← Digital Disruption with Geoff Nielson
Digital Disruption with Geoff Nielson
What AI Can Never Understand: Futurist Jonathan Brill Explains
00:59:13
Loading summary
Transcript102 lines
- [00:00]
Jeff
Hey, everyone. I'm so excited today to be sitting down with business futurist Jonathan Brill. Forbes has named him the number one futurist in the world. And he's been a futurist for organizations like hp. He's been a board advisor for Frost and Sullivan, and a strategic advisor from everyone from Amazon to IBM to Samsung to Pepsi. I want to ask him what organizations need to do to rewire themselves to be more resilient in the face of the coming wave of disruptions. Whether it's AI or even disruptions that we can't predict, it should be a great convers. So let's jump right into it. Jonathan, thanks so much for being here. I wanted to jump right into it and I'm really interested in getting your perspective and just what you see as being the current state of technology, especially in an enterprise sense, in 2025. And I guess maybe I'll preface that by saying one of the things I'm seeing from my perspective and just here hearing, is that the world we're seeing right now in 2025, in this space, is just fundamentally different than we were seeing even in 2023, 2022. And I wanted to get your perspective on that. Do you agree? And what do you see as kind of being the big differences?
- [01:14]
Jonathan Brill
Yeah. So I think obviously AI is the big talk of the town. I think that the real enterprise plays are probably another 18 months out, but that's obviously the big thing. We can talk about what's going to happen, why, what the limits are that cause us to know that the other issues are that CIOs, CTOs are being asked to move beyond the technology stack, to start thinking about how it impacts things like governance, how it impacts things like go to market. And so technology isn't just about technology anymore, it's about the entire organization, how we structure our firms. The third thing that's, I think, going on and it hasn't quite happened yet for most firms, but we're starting to see AI enable people who don't know how to code, do conversational coding. Right. Talk to an interface that writes code for us. It doesn't really work yet. Yet. Right. But what we're going to see over the next 12, 18 months is that it does. And so much like the early 2000s when we went from having people with master's degrees writing HTML to all of a sudden Blogger, and Now Squarespace and WordPress, those things that used to require technical support, suddenly don't. That doesn't mean technical support goes away or that it's decreasing. It means that the amount of code that runs our organizations is about to go up geometrically. And that's really exciting to me. So we have complexity with that complexity, we have security issues that are going to go up geometrically too. We have the more strategic nature of IT in organizations and then we have AI really shifting the game and starting to see it for the first time, maybe in 10 years, 15 years, really start to become central and strategic to the growth trajectory of organizations.
- [03:22]
Jeff
So tell me a little bit more, Jonathan, about that, that 12 to 18 month time horizon. Because if I, if I understand the implication of some of the other trends you're seeing, is it that that's how long it's going to take for these organizations to kind of rewire themselves to get to value here that take into account some of these low code, no code trends or expanded role of it? Is that the case? And what is going to separate the winners from the losers here?
- [03:51]
Jonathan Brill
So historically it takes about five years from something to get from research white paper to software business. So we saw the first round of this 2017, we saw the transformer model in a white paper. 2022, 2023, we start seeing that start to build out, right? Lots of people are talking about AI and they've put in stuff in chatbots in their organization. But looking at Salesforce, right, looking at SAP, looking at Oracle, they're starting to put out product. It's still early stage and it'll take that about 18 more months. And if you're an enterprise buyer, you never want to buy the first version of any, right? So that's why I'm saying, okay, well you know, we started building this thing out, there's a strategic priority in putting it in our organizations. These companies are now playing the large players, are now playing catch up. And then in about 18 months there will be enough maturity there for us to start putting this in our organizations in a deep way. You were talking about change, right? So here's the reality, what's going to happen? And I talk to senior leaders, I talk to operational leaders, I talk to executional leaders, and everybody thinks the other ones are going to be impacted.
- [05:17]
Jeff
Interesting.
- [05:18]
Jonathan Brill
So the answer is probably all of the above. But what we look at when there's a recent meta analysis by the National Academy of Sciences led by Erik Brynjolfsson, and it said about 14% of the tasks in professional services will be automated by existing AI tools, I think that's about right. What I'm hearing when I talk to people who are selling this stuff is, hey, we're getting 50 to 70% efficiency bumps for our software developers. And then I talk to my manufacturing clients and they're like, yeah, we're getting 15 or 20% efficiency bumps. Well, that's like, that's nothing that's investable. But we're seeing changes at the scale that will shift the nature of firms into what I'm starting to call an octopus organization. And when you think about the way that organizations are structured today, we built them, the current architecture in the 1850s and 60s based on the railroad, and there were several problems with railroads. One is you needed to get, you know, some box from this place to that place across a couple of different rail lines. And all of this stuff needed to be synchronized. And the tools that you had to do to do that were a pocket watch and a telegraph that went one direction. Right. So you didn't have telephones, Right. You didn't have bidirectional. Right. And when you take a look at Thomas Edison, right, His big breakthrough was, was the bidirectional telegraph. The reason that he invented that was he was involved early on in his career working in a telegraph station. And because someone's pocket watch was off by four minutes, two trains, I believe collided, there was a train accident of some sort. And so he said, okay, this, I see an opportunity. And the kind of multiplexing is what was, was a result, was invented as a result of that. So my point being, we in, we, we had that going on and we also had, you know, high levels of, of illiteracy, you know, in like almost no one going to college and no one having good decision making skills. Right. So that's why we built the organization like it is today. You had to have a leader who didn't know what was going on on the ground and people on the ground who weren't capable of making good decisions and had no context for what was going on up here. When we think about what's going on today, large language models are sucking up all of the information in the world. All of us have OTTER or a transcription tool, hopefully in our businesses. As a result, we can build models and we can build agents that are looking at all of the new information every day and giving it to exactly the right people across a 50,000 person organization. So that context problem goes away. The second thing, and I think that we're getting where chatbots are going to be useful all wrong, which is we think we're asking these things to answer questions for us and pretty good at it. And the new stuff, like, you know, you look at 01 Pro and you look at the, the new version of Gemini 2.5 that came out like 48 hours ago or something, you know, and, and they're, they're pretty good, you know, compared to two years ago. But what they're all really good at and better than most humans at is empathy, creativity, right? And frameworks. They have every business framework that has ever been developed. They have access to every code library and every, every code pattern that's ever been developed. Right? And so they're going to be better, not necessarily, I believe, eventually at making decisions than humans, but they're going to already be better at helping us make better decisions, at coaching us. And so we're using these things, I think, for the wrong paradigm right now. They can dramatically improve our executive judgment instead of us trying to offload our executive judgment. That's massive, right? Because we have this situation where all of a sudden every person in your organization, every intern in your organization has context of a CEO or better today and executive judgment of a senior leader today that radically shifts the nature of an effective and innovative and agile firm. Because information, knowledge, decisions, governance, doesn't have to come from the top down, can start to come from the bottom up. And that transformation is profound. And it looks much less like the railroad, and it looks more like the neurology of an octopus, where instead of having one big brain, it actually has nine. It has a big brain that deals with executive function, and then for each tentacle, it has a smaller neural cluster that looks at that, looks at that tentacle, directs it, gets all that information, and then a neural necklace that communicates with all of those smaller clusters to figure out what's going on, to come up with an idea to take action faster than the big brain can even comprehend what's going on. And so I think we're moving into this world of what I'm calling the octopus organization. It's a radically different place. And the reason this is important for IT professionals and for CIOs in particular, is all of a sudden we're being asked not just how to rack and stack servers, but. And how to implement some SAP instance. But why? Why are we doing this? How does this impact our human resources? So on and so forth. And as the leaders of code, capability and organizations, we are going to get flattened first and we are going to build the culture first, and we are going to make the change first and culture change independent of what people might say. It takes five to seven years, an entire shift of a leadership team. And so if you are a year or two early in that shift, your potential to be a leader goes through the roof. We're moving into a radically different world, like you said, than 2022 or 2023. And it's not just technology. It's that CIOs are being asked about the architecture of firms, about the processes of firms and having to lead the culture change of firms. And this isn't what we've been trained to do historically, but it's our opportunity to lead now.
- [12:27]
Jeff
Right, right. So lots of, lots of things I want to, that's super interesting and there's a lot of places I want to take that. I love the octopus model. And if I'm hearing you correctly, we need to change, if I can call it this, the physiology of the organization. Right.
- [12:48]
Jonathan Brill
We need to evolve the physiology.
- [12:51]
Jeff
We need to evolve the physiology. And it sounds like that rearchitecting that CIOs that what we previously have called the technology organization, we're going to be at the vanguard of that. Is that necessarily true? Is that a good thing? Is that a bad thing?
- [13:09]
Jonathan Brill
I don't know if it's a good thing or a bad thing. It is necessarily true.
- [13:11]
Jeff
Sure.
- [13:13]
Jonathan Brill
Because you're, you're, the, the, the, the demands on you are going to be far greater than, than, than the number of people you have. So you're going to have to automate first.
- [13:20]
Jeff
Right, right. And, and technology.
- [13:23]
Jonathan Brill
And it's not a bad, like you said, it's not, it's not a bad thing. Coders are freaked out. Like every, every professional services, you know, code integration shop I talked to is freaked out. The amount of code we are going to have to produce and qualify is going to go up geometrically. I am not concerned about the future of coders. I am concerned about the security implications. I'm concerned about the efficiency. I'm concerned about the amount of legacy code we're going to produce and making sure that we're able to either take it out of our organizations or, or figure out how to just not keep it running. Because the energy cost is going to be massive.
- [14:08]
Jeff
Right. One of the things you said a couple of times now is this five to seven year time horizon to really see cultural change and see this take off. I'm sure you're very aware, like one of the buzzwords we're hearing now more than ever is accelerated pace of change. Right. Like oh, things are moving faster than ever. Can large organizations afford five to seven years here? Is there, is there an Existential risk that organizations that are wired from day one with an octopus model are going to come in and, you know, eat their lunch and, you know. Or is that an overblown risk?
- [14:46]
Jonathan Brill
I think it's highly likely. The question is how quickly you can move. So in 2012, I got a call from the Yellow Pages, and the Yellow Pages wanted to build a search engine that they could put local advertising on because they had, you know, it's what they do access to. They, they had 10,000 salespeople across the country that could go and sell advertising to your mother's hairdresser. And they thought this was a really great plan. The only problem was this was a really great plan in 1996 or maybe 2006, but in 2012, it was too late. And so there are two things going on here. One is certainly in 2002, 2003, the yellow pages could have caught up. They could have maybe bought Google or at least and also ran and had the capability they could have and done something they didn't. They continued doing what made the money today. And so that's the bigger risk is that we look at what our investors are asking of us instead of remembering that what our investors really want is to know that our company will be worth more five years from now than today. And so there's this incentive misalignment, often at the C level or the board level, that causes these challenges, that if no one's being paid out 10 years in advance, which is what Tim Cooks, I believe, at Apple, his big payout was, was after 10 years after, after Jobs. Right. Well, there's a reason Apple's done what it's done. Right. Has never been concerned about it this quarter.
- [16:28]
Jeff
Right.
- [16:29]
Jonathan Brill
And that's a rarity. You know, so, so I think it's a big issue for, for large firms. I think the second question is, you know, what moats do you have? Right? So if you are a law firm, guess what, you know, three quarters of the Senate or something are lawyers. They'd really love to have law firms when they leave the law, you know, when they, when they leave Congress, right There, there are these moats that, that companies have that will slow the pace of change. There is, you know, within law, the, the, the, the bar, the American Bar association has said that you can't, you know, that the AI can't provide opinions for customers. They said it's an ethical issue independent of the fact that I, I think at this point it might be almost as good as my lawyer giving me advice. I also want to ask my lawyer. You know, it's just like, it's just like, you know, intuitive surgical. You know, they make the surgical robots for brain surgery. Well, the reality is the robot is doing the brain surgery. The surgeon's there just for the liability. And in case something goes bad, I'm totally glad to pay for that surgeon. Right. May not make sense. Just like I'm totally glad to pay for my financial advisor for the three weeks a decade that he provides Value.
- [18:00]
Jeff
Right. It's plus AI. It's not instead of. It's not.
- [18:04]
Jonathan Brill
Yeah. And so, and so it's, it's really, where's the value? And what we're going to see is the question moves from what are the tasks that I do? To where is the value that I provide? Right, right. So the, one of the things I like to think about is in the last election cycle, presidential election cycle, Andrew Yang was taught going on about how truck drivers were going to go away. And what he forgot was we used to have stagecoach drivers and they did the same thing. Right. It wasn't that they spent 12 hours a day turning a steering wheel instead of a horse. It's that they protected the gear, the people, the material, and back when something went wrong.
- [18:45]
Jeff
Right.
- [18:45]
Jonathan Brill
The value has nothing to do with how you spend your time. And so what we're going to become really clear about in the next five years is where's that value? So as a coder, right. Your value isn't encoding.
- [19:02]
Jeff
Right.
- [19:03]
Jonathan Brill
It's an understanding the unspoken challenges that your customer faces, surfacing those and turning those into code. AI will not necessarily understand the unspoken challenges because guess what, there is by definition, no data.
- [19:25]
Jeff
So again, lots to think about there. But as we think about finding that value, as we think about rethinking our business models in terms of how we get ahead versus trying to play catch up, you started to answer the question of what do we as leaders need to be doing differently? What do organizations need to be thinking about? And as you were answering that, I was thinking about the kind of adjacencies with, with rogue waves. You know, some of the guidance in, in the book you wrote on this a few years ago, Is it the same playbook? Is it a similar playbook? And, and you know, for those who, who haven't gone through the book, you know, what do you kind of recommend are kind of the big elements here?
- [20:10]
Jonathan Brill
Unexpected plug. Yeah. So I, I think there are a couple of things to be thinking about. The first and most important thing is that for any of this technology change to work, there will have to be deep psychological safety on your team. What we know about Transformations is people get freaked out, people don't use the tools. Right. So you spend the money and you get none of the value. So how do we create a situation where we. Where we make sure that people feel safe enough using the tools to create the value? So I was talking to the CTO of a big four consulting firm, and he said, yeah, I had to just make my people use AI said, you know, in your 360 reviews, you know, you're going, you know, in your annual review, you're going to tell me how you used AI in your thing or you will not have a job.
- [21:19]
Jeff
Wow.
- [21:19]
Jonathan Brill
Right. I don't recommend that. But he wasn't able to get people using the tools. So there's. There's kind of that. That stick part, but then there's the carrot part. And we did. We recently worked with the Harrison Assessment on a survey of 2.7 million managers to figure out why some are dramatically more effective in times of uncertainty than others. And these are going to sound relatively obvious when I say them, but they were dramatically better at leveraging help from their peers than their peers. They made unexpected connections. They had much broader social networks than their peers, both inside and outside of the organization. So when they needed some informal advice, when they needed a new way of thinking about something, when they needed to shortcut red tape. Right. They were able to do it. In a world of optimization, neither of these things sound like things we want to be doing, but they're critical in the age of AI. The third piece was. Is about controlling chaos. And this is really what Rogue Waves is about. It's about when the world changes. How do you turn that into leverage? And the reality is that we build these standard operating procedures in our organization to manage risk for one of two reasons. One is the people in the organization don't have the context or the executive judgment, or two, we're just too lazy to manage it ourselves. And in a world of significant change, when the rule set changes, when the playing field changes, standard operating procedures will often kill us. Right, Right.
- [23:06]
Jeff
It's bureaucracy. Right. It's what holds us back.
- [23:09]
Jonathan Brill
Yeah. And so. And so how do you get to the situation where you're teaching people to think in from a first principles perspective about their challenges, about their risk and how to resolve it? Right. AI is going to be critical in helping people do that. Right. It's going to be critical in helping us overcome our decision biases, overcoming our lack of knowledge of the situation, our lack of history, our lack of contextual awareness, our lack of understanding of who else is impacted by our decisions. It's going to be incredibly important. And then the last piece, you know, is about taking the time to know what's missing. Right. We spend so much time, right, like, looking at what's right in front of us and solving for that, instead of taking back and saying, okay, well, what's causing this? Yeah, right. How do I solve for that? Am I still creating value by doing the thing I'm doing? Because the world's changed. So when you take this time to leverage help to create those unexpected connections to control chaos and to know what's missing, you have this really nifty consultant acronym, luck. But you also have a tool to rapidly teach your people how to create more in your organization. Because in a world of probability, if you can shift possibility, you create dramatically more luck for yourself, for your organization, in your business, in your life.
- [24:40]
Jeff
Right. And that list is so interesting to me because everything on it is things that are just. They're not replaceable by AI today. And I feel like even the things we talk about AI doing five years from now don't come close to that. And so it's interesting to me. It's so interesting that that's what the study came up with, because it highlights what's indispensable about us. Right. And about what we add to the equation.
- [25:10]
Jonathan Brill
Yeah. If it's in the database, you don't need to know it. Right. In five years, if it's in the database, you don't need to know it. And this is all about dealing with stuff that's not in the database, dealing with stuff that's not in the system. And when you look at. I get to teach once in a while at the Army War College, and they have this great model where they talk about executional, operational, and strategic leaders. There are three levels of leadership, and the things you need to know are essentially different. And when we were talking earlier about the new role of the cio, it's historically been a relatively executional or operational role. What I'm suggesting is it's suddenly a strategic role. And that is, in layers below the CIO are suddenly strategic roles, because we're dealing with architecture, we're dealing with process, we're dealing with culture. How do we put those new methods of governance in place to shape the firm, to shape its agility instead of just taking orders? And so that's a very exciting time. But it requires this stuff that is not kind of traditional. Project management, scientific management, PMI type of stuff.
- [26:28]
Jeff
So I'm curious in your mind, Jonathan, when we traditionally talk about even in the realm of cio, what's strategic and we say, oh, they're going to put together a technology strategy for the organization. We've talked about that. Cascading down from the high level business strategy. Is that going away? Is it just now? It is the business strategy.
- [26:48]
Jonathan Brill
The technology strategy is the business strategy. Yeah, absolutely. And certainly in five years.
- [26:55]
Jeff
Yeah.
- [26:57]
Jonathan Brill
The snarky thing people like to say is every leader is now a technology leader and every company is now a technology company. I don't know if that's true today, but I take a look at. We were talking about law firms, law firms in five years. Right. I have a real strong question about is Westlaw, which owns the databases and the software side of things, the law firm and the law firm is simply a sales mechanism for Westlaw or what's the value add there? Because it's not the paralegals, it's the relationship. It's that ability to deal with the things that aren't in the database.
- [27:39]
Jeff
Yeah. And it's really interesting to me that you brought up the snarky one line about every company is a technology company because as we're talking to me, it's almost become the inverse. It's every technologist is now a business person. That absolutely right. Like, like, because when you talk about those, those, you know, big four traits, my, one of the places my mind went and we talked about it is like what's the inverse? What are now the commoditized skills and are those the technology skills? Like is the, what's traditionally the technical skills being commoditized? And what does that mean for what's today or yesterday are, you know, technical organization?
- [28:23]
Jonathan Brill
I, I think that's absolutely true. You know, the, the thing that you know. And I think it was a mistake for Jensen Huang at Nvidia to say that the codings are relevant in the same way. It was a mistake for, for, you know, Jeff Hinton, the kind of the creator of the transformer model and deep learning or really deep learning to say that radiologists were irrelevant because, you know, machine learning is going to solve this over, I think, you know, your average medical doctor does, I think radiologist does over 40 tasks. One of them is looking at radiology images.
- [29:09]
Jeff
Right.
- [29:10]
Jonathan Brill
The reality I think of learning to code is it's learning how to make really good lists. Really, really, really precise directions that get you from one place to another. That skill is going to become incredibly valuable moving forward because we're going to have to direct, you know, a thousand agents, a thousand chatbots, a thousand AIs, to do something, and that ability to make a really good task list, that's the future.
- [29:48]
Jeff
So when you say task list, I just want to make sure I understand correctly. This is like, sort of like operational parameters or guidelines, like what's the objective and what are the things you need to do to get there? Or do you see it differently?
- [29:59]
Jonathan Brill
Sure. Well, I think it's being clear about the right level of. You're digging a little deeper than I've thought about it. And it's a great way to think about it. So in, in the military, they have a thing called an op board, and it's, you know, it's basically says, hey, here's what we understand about the situation, here's what we think might not be true, here's what we think might change, here's the outcome that we want you to deliver on. Here are the limits that we are giving you that you can fire up if fired upon or whatever. Here's what to do if things go wrong. Right. And here's what to do if I don't come back. Right. Like, that list is incredibly powerful because you give people all of the information to be successful without telling them what to do.
- [30:46]
Jeff
Right, Got it. Right, yeah.
- [30:48]
Jonathan Brill
And so that's one level of abstraction, but the next level down is, okay, well, we want to put the processes in place. We want to be that operational leader. Right. That's the strategic leader. We want to be that operational leader. Okay, how am I going to execute this? And what types of software, what types of AI do I need to execute this, what amount of data center access, what network bandwidth, what, you know, whatever do I need to execute this? And then getting down to the next level about the executional leadership, I think that's increasingly going to go away. Outside of the stuff about how do I get access to the software, how do I get access to the material, how do I get access to the data center, how do I get access to the energy, how do I get access to the data, do we get access to the electricity? That's still a human problem. And it will be, as we understand that those are increasingly valuable things, but you're going to still need all of those skills. But I think that ability to say, okay, well, what is this person really asking for?
- [31:52]
Jeff
Right.
- [31:53]
Jonathan Brill
And then figuring out how to make a really impossible to go wrong list of how parameters, as you were saying, how to deliver on this, yeah, hey, that's still a rare skill, man. And it's the thing we teach you as developers.
- [32:18]
Jeff
Well, and that's exactly it. Right. Like in, you know, having done a bit of development and worked with developers, I mean my perspective is that's been the value of developers from the dawn of time. Right. Like if I think about your, your, your trucker analogy. If you think that a coder's job is to hands on keyboard write lines of code like yeah, you're going to think it's completely disrupted. If you think their job is to structure a problem and understand a scenario. Right. That's, that's not necessarily going anywhere.
- [32:50]
Jonathan Brill
And, and, and I think that this problem structuring may start to go away.
- [33:00]
Jeff
Okay.
- [33:00]
Jonathan Brill
But the elicitation and the looking at information that's not being given, I think that that will continue. You know, I think about the, the, you know, the big game that the AI kiddies are playing right now is, you know, competitive programming. It's really good at competitive programming. It's like, well, yeah, but there was a really clear objective.
- [33:29]
Jeff
Right, right.
- [33:32]
Jonathan Brill
So that we could judge whether you did it or not.
- [33:36]
Jeff
Yeah. What happens when there is no objective or the objective is unknown?
- [33:39]
Jonathan Brill
What happens when there is no. Yeah. What happens when there is no objective? And so I think there are two things here that we need to kind of keep in mind. One, we've gone from, you know, a year or two ago, you know, ChatGPT 3 or 3.5 was, you know, on the 60th percentile of programmers at competitive programming. And then 01 Pro, I think if you give it a million dollars of tokens is at the top.02%. Right. We can argue the economics of that. Right. But let's ignore that for a second. What we have seen is that the software has gotten effectively better than most humans at this task. And the question is, did it in fact get dramatically better or just better than humans? It's one of the things I think about a lot is there are cognitive limitations. So if I ask you to remember three numbers, 200, 100, 123,624, you can maybe do that now. 1002-436247-37263,000, you probably lost the first one or two numbers. The reality is we can remember three, maybe four numbers at a time. That we remember seven digits is actually not true. A whole bunch of psychology, psychological testing around that we remember three to four and we can block that into chunks of A couple of chunks. And that's how we do it. My point being that there are cognitive mechanics, just neurological mechanics that limit how well we can do. And so as these tools get just 1% or 2% better than us at some of these cognitive mechanics, they become dramatically better than us at whatever the downstream task is. And so what we're going to see over the next couple of years that's just really exciting to me is that there are these things that seem like, yeah, only humans can do that, and these tools aren't going to get a hundred times better on an absolute scale than us. They're going to get 1% better than on an absolute scale. But on a relative scale, they're going to move from, you know, 60th percentile programmer to, you know, whatever, 99.98% percentile programmer just by getting a little better.
- [36:14]
Jeff
Yeah, you mentioned, you mentioned coding specifically for that. Are there, are there other use cases you've seen or you're starting to see emerge where this is the case where AI is starting to enter that 99th percentile of just human capability?
- [36:29]
Jonathan Brill
I, I think with prompting, you know, it's, it's getting pretty good at some things, you know, and, and I think that there, there are a number of domains that we can judge that on, right. Is it faster, you know, can it do 40 hours of work, you know, at the 80th percentile in five minutes?
- [36:46]
Jeff
Well, that good enough, right?
- [36:48]
Jonathan Brill
Economically, really, really valuable, right? Better than my intern in five minutes. Economically, really valuable, right? So that I think is happening really quickly. So I'm working on research for forging, for the forging industry, right. These guys have giant, you know, 2,000 ton hammers that just bang massive pieces of metal. And they're trying to figure out, okay, well, what does AI mean for them? And you know, I could have gone and read, you know, hunted down in read, you know, 5,000 paper pages of papers, or I could have AI go and hunt down 80% of those and point me to exactly the sentence that matters and then I can make a judge. Yeah, right. That's super valuable. And is it better than me? It's better than me that. In that I would never do that work, right? Like economically, economically I could never do that work. And so it makes me dramatically better. I think the other thing we need to think about here is like the AI kitties. And sorry, if as a listener you're one of them, but the AI kiddies are obsessed with trying to create artificial general intelligence. And I think that's A really bad idea to try and mimic humans. I think we should find the things that humans are terrible at. Right. Or humans don't pay attention to and get better than nothing at those. Right. That's the easy way to get superhuman.
- [38:31]
Jeff
Right?
- [38:32]
Jonathan Brill
The.
- [38:33]
Jeff
Yeah, the kind of the negative space around humans versus just competing directly.
- [38:37]
Jonathan Brill
Yeah, yeah, yeah, exactly. And I think to me that's. That's where the real power is. You know, we have all of these cognitive, you know, even within the things we're able to do, we have all of these cognitive biases. Right. Like if it just checks those biases for us. Totally able to do that today. Totally able to say, like, hey, research Daniel Dennett, here's my paper. Where am I biased? Like, it's able to do that counterfactual thinking because there's a framework.
- [39:08]
Jeff
Right.
- [39:11]
Jonathan Brill
Right. Stuff we're terrible at. And by the way, I can do it for you, you can do it for me, but I can't do it for myself, right. Because of the. Because it's already in my brain. I am the system, so I can't look at it. So that kind of thing, I think is going to be incredibly powerful.
- [39:33]
Jeff
So on that note, I'm thinking again about the interplay between AI and people. And we've talked about this an awful lot and you know, there's some conflicting opinions here and I've heard, we've almost talked through implicitly both sides of it. But you know, I've talked to a number of futurists in this space, Jonathan, and one of the prevailing schools of thought is we're gonna see the rise of the generalist and the technical skills are gonna go away and it's rise of the generalist. Now you've got. I think I'm hearing a little bit of a different perspective.
- [40:05]
Jonathan Brill
I just made a face for anyone who's. For those in our radio audience.
- [40:11]
Jeff
So. Well, so, so why don't, you know, why don't we respond to that before I even, you know, add any more color?
- [40:18]
Jonathan Brill
So I kind of. Everybody cooks a little bit, right? And what we discover when you read a recipe is that there is so much information that is tacit and never in a recipe, right? About, like, if you're baking, what's the humidity today? How does that impact my pie? The way my bread rises. Rises, Right. And so the question here is the generalist will be able to do a lot more, but they will never bake a great pie because they don't have the feeling of what's going on. And I think that's incredibly important to understand that having depth, intuitive depth in what you do is incredibly important. When you take a look at biology papers, what's the replication rate? 50%, 30% of them can be replicated. Either everybody's lying or there's parts of the recipe that aren't written down. And it's going to be some time before the whole recipe gets written down. And so I think that's true for almost everything. And so I don't know that the power of the specialists is going to go away, but it's. And it may be that once you become a specialist in one thing, then you can move that way of thinking to other things. But I don't know that just being a dilettante will create outsized value in the future because everyone can do it, right?
- [42:09]
Jeff
So it's not, it's not necessarily AI is turning everyone into a great baker. It's AI is making the great bakers that much better.
- [42:19]
Jonathan Brill
Well, it's making everybody into a good enough baker, right? So. So I think about PowerPoint, right? So up until the early 1990s, you know, you had, in many cases, you know, some guy who, who would literally cut up graphics and, and paste them. He was called a paste up artist and paste them up. And then you had another guy who would shoot film of that thing, and then you had someone who would run the printer and maybe someone who would run the ink in the printer. And then you had a graphic designer, right? And then you had the customer. Today you're doing your own PowerPoint. Now, are you as good as all of those craftsmen stacked up? No. You just aren't. You will never be Saul Bass, you will never be one of the great graphic designers in the world, and your graphics will never be in the Museum of Modern Art, I guarantee you this. But is it damn good? Yeah. So I think the question here about the specialist is where is that needed? Right. Graphic designers, right? We were becoming an information economy. We were all pumping out text for a living, and all of a sudden, good enough graphic design became fine. But, you know, do I want a good enough brain surgeon? Do I want a good enough, you know, car mechanic? Do I want a good enough, you know, truck driver? Probably not like good enough truck driver that doesn't run into the bridge embankment. Big bridge embankment, 99% of the time ain't good enough, right? So I think there's that need for specialization and that need to have a cognitive intuition about the subject is going to continue to be important.
- [44:23]
Jeff
Well, and I think that's incredibly powerful and really interesting to me. And the reason I asked is because you said something earlier about actually implementing this organizational change and the importance of psychological safety, about bringing employees along with you. And my sense is there's still an awful lot of angst about are we being replaced by AI? Do I still have a job? What is my value? Is AI going to prevent food being on the table for my family? And my sense is as leaders, if we're going to move our organizations forward, we need to be able to address some of these questions and probably be truthful about it. But I mean, do you buy that or is, or is there a key piece of the conversation that we haven't touched on yet?
- [45:14]
Jonathan Brill
So what you're asking about foundationally is how fast will this happen? And the kind of standard line is, well, these are generational shifts and they happen slower than you expect and everybody does fine and the economy grows. And the reality of the 1980s and early 1990s is that this happened slow enough. The 50 year old plus all retired out and it was all right. So if this shift happens in 10 years, that will not be true. The second thing to think about is that unlike the 1980s and 1990s, we are not in a population boom, we are in a population bust, right? So when we take a look in the next five years, I forget what the boomer population dropping out of the workforce is, but it's like 75%, 80%. And so, you know, we're going to see that this generational knowledge, this generational way of working happen goes away at a much faster pace than it did in the 70s and 80s. So that's happening. There will be a dislocation. These demographics will help us. The next piece is like, what do we do? And that's kind of the question. And part of my answer is our economy is so overproductive at this point that I have a person who makes my coffee. I go to Starbucks and I have a person who makes my coffee. And the hundred year thousand year historical perspective, think about that. I have a person in Jamaica that grows my beans. I have a person who puts that on a boat. I have a person who roasts that coffee for me. And then I have, then I drive down in my car with 5,000 parts in it that was made in Germany to go and get my coffee every morning from my person that makes my coffee. And if I'm really lazy, that person, there's another person who will bring my coffee on my croissant to me. I don't do that very often because that's snotty. But my point being, you know, we will come up with new things to do. I'm not concerned about that in the US where we are strategically trying to bring technology and manufacturing back to the US for I think some pretty good reasons. You know, there will be transferability into other industries and other roles moving forward. Like I said in tech, I don't think those roles are going away. I think we're going to figure out that there's value here. I think, like I said in professional services, we're going to see a 15% or so efficiency bump in the next few years here. That's pretty massive in an industry that doesn't really see efficiency bumps and that's an outlier. But we're going to see that in a bunch of other industries too. We are going to see dislocation and we're going to see the job roles shift as one person is able to do 1 point whatever, 1.07 jobs. But I think it's going to take a little longer than people think and a little longer than the CFOs are dreaming.
- [48:44]
Jeff
Right. So I want to ask you a slightly different question about that, which is, you know, you're, I think, doing something very important, which is you're debunking some of the, some of the hype around, you know, oh, this is going to change everything, you know, overnight or some of the, you know, some of the hot air that comes with AI.
- [49:04]
Jonathan Brill
It will repeatedly change everything overnight. It's just the scale at which large organizations can absorb that.
- [49:11]
Jeff
Right.
- [49:12]
Jonathan Brill
Is in, in the scale at which governments will allow that to happen because of the implications for their tax base and the monetary system. You know, there are governors beyond the technology.
- [49:25]
Jeff
Right, Right. There's the human pace of absorption and the societal and the. Yeah, yeah, I guess the organizational pace of absorption. But it's like, can we control that or, or is the pace of change at some point just going to, you know, run us over? And as consumers become, you know, more impressed with what this can do, even if it's not quite as good, you know, it just, you know, I mean.
- [49:50]
Jonathan Brill
It'S changed my life. You know, I mean, I was doing econometrics on this in 2018 and 2019, and I saw, you know, and we were in 2016, 2017, building transformer models, or what came to be called transformer models models. So, you know, I've been at this for a while and I knew exactly what was going to happen from, you know, a Task efficiency, perspective. And by the way, this is not a surprise. It's kind of like, tracked exactly what we thought in 2018, 2019. So it's not magical. Like, this was all very knowable, but it still blows my mind every time I'm suddenly able to do, you know, 40 hours of work in five minutes. You know, it's pretty magical. So I'm not saying that's going away. What I am saying is that when Covid hit, I pulled all my money out of the stock market. I said, hey, you know, I don't believe this White House is going to be able to deal with this quickly enough. Things are going to go to heck. And. And I pulled all my money out of the stock market. And what I didn't understand was that the government and the Fed can just change the rules. Right, right. Like, every company in the United States should have gone out of business, and they just changed the rules.
- [51:15]
Jeff
Yeah.
- [51:16]
Jonathan Brill
And. And this is not the first time that's happened. They did in the 1970s when we got off, you know, late 60s, 70s, when you started getting off the gold. Gold standard. They did it, you know, in Bretton woods in. In the 40s. Right. We can literally just change the rules.
- [51:31]
Jeff
Yes.
- [51:31]
Jonathan Brill
And the Trump administration is actively just changing the rules right now. Right, right. That's what this is all about is like, you know, Peter thiel, Elon Musk, J.D. vance, people who actually understand technology, saying, hey, we need to kind of change some of these rules. Right. You might agree with them or not. You might think that this is the right answer or not, but that's what's going on. And so what we need to understand is that the game, the playing field will shift along with the technology.
- [52:05]
Jeff
Right.
- [52:07]
Jonathan Brill
And so this isn't all going to blow up on Tuesday just because there's some AGI thing that wins Nobel prizes every 30 minutes.
- [52:17]
Jeff
Right. So I did want to ask you, Jonathan, before I lose sight of it, one of the things you do in your role as a futurist is you separate kind of the signal from the noise, if I can call it that. What is some of the noise you're hearing right now? What is some of the hype that you're like, that is bs that is not going to come to pass. That's on your radar, is like, we should be tuning out.
- [52:40]
Jonathan Brill
We're starting to see some reality testing happening in. In the market. So the first thing that was happening was, hey, this is all gonna just explode, and it's going to be great. And that because we're seeing thousand times increase in efficiency gain over 18 months or whatever it is, that this thing's just going to run. And the reality is, because we're talking to a mildly technical audience. A technical audience I'll get mildly technical. We were looking at transformer models, now we're looking at reasoning on top of transformers. Now we're looking at multiple shot on goal, like running the thing 10 times and finding the middle ground. Right. Every time you do that, energy use goes through the roof. But what we're still doing is basically, and we've gussied it up, but this is a database search tool, right. And searching a database, the second you get into agentic AI and deep research, right now you have an agentic tool that's doing all of this stuff with its database and pretty quickly going to another agentic tool that's doing the same thing. And maybe those two agents say, hey, there's another database with some other information. And doing that again, you can see how the energy use goes geometric here. And no matter how efficient we get, we're going to see a dramatic increase, increase in energy use. And that's going to be a massive limiter. And the reason it's a massive limiter is that A, we hadn't planned on this in the US electric grid or anybody else's electric grid. B, the gas turbines that, you know, people like Elon Musk are using, he bought, he rented half of them in the country just to run his, his, his new data center and I guess in Nashville, I guess Kentucky, Tennessee. But you know, he took half of the stock just to run one data center. And then you look at those turbines or that class of turbine and there's a 90 month lead time to buy another one.
- [54:55]
Jeff
Wow.
- [54:55]
Jonathan Brill
So we have some real limits, we have some real limits to how quickly we can increase energy use. And so that's a governor on what we can do and what's cheap. So you go out to 2035. Yes, problem starts to be solved. We have small modular nuclear reactors coming online. We have a 10 year backlog of purchases of those things, you know, but up until 2030 or 2032, you know, it's going to kind of look a lot like what we're, what's at the leading edge now in the Enterprise space right now, what we're doing in the Enterprise space. But you take, you know, you take three instances of O1 and you get them talking to each other or you take, you know, Sonnet 3.7 and 01 and Gemini and you get them talking to each other with different databases. You know, that that's about where we're going to be for the next five years.
- [55:53]
Jeff
Right.
- [55:54]
Jonathan Brill
You know, and there will be government stuff and there will be, you know, hedge fund traders and, you know, like whatever that want to spend, you know, $100,000 a month and that'll be different. But for you and I, that, that's the outer edge.
- [56:08]
Jeff
So, you know, my natural response to that, Jonathan, is, and I've heard people say this before, that as, as the need for energy grows and it becomes, you know, a more and more valued commodity, sure, you can increase the supply of it, which you're talking about as being, you know, kind of 10ish years out, but it could potentially also drive a race for efficiency within these models that, you know, can become, you know, not, not geometrically more efficient, but exponentially more efficient. Is that on the table?
- [56:41]
Jonathan Brill
Absolutely. What we're seeing, according to Satya Nadella, and you know, certainly one of the bubble blowers in this whole thing, but the, the, the least, the least smoke blowing of the bubble blowers, you know, he's saying, hey, you know, Wef, if you multiply out what he said, he says, hey, I think there's about 100,000 times increase in energy efficiency between now and 2030. And you know, when you just calculate out, you know, in terms of performance improvement, that's about a 10 times increase in performance improvement between now and 2030 or 100,000 times decrease in energy efficiency. So in that is the range of math. Right. So we'll see things that are stunningly efficient that are ChatGPT 3.5 that will be free, or ChatGPT 4, maybe that will be free. And then we'll see more advanced tools that I would actually use to do work.
- [57:42]
Jeff
Right.
- [57:44]
Jonathan Brill
That will be not free.
- [57:45]
Jeff
Right.
- [57:45]
Jonathan Brill
And the cost of those will go up on a geometric curve. How long will, you know, ChatGPT's or OpenAI has been threatening to put out a $20,000 a month tool. How long will that stay? $20,000 a month? That's a really good question. I think looking at how quickly, you know, everybody's nipping on their heels, you know, six months and then it's 2000 and then it's 200 in a year. I mean, that's pretty incredible.
- [58:19]
Jeff
Oh, absolutely. It's exponential, Right?
- [58:24]
Jonathan Brill
Yeah. But the demand for that tool will go through the roof and so you'll be limited in your ability to access it.
- [58:34]
Jeff
Right.
- [58:35]
Jonathan Brill
Because the value will, I think the value of the tools I'm using now are so high, you know, and I'm spending hundreds of dollars a month. And they'll get easier to use. And every executive will insist on it. I mean, it's just kind of how it's going to work. And every technician, technical person is going to insist on it. Right.
- [58:56]
Jeff
Jonathan, I wanted to say a big thank you for joining us today. I've really appreciated your insight, and this has been a great conversation.
- [59:03]
Jonathan Brill
Absolutely. Thank you. Jeff, it's wonderful to meet you.