, we explore how technology moves business forward — what they’re seeing, what they’re building, and what it means for the rest of us. I’m Scott Kinka, and this is the Bridgecast, presented by Bridgepointe Technologies.

Hi, and welcome back to another episode of the Bridgecast. I’m super excited that today we’re joined by John K. Thompson. He’s an AI and data leader, author, educator, and consultant. John has led AI and analytics groups at major organizations, and he now teaches AI entrepreneurship and product management as an adjunct professor at the University of Michigan. I think, super importantly, he’s also the author of several influential books — one of which I’ve tattered up here, called *The Path to AGI*. He’s also written *Data for All* and *Causal Artificial Intelligence*, which is a topic we’re going to get into today. Across that work, he challenges hype-driven thinking and argues for a more disciplined AI roadmap built on strong data foundations, causal understanding, and enterprise-ready systems that solve actual real-world problems. John, welcome to the Bridgecast.

[00:02:06] John Thompson: Thanks, Scott. So excited to be here. This really — I have been waiting for you to invite me on. I was actually going to reach out and say, “Hey, dude, where’s my invitation?” But finally it came. So, patience was paid off.

[00:02:20] Scott Kinka: Can you just give us a quick and dirty rundown? How did you get started in this business? Where did you land? I mean, some of it’s covered in the book, but you can spoil it a little bit right now.

[00:02:28] John Thompson: I’ve been doing this for thirty-eight years. I originally was an auto mechanic — I was never going to go to college. I left high school and worked for a year as an auto mechanic, and then I thought, “There’s got to be a different way to go.” So I ended up doing an undergraduate degree in computer science and then ended up in Chicago. It just seemed obvious to me — I think it was 1988 — that everything was going to be about data. Just everything was going to be about data. I started talking to people, and they all just looked at me like I was nuts. Like, “No, it’s the code. It’s the programming. The data’s nothing. The data’s garbage. Data’s everywhere — it’s useless.” And I’m like, “I don’t think that’s the way it’s going to go.” So then over the next thirty-plus years, I’ve just been working with data, analytics, and intelligence. I got involved in AI in about 1991 or so — IBM wrote their first neural network utility, and I was working at IBM, and I got my hands on it. From then on, it was just AI and intelligence all the time.

[00:03:30] Scott Kinka: You spent four decades — I think you said something along those lines — in data and analytics, and ultimately that leads you to AI. So let’s take that vantage point for a moment. What’s different about this AI moment? And I don’t mean just specific to AI, but this moment in general — versus every previous wave. We had cloud, we had the internet, we had networking. Is this just the next wave, or is there something intrinsically — humanity-altering, maybe — about this moment? How is it different?

[00:03:59] John Thompson: Well, I think it’s different because we’ve gotten far enough in the seventy years we’ve been working on AI — with predictive AI, neural networks for the most part, symbolic AI, neurosymbolic AI — that it’s all come together in a way that it is now ensconced in our infrastructure. Enterprise corporations have lots of predictions for marketing and sales, and Amazon has a ton of predictions on what we should buy next, and all that other kind of stuff. Predictions are part of our life and part of our technology. And now, lay on top of that generative AI, which has been here for about three years or a little bit more than that. Soon we’re going to have — well, we have causal AI, but soon we’re going to have physical AI. So we’re going to add another pillar to the AI world.

Up to now, it’s always been touch and go. Is AI going to fall off? Are people going to use it? Is it trustworthy? Can we make it enterprise-scale and responsible and ethical and all that kind of stuff? And the answer’s been kind of, “I don’t know, maybe” — but the answer is emphatically yes now. We are past the point of going back. There is no going back. I still get, oddly enough, questions from executives. Look, I teach at Michigan. I’ve talked to over 500 companies in the last four years about AI, and invariably someone will grab me by the arm, kind of pull me over to a corner, and whisper in my ear, “Can I wait it out? Is this just going to go away? Is it going to fade off into the background?” And I’m like, “No — it’s just like math. It’s like asking when math will end.”

[00:05:44] Scott Kinka: You know, every one of these waves built to the next one. We don’t get AI without cloud, because we have the infrastructure to support it, and we don’t get cloud without the internet. We don’t get the internet without networking. So this is just an accumulation, I guess, of where we were headed?

[00:05:57] John Thompson: Yeah, it’s just stacking on top of each other. There’s no doubt about it. We talk about it in the book — I talk about the AI winters, and those were all based on overselling what symbolic AI could do. We’re past that now.

[00:06:11] Scott Kinka: I want to explore a little bit of that history in a moment, but let’s park that for one second. I want to go back and reorient around when you say, “seventy years” in the book — I think people assume this is a three-year-old technology, and I do want to walk that out. But before we jump into the history, let’s get controversial for a second. I’m sure it’s the first question everybody asks you, and I know you do a lot of interviews — you did one yesterday. You argue that true AGI is likely more than a century away, and I think many of your peers don’t agree, or at least are thinking about it differently — maybe their definition is different. So, A: back that up. But, B: while you’re doing that, why do you believe the industry has become fixated on something that, in your mind, is so distant?

[00:06:54] John Thompson: Let’s tackle the definitional part of that question first. Elon Musk, Sam Altman, Ray Kurzweil, Rodney Brooks, and I are all out there talking about artificial general intelligence, and the one thing we agreed on — maybe two years ago, I haven’t asked those gentlemen what their definition is now — but at one time, the definition was: when AI acts and interacts with you in the same way that an above-average undergraduate student does, we have AGI. Pretty simple definition — empathy, sympathy, understanding, memory, awareness, thoughts, all that kind of stuff. I don’t see any AI engines doing that. I teach undergraduates, so I’m in the room with them all the time, and I don’t see any AI engine acting that way. Now, that was the unified definition that all the people I cited, and myself, agreed to. So when you start tearing that apart and asking, “Okay, when are we going to get sympathy? When are we going to get empathy? When are we going to get novel thoughts? When are we going to get sentience? When are we going to get psychological awareness?” — I don’t see that happening in another hundred years. I mean, at least a hundred years.

[00:08:15] Scott Kinka: How would we even know sentience at the end of the day, John? The systems are so good at mimicking — and I guess that’s a lot of, frankly, what you’re talking about as we get through these stages, which we’ll get to in a minute. We’ve been talking about the Turing test for decades now. Would we even — let me park the question for a second. I talked to my mom, and she’s like, “You talk about these AI things, I don’t know what you’re talking about. I just know I saw this video of the robot saying that the way to save the Earth is to kill the humans.” And I’m like, “Well, Mom, it was informed on every movie script we wrote over the last forty years. That’s just mimicry.” And that’s an extreme example. But give me some thoughts around sentience — would we even know when AGI happens? There’s not going to be some moment where it’s like, “Hey, I woke up, happy birthday.” So when will we know?

[00:09:02] John Thompson: It will be difficult to know, that’s for sure — because AI probably won’t tell you that it’s sentient. There’ll be all sorts of tests. But the thing I talk to people about now is, okay, you’re working with Claude, Claude Code, Mistral’s models, OpenAI, whoever — and most of the time, you end a session, you go out and have a coffee or go for a run, and when you come back and start another session, it doesn’t even remember you. So memory alone would be a good thing. I know Apple’s working on this — they’re trying to fix it. But how many times does Siri get you to the right Starbucks? So we’ve got to work towards this. We’ve got to work towards memory. We have to work towards ideas, concepts, and higher-level views.

Most people — like my sister, your mom, whoever we want to talk about — they’re not in our ecosystem. They don’t understand that it’s not a model. Most people are thinking, “Oh, I’m prompting a model.” Well, you’re not anymore — you’re prompting an orchestrator, which is prompting an ensemble of models behind it. That’s how we’re going to go forward. That’s how we’re actually going to get there. If you look at many of these people — Kurzweil and all these other folks, Bostrom — they’ve all come off the AGI track. Now they’re talking about superintelligence — we’re going to have superintelligence in these domains, or in these small areas, or something like that. I still don’t like the tag, and I still don’t think we’re very close to getting there, but I like the idea, because that’s how we’re going to get there. I’m an AI optimist — I think we’re going to do all the things we say we’re going to do. It’s going to take a lot longer than most people believe, but we are going to end up with a mixture of experts, which is just a technical term for a bunch of models behind our orchestrator that are focused on trigonometry, calculus, algebra, mathematics, biology, string theory, physics, economics — getting your kid into the best college, you name it, we’re going to have models for it.

So the idea that most people have in their head — that AI is a straight line, that we started at Dartmouth in 1956, and here we are seventy years later in 2026, and it’s been a straight arrow from one point to another — that’s not the case. It’s been a very winding ride. Marvin Minsky told a PhD student, “Figure out computer vision in the span of a semester.” Well, he didn’t get it done. This idea that AGI, or artificial general intelligence, runs through large language models or runs through GenAI, is completely misinformed — because GenAI is nothing more than a large neural network that we jammed the entire internet through, and it matches patterns. Do you remember, like, two and a half years ago, we were talking about emergent behaviors? People were like, “Oh my gosh, it’s doing interesting things. The neural networks are coming alive, they’re waking up, they’re having new ideas.” And I’m like, no — what it’s doing is finding dark alleys you’ve never gone down and illuminating them. It’s just matching patterns and giving you patterns. Someone called it a “stochastic parrot,” which I think is probably a good description of it.

[00:12:30] Scott Kinka: I want to move my personal conversational assistant to, like, Claude or Gemini, and I can’t get off my original ChatGPT model — because it agrees with me all the time, and I’m so smart, after all the time I’ve worked with it. By the way, I said to you the book’s going to end on the hall of fame — I talk about it here on the thing. I just figured out where you’re going to go, because I have Ray Kurzweil’s book, *The Singularity Is Near*, back here. I’m just going to slide you right next to him, since you two know each other, and that book’s come up a handful of times on this show.

I want to get into the history, and then I want to get into the disciplines you talk about — foundational, generative, causal — because I think it gets us to your endpoint, the way you think about it, which I think is novel. You probably don’t feel like your approach to it is novel, you know what I mean, but you’re talking about it in a way that’s accessible to most people. And that’s no knock on your friend Ray, but that’s some heady stuff when you read it — you’re making it very practical, so I want to dwell there and probably spend most of the rest of our time together on that.

But before we get there, I have one question I wanted to come back to, because you were talking about executives earlier and this rush to do something — “Am I going to be left behind?” If you’re an executive right now, whether that’s on the IT side, or worse yet you’re a CEO or a board director pointing at your CIO and saying, “Go do something, so I have some fun story to tell our shareholders” — if you’re that person who’s worried you’re not moving quickly enough, how should you balance the pressure to innovate with the discipline required to invest in capabilities that are actually going to deliver something to the business? What’s your simple advice when somebody sits down over a cocktail — not in a boardroom — about how to balance pressure versus outcome?

[00:14:19] John Thompson: I always tell people — whether they’re on the board, the CEO, the chairman of the board, CIO, CTO, whatever they are — I tailor my advice to their situation, but the theme is generally: you’re not behind. This is not a sprint, this is a marathon. This will go on for a long time. Now, I understand your sixteen-week quarterly pressures, I get that, I understand what you’ve got to do. So, start out by enabling your entire organization to be AI literate. That’s what I tell them to do — get these tools in the hands of everybody, and don’t worry about what they’re doing. A lot of people say, “I’m going to give you AI, but you can’t do anything with it but work.” Well, now you’ve just cut off all their motivation and interest in being engaged.

When we built one of these systems at a very large corporation, we told people, “We’re not watching what you’re doing” — because we weren’t — “and we don’t care what you do.” Now, we did spot checks, and I saw there were places where people were writing sonnets to their puppies and planning their three-day vacation to Vermont, and things like that. And that’s fine — do it, understand how this stuff works. So, number one: get everybody involved. Number two: start looking at your processes, because that’s really where the friction is. If you don’t reorganize your work processes around agentic workflows and AI capabilities, you’re never going to really realize the value. That’s been the problem we’ve had for the last year — boards and C-level executives were sold on AI for personal productivity, but they were expecting strategic change.

[00:16:01] Scott Kinka: I walk into businesses every day in my current role where the first thing we talk about is, “Alright, great, I love the idea of that agent — what are we going to train it on?” And they’re like, “Well, we train people over the shoulder.” I mean, you’re not going to inform it off of call recordings, right? That’s where you’ve got to bridge that gap, often — I think that’s where businesses struggle. Can we jump into the history? We don’t need to go super long given where we’re at, but most people, when you make a statement like “we’ve been at this for seventy years,” probably raise an eyebrow. We don’t need to spend five minutes on it, but walk us through the beginning, where you think we are today, and maybe a little bit about what that looks like in the decades to come.

[00:16:46] John Thompson: If you’re listening and you’re interested, and you have a browser, go search for the Dartmouth Summer Symposium — that’s where the phrase “artificial intelligence” was coined. It was Marvin Minsky, Vannevar Bush, John McCarthy, and all these computer science whizzes coming together to define what artificial intelligence should be. They were supposed to have a summer-long symposium arguing it all out, but it didn’t transpire that way. What you should read is the document that came out of it — it’s the seminal definition. I hate to use the word “bible” because it’s so loaded, but it is the tabula rasa of AI. It defines what AI should be, and it took us seventy years to get there. So if you really want to know where it started, that’s it.

Then we jumped forward into the sixties, and it was operational research and all this other kind of stuff — people were trying to figure out how to play chess and checkers and things like that. Then we jump forward into the eighties, and we had, as we talked about, the AI winters — it was all about symbolic reasoning, rules engines, and things like that. Those are great — I love symbolic AI, I think it’s a fabulous technology, but we overloaded it back then. And then it didn’t deliver — it couldn’t deliver. It was impossible for rules-based systems to deliver on what we said it would do. So investment dried up, research stopped, all this kind of stuff. Then three guys ended up in Canada — Hinton, LeCun, and Bengio — and they really drove the whole idea of neural networks forward. I was at IBM at the time, and I was working on neural networks as well. So neural networks have really taken us from the eighties to where we are now, and generative AI is really just another implementation of a neural network. But as I said earlier, we jammed the entire internet through it. Now people seem to think that’s magic — no, that’s just scale. And then the compute power, the memory, and the interconnected nature — like you said, it all stacks on top of each other. Without these substrates, this couldn’t happen.

So here we are, and like I said earlier, we’re at the mixture-of-experts stage. It’s not you prompting Opus — you’re prompting an orchestrator, which is then tearing your prompt apart, sending it to a bunch of models, which then come back. The orchestrator reassembles it and sends you an answer. That’s the way the world is going to go for about the next thirty years — the mixture-of-experts moment is going to last a few decades. Then what we’re going to see is, like you said, predictive AI or foundational AI, generative AI, and causal AI all getting integrated together. We’re going to have composite AI in about forty years from now, and then we’re going to start seeing physical AI come into it — maybe even a little sooner — and then we’re going to work on having all of these different engines working together behind the scenes. So when that recedes into the background and those engines are all integrated together, then we have the promise of AGI.

[00:20:06] Scott Kinka: You are listening to the Bridgecast, proudly sponsored by Bridgepointe Technologies. When it comes to customer experience, Bridgepointe delivers innovative solutions that empower your business to stay ahead. Visit bridgepointetechnologies.com to learn how we can transform your customer experience today.

I love the way you look at this, because you’re right — you can’t get here otherwise. We think of this like a seminal moment right now where we’ve moved to prompt engineering and it can just do stuff. But the reality is you did have this decades-long, rules-based moment, which we didn’t consider intelligence because we had to give it the instructions — it still took the rules and made correlations and did things. Is it fair — I’m saying this from your lens — is it fair of me to say that when you go from rules-based to the generative model we’re in right now, the difference is that we’re prompting versus having to build the sub-rule sets, largely speaking? And now we can at least get there without creating a bunch of rules-based choices. We still have to prompt it, and it still has to ultimately hit a rule set that hits a set of data that creates correlations. So you’ve got to feed it something to get something out of it — that’s ultimately the difference in terms of us getting to general.

[00:21:26] John Thompson: I don’t know if you’ve heard of Gary Marcus or not — Gary is the godfather of symbolic AI, and he’s always out there picking at Yoshua Bengio and Yann LeCun, saying, “Neural networks aren’t everything, symbolic AI is great.” He put out a post the other day that someone had found that Claude Code — one of its back-end processes is actually a symbolic rules engine — and he was like, “See, Gen AI isn’t everything.” And I’m like, “Gary, relax, man.” The whole world is a stack of models, a series of models — it’s not just one model. Anybody who thinks we’re going to get to where we need to go on one architecture, or one model, is sadly mistaken.

[00:22:11] Scott Kinka: I don’t even know that we have solid, human-based definitions of original thought. There’s this giant debate in music right now — you’ve got Suno and these derivative engines, and musicians are saying, “Well, you’re not musicians.” I get it, but listen — if you write a rock song right now, you’re doing it because Paul McCartney was the best bass player in rock ever, and everybody plays like him now, and because we’ve got a four-on-the-floor trap-set beat, and that’s been happening forever. Everything’s ultimately a derivative of something, including human language. I think that’s going to make it harder for us to get to that point where we define sentience too — because we only have our current patterns to make the next best definition, so everything’s going to be derivative forever, really, at the end of the day.

[00:22:59] John Thompson: It’s very interesting — people try to talk about what is original and what is derivative, but, like you said, pretty much everything’s derivative. But there’s been a really interesting resurgence in philosophy majors. Computer science has had a tough year too, because of all the “we’re never going to need another developer” talk, which is a silly position. But they are hiring people who talk about metaphysics now — what is real and what is not real, what is original — all this other kind of stuff. So there’s not going to be an end of developers, and there’s not going to be an end of entry-level jobs coming out of colleges. But I am heartened by the fact that they’re hiring more philosophers.

[00:23:44] Scott Kinka: I can’t dismount from your seventy-year history before I say — I often, in these kinds of talks, will go further back and note that we didn’t have the tech yet, but Asimov was writing about this two decades before this time. I’m waiting to be able to apply the three rules — that’s going to happen, I don’t know, by your point maybe in another hundred years we’re going to be worried about doing no harm.

I wanted to nail down two things, and then I think we can jump into a little bit of fun. You’ve gone foundational, generative — I think people get that. Can you define causal, and then ultimately composite, a little bit better? If you were to do it in a couple of sentences — give me causal, and then give me composite, in a bit more detail.

[00:24:30] John Thompson: Causal is fantastic — I wrote a book on causal, as you mentioned in the intro. Probably two and a half — no, three years ago now. It was before Gen AI exploded, because I was thinking, “Oh man, causal’s next” — and I was wrong about that. So I’m wrong about a lot of things. But anyway, causal is a really interesting branch of calculus. Judea Pearl wrote *The Book of Why* — he’s a professor at UCLA, been working on this pretty much his whole life. He created a new branch of calculus called do-calculus, so you can go in and define an environment.

One of the great things about GenAI is that causal AI was held back because it was hard to formally model causality. Now Gen AI has ingested all the information about causality that we know in the world, so you can use Gen AI to create what are called directed acyclic graphs, which is what causal AI is based on. You can use GenAI to generate causal graphs that you understand, and then ingest them into a causal platform and model it — A causes B, but it’s confounded by C, and it’s influenced by D. So it’s really a branch of mathematics we use to model something quantitatively that we understand intuitively, but it’s hard for people to get a grasp on it from a mathematical perspective.

[00:26:03] Scott Kinka: In a lot of ways, it’s the generative-AI-developed version of the rules-based engine you were talking about, right? We wouldn’t get causal without generative in the middle to help build effectively correlations past what we can process individually. Is that a fair way of putting it?

[00:26:20] John Thompson: That’s a fair way to put it. Absolutely.

[00:26:22] Scott Kinka: And then you’re talking about getting to composite before we get to general. Explain composite.

[00:26:26] John Thompson: Now we’re starting to see overlaps — smart developers are doing things with open-source platforms and proprietary platforms. Like I said, we see people taking GenAI models and generating causal graphs, or DAGs, and then driving causal AI. That’s hand-knitted, hand-rolled — rolling your own, or whatever you want to call it. But what we’re going to see over the next thirty years is that companies like Mistral will bring causal engines into their GenAI platforms, and other companies, like Tibco, will bring in their predictive engine and marry it with causal, or marry it with generative. So we’ll see different engines being integrated together by professional software developers. Right now it’s just roll-your-own, but over the next thirty years, composite AI is going to be a professional integration of causal, predictive, generative, and physical.

[00:27:32] Scott Kinka: At some point in that window, the lines blur and we don’t really know what “general” actually is. But if you just follow the source — the thing just keeps getting deeper and deeper — I don’t know if there’s really a “happy birthday” moment. I don’t know why we

[00:27:49] John Thompson: want it so bad. We shouldn’t want it so bad.

[00:27:52] Scott Kinka: Ultimately, this whole thing started with rules. If we could get this really working to the point where it feels sentient, but it’s still built on the rules base, then we have what we want. I don’t know that we really want to get to general, at the end of the day.

Let me hit you with two fun ones, and then we’ll get into a couple more interesting questions. There were a few things I really enjoyed at the beginning of the book. First off, I’m going to give you a crazy recommendation on this, because we’re talking about some of the tomes around AI — and I’ll just say to our listeners, our CIOs and directors of IT who are dealing with practical things day to day: John does not mess around. By page eleven, he’s in step one on what you should be doing with your data. So there’s a lot of theory in here, but it’s also a really good roadmap, which I enjoy — and we’ll get to that at the end, chatting a little bit more about it.

But getting to practicality — you mentioned, really early on, maybe page six, “Never propose, build, or support tech for a manager or a company where the technology you’re building isn’t at the core of the organizational mission.” And you were saying that while bemoaning your life, fixing some eighteen-hour networking challenge. But put those two things together — there’s somebody listening to this who’s probably enjoying this in the morning after fixing an eighteen-hour networking problem. You know, isn’t the network — and maybe you mean this — but isn’t the network still core to the organizational mission, or at least the plumbing on which the house has to be built?

[00:29:34] John Thompson: Every company needs a technology infrastructure — databases, DBAs, network administrators, all that kind of stuff — and I think that’s fine. My career has always driven me to be in a technology organization, and an organization that’s driving innovation through technology. And through that, my career has ended up at different companies that are not technology companies, and what I’ve always found is that my personal mission to be involved in leading-edge technology and driving innovation is always thwarted by someone who cares more about the trains running on time — which I’m not that kind of guy. I’m an innovator, I’m a driver, I’m a technologist. I’m not an operator. So if you work in a company like GATX, which runs the US’s largest network of train cars, it’s important to get the train cars in the right place at the right time, but personally, I don’t care too much about that.

[00:30:36] Scott Kinka: You’re in front of the next generation of tech leaders right now at the University of Michigan, and you’re talking about AI entrepreneurship. What are the questions they’re asking you about AI that the current generation of enterprise leaders you consult with and work with and talk to aren’t asking? What can we learn from their curiosity that maybe we don’t have?

[00:30:58] John Thompson: They’re fantastic. I teach mostly graduate students, but every now and then undergraduates slip into my class — they get special approval or dispensation to be in there. It’s great — the undergraduates will say things that are just ludicrous, but they say them because they’re young, and that’s fine, and I applaud them for it. The first thing they always ask me, almost every semester, is: when is the apocalypse coming? When are we all going to die? When is AI going to take over, and Arnold is going to say, “I’ll be back,” and all that kind of stuff? And I’m like, “It’s not happening, don’t worry about that, it’s all fine and good.”

The graduate students are more aligned with these themes: “Okay, we have all this information — how do we harness it in a way that we can disrupt different industries? How can we make sure that we’re doing things…” You know, we have a billion-dollar company with two employees and 6,000 agents — so they’re really interested in understanding how to scale this to global levels and have some level of control. They don’t really care too much about control, because they’re young and they’re looking for fast, fast, fast, but they do have some notion of control in there. So generally, they’re looking for ways to use these technologies to disrupt the existing world order.

[00:32:21] Scott Kinka: Fair to say they’re less enamored with the tech and more enamored with the outcomes, maybe, than some of the executive-level conversations you’re having?

[00:32:28] John Thompson: I’d probably change the word from “enamored” to “focused.” They’re like, “So what — the tech is there, we can use it, it’s something we’ve had forever. How do we actually turn it into a competitive edge for us?”

[00:32:40] Scott Kinka: I think I know the answer to this question already, but I’m going to ask it so we have it on record. What is your most contrarian opinion about the current state of enterprise AI?

[00:32:50] John Thompson: That Gen AI is going to be the panacea that solves all problems.

[00:32:54] Scott Kinka: Besides *The Path to AGI* — and I know you probably don’t read your own work, or Ray’s work — what’s on your nightstand right now? It doesn’t have to be tech. What are you reading?

[00:33:02] John Thompson: I’ve been going back to it — it’s right behind me here — *Man’s Search for Meaning.*

[00:33:09] Scott Kinka: It’s interesting that a scientist and thought leader around AI is reading that as well. I think it says a lot about the way you’re thinking about the next stages here. I love that.

Tell me — okay, this is a fun question, one I’ve asked in every single episode we’ve done, and this is, like, episode 106 or something along those lines. We get to the end of the world your students were talking about — you can pick whichever dystopian future you want — but there’s one app that’s still working on your phone. If you could choose it, which one would it be?

[00:33:42] John Thompson: Spotify. Why? I like music — it soothes me, it gives me comfort, it allows me to while away the hours and my thoughts. I don’t really need to connect with other people. And if it’s a dystopian future, then I kind of want to be in my own little bubble.

[00:33:59] Scott Kinka: We’re going to do two segments — some snippable moments. Here’s what we’re going to do: I’m going to ask you some questions, we may have already hit on them, but that’s okay. What I’m challenging you to do is answer them in a sentence, without extra clarification. Alright — harder than it sounds, maybe. Okay: what do most leaders misunderstand about artificial general intelligence?

[00:34:22] John Thompson: They get caught up in the fancy idea of it, not the practicality of it.

[00:34:27] Scott Kinka: What is causal AI?

[00:34:29] John Thompson: It is a way to understand the world we live in, as opposed to the world we think we live in.

[00:34:34] Scott Kinka: In ten years, what will business leaders wish they had done differently today?

[00:34:40] John Thompson: Gotten started sooner with AI, rather than worrying about preparing for AI.

[00:34:44] Scott Kinka: Define AI literacy in a business for me.

[00:34:47] John Thompson: The ability for every employee to sit in front of a prompt window and drive AI in a meaningful way.

[00:34:54] Scott Kinka: I like it. Okay, now I’m going to put you on the spot even more — new segment, we’re calling it This or That. Some fun ones, and then we’ll get back into tech. The World Cup or American football?

[00:35:03] John Thompson: American football. I’m from Chicago.

[00:35:06] Scott Kinka: Podcast or audiobook?

[00:35:07] John Thompson: Podcast.

[00:35:08] Scott Kinka: University classroom or corporate boardroom?

[00:35:10] John Thompson: Classroom, every day.

[00:35:12] Scott Kinka: ChatGPT or Gemini?

[00:35:13] John Thompson: Gemini.

[00:35:14] Scott Kinka: Build it or buy it?

[00:35:15] John Thompson: Depends on what you need.

[00:35:16] Scott Kinka: Okay, fair. Enterprise software inside a business that’s not a tech company — your CRM, your ERP, whatever — buy it or build it?

[00:35:24] John Thompson: Buy.

[00:35:24] Scott Kinka: Correlation or causation? You can’t answer, and I already know the answer — I just wanted to have some fun with it. Alright, last one, here we go — I’m not even asking you to answer this one: an impressive AI demo, or a boring agentic AI that actually does something and produces ROI inside the business?

[00:35:44] John Thompson: The latter.

[00:35:45] Scott Kinka: I love it. John, for listeners who want to learn more about your work, your books, your teaching — what’s the best way to find out more or connect with you?

[00:35:52] John Thompson: Amazon is where all my books are — you can see them behind me on the shelf. The white one next to *The Path to AGI* is the Chinese version of *The Path to AGI*. My books have been translated into Chinese, Polish, and Spanish, and I think some into French too — so, all over the world. Amazon for the books. If you want to see what I talk about, LinkedIn — go connect with me there. And if you want to understand what I do in totality, I have a website called The World of Analytics, all smashed together, and it’s all there.

[00:36:25] Scott Kinka: Again, I’m going to give a pitch for *The Path to AGI*. John was kind enough to send me a copy — I started on it over the weekend, about a third of the way through, and you can see it’s tattered. I’ve read it on the beach, I got it wet, and I’ll also say it’s pretty good. It’s 400 pages, so it’s great for swatting away aerial rats on the beach too — great job there.

Okay, last one, John, and then I’ll let you go — I can’t believe we covered as much as we did, as quickly as we did. If you had to summarize your leadership philosophy in one sentence, what would it be?

[00:36:56] John Thompson: Laissez-faire.

[00:36:57] Scott Kinka: Okay, I’ll let you clarify that.

[00:36:59] John Thompson: I tend to hire people who are highly independent, very intelligent, and self-driven. So I give them a lot of rope, I give them direction, I swat away the aerial rats and give them air cover. Other than that, I let them run — it’s pretty much hands-off, and you get some brilliant discoveries and brilliant innovations that way. I’m just about as far away from a micromanager as you could get.

[00:37:30] Scott Kinka: I think my biggest takeaway from this conversation is that for somebody who spends so much time thinking about AI and artificial minds, you have a keen focus on the human mind, which is the basis of all of it — which I love. Well, thank you so much for the time, John. This was one of my favorite episodes ever — and we’ve done a hundred — so, great stuff.

[00:37:50] John Thompson: I love it. I really, truly enjoyed it. Hopefully we’ll be able to do it again soon.

[00:37:54] Scott Kinka: Looking forward to that. And that is a wrap on this episode of the Bridgecast. If today’s conversation gave you something to think about, please hit subscribe, leave us a review, share it with your team. More importantly, don’t miss a future episode on your drive, or on the treadmill — we will catch you soon on another episode of the Bridgecast. Thanks so much.

Listen, if you made it this far, there must have been some good content, or you’re just a fan or a relative of mine. Either way, we appreciate you spending your most valuable asset with us — your time. We don’t take that lightly. We’d also like to say, one more time, that we’re appreciative of Bridgepointe Technologies and their belief in and sponsorship of this show. If you, or someone you know, is thinking about your company’s digital transformation, or simply the next IT project you may not have the resources, budget, or time to get to, I’m sure that Bridgepointe — one of the country’s fastest-growing technology advisory and procurement firms — can help. Check out bridgepointetechnologies.com — don’t forget the “e” on Bridgepointe — or simply reach out to me at skinka@bpt3.net.

Also, take a minute, please, if you would, to give us a five-star review on your favorite platform — it helps give us the visibility to reach other people like you. Thanks for listening to this episode of the Bridgecast. I’m Scott Kinka, and until next time: there’s a lot of noise out there in business and in life. Do what you can to be the signal.