The Competitive Advantage No One Is Protecting
Episode Summary
Most companies track financial capital, human capital, and physical assets. John Sviokla, co-founder of GAI Insights and an Executive Fellow at Harvard Business School, argues the asset nobody is protecting is cognitive capital: the accumulated know-how of how your company decides, operates, and works. He traces where it came from, why most organizations are giving it away without noticing, and what to do about it in a strategic planning cycle. Along the way he introduces financial disruption, which arrives well before the operational kind, and asks the question executives rarely make explicit: when you automate something, who gets the value.
Key takeaways
Digitization created three new forms of capital alongside the traditional three: behavioral capital, which Facebook sells, network capital, which LinkedIn sells, and cognitive capital, which is how you do something. Most of us traded the first two for cheap services without noticing
Start strategic planning with two questions: what is your ratio of human workers to digital workers, and where are you on the maturity curve, from educating yourself to islands of automation to transformation to AI helping build the AI
Financial disruption comes before operational disruption. The market decides your model is wrong long before a competitor takes your customers. Sviokla cites Gartner, down roughly 50 percent in market capitalization while still growing
The leading companies are hiring people who have already built their own tools. Making a GPT or a Gem is now easy, and vendors spent years keeping professionals out of building anything
Value from automation can go to exactly three places: the customer through better or cheaper products, the investor through returns, or labor through wages. Sviokla argues too little has gone to the third for four decades
His one piece of advice: ask a robot. He says he does nothing of consequence without one, from reviewing proposals to checking any important email before it goes out
About John Sviokla
John Sviokla is a strategist, educator, and co-founder of GAI Insights, and an Executive Fellow at Harvard Business School. He began his career at Harvard Business School studying the economic impact of expert systems, and has advised Fortune-class firms across economics, technology, and operations for decades. His current work centers on owning your intelligence: protecting cognitive capital, personalizing work with targeted assistants, and scaling agent-based automation with clear governance and measurable outcomes.
In this episode
| 00:41 | Welcome and guest introduction |
| 01:40 | From expert systems at Harvard Business School to now |
| 02:45 | Own your intelligence |
| 03:21 | Six kinds of capital, and the three we created by digitizing |
| 04:22 | How companies are protecting intellectual capital and resilience |
| 05:21 | Strategic planning for 2026 |
| 05:54 | Human workers to digital workers, and the maturity curve |
| 09:26 | Models suited to deterministic work |
| 09:49 | Marvin Minsky, The Society of Mind, and specialized processors |
| 11:17 | Mixing model types |
| 13:07 | If your business makes physical things |
| 14:23 | The baseline nobody can skip |
| 17:20 | When models become the front door to your software |
| 18:43 | What boards need to understand |
| 19:16 | Financial disruption before operational disruption |
| 20:11 | How to detect it |
| 21:42 | Summarizing the three risks |
| 22:42 | Hiring people who have already built their own robots |
| 24:14 | What executives say about the workforce |
| 25:54 | Who should get the value from automation |
| 28:08 | Reinvesting, and where it actually goes |
| 30:10 | Public investment after the war |
| 32:57 | Returning to the common good |
| 34:05 | Why he stays optimistic |
| 36:14 | Leadership: the rational and the emotional |
| 39:02 | The one thing to remember |
| 39:33 | Resources |
| 41:00 | How GAI Insights works with businesses |
| 42:02 | Wrap-up |
In John’s words
“Traditional capitalism was built on three kinds of capital. When we started to digitize, we birthed three more: behavioral capital, network capital, and cognitive capital.”
— John Sviokla (03:21)
“What’s my ratio of human workers to digital workers? If you’re not asking yourself that question, that’s the first thing.”
— John Sviokla (05:54)
“There’s a step way before operational disruption, which I call financial disruption: when the market thinks that you have the wrong model.”
— John Sviokla (19:16)
“When you automate something, there are only three places the value can go. The customer, the investor, or labor.”
— John Sviokla (25:54)
“Ask a robot. I don’t do anything of consequence without asking a robot.”
— John Sviokla (39:08)
Resources
John Sviokla and GAI Insights
• John Sviokla on LinkedIn: linkedin.com/in/jsviokla
• GAI Insights: gaiinsights.com. Executive briefings, AI strategy work, customized trend research, and conferences
Reading he references
• The Society of Mind, Marvin Minsky: The 1980s book behind his framing of modern models as many specialized processors working together. Minsky was at the Dartmouth conference where the term AI was coined
• Clay Christensen on disruptive innovation: His late colleague’s work, which Sviokla extends by adding financial disruption ahead of the operational kind
Companies and tools discussed
• Gartner: gartner.com. His worked example of financial disruption, down roughly 50 percent in market capitalization while still growing
• HubSpot: hubspot.com. His example of a model becoming the front door to software
• Custom GPTs and Gemini Gems: His example of power tools that professionals can now build for themselves
Next: the related episodes. The titles below are already live links, so select from the heading down, copy, and paste. Nothing in that block needs deleting.
Related AI Realized episodes and events
• Smaller Models, Bigger Wins: Verify Before You Answer: Jason Williamson of MythWorx on deterministic systems that verify before they answer.
• AI Search Visibility: When AI Says Your Company Is Dead: Curtis Sparrer of Bospar on brand visibility when answer engines mediate discovery.
• Agentic AI and Revenue Work: What Actually Pays Off: Christopher Penn of Trust Insights on measurement, and proving lift to a finance team.
Frequently Asked Questions
-
John Sviokla defines it as how you do something: the accumulated know-how of how a company decides and operates. He places it alongside two other forms created by digitization, behavioral capital, which platforms like Facebook monetize, and network capital, which LinkedIn monetizes, and contrasts all three with the traditional trio of human, financial, and natural capital. His argument is that consumers and companies largely traded away behavioral and network capital in exchange for cheap goods and services, and that cognitive capital is now going the same way without anyone deciding to let it.
-
Ask two questions: what is your ratio of human workers to digital workers, and where are you on the AI maturity curve. John Sviokla says the first is fundamental whether you run a bank, an HVAC company, or a school. The curve runs from educating yourself, to building islands of automation, to transforming and scaling systems, to what he calls emerging intelligence, where AI helps build the AI. He notes the more sophisticated companies are already at that last stage.
-
Clay Christensen described operational disruption, the point where a competitor starts taking your customers. Sviokla argues a step comes earlier: financial disruption, when the market concludes your model is wrong. His example is Gartner, whose market capitalization fell roughly 50 percent while the company was still growing at 3 to 4 percent with 24,000 employees. The market decided AI would crush the model and that transformation was not visible fast enough.
-
As many specialized components rather than one model. Sviokla credits Marvin Minsky’s The Society of Mind from the 1980s, which argued that real intelligence would come from specialized processors working together, using the example of the eye pre-processing signals before they reach the optic nerve. He points out that querying ChatGPT or Gemini is not hitting a single model: the query is parsed and routed across components.
-
Universal education on using the models, identified user champions, a secure chatbot environment, and custom GPTs or Gems that make specific teams more productive. Sviokla stresses refreshing it constantly, because capability changes roughly every three months, and cites tasks that became possible in the space of six months which simply were not before.
-
Sviokla frames this as a question executives rarely make explicit, with only three possible answers: the customer, through cheaper or better products, the investor, through higher returns, or labor, through wages and compensation. He argues labor’s share has been in decline for four decades and that this explains a good deal of the anxiety around AI. He is stating a position here rather than a consensus, and the episode includes Christina pushing back on parts of it.
-
People who have already built their own tools. Sviokla observes that creating a custom GPT or a Gem is now straightforward, and that for years both software vendors and professional norms kept practitioners out of building their own software. His analogy is a house built in 1896 with three courses of granite in the basement, because the builders were digging by hand: constraints shape what gets built, and this particular constraint has lifted.
-
Working across two dimensions at once: the rational case and the emotional reality. Sviokla points out that every Western myth about advanced technology is a warning, Prometheus, Icarus, Frankenstein, the Terminator, and that AI touches something fundamental about people’s place in the world. His view is that leaders have to make the productivity case and be genuinely prepared for the emotional response to it, rather than treating the second as an obstacle to the first.
Full Transcript Episode 34
-
[00:41] Christina Ellwood: Welcome to AI Realized, the podcast for enterprise executives leading AI deployments. From tackling security, data, and operational challenges to navigating organizational transformation, AI deployment offers a unique opportunity to redesign our organizations from the inside out. I'm Christina Elwood, your host for today's episode, and we are talking today with John Sviokla, the co-founder of GAI Insights and Harvard Business School Executive Fellow. John, welcome to the show.
[01:11] John Sviokla: Christina, it's lovely to be here. Great to be with you as always. There's so much going on. Great time to be alive.
[01:16] Christina Ellwood: That is for sure. I c- I absolutely agree with you. I remember hearing you say at the GII World 2025 just a month or so ago that this is the time in your career when you are learning more and doing more every single week than you have a- at any other period in your career. Yes. I think that's an important statement. Maybe you'd like to tell us a little bit more about that and why you find generative AI so exciting.
[01:40] John Sviokla: Sure. Yes. I started my career, a professional career, at Harvard Business School in the area of expert systems and looking at what kind of economic impact they had. And then we did a ton of stuff at the combination, intersection of strategy and technology and innovation. And, and I'll tell you, what's happened now is that fundamentally we're at a place where the ... it's as big as the Industrial Revolution, and the internet wasn't. The internet was a great increase, just like jet planes or something like that, right, which has really changed the world. I'm not saying it didn't change things. But this is fundamental, and the reason is, what a lot of people don't realize about the Industrial Revolution is it had three parts. It had the engineering of how to redo the work. It had addition of capital, for capital substitution for labor, for horses and people and so forth. But the thing that a lot of people don't understand is how much of the Industrial Revolution was really about knowledge management. And the reason that this jacket and this shirt are high quality and cheap, yes, it's a combination of industrialization and process, but it's also knowledge, how to do it. And that's where we are. We're at a whole new ec- economics of knowledge that have profound implications.
[02:45] Christina Ellwood: This is one of the reasons that I love your model at GA Insights that you call own your own intelligence. Because to my mind, a significant element of every business is the unique aspect of their, what you call, cognitive intelligence. How they work, how they decide, how they interpret, operationalize, engage, serve their customers is unique. That's what makes businesses different and competitors is that they do not approach everything the same way. What do you see people doing to protect their corporate intelligence today- Yes especially since you've been calling for this for as long as I've known you?
[03:21] John Sviokla: Yes. Yeah. It's super important, and you see basically three kinds of activity. First, you have folks who just aren't paying enough attention to it, and that's really a mistake. Because if you look back in the history of digitization, traditional capitalism was built on the back of three kinds of capital: human capital, financial capital, and natural capital, steel, water, energy, and so forth. When we started to digitize things, starting in the 1940s, we birthed three new kinds of capital. We birthed behavioral capital. That is, think about Facebook. They're selling my behavior. You- there's network capital, whom I'm connected to, and LinkedIn sells that. And then we've got cognitive capital. How do you do something? And the, as a country and as a consumer group, we've traded, in large part, we've traded our behavioral capital and our network capital for cheap goods and services. Email, inexpensive or free LinkedIn. To do that with your cognitive capital, that is, how your business creates a value, is a huge mistake, because you're inviting people into your business with a better model than the way you're doing it today.
[04:22] Christina Ellwood: So are you seeing businesses protecting their, their intellectual capital in predictable ways?
[04:29] John Sviokla: Yes, definitely seeing businesses protecting their intellectual capital and also their operational resilience. A number of the large organizations, large banks and so forth, first of all, have different kinds of supply, so they don't just rely on one hyperscaler. The second thing is that they're very clear on the contracts in terms of what they own and what the supplier owns, because you don't wanna just kinda leak out your intellectual capital that way. The third thing is, for certain activities, drug manufacturers, certain kinds of insurance operations, they're doing-- they're thinking more like the military. They're thinking air-gapped, "Hey, yes, we'll bring a model in, but we're gonna train it, or we're gonna have a partner, like a SambaNova Systems or an A21 or somebody who's gonna help us do it ourselves or do it in conjunction with us so that we actually own our own intelligence." So the more sophisticated shops are really a hybrid of buying off the shelf and also building their own.
[05:21] Christina Ellwood: So as we face 2026, we're heading into strategic planning season for every business. Many of them are already deep into it, and we're thinking about what is-- what are we going to face in terms of complex problems that we need a strategy for? And what is going to continue that is already in flow today?
[05:45] John Sviokla: Yes.
[05:45] Christina Ellwood: And how do we align our resources to address both? What's your- Yes ... thought about, like, how businesses should be thinking about those three questions?
[05:54] John Sviokla: Yes. First of all, every business needs to understand where are they on their AI journey, because this is fundamental. Whether you're running a bank or an, an HVAC company or an education institution, you know, you're gonna be using robots. I can't imagine a 2030 where people aren't interacting with robots all the time, from customer service to marketing to onboarding people to training and so forth. So that's just a fact of life. What's my ratio of my human workers to my digital workers? And if you're not asking yourself the question, that's the first thing. Second thing is, where am I on my maturity curve? Am I just starting and educating myself? Am I doing islands of automation? Am I actually transforming my system and scaling, or am I doing what w- we call emerging intelligence, where you've got the AI helping to build the AI, and the more sophisticated companies are doing that. And we have this RISE adoption model we call Research and Education, Islands of Innovation, Scaling, and Emerging Intelligence. So ask yourself where you are on that. That's the first thing. Second thing is, how does that relate to where I wanna get in 2026? So a more hybrid organization is gonna have a bunch of interesting attributes. It's gonna have higher revenue per employee. It's gonna have faster innovation. It's gonna have more ability to distribute knowledge in terms of customer service, so they should have higher net promoter scores and things like that. So you're gonna see real operational capability in 2026. There's gonna be three things I think people are gonna-- three predictions I'm pretty confident on. One is, in more and more markets, it'll become apparent to the people in that market that they're competing with an AI compe- an AI-enabled, a hybrid competitor. So JPMorgan Chase in, in financial services, Goldman Sachs, they're totally hybrid now. They've trained everyone. They've got robots. They've got their own platform, the whole routine. That's showing up already, but gonna show up even more in '26 in relative economic and operational measures. And that's, that should scare everybody who's competing with them because it's like the old joke about the bear. I don't have to outrun the bear, I just have to outrun you, and that's gonna happen. You're gonna have a faster competitor. That's the first thing. Second thing, we're gonna see a huge adoption of AI on mobile because Qualcomm and others, just about every major manufacturer, is now putting neural chips into these devices, and intelligence at the edge makes it super easy. Right now, if you haven't tried it, just pick up your phone, point it at anything. How do I use this microphone? How do I repair this washing machine? What is this thing I'm-- what's this painting I'm looking at? And 99 times out of 100, the model comes back with the right answer, and people are gonna start to wake up to that intelligence at the edge and how it enables service and sales. And the third thing Is we're gonna see, and we're starting to see already, an explosion in science. The thing is about these models is, look, they speak mathematics and they speak every other kind of language. And the beautiful thing about math is it's incr- and well-specified problems, is they're incredibly susceptible to reinforcement learning. So the, and this is where a lot of the early neural network stuff took off with AlphaFold and AlphaGo, which was the model has a model of what's better. It then generates options either from existing real data or synthetic data. It then improves it, so that reinforcement learning is just learning faster and faster and faster. And the thing you have to remember is that's learning a billion times faster than people are, right? They think in nanoseconds, we think in human seconds. That means science is gonna take off like this, so we're gonna see incredibly new everything from trading strategies to molecules to, to new kinds of material science and so forth, and that is just beginning. So those are gonna be three biggies in '26.
[09:26] Christina Ellwood: Are you seeing any new emerging, for the purposes of enterprises, not on the cutting edge, but-
[09:33] John Sviokla: Sure ...
[09:33] Christina Ellwood: commercial- Yes ... versions of models that are even better suited to these kinds of complex deterministic use cases using things like satisfiability or other kinds of AI strategies that are not specifically generative AI?
[09:49] John Sviokla: Oh, sure. Yeah. Look, the models are already a combination. I'm inspired by a book that was written by Marvin Minsky back in the, in 1980s called The Society of Mind, and Marvin was, the, one of the original people at the Dartmouth Conference where they birthed the, the term AI and the whole notion. And Marvin has this, had this idea, he said, "Look, the way really intelligent AI is gonna work is we're gonna have a bunch of specialized processors. They're gonna all work together." And he gave examples from the mind. My eyes are a pre-processor for my, for my visual cortex and so forth, right? There's a bunch of processing that happens actually in my eye with the cones and rods and so forth before it gets to my optic nerve. The same thing's happening with these large models. You go into ChatGPT or Gemini 3, and that's not one model, right? It's taking your query, it's parsing it, and saying which model is best to serve here. So what we've already started seeing and gonna see more of is this notion of a hybrid model. Some of it's gonna be a large language model. Some of it's gonna be traditional AI. Some of it may just be a database lookup. Others could be traditional regression analysis. But that's where we're going with the big models. So we have that, and then we have small models that are very specific to a given task or domain. And for example, we have models now that are getting built in the research area that are gener- started not with words, but start with things like proteins when you're gonna look at a new kind of protein evaluation. So I think we're gonna see a lot of that in '26. The big models are already hybrid, if you will. They're not hybrid in the organizational hybrid or multimodal models, and I think we're gonna continue to see that growth.
[11:17] Christina Ellwood: What about multiple kinds of models too, like a neurosymbolic model with a generative model, with a predictive model, with a... Are you also seeing that sort of mixed types rather than just multiple in the same type?
[11:31] John Sviokla: Yeah, absolutely. Yes. We're seeing a lot of, a lot of mixed-type models. In addition, we're seeing real world models really informing the, the use of all kinds of things from robots to simple understanding. And the, the two vectors for real world models seem to be from games, right? So there's mu- much, much better models that help you generate in real time if you wanna create a character or a gaming world or whatever. Those then map pretty well onto reality reality. And then you have the other side, just people working on real world models, and where I think a lot of the advanced work in this is classified because it's, it's so central to the military industrial complex. I know some of the very early AI terrain models and so forth were used for cruise missiles and such, and you can imagine both the Ukraine war and the Israeli war have really shown a totally different doctrine based on the use of AI. So the real world models are super important. So yes, I think we're gonna see that. The cool thing about the real world models is that with a very small sample, I could take a picture of it, of a room or something. I can, with a real world model, I can tell you, okay, what kind of room it is, who's likely to live there, what can be done there. What's the physics? Can I actually have my robot go this direction? No, I'm sorry, there's a table there, and you can't go through a table. But you can look out a window. All those factors, just with one quick picture, I now have a model of that reality, and that's profoundly powerful when you wanna have an agent go into a novel situation. If they can just look around, take a picture, it's, "Oh my goodness, this is how this works," and then can execute.
[13:07] Christina Ellwood: Yeah, that is really, that's, uh, just incredible. Really, truly incredible. So if you were running a business that has physical nature to it, you d- you build a product and ship it, or you have warehouses or things like that, those physical world models are really important. It's not just the robots that run inside of them, it's all of the systems around management and simulation and things like that you need to be able to do. Those small language models also relate to what you ment- mentioned earlier about mobile, that those models need to be small enough to fit on the mobile phone and be able to do certain things. Also connects to the own your own intelligence because we need to decide, do we allow that data to be used, personal data to be used to train models? So there's many areas here that are ch- changing at the same time. So if you're doing strat planning for your business for 2026 Zero in for me on what are the things that you feel are perhaps we- are- that I think you've identified the ones that are most likely to be talked about. What are the ones that they need to consider in their strategy that they may miss because they don't realize it's coming so fast to them? Mobile, I think, is a good example of that. They may not realize that mobile's coming into their world in such a t- short time period.
[14:23] John Sviokla: Yes. I absolutely. I think that the-- first of all, the- don't-- y- you can't ignore the baseline stuff. Is everybody in my organization educated on how to at least use these models? Then do I have user champions? Am I starting to build GPTs and gems and so forth so I can make my organization more productive doing that in the s- have a secure chatbot environment that I can teach people and to make sure that you have that, and that you're refreshing it. Because every three months, the capabilities are very different. What Gemini 3 just came out with now, you couldn't do six months ago in any of their models. And so now you can attack PowerPoint, you can attack Excel, things like that, which you couldn't do just a short time ago. So that's really important. The second thing, in terms of what to watch for, yes, I think that mobile's gonna be huge. And, and the really important thing about mobile is that it will happen in physical businesses and non-physical businesses. So for example, l- let's say that you're running a parts distribution company. We're working with a private equity firm that owns one of those. Okay. Uh, we're, we are encouraging them to begin by just, and they operate in a bunch of different languages across the United States and across, uh, North America and some in South America. And parts reordering is often hard, right? 'Cause they serve a long group of trucks and other industrial equipment. So we're saying, "Hey, look, let's just do an experiment and find out how much do each of the models know about your product and service, just generically," and to help folks who are out in the field doing service or repair and just take a picture, and can it recognize it? Does it know how to fix it? Does it know where to, does it know where to get it, and how soon it'll come and so forth? So really just using, I think of it as use the world as your barcode, right? Everything is explainable with these models. I think people are gonna start to understand what that means for service, for sales, for delivery, for repair, just all kinds of stuff that's at the ends of the network, and everybody's got a cell phone, so a smartphone, so what the heck? So that's gonna be a biggie that I think not enough people are experimenting with. Second thing is education. There's one of the, uh, we looked at one of the national labs and one of the case studies we looked at, and they were doing very high-end genetics work, okay? And the challenge was a lot of that's very... A lot of the equipment's pretty funky and sophisticated, and so if they had a downtime on some of the equipment, they would have to wait until the expert person came in at the normal 9:00 to 5:00 shift. But a lot of these reactions and so forth are being done 24 hours a day. What they started to do is they simply took pictures of how the experts did it. They used that as the basis for how they then did the instruction so that they didn't need the top expert to be able to keep doing things. Well, that's just sampling reality using the semantics that we have in the models, using that to then train the rest of everybody else so that you can have 24 hour a day, seven day a week uptime as opposed to eight hours a day, five day a week uptime. And that's huge in terms of the productivity of those labs. So I think that's gonna be a big thing. The merge between the world as your barcode and the phone, that's huge. Do you
[17:20] Christina Ellwood: have an example of that for companies that are not doing physical goods, financial services or software companies, or companies that are not selling physical products?
[17:30] John Sviokla: Not so much the mobile part, but using the model part. The models now are the front door for many people for any software and so forth. So when I'm using HubSpot, which we use, I never go to HubSpot's information. I go to the models because the models are better at explaining how to use HubSpot than HubSpot is. I think that's true of every integrated development environment that I'm familiar with. With-- And then the meta products, Jira, all of Atlassian products, right? Jira and Convergence, I think, is one of the other ones, how you use those. So I think in tho- any of the services and software activities, for sure they're already using the models to do the things that the companies used to do themselves. So that's huge. I s- the other thing, and I haven't seen evidence of this yet, but it's obvious that people will do it. When satellite information became more granular and cheaper, a lot of people started using it for trading strategies. So they'd start counting cars in the Walmart parking lot at a particular time of day and so forth, and that allows them to create a demand model. We're gonna have more and more of that ability to do it, right? I can actually look and see, oh, I can analyze the gate of people going into Walmart. Is that, is that a predictor of how much they're gonna spend? What-- So the ability to sample reality and use that for financial trading I think is gonna go up, although we haven't got specific examples of that.
[18:43] Christina Ellwood: So that's a challenge for boards too, right? The boards of directors are trying to understand how AI is-- it needs to be used in the businesses they're in and how to, uh, mitigate risk, and sometimes that risk is evaluation risk, right? And so their concerns about if we adopt it, our valuation will go up. That's one thing that you've been talking about for a while. But what are the things that are driving the assessment of the investors in our business? Do you think that's just like the physical example you gave? Do you think that's changing in other areas besides the physical world monitoring?
[19:16] John Sviokla: Yes. Yeah. My friend and colleague, Clay Christensen, of course, got very famous talking about disruptive technologies and did a great job articulating that. It's helped a lot of people. The before-- But he talked about operational disruption, like when does Southwest Airlines take away market share from United Airlines? There's a step way before that, which I call financial disruption, which is when the market thinks that you have the wrong model. So if you look at Gartner Group right now, Gartner Group's market capitalization has gone down about fifty percent. They've got twenty-four thousand people there growing at about three to four percent per year. Why are they disrupted? Because the market has looked at them and said, "Your model is gonna get crushed by AI, and you're not showing us that you're transforming fast enough. You may have the customers. You may have the margin today. You may have the revenue. But we don't believe that you're gonna be successful going forward given what you're showing us." So-
[20:11] Christina Ellwood: Is the primary indicator of that a, a devaluation of the company, or are there other ways to detect that?
[20:19] John Sviokla: Uh, primary is, uh, devaluation of company, but it doesn't just affect finances. It affects talent. Would you be happy if your kids went to work at Macy's right now? I don't think so. Macy's is a walking zombie. It's been around forever. It's been financially disrupted for three decades, or two and a half decades. The, so that's, so you lose the talent, you lose the money, and the other thing is you lose lead customers because a lot of times your, some of your best customers are the ones who really want the better solution first. And those are three hard things to do. Financial disruption, uh, very few businesses are gonna be completely out of business next year. But there'll be many industries where people will get marked against an AI-based competitor, and they're gonna look not good, and they're gonna see, like Chegg has just had a huge layoff. Gartner. If S&P Global didn't convince the markets that they were in fact on top of it, they would have a disruption. So I think a lot of companies are gonna face that. And by the way, we've did, we did research in one of the recessions in, uh, the 19- the 2000 recession, and we showed that those companies that invest during those downturns pull ahead of the competition. So if you're, if we go into a recession or an AI winter or whatever, if you keep motoring through and making your organization more productive, you're gonna win in the near and long term.
[21:42] Christina Ellwood: Okay, so that's really good input, I think, for strategic planning because first of all, we need to evaluating those three risks in our business. Are we at risk to be downgraded because we are not sufficiently responding to this AI opportunity? And there's the areas that we could be detecting whether or not that's the case. Are we losing customers because of it? Are we losing employee hires because of it or critical staff because of it? And is the market not giving us full credit for what we're doing otherwise because we are not sufficiently So those are three very good things, I think, to put into that strategic planning hopper, and then in looking at what we wanna consider for adoption, looking at the, the vectors you described and things like- Sure ... your hole in, in physical world, and then perhaps in upscaling and developing your talent and making sure that you have a way not just to train them once, but to keep them up to speed and keep the agents or systems that you're using to enable them with the AI, keep those systems up to date on both their capabilities and on the data that you're feeding them. Is that a pretty good summary?
[22:41] John Sviokla: Excellent summary.
[22:42] Christina Ellwood: Did I
[22:42] John Sviokla: get it? Exactly. No, I, not at all. I do think there's a whole new thing. The leading companies we see are hiring people who have already built their own robots to help them. Just think of in simple work, we've been... Many people have been so afraid to build their own software, do their own tools and so forth, and the vendors have successfully kept professionals out of doing a lot of that. And the profe- It's not only the vendors, some of the professionals themselves too haven't learned it. Well, now, you know, creating a GPT is easy. Creating a Gem is easy. These are power tools, and just, this is my, uh, one of my houses was built in 1896, okay? And the, the... I remember I was in the basement one time and I was like, "Why did they build up three courses of granite here? Why didn't they just dig down more?" And it occurred to me that, "Hey, they were digging by hand." And maybe they had a horse involved or something like that. And then when you look in my living room, it had curved windows. Why did it have curved windows? 'Cause they had glazers on site and they had steamers and they could bend stuff. So none of that stuff happens anymore. And my, my grandmother, who lived to be 105, told me about my grandfather who was working doing hand polishing in, in the Quincy Granite yards for, uh, headstones, right? It's like none of that stuff happens anymore. That's what's happening now. We're going from hand tools, hand saws, hand drills, hand polishing to power tools for all of knowledge work. And you, uh, who can compete? Now you may lose a, like... It cost me a fortune when we had to recreate the curved thing, right? Because nobody does that anymore. So we'll lose a few curved windows, but in general, things are gonna be a lot more productive and a lot cheaper
[24:14] Christina Ellwood: Yeah, for sure. And I saw an interesting report, John. I wish I could cite for you right off the top of my head who did the report, but they interviewed C-level executives in enterprises and asked what their plans were, career plans were, and 30% of them were leaving their jobs. They were retiring. Wow. Yeah. And now many of them were retiring to something else, but they were leaving the jobs that they were in because... Now, they didn't say why, but one- Okay ... speculation by the part of the company that did the research was that they, that what was needed in that job, they could see- Sure ... that what was needed in that job was not what they brought to the party. That someone- Yeah ... with different skills and perspective was necessary in order for that role to be properly contributing to the company.
[24:59] John Sviokla: Sure.
[25:00] Christina Ellwood: I think it's really i- in alignment with what you just described.
[25:03] John Sviokla: Yeah, what's fascinating is we don't teach labor history anymore- Yeah ... if we ever did.
[25:07] Christina Ellwood: Yeah. It is a shame that we don't teach that l- that law be- or that history because there are, and the laws that go along with them. Sure. Because I think we take for granted a lot of the operating systems that exist today, and if we understood how quickly they could be changed, we would have, we would have a different approach to, uh, looking at the future. We have a lot of despair right now, and I don't understand that because I've, I, it seems like such a, an op- such a hugely fertile time that we are living in. Yes. And so to be despairing, to me, is anathema. But perhaps that's the way it was in the Industrial Revolution as well, that people were despairing over the end of their, of the lifestyles that they had known or what have you, as well as being excited by the opportunity of fundamentally changing what their own personal future looked like.
[25:54] John Sviokla: I think that what we've lost, I think, and I think a lot of executives should think about this in a very practical way, is who should get the value? Like, when you automate something, when you create a better way to do something, there's only three places the value can go. It can go to the customer in a cheaper and better product. It can go to the investor in a higher return, or it can go to the labor in higher wages and compensation, whether it be long-term or short-term compensation. Okay, those are the only three places it can go. Right now, I think a lot of the people have despair because the power of labor has been on a 40-year decline, and you can see that by the percentage of GDP that goes to labor in terms of economic return. It's gone from about 9% down to about 6%, if I remember right. Anyway, it's gone down, right? Meanwhile, the top has gone very high. And then you have this fanciful thinking. You have somebody like Elon Musk, who I think is an incredible entrepreneur, but I think an idiot of a human being, and he says stuff like, "Oh, we're gonna have universal basic income." At the same time, that same guy is gonna fight tooth and nail to have any taxes or any compensation come from his fortune. It has to come from someplace, okay? It's, what are we gonna just make it out of thin air? It's gonna come from someplace, and it's gonna come from, I believe there's plenty of surplus to go around if we think about what we want to have for the society. And within an organization, I'll tell you, from talking to some friends of mine about how we would-- We, I used to be part of a team running different consulting companies. We had a company, Diamond, which was publicly traded. We sold ourselves to PricewaterhouseCoopers. And at Diamond, we believed in very wide ownership of equity. We're pro-- because we thought that for lots of reasons. I'll tell you, I'm looking at a company that we, I don't think I'll do it in the near term, but might buy or be part of a team buying, and the first thing I would do is cut all the professionals in on the upside of the labor automation. If I wanted the, uh, this particular company has about 6,000 employees, I would take the top thousand people and cut them all in and say, "Look, we're gonna go from labor-driven to capital and labor. We're gonna go from services to products and services, s-services, software, and you're gonna not just get your salary, but you're gonna share in the equity upside, 'cause I want you to accelerate this transformation for me."
[28:08] Christina Ellwood: Well, that's very smart. And of course, y- I completely agree with you that we need to reinvest the money. And I have to say, I think most boards and leadership teams are thinking the same way. I know we criticize them when they put it in their own pockets or only give it to the shareholders. But a significant percentage of the, it, o-over all of the, the transformations we've seen in the market, the vast majority of the investments in savings have gone into growing the business. Not in putting it in shareholder pockets. Some fraction goes there, some fraction goes to the customer, but a significant fraction of it goes to growing the business, which means new products, it means new markets, it means acquisitions and grow- growth by acquisition, and things like that. It's never zero that goes into the business itself.
[28:55] John Sviokla: Well, but going into the business doesn't necessarily mean going into labor.
[28:59] Christina Ellwood: That's true.
[29:00] John Sviokla: I think that if I look at the way Amazon treats its people, terrible. They don't give a hoot, and they're automating the living daylights out of them, and they, they push down wages as much as they possibly can. They fight unions like crazy. So I don't... And I think they're very successful in terms of being an AI first company, so I don't think they have any concern for that. And I- Are they the
[29:19] Christina Ellwood: majority, John, or are they the exception?
[29:21] John Sviokla: They're the majority by far. The, you look at what private equity's doing, they're not giving it to the labor. Private equity has a trillion dollars on the sidelines ready to get invested, and they already control a significant portion of the economy. There's no question it goes straight to the investor first, and the customer, um-
[29:38] Christina Ellwood: That's the business they're in though, right?
[29:40] John Sviokla: Yeah, but they're-- More and more businesses look like that. I just look at the numbers. The numbers are that labor has been getting less and less of the value.
[29:47] Christina Ellwood: No, your point about labor is absolutely right. I, when I refer to the business, I am not just referring to labor. You're absolutely right. It's investing in systems, it's investing in operations, it's investing in the types of businesses that they're in and the diversity of the business they're in, which does not necessarily mean labor. Yeah. So you're, you're correct that the labor part is losing on every front, and it really has lost in every generation of technology.
[30:10] John Sviokla: Except right after World War II. World War-- Right after World War II, we made huge investments as a company. First of all, about 10% to 12% of the GDP was invested in looking stuff like new science and capability, the kind of stuff that Musk and Trump cut, the leading science and things like that.
[30:25] Christina Ellwood: You're saying the country invested.
[30:27] John Sviokla: Yes.
[30:28] Christina Ellwood: Not the company, you meant the country.
[30:28] John Sviokla: Yeah. The stuff that DOGE cut, I'm not talking about USAID, which is another whole conversation, but the research stuff was the stuff that this country ha-- We used to spend twice as much on a percentage GDP basis than we do today, and we all benefit from it. Like, why do we have great medicines? Because of that. Why do we have the internet? Because of that. Why do we have self-driving cars? Because of those government investments. Musk's company sits on the top of a bunch of government investments, right? We literally bailed him out with $100, $100 million loan from the government that the government didn't charge, take any equity for, kept them out of bankruptcy, and then the contracts on top and the science he sits on top of was all publicly created. So, you know, NASA mostly. So, you know, that public good, we need to continue to invest in. Then there's issues, then there's distribution of that knowledge. So back after World War II, we had the GI Bill, which increased labor liquidity, right? That made it easier for me to buy a house. It ma-- I could, for less than a year's salary, I could pay for my whole college education. I got into all these colleges and, and that kind of capability, that we built infrastructure so at the government's expense so that people could just buy a car and then have transportation around the whole country and have incredible labor liquidity. And there's this wonderful quote, 'cause they were worried. They made a bunch of these investments because they were worried about the tens of million, the over 10 million guys who were coming back from Europe. They'd been in wars. They were afraid. They'd seen communism. They'd seen fascism. They were worried about what their politics was gonna be, and so they wanted to make sure they put them back in a capitalist lane or make sure they stayed in the capitalist lane. And there's this wonderful quote from the guy Levitt, who built Levittown, which were the manufactured housing in the New York area and then all over, inexpensive housing. That was part of the GI Bill, was part, right, that whole thing. Levitt said, "Hey, look, if a guy has a wife, three kids, a dog, two cars and a house, he's gonna be too busy to be a communist." Okay? And-
[32:18] Christina Ellwood: That's a good quote. Yeah. But what you just described, I think illustrates my point, which is that it was government that made those investments in labor. An incentive business to invest in labor, not business itself. So Uh,
[32:33] John Sviokla: uh, yes, I think it was a combo, but government... First of all, in the business community, there was an, there was a concern about the common good. When was the last time you heard Jeff Bezos, Elon Musk, Eric Schmidt, Sundar Pichai talk about the common good? I, like nobody, right? And so there's that
[32:57] Christina Ellwood: So now we're back to talking about the social impact?
[32:59] John Sviokla: Yeah. I think it's a huge deal. And I think those-
[33:02] Christina Ellwood: An opportunity to talk about the social good again, right?
[33:05] John Sviokla: Yes. I also think it's gonna be the fastest way for businesses to transform, that if they are planning on giving it all to the investor, it's... Look, we talk about the Luddites, and Ned Ludd never really existed as far as they know, but it was the name of the movement, and the people who followed it, they were violently put down in England, shot to death, hung, stuff like that. But they weren't wrong. Their whole way of life was changing, right? If you're a weaver before and after the Industrial Revolution, the way you ran your family, fed your families, ke- taught your kids, everything was blown up completely. And they weren't stupid. They just didn't, they weren't welcoming of the stuff, and it was forced on them by the government and by the military in England at the time.
[33:46] Christina Ellwood: So we have a chance to talk about the common good again. Right. And an important topic, and it's one that hasn't been on the agenda, social agenda for a while.
[33:55] John Sviokla: Yes.
[33:56] Christina Ellwood: Who do you think will create the- Conditions in which the conversation persists rather than just get started
[34:05] John Sviokla: I'm-- First of all, I'm incredibly optimistic. I have to be. I have five kids and two grandkids. The, I can either get super depressed or be optimistic. No, but I think that what we're starting to see is that I, I think that, first of all, the generations behind us, I'm the last of baby boomers, and the generations behind me are... do have a concern about what the common good is. So I think that's a good thing. We have folks like A21, the Allen Center, they're doing fantastic work where they're doing open source models, everything from the data, 'cause you can't get the data from the Chinese. You can get the data from the Allen Centers through the training routines, through the weights, through, right, so fully open. So there's more and more emphasis there. I'm going down to some friends at Carnegie Mellon in the beginning of the year. We're gonna be talking about how they're trying to create an incredible learning teaching platform, and as they educate their students, the students are leaving with robots, but also with open source libraries and the ability to build these robots to share and so forth. So there's great efforts going on on that stuff. I think there's that. I think the other thing is that it's not gonna take a lot of money, I think, to begin to spread knowledge more fully out throughout the world. We're involved at GAI in a, an initiative in the Caribbean where we may be partnering with a major telecommunications company to actually bring all kinds of skills, everything from technology skills to financial skills to basic life skills and medicine and medical knowledge and so forth, throughout the Caribbean through the modality of the cell phone using the models. So I think that there's unbelievable work that we can do. I just hope that we as a country will do what Ben Franklin did. Ben Franklin started the public library system in this country, and we need the Ben Franklin of AI.
[35:47] Christina Ellwood: All right. Let me know when you find him, and we'll, or her, and we'll support them. I think that's a, that's a good note to wrap our conversation on. Let me ask you this, Jon. You've been a leader your entire professional life. What leadership skill do you find is most valuable to you today as you work with companies and institutions who are trying to make this transition into the AI era?
[36:14] John Sviokla: I think it's the, the ability to go across two, two dimensions. One is to make sure that you can deal with the rational stuff that people are interested in, but also the emotions, 'cause this is hugely emotional. And AI, all of the West- all of the myths about advanced technology in the Western coda are negative. Prometheus, Icarus, Frankenstein, Terminator. And I think people And I think there's a very fundamental thing that this is hitting on, what's our place in the universe? I don't mean to get all cosmic with you, but that's huge, and it's emotional, and it's real. And we can't just ... So leaders, on the one hand, have to say, "Think of all the fantastic thing you can do. Demonstrate, be more productive, beat the competition, serve the customer better." But at the same time, to be fully ready for the emotional implications of that, just the way I think the great leaders in the Industrial Revolution did as well. So Henry Ford doubled the going wage for his workers, right? So both because he wanted to get the best workers, but also because he wanted to make them into customers. He was a Keynesian before Keynes came up with his formulas, right? So I think that ability to go rational to emotional. The other thing is from Miyamoto Musashi, who is, legend has it, the most successful samurai that who ever lived. He had 57 duels, and obviously didn't lose any, 'cause he was still around after them. The ... And so he had this fantastic phrase, which was, "The close view of distant things and the distant view of close things is the essence of strategy." And so how does that translate? The close view, the distant view of close things is I better get my people unskilled in using this stuff and understanding it, because talking about Frank Zappa had this fantastic quote where he said, "Writing about music is like dancing about architecture." Okay? And I think writing about AI is like dancing about architecture, right? It's you can't experience it till you experience it. And so that's the close- that's the distant view of close things. What am I doing every day? What does that lead me to? Then the dis- then the close view of distant things, when I think about if I go back, if I go out to 2030, four years from now, can I imagine an organization where people aren't gonna be using robots for different parts of things? Reviewing a legal contract, sending a picture back to a customer, putting together a proposal, doing an engineering drawing, teaching my kids how to, about mathematics. Am I gonna do any of that without a robot? I doubt it. So I have to take a close view of that. That's not there now, but that's a distant thing. It's not that distant. So I would say those two things, being able to do the rational and emotional and to be able to see what is it you need to do today for your strategy and what, how can I see the thing that's gonna be, that's in the future but gonna be important?
[39:02] Christina Ellwood: Great. If our listeners remember just one thing from today, what should it be and why?
[39:08] John Sviokla: Ask a robot. The, the, I don't do anything of consequence, give or take, without asking a robot. I have it review proposals, any important email that goes out. As I'm thinking, as I'm teaching my granddaughter about gravity and oceans I'm just asking the robot all the time. So I think that's super-duper important. It's a totally different way to work, and everybody should embrace it.
[39:33] Christina Ellwood: What resources do you recommend to listeners who wanna learn more about you and your work?
[39:38] John Sviokla: Yes. We c- folks can feel free to reach out to me directly at john@gainsights.com or john@svilko.com is my personal email, which is fine. The, we have at GAI Insights, our whole reason for being is to help organizations understand the power of AI and transition to hybrid organizations as fast as they can. And what we, in order to do that, we have a daily AI news show that starts at 7:00, 7:30 to 8 o'clock Eastern every morning, and that's available on YouTube as well as X, and also on LinkedIn. And you can also, if you don't wanna watch the show, you can sign up for the newsletter. We send it out every day, five days a week. That's huge because that's moving-- And we rate the articles essential, important, or optional. So we do our editorial. Then we have a learning lab every Monday night that's open to everyone. Last week we did how to build GPTs, and so very practical hands-on. We've got over 3,000 people in that learning community. And then the last thing is that we have a tremendous amount of things that we participate in that are, that we publish all the time at gainsights.com and, for example, we're coming up in January, we'll be going to Davos at the Imagination in Action event. In there, we're gonna be talking a lot about the hybrid organization and what does it mean to build an AI-first organization from a leadership perspective.
[40:56] Christina Ellwood: How do businesses that wanna work with GAI Insights get started?
[41:00] John Sviokla: Generally, we do, we start with businesses on one of four areas. We help them with executive briefings, think of that as a top-down understanding and training, so top-down, bottom-up, the, and skill people on what's going on. We help them with strategies. What does AI mean to me? How quickly do I have to worry about it or not? We help people with news and customized news so that, for example, we're working with a private equity company, and we, every two weeks, we summarize trends that are relevant to their businesses, and we give them important case studies 'cause we like to stay practical. And then the last thing is we do conferences and communities. So we just had our first ever group of folks who are worried about optimizing from answer, from search engines to answer engines across eight different organizations. And we have our yearly conference. We also have a number of other conferences we participate in, Davos, Imagination in Action, and others. AI Realized, and thank you for that, that we participate in. So the easiest way is you need strategy, you need training, you need briefing, you need to understand what the heck's going on, what does it mean to me, and how can I get started?
[42:02] Christina Ellwood: All right. Thank you very much. John Sviokla, the co-founder of GAI Insights and the Harvard Business Review, or Harvard Business School executive fellow. I am addicted to your newsletter. I read it every morning with my morning coffee. It is fantastic, especially since they are categorized. So if I'm short on time, I might only read the essentials and save the rest of it for later in the day. But it's a fabulous way to stay up on what's going on, and I know many people who listen to the show later in the day when they are walking their dog or taking a break from, from work and listening to the YouTube. So it's a great gift that you do that for the community. It's 7:00 AM Eastern Time, so it's a little early for me. That's why I'm not listening to it live. And the Learning Lab is at 4:00 PM on Pacific Time, 7:00 PM on Eastern Time. Yes. So all of those resources can be found on the GIA website, and we'll include them in the show notes as well. Our listeners can benefit from that. So John, thank you so much for joining me today.
[43:02] John Sviokla: Christina, thanks for all the great work you do. It's always a pleasure to be with you.