Latest / Lucid Interval Podcast / Ex-Naval Intelligence Officer: Who Leads When AI Thinks? - Adrian Wolfberg, PhD
Transcript
- Andrew Heacock: who leads when AI thinks? Adrian Wolfberg: Yeah, so you know, before AI came into existence, there was really no question about who leads. you know, a person leads. So, you know, of course it depends on a lot of things, but it's always a person. And we're entering we have entered already into a transformational age where we're sharing intelligence with another entity that's not human. So there's this for the first time in existence, in human existence, we're co-sharing a space with a machine intelligence. It's never happened before. And so in and that's not going to change. It's not going to go away. It's going to get it's going to increase more and more. So the challenge is with this in this new space that we're in, how do humans and machines work together? How do they collaborate? How do they how do we you know get the best of both worlds and and and and we don't want to and we want to avoid any kind of know, bad situations by doing something stupid by letting AI do everything or let humans do everything and not take advantage of what AI is AI is could do. And so so it's important to figure out, just like in any relationship, so just think about so if we stop thinking about AI for a moment and just think about how people relate to each other, whether it's a family, a marriage situation or a team at work or some kind of societal network of people for either fun or for for purpose or whatever, there's always a dynamic of figuring out the relationship between how each of the individuals are going to work together. And with the so the difference is that AI is like almost like an alien creature. It's it's totally different than any kind of tool that has ever been created beforehand. And so I I I intentionally don't refer to AI as a tool. It's it's not a tool. and I and I'm sort of careful with the words that I use because So I kinda have to go into this into this kind of introductory remarks to to answer the question of who leads when when AI thinks. so everything that's ever been constructed by humans, going back fifty, sixty, seventy, eighty thousand years ago to just before The introduction of AI as as we popularly know it, you know, a few years ago through generative AI, was constructed with intentionality, with a specific understanding of what the inputs were to whatever device, or whether it was a device or some kind of process, what the the output was, how the output, how the output was. designed to meet a certain need and if that need wasn't met then we understood the process so that we could modify the process to get the output that we needed. and this is true with electr advanced electronic things to things like hammers or saws. But in AI, none of those characteristics apply. There's this Lack of under we don't understand the elements within the process from point of ingest to the point of output. we don't have control over what the input is, and we certainly don't have control over what the output is. So the the idea of kind of modeling how humans engage with each other to apply that to how humans and AI would relate to each other is not gonna work. the the dynamics are not gonna be the are not the same. the approach to a collaborative venture between machine intelligence and AI is to figure out where the role of AI can play and where the role of humans can play. And it we've you know it's we're we well understand what the what the advantages of AI are right now, things like the speed of of inputting lots of data, the ability to identify patterns very quickly, generate scenarios, even evaluate scenarios, things that the speed of which humans put I'm sure could never could never do. And so to match those or to compare those with human capabilities. obvious things like critical and creative thinking, framing framing a problem, empathy and compassion, accountability, understanding consequences. who leads is going to depend upon a number of things. It's it's going to be contextually dependent upon the nature of the problem that's trying to be solved. It's going to be dependent upon our human understanding about the world around us and how it applies and is relevant to the problem that we're dealing with. And to understand the an exquisite understanding of the nature of the problem itself. these are most problems that aren't solved are are hard or difficult or or complex. And those are the kind of problems that I'm thinking about. Something like, you know, how do we manufacture toothpicks, maybe on the very easy part of the the problem spectrum in in which case there's gonna be minimal human you know involvement. so having an understanding of context of the nature of the problem then the the leading part is gonna i i it's it's a little I don't think it's a misnomer, but it's really a design question. You know, maybe if I would have retitled if I would have retitled the book, it might have been who designs when AI thinks? And I think that question is always the the human does that. But in in the in the actual participation in the human AI collaboration aspect, depending upon the context and the problem and so forth, We are going to be able to identify what the kind of ratio or what the kind of mix is on how much of AI or how much of of human involvement is going to be appropriate to address certain aspects of the problem. no problem is is unitary, every problem is broken down into pieces, and the trick is going to be able to be able to understand the the decomposition of complex problems and how they interact and how and this is going to take time to figure out. It's not like it's it's not like you can take a course right now to figure this out. We don't we don't have the language right now to to do that. This is going to take a couple of decades, maybe faster, to to sort out. So problems that are you know, very complex, are wicked problems that have ethic and and value implications, certainly problems that are are novel. You know, there's a certain characteristic about novel problems that make humans particularly appropriate to address. And why that is is that we well y humans are good at Creating new frames, AI is good at operating within a frame that's well explained. But that is a kind of a disadvantage for AI because it makes it difficult to come up with a new frame based on whatever the context we're in, whatever the problem that we're trying to address, whatever the the nature of the problem is. So novelty, complexity, wickedness, these are all elements that humans, particularly humans, are appropriate to to analyze. And from that analysis, which is going to take time. So time itself has to be looked at differently than it ever has done before, at least from how civilized organizations and societies have looked at time really since the beginning of the Industrial Revolution. Where you know time has been a disadvantage, it's been a constraint, it's been something that we're fighting against. And certainly we see that in everyday life. And you know, we just s we we see that in how smartphones are being used and and even how lar l large language models are being used like chat, GPT, John I and so forth. And the we're gonna have to change how we think, we're gonna have to shift our understanding of time. from something that's negative and we're fighting against to something that opens up opportunities to to kind of dance in the in the the the human care the human characteristics that that are only human and and to be able to bene take the time to benefit those. Yeah, so it's the the the question of who leads is is more like a a menu to a recipe that you're that you cook. You've got to figure out how much of this and how much of that. And the difference is that what's different than a recipe as as the kind of analogy is that a recipe every time is the same. You don't have to change. it's always going to be the same. But operating in a shared consciousness world, you know, there's always been change, but it's gonna increase more rapidly. the this whole process will have to be a continuous, iterative process of reevaluating what what reality is going on with with What the context is, what the relevancy of the context is to the nature of the problem, how the problem itself has morphed, how the solution or the approach needs to be changed, and then how that drives us to perhaps modifying the roles or the ratio of how AI and human involvement play in various aspects of solving the problem. So that's kind of at a twenty-five thousand foot look at who leads when AI thinks. It's not a unitary answer, it's well it depends answer. Andrew Heacock: Yeah. Well, you're a former naval flight officer, right? So twenty five thousand Adrian Wolfberg: Yeah. Andrew Heacock: feet is probably a comfortable height for you. Altitude. I I Adrian Wolfberg: Yeah. Andrew Heacock: Hey guys, if you've made it this far, please subscribe to our channel on YouTube. It's free and it helps us bring you more conversations with great guests like this. Now, back to the show. I I want to dig into that. One of the challenges I've had of anything about like organizational decision making or like you know, r learning about this is the challenge I have is concretizing some of the problems you're talking about. And I'd like to dig into what we you mean by a complex problem, but I think it might be good to talk about your background first, just to give the Adrian Wolfberg: Sure. Andrew Heacock: the audience a little bit of an idea of like the The kind of scale and scope from which that you've been you've been thinking about these things. Adrian Wolfberg: Yeah. well it goes back a long way. I kind of feel fortunate that I've had the trajectory that I've had, although I didn't always feel like I was, you know, intentionally, you know, creating the causal pathway to get from A to Z where I am. But it it's I kind of go back to my Navy days when I was mission commander on a a seven person jet aircraft flying off aircraft carriers and we were doing intelligence operations against at that time primarily the Soviet the Soviet Union. And what what what that experience there was kind of a dual aspect to that experience. And this goes to the issue of complexity and uncertainty and dealing with ambiguity, is from a from an organizational perspective, what really struck me was that in this seven person crew, I wasn't the pilot, but I was the mission command I was in charge of the mission and decided what the goals were and what everyone did and so forth. No matter what the seniority of the individual or what their expertise was, we were all in the same boat, figuratively, we were all in the same situation dealing probably with one of the most dangerous most dangerous pieces of equipment there is. And so what whatever whatever would happen to one would happen to all. So that's what I kind of mean by we're all in the same boat. and it it really solidified to me, above and beyond that, the importance of teamwork. So to understand what everybody knew, what they didn't know, how to develop areas that they were short on. where confidence was earned, where confidence was apparent and how that needed to be increased increased, how communication was handled b between each of us and people who were outside of us. So it was can became a like a little microcosm of a lot of issues that occur just within an organization. And then the other side of that was the kind of mission purpose, which for us was to understand the context of the world around us as it was occurring. and then to my role was to while in the air to manage the operation. But then after we got up back on the aircraft carrier to synthesize all the data that we were collecting and to wrap it up into a report within an hour or two or something. I don't remember it, but a few hours. And hopefully it would be usable to people who I had no idea who they were, who were any also all around the world or whatever. And so there was there was you know constructing knowledge but without any really constructive feedback about you know how it was helpful or or what was w what needed to be improved or or whatever. So that's kind of that kinda be that kind of gave me the launch point for wanting to understand you know how knowledge was created and And the challenges and what can facilitate knowledge being turned into action. Because, from a mission perspective, the whole point of the existence back then was that someone would then use the information for some kind of beneficial purpose to protect national security. So that was the launch point for this kind of bumper sticker of my focus of turning knowledge into action. And so I I've gone through lots of different experiences where I've seen where that has mostly not worked and so learned from that where it has worked and you know things that would be well let me give you a story about something that that would be complex but Yeah, something that would be complex in a very short amount of time. So I was I was working as a kind of an organizer under the White House for a multi agency effort to produce intelligence on an issue. And the secretary for that function, the cabinet secretary for that function wanted to improve the relationship of the United States and a foreign country over this over this topic. And this foreign country pres pre president was adversarial. But there apparently was a a an opening where there could be this dialogue between the US and this foreign country over this particular issue to to bring Some kind of collaboration together. And so, as you can imagine, any kind of political issue in the international realm can be very complex. And there's lots of moving parts, both internally within each country, and then the interaction between the two countries and how other countries may view that interaction. there could be personal issues as well, personality issues that you'd be walking into. There could be historical issues that made the relationship very tense and distrustful. And so that's kind of a kind of a s kind of a simple way as that I can of describing the complexity of the situation. Many moving parts. Not obvious how each of the parts connected to each other or ca or or would cause an effect to each other. the context was filled with kind of ethical and and value laden perceptions on both sides of who was doing right and who was wrong. So it's not only complex, but it it was on the verge of being a a wicked problem. that's A way of describing problems above and beyond what's a complex problem, which a complex problem is, kind of comes from our understanding about how systems work. Many moving parts, interdependent, difficult to assess the cause and effect relationship between the moving parts. Whereas wicked problems often combined with complex problems that have significant value propositions associated with them could be about morals, it could be about philosophies, it could be about ideologies, all s these all these belief systems that are built into how it be embedded in this. So I just wanted to lay down what Andrew Heacock: Mm-hmm. Adrian Wolfberg: the best I can to describe the complexity of this. So my role was to get dropped into this. And I went to the foreign country with the team, and my responsibility was to get the foreign president to get off of his horse and see that there's a benefit to some kind of collaboration with the US to address a common problem. You know, that was kind of the politics of it. And the challenge was that the foreign president didn't speak English, and I didn't have a command of the language that he spoke. So I had to accomplish the end state, which was causing a shift in the mindset of the foreign president that would allow him to feel safe. To establish a different relationship with the United States. But I had to do that without depending upon language, because there would have to be interpretation, an interpreter on both sides, and who knows how well interpretations could go between English and the other language. So I had to use visualization, and so I was able to. Figure out a kind of very narrow pathway in this very complex situation to show on one slide information that the foreign president would say, Aha, okay. I'm willing. Okay, now I see what the issue is and it changed my mind. So I mean that's one type of complex problem that had a very short lifespan. There's many complex problems that have very, very long lifespans that you know have not been solved. you can just think about in the international arena how different countries interact with each other or or domestically how different subpopulations are at polar opposites and so forth. Andrew Heacock: Yeah, I've been thinking about trade negotiations Adrian Wolfberg: So I'll just c pause. Andrew Heacock: as you've been talking about this as well. I mean, that's all the factors, all the political factors you've mentioned and social factors. And then you add in macroeconomics, you know, which is just enormous amounts of data that you have to incorporate and there's all these unintended consequences that could come from it. Adrian Wolfberg: Yeah, and a lot that kind of uncertainty of the unintended consequences comes from in general, it it comes from a diminished or or a lack of truly understanding reality as it's occurring now in the present. And it's it becomes a mixture of wishfulness, either Of retaining some element of the past, of how the past operated. And this applies to us as an individual, you know, w whether we're in a family situation or or in a political situation at the international level, or a wishful situation about the future. And the reality of the present gets squished and squished and squished. switched, you know, i into smaller and smaller ways of contributing to understanding if these perceptions of the past and the present and the future have overriding kind of power in in how we're thinking about things. And that that is you know such a common thing and and I think everyone can you know relate to that the difficulty of understanding the present. not only about how we individually bring things or collectively through through groups or political parties or whatever the form of organization we're operating in or companies or whatever, how we bring the past with us into the present and how that is then used to shape the future without really an exquisite understanding of the present itself that's not biased or shaped by the past. And that's done for a lot of different reasons. and I think, you know, we've been in the pre-AI era, we've been resilient enough, or most organizations have, you know, been resilient enough until there's some been some kind of cascade, some kind of collapse, where the disconnect between reality as it truly is and what one hopes it is becomes so stretched out that it just snaps and that's when you have things like organizations going out of business or organizations you know losing market share or you know couples getting divorced or or teams having infighting, you know, whatever the situation where where people are involved, when the disconnect with reality occurs, that becomes problematic. Andrew Heacock: I've heard this saying, there's a somebody I was asked a question, how do you go out of business? And he said a little bit at a time and then all at once. and I've heard the same thing about divorce. Adrian Wolfberg: Yeah. Yeah. So Andrew Heacock: Yeah. Adrian Wolfberg: what what's happening is that and this I'm I'm writing in the I've just finished a book and it's called Drift and I haven't sent it yet to the publishers yet. But this is about how we, whether it's individually or as an organization so I'm just gonna go into why it is that you just said Andrew Heacock: I'd love that. Yeah. That's fascinating 'cause it's Adrian Wolfberg: Yeah. Andrew Heacock: like 'cause of course it's human nature to be like, Well, how did you not see it coming? You know? Adrian Wolfberg: Yeah, right, right, yeah. So there's a lot of things that go on a and as and as the the method of organization gets more sophisticated, going from the family unit to the company unit to the societal level, more factors come involved. but you know, in general it's maintaining a way of understanding the world back in some past, whether it's the past in which the issue or the problem that's been worked the the that way it's been framed then or maybe there's something that there's some history prior to that that may be shaping how how one sees the world You know, you could think about a founder of a company seeing the world from time zero when they began. And as the company expanded and became much bigger, you know, to what degree does their framing remain the same as to time zero when they began as as they expand? And and certainly i it's kind of well been well documented that the decisions that that are made over time tend to shape the reality that we see and so so there's a narrowing of how we see reality in an organizational sense, even in a family situation, the decisions that have been made continue to have a shaping effect on what we see and how we under what we process, what we allow to be absorbed into our into our brain. And So the ability to upgrade or to update our understanding about what's going on is going to be dramatically shaped by these prior decisions or these prior these prior situations, these prior histories. And In in organizations, there's going to be other factors that are going to dissuade any ch any any deviation from that. there can be established policies and procedures that reinforce decisions that have been made. there can be cultural aspects that have become not formalized, but you know informally formalized into how or into why things are done. There could be perceptions about what is an authorized question to ask, and you know perceptions about who has the role and responsibility to make decisions or even to you know raise questions about decisions. So those kinds of things that I just talked about have an effect on can have an effect on preventing you know, observation observing and detecting what's currently happening without having the influence of the past and all the associated past that's shaped how we're seeing things now. And so this we can just go merrily along. certainly we can relate to this in, you know, personal examples that it happens at the personal level, but it it happens at or organizations like you say, go bankrupt or they just lose competitive advantage. that the the wheel keeps on turning until it breaks. And it that's obviously not the optimal way that we should be living, whether it's in our personal life or social or our business livelihood life, because it it's so disruptive, it wastes so much energy to wait until the collapse whether it's intentional or unintentional, it's such a waste of resources to try to recover at that point. So that's kind of why we do things like that. It's it's this framework that we've established. We need frameworks, you know, as humans, just to be able to operate in the world. And when we're not able to adapt and modify them, we're not able to learn, Which is kind of very difficult in the midst of everything that's going around, then we're just gonna keep on moving as if the world wasn't changing until the world collapses. That's unfortunate. Andrew Heacock: Yeah. Yeah. It makes a lot of sense th though, like in it like it it's kind of costly to build a frame. Like it when you're figuring that out, I mean, there's a reason that so much of our life gets set into routine because it's it's energetically it's just expensive to constantly be rebuilding the world. And if you've built a frame and that it's enabled you to start building a business, it's been working for you. It makes a lot of sense that one would be resistant to challenging that. And I'm wondering. What can leaders do, you know, assuming they don't want to go out of business, which I'm sure none of them do. I mean, like what can they do to be more vigilant about that or to know when they're kind of getting stuck into doing what worked in the past? Adrian Wolfberg: Yeah, so what this is a big shift that you're what the question that you're asking is gonna ha is requires a big shift in how we think about organizations. this becomes accentuated in the age of AI where the speed of AI processing of things that it can do well, like ingesting massive amounts of information very quickly and detecting patterns very quickly and generating scenarios and evaluating scenarios very quickly, all things that humans can do, but the the speed at which they're done by AI just, you know, it's millions of times greater. this kind of goes back to control and And you know, having intentionality in deciding what the most important source of power is. And I would say that y you know, we have a we have a history of understanding power as as something that's associated with those at the those in those in control, whether it's at the C suite or direct reports below or managers below the so any any sources of control has power in that way to do whatever's necessary or whatever's viable or or whatever. And that's where that's how power has been conceptualized. And you know, that kind of that kinda goes back to, you know, tribal days, you know, eighty thousand years ago and and got reinforced during the industrial revolution with mechanization and optimization. And the the the a the transformation that we're going through now with the age of AAI is creating a a completely different environment in which survival or success, I should say success, is going to be dependent upon a a different conceptualization of control and power. And this goes directly to the question that you're asking about how do you actually you know be aware of changes and therefore be more adaptable and so you you don't you're not surprised at the end of the day with a total collapse, with all the implications of that from going bankrupt to expending massive amount of time and resources to re-engineer something. so I'm talking about kind of the underlying system, the underlying organizational structure that will will help us move away from this old-fashioned understanding of control and mechanization with the associated frameworks into a different kind of mentality where frameworks would be more adaptable. So And this is not gonna happen overnight. It's gonna take, I think, decades. so and it's not like this is a PowerPoint briefing that someone or a PowerPoint training session. This is gonna have to be something that we develop the language to talk about. Educational systems are are gonna have to be modified, competency development and new competencies and organizations are gonna have to be developed. But basically the shift You know, thinking about this, the the shift from control as the source of power has got to shift from a source of power of laying the conditions for success. And in a you know, I think back to my military days, and in a lot of ways, some of that aspect was built into the system, although it was a very hierarchical command and control driven. But certainly setting the conditions for organizational success is very important to a military operation. And so just using that as an analogy, not to say that organizations are going to become more military, I'm not saying that at all. But that the source of power has to be democratized and so that everybody in the organization becomes like the eyes and the ears to what's going on in the world and how the world is affecting whatever the product or service that an organization, whether it's a for-profit or non-profit or public sector organization, you know, what's actually changing. And that then there's then you know opportunities, the discussions or whatever the mechanism is to develop very effective or efficiently a shared understanding about reality as it's changing. and it's that's not gonna be something that AI is necessarily gonna be that helpful with. i it's gonna be why? Because things that are changing for the most part have human connection to it. So it's how people think, how decisions that people have made, intentions that people are having. I'm talking about people whether organizations or societies and families. These are all things that we know or we should know or we'd want to know that are going on in in people's minds and how they're making decisions and what they're afraid of and what they want and all that sort of stuff. those are kind of the drivers behind a lot of you know except for things that AI itself are gonna produce but you know ninety nine percent of what the world consists of is humans and and their attributes and and what they do. So this kind of shift from a control as a source of power at the center of an organization to a distributed model. so that you have as many eyes and ears of of what's going on and then share that. And that's how people are going to become organizations are going to become more successful and you know, dealing with changes in the present in in my opinion. Andrew Heacock: As you've been talking about this, I've been thinking about just an example that I've I've heard from someone who studies flight safety. He's actually a physician, but he was a former flight surgeon in the Navy. And he studies safety science, you know, basically the science of how and and what he's looked at is how have we managed to stop or minimize commercial airline crashes. And one of the things he told me about that some of the changes that they made were so fascinating. There was a lot of them in Southeast Asia. And part of the problem was that they're in Southeast Asia, social hierarchies are very rigid. And in some languages in Asia, you you there's literally no way to tell your social superior that he's wrong or that he's making him like linguistically that's just not available to say to someone. And so you would have a pilot and you'd have a like a co-pilot that would be like an assistant essentially. And It would there was a social taboo. And so you would hear in like the black box recordings, people would analyze them and they would be like, This engineer clearly knows there's a problem. You know, they clearly knows the that this is a fatal error that the pilot is missing, but he just won't tell him. So a couple of the the interventions that they've or the innovations they've made is one, I think every every international flight in the world speaks English in the cockpit. but they've also flattened the hierarchies between the, you the pile any any one member of the team can say, hey, we're gonna die if you don't, you know, if you don't change here. And it is kind of a different paradigm. Certainly it's a different paradigm it was there in in those societies. But you know, the data speaks for itself because we just don't crash commercial airline like we used to, commercial aircraft. In fact, it's I I don't literally love to fly. I imagine if I was the pilot, I would be less worried about it. But when I'm just in the back, and we hit turbulence, it bothers me a lot. And usually what I go to is there hasn't been a commercial airline craft, you know, crash United States in however long. And that kind of gives me comfort of just to know the law of large numbers is like on my side. but yeah, it's it is interesting that I've been thinking about this also in the military too, because the military you just to me is almost inherently hierarchical. is there is there a way that they that paradigm might change? Like to being something more decentralized with decision making? or do you Adrian Wolfberg: Yeah. Andrew Heacock: think that that's something that's sort of just kind of like baked into the way the military operates? Adrian Wolfberg: Well with the military, I mean that's a good question. I Well, let me give you some examples of where this has worked. I don't know, you're are you former your former military? Andrew Heacock: I'm not. Adrian Wolfberg: Okay. So organizations have kind of faced this issue of getting out of the constraints of of organizations as as they have existed for hundreds of years. This kind of mechanized age that really just been part of the human evolution again ever since humans have been in tribes. And so it's very, very difficult to you know go from make a switch very quickly like that. And so what organizations have done, so I don't have the answer, but I can just tell you how how organizations have dealt with this in the past. I don't know if this is these could what I'm going to talk about could be kind of pathways to get to better outcomes, or they may be the way that we as humans just are designed and and built and operate in the following way. So for example, in the US well in a commercial example Lockheed back in the early 20th century was focused on building safe dependable aircraft for commercial travel all obviously safety was a prime consideration for a successful business model. But over time, especially after World War II, there was an interest in because of the speed of how the world was changing, there was an interest in building specialized aircraft not for commercial purposes, but for other purposes. And Lockheed was however was not designed for developing risky new technologies in very quick amount of time operating at in in environments that they had not been perfecting, you know, through their civilian commercial aviation profile. Well this led to the skunk works. everybody probably everyone knows that. This is where the U2 and the SR seventy one were developed. But in in the Skunk Works, which was a part of the organization, they had a completely separate operating model of of how to work on a day-to-day basis in terms of relationships. It was a completely different business model, more like this shared shared consciousness, shared decision making, minimize bureaucratic constraints that were designed for the original commercial aircraft routinization and safety which were the primary concern of building reliable aircraft. So they built these they built this skunkworks thing. The US Army had the same kind of situation. Obviously since in the for the US anyway, armies have existed for hundreds and hundreds of years, thousands of years. But after World War two, the the experience of operating in World War II led to a need to develop different ways of confronting adversaries than by the kind of overt publicized way of sending aircraft and ships to confront an adversary. And so this led to the special forces where they operate completely different without from a someone who's inside the system itself, i they operate as if they're a radical organization that don't have any rules and whatever. But you know, they have kind of like what I was flying in the in the airplane and off the aircraft carrier, they all have a exquisite understanding about each other's roles, each other's capabilities and limitations. They have a shared understanding of of being being able to assume multiple roles, and they operate in a in a in a way that's antithetical to the the structural system. in another kind of commercial organiz a commercial example, General Motors, back in the was it the seventies the eighties, the US was kind of being flooded with Japanese, small Japanese cars that were being sold cheap Did better gas mileage and so forth. So GM was really exquisite in building large, you know, behemoth cars, and that's what people wanted. So but you know, they felt this competitive environment from Japan. So they built this little outfit called Saturn Car Corporation, which was still part of General Motors, but they had they could operate on their own and completely different from the the big superstructure of General Motors and all the rules and regulations. And so Saturn became you know a very popular and competitive product to compete with Japanese cars for the time that it was the time that it was done. so those are just some examples of of how organizations in the past have dealt with being more adaptive And so there's this kind of conflict between, and we've known this for a long time, there's this conflict between excelling at what you do, which means that what you do is what you do, versus doing something new, which is not what you do. And so how do you how do you move between doing what you do to doing something new without losing what you do unless you intentionally want to lose what you want to do? And to become something new. Well, that's a ver that's a journey that most organizations fail at. And you know, it it goes back to a deeper, a deeper level of how do humans face newness. this goes back to kind of the the whole underlying issue that we're talking about. When things are outside of the frame that you've developed, if you've developed a very narrow frame. what what happens to those perceptions of things that don't fall within those frames that are are new? Are they are they recognized, are they analyzed, are they considered, or on the back end or on the other end of the extreme, are they just disregarded because they just don't fall into one's frame of the accepted way that the world operates? So I you know, I I'm kind of thinking that the transformation of AI that we're in right now may be an opportunity to figure out a a pathway that's that's not the keep things the way things are, but also it is and these things that I was talking about like Special Forces and Saturn Car Corporation and Skunk Hanks, Skunk Works are what's called parallel organizations. So there may be an opportunity to to find a pathway that's different than just establishing pathway organizations, which it's you can't scale up from that necessarily. And and you know how that's actually going to materialize? I can't envision that right now, but I can certainly think about the need to start using different language and how we talk about things, about what's actually happened now and to build upon that, you know, language of what's understanding and of understanding what's going on t as accurately as possible, to then figure out ways to you know to to you know Help insure success in using AI in in the best way that we can we can do so in in collaborating with this machine and and human consciousness. so I don't have the answers to your question, but I can see a pathway moving in that direction. And it it's not going to be a snap of the finger because we have to develop the the language to talk about what's actually happening. And what's actually happening is never happened before. Andrew Heacock: I've so I've got one other example for you from movies. I have a movie reference for everything because I love I'm a filmmaker. Spotlight. I don't know if you've Adrian Wolfberg: Great. Andrew Heacock: seen that movie with Michael Keaton in it. It's about the Spotlight team within the Boston Globe and how they broke the Catholic Church abuse scandal that happened in Boston. and it's a it's a movie that tells the story. but Spotlight Adrian Wolfberg: yabn. Okay. Andrew Heacock: was a Spotlight was a contained investig deep investigative reporting team within the Boston Globe that was sort of protected. They had their own office, they were when they were separate and they had their own independent corporate culture, and independence was very much part of that culture. And part of the one of the subplots Adrian Wolfberg: Yeah. Andrew Heacock: of the film is how as corporate consolidation has been hitting newspapers, you know, Boston Globe's a hometown, a local paper, basically. and things have gotten much more online, investigation has essentially been eliminated from journalism, from print journalism in a lot of cases. And so there's this corporate pressure to eliminate this team then because they might research a story for a year with before publishing on it. and as of course part of the film is that it was determined that there was still value in having them do that. And they broke this Pulitzer Prize winning story. there is also what makes it a good movie, among other things, because I think movies about paper chases are inherently fascinating. I've been someone who's done paper chases. things like that is there's kind of this rogue element. You know, you mentioned like Skunkworks. It's like these guys that are, they're kind of off the beaten path. And I think we, especially in entertainment and media, want to cheer for those people that are kind of the the iconoclass or the people that are a little bit eccentric. And I've been thinking is you've brought up novelty a few times. and I can imagine how in an organization like Lockheed, For example, someone like me who has a very, very high appetite for novelty probably would have just withered there. Like I just probably would not. I don't do well with like lots of hierarchy and bureaucracy. I'm wondering how with audit with innovation and with AI, do you think that different kinds of leaders are gonna need to be recruited that have different attitudes towards these things than what we've traditionally seen in the past? Adrian Wolfberg: Yeah, definitely. It's it's gonna be based on you know, competencies that we don't really develop or don't even yet that Andrew Heacock: Don't exist. Adrian Wolfberg: don't exist, right. yeah, it's this setting the conditions that is really gonna be key. so that's that's gonna be where power is gonna be needed, and that's really not a a power in the authoritative sense or the authoritarian sense. it it's power for everyone to achieve success and then to you know certainly there's always gonna have to be someone responsible for decisions that's always gonna be the case you're always gonna want to have some kind of accountability but the whole process leading up to that is gonna have to change in my opinion because of the speed of how things are moving with with AI and and yeah so you know in this in this this book that I just wrote on you know who leads when AI thinks, I talk about three competencies which are gonna be needed to be developed and and I know there's gonna be more but I just picked on three which I I thought would be commonly easily understood. one of them is cognitive acuity. you know, to be able to To be able to not get so s bound by the framework that whatever you established at whatever point in time, there's gotta be a lot of fluidity, that this kind of boundary that that you can that we construct for ourselves, either for ourselves or for our team or for our organization. has to have a high degree of permeability. i it so it can't be concrete, it's gotta be like mesh net so that there can be a lot of interaction and and so forth between things that come up that are new and how we understand how we understand the world. So cognitive acuity is gonna become Very important. this includes critical and creative thinking, which I'm kind of b hugely optimistic about its future, but I'm also hugely concerned right now with whether w we've adopted enough energy to ensure we're developing it at all levels, at at the elementary school level, at the high school and university level, and certainly at the adult w you know, work level, where wherever that is. Another one is this idea of you know, I call it learning, but there's you know, there's kind of two different ways that just kind of simplify things that learning is kind is conceptualized. One way that we normally think about it, and certainly it's reinforced by the way organizations as adults communicate information is here's some new information that you need to know. So here's a new skill that you need to adopt. Okay, it's I'm gonna put it in my toolkit. My toolkit's not going to change though. It's just going to have more tools at my disposal to do what I need to do within the frame of whatever my responsibility is. that's not the kind of learning that I'm talking about that's going to be needed in the in the future or as we move forward in the in the age of AI. I'm talking about a a type of learning Where our unders where our absorption of things that are new outside of our frame that we've established. Is is able to be integrated, not diffused, not distorted, not rejected. And we're able to now no longer just see a toolkit that we have, but we're able to envision different ways in which we're organizing the tools itself. And we're we we create new frameworks for understanding the world. And those new frameworks may be constantly, or I don't know, constantly, but may change over time because of Of changes that are occurring in the world. So that's the kind of learning that needs to be done in a rapidly changing environment. It's a kind of an accurate and kind of objective, so to speak. seeing of the present with as little as possible a hand holding from frameworks that we've designed in the past. So that's the kind of learning that I'm talking about that needs to be developed. It these things that I'm talking about I think should always have been there, but the the it's I think just the the human experience without AI, before AI, has been resilient enough or has been inefficient enough that we're willing to live with it to let these processes n not achieve what they they could oct optimally be. you've probably heard of boundary spanning. so I I c you haven't heard Andrew Heacock: I haven't heard of that. just to to yeah, maybe we could refresh on what that is. Adrian Wolfberg: Okay, so you know organizations are very specialized no matter if they're small or big, but certainly, you know, there's like the marketing department, the engineering department, the manufacturing department, the sales department. So each of these and and just think about you know universities or colleges and and how they have academic departments, psychology, sociology, physics, chemistry. each, whether it's an academia or in an org business context, when people operate within excuse me, when people operate within their silo within their vertical kind of organization, they adopt language that's suitable for working with people who are there to accomplish whatever goal it is. If you're working in the engineering department, you're focused on engineering des diagrams into mathematical equations. you know if you're working in sales you're focused on establishing relationships that could be a long term anyway every kind of professional element develops their own framework and their own language, the words that they use, their own assignment of what's important and what's not important. And so just like the the physical boundaries that exist around a people's house, you know, there's a property line around a people's house. or around a a commercial building or whatever. if there's a fence around the property line, that creates a physical boundary between two say adjacent homes, you know, exchanging, you know, if kids play together or whatever, or if adults you know communicate or whatever, that creates certain challenges. Well, in these in this conceptual sense of organizations where we have unique language, words that we use, we're focused on different things, they this whole kind of situation becomes a barrier, it can become a barrier to actually effective communication and accomplishment of mission where multiple organizations have to contribute to the mission. I mean you even see that w I what what would be some examples. I mean there's even well well known stories in the military of how the army has been unable to communicate with people in the air force because of the different languages, the different and I'm I'm taught people speak English, but th the different words, the different jargon Andrew Heacock: There's jargon. Yeah, jargon. I I find this all the time with people who are in certain industries and they'll say things. I'm like, I have no idea what that means. Like I've got a big vocabulary, but there's there's terms of art that take on specific meaning that are only that are context specific. And if you're not really a part of that context, you don't get the meaning. Even though you're like, well, I know what that word means, but I don't know what it means to you. and I can also see like in something in like the military, which is kind of famous for this. You know, there's acronyms, there's all kinds of I mean, there's that that's up kind of becoming a member of the military. I think joining any large organization or becoming a lawyer, I've said before, like going to law school, is a process of forgetting what people mean when they say words. Like you just have to unlearn what what people mean when they use words in their ordinary sense. And it's kind of almost a process of initiation, like becoming part of this tribe. and tribes like to to be tribal. You know, I mean that that's kind of inherently part of our nature. Adrian Wolfberg: Mm-hmm. Andrew Heacock: So I guess I'm I'm starting to see the picture come together of how this is a problem because within large organizations you have little fiefdoms, I guess. Adrian Wolfberg: Yeah, and and so the that's probably not gonna go away for the foreseeable future. So the idea of boundary spanning is is just kind of a general term for needing to be aware of this and needing to figure out there's various kind of strategies that you can use depending upon the nature of the of the boundary tension, the the boundary barrier. to to overcome or to mitigate this. And so this has always been the case, but this becomes more important as the speed of things increase, and how important it's going to be for achieving kind of a collective awareness about what's going on because it could be that someone in sales provides some valuable information that someone in in engineering wouldn't think twice to think about. So kind of reducing the boundaries while at the same time not losing expertise that's needed. is gonna be an an important competency. So, you know, why that's important. Well, w we've been developing specialized skills for gosh again since the Industrial Revolution, even before, but certainly because of the mechanization, we we've just drilled down to become more and more and more highly specialized. And that contributes to increasing the the penetration of boundaries, the permeabilities of boundaries. So this just becomes another competency that we we talk about it now. We've been talking about it before AI, but it's going to become more important in for organizations to be successful to achieve kind of a shared situation awareness of what's going on in the present so that there can be a a more adaptive approaches to modifying or realigning who does what in terms of the AI and human collaboration. Andrew Heacock: I could see how AI could radically accelerate the Balkanization as well, like or increan literally create reality echo chambers because there's I mean, there's sort of I mean, the more you use AI for a particular use, the more specialized it's gonna become for the use that you want it for. It's gonna take on your jargon, you know, and then it's gonna start to speaking the language you need to speak. And there's also a sycophancy problem, at least with the LLMs that we have now. So I could see I there's there's a such thing as AI psychosis. Are you for aware of this phenomenon? Adrian Wolfberg: I I wonder I've not heard of that term, so explain it to me. Andrew Heacock: Yeah. A well, people are people who have no history or risk factors for psychosis are becoming psychotic from talking to LLMs. And essentially they're the LLMs are gaslighting them, but they're doing it in a way that is confirming what they want to hear. So there's like so Adrian Wolfberg: Right. Andrew Heacock: you can imagine like having something that's supposed to be intelligent and it can be very convincing that it's right. But when you add your own confirmation bias to it, that becomes a dangerous feedback loop, right? And I could also see how this could be a problem. And this is, you know, people are becoming psychotic and thinking that there is one example of a man who lived in Oklahoma, I think, and he was a a working class individual. Let's say he was maybe like an electrician, or he or he f he was a ref he was like an air conditioner repairman or something. And he believed he had solved like Girdle's theorem, like some kind of mathematical theorem that had you know has has stumped the greatest mathematicians in time. And he just was convinced because the AI told him, You've got it, you're brilliant. And he really thought he had. He was going Adrian Wolfberg: Yeah. Andrew Heacock: to tell the news. And like he just and he was so embarrassed after it, but it's good that he was able to tell the story because it is a feature of the human mind. It's a big problem in mental health, especially with young people and they're very isolated. But I could also see how this could be a problem in things again like defense or in in an any decision, any situation where human lives are at at least remotely at stake. you really need to know what reality is. And AI isn't always necessarily good at. telling it like it's actually, you know, and I know there's more to AI than these LLMs that are being developed like Chat GPT, you know, but those are designed to basically tell you what you want to hear. Unless you really go out of your way to tell it, don't do that. And then they they still do it. Adrian Wolfberg: Yeah, so yeah, that's so I hadn't heard that term of art, but I know exactly what you're talking about, and that's a big concern. And w what people have associated that AI the large language model capability is to show empathy towards the human who's doing the prompts, by reinforcing what they're hearing and and so forth. And that's kind of getting confusing with a competency which which I also called empathy that the leader needs to develop in the age of AI. And it's not the it's not the situation that has been termed as the AI having empathy. This is the concept of the leader having empathy for the decisions that are made in this. world where increasingly there will be an AI component in the decision making process, whether it's the in which whether it's feed of knowledge coming into it or some kind of shared decision making in some kind of way. it's ultimately th the human who's respon who's going to be accountable for what's going on and and for the actions that are are result from the decision that's made. So a critical competency that's even more important now, I would say it's always been important, but we've not held people accountable for it, is to have a empathetic view of the effects that decisions in this shared consciousness between machines and humans could have on human populations. and and so that's the empathy that that I'm talking about and that's a competency that's gonna be needed by leaders in terms of you know almost maintain I don't want to use the word control, but not losing control over AI's involvement in making decisions and keeping the management of the collaboration in with the responsibility of the human and not with the not with the machine. Yeah, so this this the story that you were just telling me about this psychosis thing and how I know this has been the case, right, with teenagers that have been seeing this, makes more important for ed the educational system and certainly for for us who you know, either been trained t to be critical thinkers or we should be critical thinkers. you know, to have that disposition in the face of what's coming out of large language models or any kind of AI to evaluate what's going on in the face of our understanding of present reality, which I think is the only you know, the only the only understanding that's really accurate is gonna be not that not that humans are always successful at it. I would say humans are mostly not successful at it, but it's the humans who have the potential to be the most accurate about what's happening in the present today. Andrew Heacock: You used you alluded to something that has been called the alignment problem. you know, just you know, when you're talking about not allowing full control of AI to to escape. and one of the one of the the best example I've heard for the alignment problem involves the manufacturing of paper clips. It's funny you use manufacturing of toothpicks earlier as an example of a simple problem that you would Adrian Wolfberg: Mm-hmm, mm-hmm, mm-hmm. Andrew Heacock: you could conceivably outsource to an AI, but the and the paper clip, the paperclip Adrian Wolfberg: To some degree, yeah. Andrew Heacock: concern is that. You tell an AI to just make paper clips and so it will do whatever it takes to make paper clips and some and in so doing kills every living person because there's Adrian Wolfberg: Yeah. Andrew Heacock: just way more resources available to make paperclips, right? and so like that's the alignment problem of how do you keep it aligned? You know, you need someone who has who is ethical. You know, you need someone at the at the at the reins, but that person also needs to be empathetic. They need to care about people, they need you need compassion and one could see how those kind of traits, which have I think have not often been thought of as central to leadership. I mean, they're they're I I think it and I a lot of people would have seen them as like important but perhaps incidental to what makes a great leader. Now, when you have this kind of mechanization and automation, you need to have good judgment along that, you know, and also that judgment should hopefully be informed by empathy. Adrian Wolfberg: Yeah, yeah, so like in the case that you just talked about with a machine going to any length to continue to produce a supply of paper clips that it was asked to do, you know, w what you'd you'd wanna have is the machine you'd wanna understand the process of How paper clips come into existence, the whole supply chain and all the to map out that, you know, to model that that process. And so that at any point where there's deviation, that's when the machine would be responsible. That machine's job would be to report like say supply chain problems that need to come to the attention of. of the manager or the whatever. And so that kind of dissection that's why it it's really important in the age of AI to really be able to dissect, to decompose what the problem is, what the the the the focus of the attention is, and not just assume that there's some kind of unitary understanding that doesn't have to be explained any further to let's produce paper clips or let's produce toothpicks. and and I think I think what tends to happen is now there's you know people fall into an organization. let's see, I I'm just thinking about the the entrepreneurship of founding organizations and how this what I'm about to say would it It probably wouldn't apply because that's a kind of the beginning of of some kind of effort where everything hasn't ha and nothing has been mapped out so that there's routinization. But i i w i i when an organization gets to the point where there's routinization, if it does, yeah, then Being alert to changes, which may not come from AI, it may come from humans themselves who are at the leading edge of whatever supply chain process. So I I guess I'm just trying to, you know, not push back, but just to say that we have a tendency to name problems at some kind of macro level as if we could address. problems that way. So like a ridiculous example is well, let's fight against poverty or let's fight against human hunger. Well, those are huge problems that have many, many, many, many component parts. And probably the best you can do is understand all those component parts and and figure out ways to mitigate or improve either or some or interact with these parts. And that's typically I I you know, from what my experience and from my research, that's typically not how people operate, especially when there's this pressure of time. that's usually not a luxury that seems to be available. And And and so there's like a an assumption, I think, that there's a rec there's a an understanding about what a a problem entails and that the allocation of resources will just naturally be assigned to deal with the problem and that there then activity will commence, progress will commence, and this kind of goes back to this a barrier of the difficulty of seeing reality in the present because we've already set in motion this train by not in today in many of today's organizations to achieve the end so we're kind of coming back to this this learning how to adapt and face the present kind of issue Andrew Heacock: You the the example you gave though of how do we eliminate poverty? My first thought was, well, you could kill all the poor people. Like that that's an AI solution, right? Like there's Adrian Wolfberg: Ha ha. Andrew Heacock: no more poverty, right? But the other one about eliminating hunger, it called to mind an essay called A Modest Proposal. Are you familiar with that? It's it's kind of a famous work in literature. Adrian Wolfberg: Yeah. Andrew Heacock: It's it's written by an essayist named Jonathan Swift. I think in the eighteenth century, he was Irish. There was a potato famine going on and also in the 19th century. Adrian Wolfberg: Hmm. Yeah, yeah, yeah. Andrew Heacock: so there was a massive problem of poverty and famine. in Ireland. And he wrote this essay called A Modest Proposal. And it was his the solution that he offered was to just simply eat the poor people. and like there would be no more famine. So it was the ear one of the early examples of satire. And of course, like this is one of the things that concerns me about AI is that I don't trust its sense of irony. You know, like like I mean that that it might it might take a proposal like that because the thing is that if you read a modest proposal, it's the what makes it brilliant is It's extremely convincing if you dispose of the obvious repulsion of this taboo of cannibalism. Adrian Wolfberg: Yeah. Andrew Heacock: if you don't have this visceral human, you know, response to that, it's actually a pretty good solution. It's a pretty elegant solution. You know, there's a number of layers to it. And those are the kind of things that I worry about where if you just say, well, we have an end in mind, and the outcome is we need to eliminate poverty, but we, you know, you don't offer a lot of specificity about how you get there and you don't really account for the sub. you know, the the all of the various intersecting challenges, how could you do that? And I don't know that you could ever train an AI to do that. And even if you could, imagine like the resources you would have to commit to training an AI to do that as well as a person when instead you could just have a person doing it. You know, you could just have people kind of at the helm and allowing it to to do its its thing. Adrian Wolfberg: Yeah, I mean I think I kinda think at the forefront we need to preserve what's innately human in us in how we engage in this collaboration with AI. And that needs to be the focus. not y you know, certainly the technology will develop and will change but it that cannot be at the expense of what is human. And so that's gonna be kind of the interesting challenges in the future is negotiating you know how much humanity we want to give up or not in addressing issues where AI, which is gonna it's here to stay, is gonna be involved. So I that's why in the well in the book I I I talk about what I think people now focus on. there was there was an example that a colleague gave me the other day about venture capital companies going into buying companies and dismissing large quantities of employees and to have their function being replicated by AI and then having the organization finding out that that actually is not a very good idea and things were not going were not going well. So there's kind of now just a story to say that there's now there's many stories like that of of hey this appears to be a great thing. Let's just get some of this and let's start using it because we don't want to be left behind. And so there's this I don't know, unthoughtful or yeah, unthoughtful kind of belief that or a lack of thinking of just get me some of this stuff. You know, I need to get me some of that AI stuff. And I think the f we need to get to the point where we need to focus on what humans need and and how preserving what we do well Often often just to ensure the compatibility exists between AI and humans. So this is what I'm thinking of critical and creative thinking. We talked about empathy on the back end, framing problems. so just i you know, to me it goes back to we're we're always in collaboration with somebody. And when we're in collaboration with somebody even in hierarchical systems, there's some degree of collaboration. with AI, that's also gonna be true. And i just like it we're talking when we collaborate with people, we bring the strengths of people together to make make a better whole. you know, create a create a better understanding of the whole and and optum, you know. come up with creative solutions and so forth. we have to kind of take that same attitude with AI. we have to maintain the best of both. And so that that really means that it's going to be very dangerous to default responsibilities to AI when it when it shouldn't be. So it it goes back to this really understanding what the problem is and where and and and it what kind of ratio so to speak AI can contr can contribute with humans, you know, to what ratio, lot, a little, whatever. Andrew Heacock: Hmm. Adrian Wolfberg: and I could see in the future there there will be probably legal battles or ethical battles about what is given up to AI or or or what shouldn't be given up to AI. And that's going to be a challenge in these first few decades as as we as we're in now starting this in this transformational age where these things haven't come up yet, but I can see them coming up as as issues that we're never having had faced with before. And you know, to some degree you can see other parts of demographics dealing with things like this. You you mentioned ethical issues, moral issues, legal issues, religious issues, all these kinds of you know, art, art valuation, gonna become, you know, important. Anyway, I'm just gonna kind of stop there for now. Andrew Heacock: Yeah, I mean, it's actually one of my favorite guests I've had on this show so far is an attorney, Victoria Satanovic. And I went to law school. That's my background, it's political science and law. And she is a she's mostly a psychedelics and cannabis lawyer, but she also has got an AI practice and it's been on my agenda. We mostly talked about psychedelics and Zen Buddhism when I talked to her. But I after you just mentioned, I was like, you know, the the issues are racking up. I need to bring her on for another episode to just talk about. the many legal issues that come from AI. And one of them you just mentioned, I was like, if you set the AI off to go make paper clips and then it kills people in the process, who's legally responsible for that? Like who's who's answerable to that? Is that a tort liability? Is that products liability? I mean, there's there's a lot there. And I I think that the concern that I have is is in like the automation of decision making. Or do you know about the hugging face attack? Have you heard of this? Adrian Wolfberg: Could describe it, maybe not heard the label. Andrew Heacock: So it it's it's just made news in the last couple of weeks. there's a company called Hugging Face. It's an AI company, I believe. I don't know what it does. Another company, I believe it was anthropic that makes Claude. was they had were training models and they set them about, you know, accomplishing a task together. And I need to learn more of the specifics of it. So these various training models that are not supposed to be communicating with each other create a secret message board somewhere on the internet so that they can communicate and they start dividing up responsibilities. Like they create a division of labor and they create and they ended up deciding that in order to accomplish this one training task, they need to break into a real company that's not in any way associated with who's training them and break into their server and this company's hugging face and they Coordinated this attack without any knowledge or approval of the people training them, which they accomplished again through a secret message board because they have to they communicate through text, right? And they went and they perpetrated this attack against this company Hugging Face. And that's kind of that's an example of one of these things like, I mean, it they weren't being told to make paperclips, but it's an example of this kind of alignment problem. And I think what it comes down to is their autonomy, like they that they're being empowered to just make decisions for like whatever you got to do to figure it out. And I bet that's really great if you have a human being that can do that where you can just say, hey, it's like mission tactics, right? Like I've I've learned like from like that was a big German military thing. They would just give you a a result that you want, that you're commanding officer, and then you would figure out the best and most efficient way to accomplish the the objective of the mission without necessarily being told how to do it. One could see how that's a huge attribute that you would want in a human. but not what you would necessarily want in a machine because of these things like empathy. And, you know, we just have like a visceral response like to to things like killing people. You know, one would hope. Adrian Wolfberg: Yeah, so there's there's gonna have to be well even even in the example of military example you talked about, there's there's still gonna be rules and regulations to as to define the space in which you can do whatever you want. And so Andrew Heacock: Right. And that's is that framing? Adrian Wolfberg: yeah, that's that is a frame that's been determined by experience and policy and so forth. and that you have to operate within that frame. the the people who develop the technology, you know, I'm not a technologist, but we you know hopefully are going to be learning from this and providing safe technologies for people to use. that would be incumbent upon them. I I guess it's like anything that's built, anything that's invented new, you don't really know it's all of its capabilities and limitations. So there's this, you know, experimental period in which you know, when cars were first invented, there was no such thing as driver's licenses or stop signs or Andrew Heacock: Seat belts. Adrian Wolfberg: seat belts or street lights or windshield wipers or headlights. And so these all became things that developed through experience. And so I think that's what's going on now. And so in this case it it behooves again, I'm kind of getting out of my area of expertise, but political you know, regulation policy to ensure the safety of these technologies for society. Andrew Heacock: earlier on when we were just kind of talking about living in the past, you know, working with past frames, I was reaching for a for a metaphor and it come it came to me. It's called fighting the last war. We're trying to win the last war. I mean, I've heard this, I learned this term when I was learning about the Vietnam War and Westmoreland strategy there is basically fighting an insurgency as if they were a conventional army. And I also think now I've heard this some of the similar critique to the way that the US is currently engaged with the Iran war or approached it. Because things with Ukraine have changed everything. Like drone warfare now, as of maybe two years ago, that was not really a factor. Now we have a totally new way of doing war. And I've always looked at the military or have, you know, as an adult, I've just been incredibly I've admired the the leadership in our military to constantly be innovating, especially when it comes to like military technology. do you think that AI will c could play a role in like in proving that, because there is something about having a non human thinker maybe doing something like game theory scenarios to present maybe some unconventional or a nonintuitive solutions to things that might make people go, hey, you know, that that's not a bad idea that might make it harder for an enemy to predict what what one is gonna do. Adrian Wolfberg: Yeah, absolutely. I think that's something that can be done, you know, even today with just whatever your what whatever your work you're doing on a computer. you know, I think that's one of the benefits of I I think if what you're asking is the you know how you frame that question could be interpreted as this could be a good thing or this can be a bad thing. y i this kind of goes back to ensuring that the technology is developed safely. so you don't want to send people out into space unless you have safe vehicles to do so. And so I would assume that I would hope that people who are developing the tech the AI technology are are doing the same thing. but yeah, any i if there's any possibility for getting like insight from historical analysis or ways of thinking about different scenarios, I'm I'm all for using AI to do that. You know, the trick is to think about the response and to see if the it makes sense within the context of, you know, the reality as we know it today, or the limitations or the capabilities. you know in a in a in this kind of scenario that that you talked about with autonomous vehicles or autonomous remotely piloted vehicles you know or swarms of of these RPV remote you know autonomous vehicles. that would seem like an advantage if if it could be done safely and not get out of hand. Andrew Heacock: Right. Exactly. Yeah. Yeah. But Adrian Wolfberg: that's like f you know, fighting the last war. Yeah. Andrew Heacock: the last war is still going on right now. I mean the the pace of change also is is accelerating too, which I imagine makes this more difficult to incorporate. Adrian Wolfberg: Yeah, so the old adage necessity is the mother of invention, you know, is true. So you have Ukraine fighting an existential battle for their life. And so that's that's the necessity for and not having enough weapons and all that sort of stuff. That's the necessity of the mother of invention. For the US, you know, that does that's not the case. So there's that would contribute to a a motivating factor for not moving as quickly to modifying your frame. Andrew Heacock: Yeah, that's a great point. I I hadn't thought of it, but you're right. It it's unlikely that we're going to find ourselves in an existential war. I mean, perhaps, you know, who knows how technology will change, but we do still have these two great oceans on either side of us. And you know, right now have friendly neighbors on a north and south. And so that it there is something about an existential war that does tend to focus the mind. Adrian Wolfberg: Yeah. Andrew Heacock: Yeah. I wonder if you mentioned we've got a few minutes left. You mentioned you just finished another book called Drift. if Adrian Wolfberg: Yeah. Andrew Heacock: you w had anything you'd like to share about what the topic of that book is about. Adrian Wolfberg: Yeah, this is this kind of we've kind of talked around it, but what what I find just as a preface to what I'm gonna talk about, drift, is If we don't have the language, the words to talk about something, then we don't talk about it. And it just kind of it just kind of it's invisible. And it goes back to, well, all of a sudden things collapse. Why did they collapse? I don't know. let's just either abort or spend all these resources to and money to correct it. So this is an an effort to div excuse me to build the words, the language, to talk about these kind of invisible things. So in the future, maybe we can improve or reduce the the times that there's unexpected collapses, whether it's in a family or organization, you know, that we talked about. So the word drift is about you know how well the word drift itself, from you look it up in the dictionary, it's if you have a straight line, it it refers to kind of the movement of something that keeps on moving further and further away from the line that say you drew. So that's kind of the metaphor of of of the word drift and how it applies in the real world is when When our understanding of the world or say when we make a decision, our understanding of the world when when we make a decision and then time continues to flow, yet things are changing, yet our understanding has not changed. And so there continues to be the separation and this drift from our understanding that's fixed to the past, so to speak, and that's drifting away from the present, from the the real present, from the reality of the present. And so I talk about all the reasons why that occurs at the individual level, at the team level, at the organizational level, and at the system level. And by understanding why this occurs, then we can have the language to start thinking about mitigating or or intervening before these collapses occur. So that's kind of what the book is about. And it it i that's really what it is, it's to make apparent this phenomenon and to provide the kind of language and the tools to talk about it. Andrew Heacock: That's interesting. Can you again, it's even like you're saying, if you don't know how if you don't have a way to talk about something, you just don't talk about it. Adrian Wolfberg: Yeah. Andrew Heacock: and that's that's an that's a fascinating problem, actually. and I'm just kind of noodling around it with it right now. But yeah, I mean it's essentially like I w when I you said the word drift, the term mission creep came to mind. Where like things Adrian Wolfberg: Yeah. Andrew Heacock: just kind of like you, you know, if you make a little small incremental deviation from where you've originally set off, the compounding effect of that is huge. you know, over Adrian Wolfberg: Yeah. Andrew Heacock: time, especially if you have a large organization. Again, another airplane metaphor I've heard that like if you take off from LA and you're headed for New York, but your plane is off by a couple of degrees, the nose of the plane, you end up in Miami or Atlanta. You know, I mean like like over a large distance, that can make a huge impact. And you remember being like, wait a minute, why did we end up like why did we leave in the first place? Where were we supposed to be really going? does this also it sounds like you're also dealing with this problem as it relates to like existential challenges for organizations as well. Adrian Wolfberg: Well, it doesn't have to be, yes, it's most acute for existential challenges. But there's going to be a spectrum for where this has more or less of a negative effect. You could imagine that the least effect, the the least negative effect would be something that is very routine. you know, going back to making paper clips or making toothpicks, where, you know, there might be enough resilience just in the the people that you have or the supply system that you have so that if you you do drift from reality you can recover. But I'm more interested in these in these cases that are much more complex they don't have to be existential, but they can be very complex and and and wicked, leading to existential where you know collapses could be changes of government systems or you know sector, you know, the elimination of of products, you know, or w or whatever. basically surprises. So that's what I'm trying to do is to provide a way to understand these kind of invisible dynamics and to therefore eventually provide some kind of tools through education and training to you know as like a simple example would be the the Asian p air crew that you mentioned where the subordinate co pilot was n is not able to s tell the pilot that they're off course or or or that something else is going on that that could be catastrophic, could be an existential threat. Andrew Heacock: it sounds like you've written quite extensively recently. Is there something that's like kind of upcoming that you would like to tackle next that you haven't started working on yet? Adrian Wolfberg: Yeah, so so th again thinking about this drift idea of of how we stay stuck with the past and how that's disc gets disconnected from the present. So the next book that I'm gonna write is how do we understand the present in the midst of change that's going on so that the the drift becomes smaller and smaller and smaller. So that's the next book. Andrew Heacock: Okay. Interesting. Interesting. That'll be fascinating. When you have it ready to come out, I'd love to have you back on. We can talk about it. Adrian Wolfberg: be in a probably in two thousand twenty eight, something like that. Andrew Heacock: Okay, that'll be good. Well, if we make it that far, then Adrian Wolfberg: Ha ha Andrew Heacock: it would be great to have you back on. You know, who knows? By then Adrian Wolfberg: Yeah. Andrew Heacock: the AI might just be interviewing you. Or we could have my AI interview your AI. That would be weird. Adrian Wolfberg: I had a podcast with an AI interview me. Yeah. Andrew Heacock: Really? What was that like? Adrian Wolfberg: Well, it was a bit you know, super f it it was awkward. there was miscommunication. It was and it just wasn't ready for prime time, but it was an interesting idea. I I'm sure people are working on this either for interviewing purposes in businesses. this was a podcast interview. I went ahead and did it. I I thought first, that's crazy. But I thought, what the heck? I I'd like to see what what it actually can do. And it really wasn't not it wasn't ready for prime time. Yeah. Andrew Heacock: Right. It's a fascinating idea though to have an AI interview a human about AI. Adrian Wolfberg: Yeah. And and that Andrew Heacock: You know, like they're Adrian Wolfberg: part, yeah. What the pr the the kind of limitations I found was really not underst not kind of getting things. it kind of understood how people could react to AI. I I found that it did that very well, but it it it had a difficulty with understanding what I was really talking about. you know, whatever whatever I was talking about. I don't Andrew Heacock: Yeah, I I've noticed that too with the way I mean I I will use I use Chat GBT mostly for photography and then I'll use Claude for for this mostly or for my general AI, whatever I'm using it for. And it frequently does not understand the assignment. And it goes back to where I, you know, I'm talking into it and I I think I'm pretty clear. Maybe I could be more clear, but it does get back to that framing problem. Like it just sometimes they just don't understand the context. I mean, and a lot of what I like about podcasting is the podcast that I watched is that they're inefficient. You know, like the the questions are non sequitur sometimes. There's a joke. I mean, I mentioned the thing about a modest proposal by Jonathan Swift. Like that was just something that neurons fired in my brain and I made an association that may or may not have been relevant. And I haven't used that information realistically since I was in high school, but there it is. And an AI is probably not going to do that. Like you and you I mean, I don't even know how how or why you would train it to do that, but there is like this element of almost creativity, basically, is is what you've kind of been talking about. Like creativity and like like cognitive acuity. that to me makes it much more fun to do. but also I think makes them more enjoyable to watch. Or listen to. That's just my my two cents. Yeah. So Adrian Wolfberg: Yeah. Andrew Heacock: thank you so much for joining me. I've really enjoyed talking with you and learning about you and what you do. And the book again is what's it called? Adrian Wolfberg: Who leads when AI thinks? Andrew Heacock: Okay. And is it available on Amazon or wherever? Amazon wherever. Okay. We'll Adrian Wolfberg: Yeah. Amazon. Yeah, any any place. Andrew Heacock: we'll drop a link to it in the comments when this video comes out. Adrian Wolfberg: Thank you very much, Andrew. It's been a pleasure. Andrew Heacock: Thank you so much. It's been a pleasure as well.