Michael Kanaan: Author of T-Minus AI Discusses Artificial Intelligence and Global Power

Informing your decisions with intelligence analysis and insight brought to you by the team@ooda loop.com hello, this is Bob Gourley and welcome to the Oodacast. Today on the Oodacast, Mike Canaan of the Air Force Air Force Captain, he's the Director of Operations of the US Air Force MIT Artificial Intelligence Organization. Correct, Mike? Did I say that right?

That's right.

Well, and you're the author of a brand new book, T Minus AI. And I want to talk a bit about that book shortly, as well as your views on AI. But Mike, I wanted to start with a little bit of your background. I noticed from your LinkedIn profile you joined the Air Force 2007. You started at the Air Force Academy, graduated 2011, and since then you had a lot of operational tours at organizations like nasic. You worked in the Pentagon for General Dash Jamison, very highly regarded general officer who's known for making big change across large organizations. And through that, you had the portfolio of artificial intelligence in the Air Force to help the Air Force understand the potential impact of this new domain. Did I get that about right?

That's about right, yeah.

And through it all, it puts you in direct contact with a lot of important operational missions where you needed to think through how AI could apply. And it also put you in contact with a lot of thought leaders across the country in artificial intelligence. And that's what really excites me about your new book. So I'd like to talk about that and ask you a few questions if that's okay.

Absolutely. Looking forward to it.

One, a lot of people use that term AI differently. We'll have academic definitions and we'll have marketing people using that term differently. And we'll have engineers who will use that term differently. And I just wanted to ask, what's your definition of artificial intelligence? How should I think about that?

Let's make it simple here. We'll use an acronym because that's what we do best and it's easy to remember as a mnemonic device now on the Internet, everything are cats. So let's use the word cat to help drive this conversation. We say the word artificial, the words artificial intelligence. We're talking about a concept now in 1956, that concept was depicted as a machine performing something that was once deemed as a human task that resided in the human domain. Well, wait a second. With that kind of definition, then surely perhaps a calculator was an artificial intelligence at some point. The Excel files we hate to be on certainly are. Tableau definitely is by that standard. Nowadays. So this definition has kicked the can down the road, which has made it very difficult for us to discuss it. And then further exasperated by that, we bring a lot of baggage into the conversation with the term artificial intelligence. We say, wait, you mean my own intelligence? Well, how could that be possible? So AI is but a concept. Now, inevitably, when we say the words AI, many people add that thing, machine learning into it. They say, AI and machine learning. Let's think of machine learning as just an application that is this, that it can change from the presentation of new data. AI itself is very simple. The way we use it and express it in machine learning is math through code, which is a language. And we can use that, those algorithms per se, that can naturally change with the presentation of new data. That's it. That's all it is. Once we had to explicitly code computers step by step by step, and in doing so, you kind of understood the output at the end of the day. Well, nowadays it doesn't need explicit code. That's machine learning application. And last, we add in, well, then how can it act like us? How is it an artificial intelligence brain? I guess. Then we say the words deep learning and neural networks. These are techniques that we use through the binary digits and bytes and bits and computer chips on your computer to roughly depict a human brain. That is neurons that can bounce back and forth. And because it's presenting new data, change over time. I think using these words precisely is really beneficial to our businesses, to the government, so that we're communicating on the same language. So AI is a concept, machine learning is an application. And deep learning, neural networks, it's just a technique. Cat, Cat.

I like that. I like that. That's very helpful. Now the book. I enjoyed the book, frankly, for a lot of reasons. I do encounter a lot of very technical things that sometimes are hard to express. And I found your book made him easy to express. You talked in a very common language. You brought us into the story. In fact, you start in September of 2017. You're getting off an airplane, you're in a car, you're tired, you've just flown across the country. You're starting to get text messages. Well, really enjoyed that. You made it real for us. You pull it in, pulled me into the book. So it read in some ways like a thriller. But this gets to. Why did you write this book? Clearly, I can tell there were so many events in that month that were occurring that must have catalyzed something and made you start to realize you needed to write this book.

Over the course of time when we're talking about this idea of artificial intelligence, this thing is going to be more like electricity in our lives than anything else. I mean, shoot, every single day it drives your 401k. It is mostly the reason you get suggested. The music that you listen to, it's the reason the ads that you see. You know, I think personally that it's somewhat compromised the game of baseball as of late. But that can be, you know, up to anyone's personal debate. It's the reason you read the news that you read. For myself, I think the whole world talks about artificial intelligence, right? Because I am being presented that because of the underlying search algorithms and everything else, it writes news now, it creates language and it generates art. And there's a good reason to say we can blame a bit of that stock market crash on early applications of machine learning as well. Yet for all of that, we're stuck communicating with the same barriers that we've always had. We too often. And when it matters most, on a topic like AI or what we're seeing in society today, we speak above, past, below, and through one another because we don't have a shared language and we don't have a common set of experiences. So what I wanted to do in the book was to write through the human narrative. This is ultimately still a human endeavor at the end of the day. And for us humans, we learn best through storytelling. AI is a completely multidisciplinary topic. You need to know a little bit about our evolution of ourselves, language, learning how numbers work both big and small, how your brain works. Then let's talk about what is AI? Those neural networks and everything else. Why do we develop them on games? How do bots and robots differ? And then the big piece. What about strategic competition? For myself, I think often about other nation states for you in business, you might think about competitors in business. Let's discuss how that plays out. So I wanted to ultimately write a book that was a human story, and we learned best through storytelling.

I think maybe that's why it resonated with me, because I have to agree, we learn best through storytelling. And that was a good narration of a lot of things coming together. And you do mention a lot of different disciplines, including cognitive science and how we think. You mentioned common cognitive biases. And you tie that to the fact that sometimes the AI gets fielded in ways that are biased. And people will quickly say that AI is biased when they really need to step back and thank the developers, the programmers, the managers and leaders who created it, had some of Their bias creep into that.

It's like looking in the mirror. Right? That's the thing about artificial intelligence. It's the ultimate depiction of us. We create data, we memorialize it all around us. Computers learn from data. That is our experience. So ultimately, it's just a product of ourselves.

Right. So another key part of AI is the fact that since it does change with input of new data, it can change over time. Some of these algorithms can drift or even become corrupt. And that also, I think, underscores the need for humans to always be in the loop and able to control this stuff and recognize when something is not behaving the right way. Do you agree?

I do agree with that. You know, you go through the experiences throughout your career and with people in private society, whether it's at coffee shops or bookshops or perhaps on many podcasts, and you want to tell the story of AI in ways that make sense. So let's think about it. Along the OODA loop itself, where is AI good at? What does it do? Its design is to illuminate and understand human patterns. Now, what's important, though, is that's not to say that it understands when those patterns are a result of biases or prejudices or are there good biases that keep me from walking in front of a car every single day. So when we think about artificial intelligence, I think about it along the observe and orient side of the OODA loop itself to illuminate things us as humans simply can't see. At the end of the day, it's this simple. Computers are good at what humans are not, and humans are good at what computers are not. Observing and orienting ourselves to find the patterns, the needles in the haystack that look like hay. That's great for artificial intelligence. Now, along the decide and act side, let's keep that for us and make sure that we are making human centric choices. Or in the infinite scales of the universe, those patterns actually mean something to the real world.

Right, I can see that, except for in, like, narrowly defined use cases where the human has decided. Okay, here I do want the AI to decide and act. For example, auto focusing a camera or auto adjusting lights based on dynamics, These kind of very narrow kind of implementations of machine learning.

And I agree with you and those narrow works. The current state of AI, why I started this book, the first sentence is, the countdown to artificial intelligence is over. I wanted that to be weighty, especially against its title itself, because the things we have now in a very narrow domain will drastically impact and alter our interactions they will have immense implications, both good and bad, for the foreseeable decade. Let's say there are no advancements in this technology. We need to understand that technology right here, right now, because it is effective in certain places. And much like a hammer, you can be effective with a hammer and pull out nails and hang frames and pictures. I sometimes use it to, you know, get the garbage disposal moving. If you put too much stuff after it, Right. You can stick it in and reset it, all sorts of things. It can also kill people. Right. We're effective with it. That's much like artificial intelligence. Right? Good.

So another question, AI is becoming ubiquitous. It's all over the place. There's some great solutions everywhere. And like so many other technology domains, it seems like we have fielded this in a way where there's going to be surprises. And I'm specifically referring to security surprises. You know, just in the day of the telephone, when telephone was first being fielded. Let's go back to telegraph, when. When the telegraph was first being fielded, there was fraud and in surprises and then the telephone and there was, you know, pranksters and fraud. There's still phone fraud today. And then there's the Internet, which of course there's cybersecurity issues and fraud and criminals and nation states on the Internet. Now we field artificial intelligence and why should anybody think that's going to be different? There's going to be a lot of bad guys using AI for their own purposes and bad guys attacking our AI. And this is where I want to get your comments, because it worries me a little bit. I think AI is more susceptible to deception and would like your comments on that. And if so, it's going to be more susceptible to adversary action. And I'm wondering what we can do to mitigate that. Comments on any of that.

It's a broad and wide ranging topic. So let's start with a little bit about trust and then we'll build on this conversation. Right. Trust is a very interesting topic, particularly related to artificial intelligence, because when we talk about it and then, right. We're putting it into application, we're worried about it in security systems. We've got to talk about this instinctual distrust that we have to begin with. So the term AI, automation, all the tech terms out there, they are so misunderstood, understood, overused and broad to the point they're meaningless. So the first question is, let's start addressing that, right, As a workforce to make a fruitful conversation. The second is an academic perspective. There is not yet for humanity, for humans, a shared or Prevailing, clear and convincing notion of trust in general. Right. So these two things very much come together. I do want to talk about trust for a second because I think this, this blends into the security conversation. Okay, what are things you and I trust? Because it is an interesting concept. I doubt that there's anyone watching this that knows everything about, let's say, cars. Certainly we have mechanics who can build a car from the ground up, but that doesn't make them great drivers. Vice versa. There are plenty of great drivers that can build a car. Or even if they could, they certainly can't code the software in each of our cars today. And you know, I, or I believe I'll wake up tomorrow. I have no reason to necessarily believe that. I hope I do. Or gravity. Right. It affects us at 9.8 meters per second squared. But we can't unpack Newtonian classical mechanics. And most people can't converse about our quantum reality or Everett's many worlds interpretation or the microwave. This one I like the kitchen appliances in our home in quarantine, we use quite often. There's a steam pot somewhere in my house over here that I like using and it terrifies me. It's just a bomb, right? But I use it every day. So to get this trust, we start with this idea of we need to build experience, right? We need to start in safe places to better understand it. So number one, that's an issue with our businesses. Now let's move to security and some real concerns that we have. There's a lot of routes to go here. The first issue is that this idea of what you and I would call unclassified, right? Or the data that you have in your business that's unprotected, right? You don't want to put personally identifiable information, if you're Blue Cross and Blue Shield onto a system that could be compromised. So you kind of reserve it for a different system. When it comes to developing artificial intelligence, we got to get beyond this idea that unclassified or openly accessible means unsecure, we can still harden things. And I bring this up because unless we bring it into a commercially viable domain in which we're all on the same software speaking the same programming languages and can deploy the publicly available technologies, you're probably never going to get there or have robust enough algorithms or those that work for most of your problems. Now on the other side, artificial intelligence can lurk and do things with patterns that we are completely unaware of or be intentionally deceitful by just changing maybe a pixel on a screen. Again, the Computers are good at computer things. Humans are good at human things. So I would say this when we think about this moment though, right? Or we could poison data. There's, you know, that's an example of how that would happen. I'd say this. For a long time, in both, you know, public government work and in private business, computer systems were something we just protected. And in only very small, certain communities around the world are in sci fi thrillers. It was something to be wielded or to make life better, by the way, for businesses and governments today. I think it's realizing that this is something you can use. A computer system is usable. It's a part of your repertoire. So, A, I need to figure out how to make it safe. B, I need to figure out how to best use it when we worry about security and to develop things. And I know this sounds perhaps a bit pedantic, but it is a paradigm shift that I think makes a difference how we discuss it. Cool.

Yeah. Now you mentioned several things there. I'd like to dive a little deeper in trust. Man, that made so much sense. And I think we need to get more and more people understanding that they're using AI right now. Oh yeah. And so that's going to be important part of the trust, I think. And another good thing that your book will do is help people understand that they're using AI right now and what it means. And over time trust will build, but security. There's stuff that I don't trust yet. The ways that adversaries can attack AI in the corporate world. For example, you mentioned in the book Microsoft Tay. Would you. What was the impact of tay? And actually what was tay? What was that example?

It was a chatbot that was developed from AI. And remember we talked about data, so we're talking about data here right now and how valuable that actually is to providers, to procuring, to having yourself to understand. So it created a chatbot that would interact with people. And you know what it did? It learned from us and some of our worst tendencies. So very quickly, within about a 24 hour period, this chatbot that was supposed to communicate with people, which by the way, AI has such a duality to it. Imagine the ways that we can address loneliness and our parents and people who might need someone to communicate with AI is good for that. At the same time, that same technology generates that thing called fake news or computer generated content. Tay learned from our tendencies on chat rooms and Twitter and all the like and very quickly became a very depressed, a very cynical, quite racist and Everything about humanity we don't like that we find on the Internet. And it just propagated that and scaled it up, and that's what Tay became. So what do we need to think about then? We got to think about the entire journey. Artificial intelligence is more like a journey than it is some end state. What do I mean by that? You can't use artificial intelligence unless you clearly understand your business. That is, most people think AI is going to replace the bottom of the workforce. Automation might do that. AI, it's meant to be used with your subject matter experts on high volume data moving at high velocities. That requires a very high accuracy. So then you start thinking about the data set because you're saying, well, I better understand my workflows, I better understand what they're doing. So for the security standpoint, you got to think about the entire continuum. You have to worry about data at rest, then data in transit, then decisions you might make, and your servers that are sitting wherever they are, whether they reside on the cloud or they're locally owned. But then you got to worry about, well, how is that data manipulated before? I use a machine learning application. So for security, the considerations have just exponentially grown in how we have to discuss security through the entire process.

Right, Very good. And you know, this leads to another thing that you discuss in the book. It's one of the few books that does this. You talk about what some other nations are doing in artificial intelligence, and there is several news stories out there. And Kai Fu Lee, who wrote AI Superpowers, wrote quite a bit about China and us and differences in AI approaches. But I think yours hit it on the head very well about other nations seeing this as a race to be first, to be dominant. And I like that. I like the way you worked up to that in the book. And so I wanted to ask, what do you think the average US citizen should know about the geopolitical aspects of artificial intelligence?

We're talking about a little bit of the ethical issues, justice and fairness and how we view society. What should we everyday people be tracking? Right. It's not just reserved for engineers. These are weighty words. And a lot of time we use these words very interchangeably. The first thing we do when discussing AI, by the way, is it's like we throw the whole kitchen sink out the window. Not even just the kitchen sink, like the kitchen and the living room itself. We just throw it out the window. We say AI and we almost forget the frameworks and the moral and ethical and legal and the valuing of human dignity and the imperatives and foundation we were all founded upon or others were founded upon. So here's the point. AI is but a useful tool in a lot of ways to achieve one's own means. And what do I mean by that? I went back to that hammer analogy. It's much like the hammer, and that's much like AI. It's simple. It's a tool that is used to achieve a party's or society's end state. Now, I am guilty of this too. We say China is using AI in meetings. That is unacceptable to us. Let's make that a little clearer. The Chinese national, the Chinese Communist Party is using it in ways that are not okay. The Russian Federation is using it in ways that are not okay with us. Right? And I think we need to really think about it, because some of the applications of AI, like using it to detect breast cancer, we're gonna say is democratically acceptable. But I think others are gonna shake the foundations of our society because by the way, that same idea of detecting breast cancer, that's used through deep fake technologies, which makes fake news, right? So there's an inherent duality happening. Some we're going to consider good and others, I think are going to be used in ways we just don't accept. So if it's viewed as a tool, then we would have the discussion at hand that this can affect us all and will then it better be representative of how we view the world and human dignity and what is fair for that's the goal. Now don't get me wrong that, you know, we're special and different because we do make mistakes, right? But the difference between society here and why this is so important to make this the broadest conversation, multidisciplinary, is when we're wrong with deployments like Tay or Amazon hiring action, we hold people to account. The conversation actually takes place because that doesn't always take place elsewhere. At the same time, we shouldn't be let off the hook, right? Because at these inflection points, we can keep trying to fit square pegs and round holes when we should probably be carving out more square holes. So what we're talking about here, and you had mentioned before, is privacy, freedom of speech. So I'll give an example of something I think about often that AI has changed in 2020. You know, the rise of consumable and believable computer generated content, and how do we talk about privacy and freedom of speech and how do the. And then how do we use this topic of anonymity with that right, freedom and speech don't necessarily and always grant anonymity. Most certainly not. Let's think about the Federalist Papers. If Aaron Burr had said that he was Hamilton, that probably would have been an issue. He is representing something factually untrue under someone else. But nowadays we perpetuate this kind of stuff in only echo chambers driven by artificial intelligence throughout the world. So it's things like this that require new square holes. But we can't make new square holes if it doesn't include teachers, sociologists, legislators, attorneys, you know, everyone else in the humanities. It's not just for engineers. And that's why I wrote the book, was to bring understanding of all of those pieces together.

You know, I think it does that very well and which is why I hope it's very widely read and in the business world. I also believe that the business who reads this book can think through their own current products and services and how AI fits into that. Either AI being used internally to make the company more efficient or AI being a piece of their product and service. And I see the book as a way to get entire teams of people on the same wavelength and have a good conversation about this topic I would.

Love to dive in on. Let's think about how to leverage it for a CEO or in the boardroom or in our everyday mid management business. So we'll dive in there and briefly burn through maybe some principles I often think about in business. Awesome. As for a strategic approach. So first what seem like the tactical choices today are the minutiae, like what software you're on, what programs you buy, the tools you pick are really the strategic decisions of tomorrow. For example, let's say you're operating on Python 3.5. You would. By the way, you would think this is never a conversation for the boardroom, right? The proverbial boardroom. Well, hold on a second. It is because in all the things you're espousing of where you want to take your organization, you will never be able to onboard the machine learning applications today because your obsolete language package can't even implement the solutions that the workforces of the future are learning right now, in school or personally and I'll address that later, or the leading commercial providers could even provide you. So it's about making sure the barriers to entry are lowest for your internal workforce and and external providers. You'd be surprised how many people operate on legacy software and hardware and say well why do we need to buy new computers? Why do I need new software? It's for these reasons the second, the data piece. Data fuels the development of AI and it requires a vast amount of IT representative data, whether that's derived for an authoritative data set that you might have or by the way, generated through digitizing your daily workflows. This is an interesting concept because in order to train algorithms for continuous delivery, which is what you're doing in AI, you're constantly monitoring and pushing out and pulling back. You should be princip and re observing. You should be principled on consistently generating training quality data. This is a key word for algorithmic development because if AI is going to influence your operations, you should be providing it and archiving it in near real time. So you want to architect the solutions for the shortest path between development and operations, that is monitoring how business is actually getting done. And maybe the word monitoring isn't so much, but it's memorializing the way it is. And we talked about the value of data. It will only be useful though if you trust that data itself, right, and how it was generated or obtained. So what you should do is set policies and standards in your companies for training quality data. You should identify, measure, build and simulate your authoritative data source or the everyday workflow. Sometimes it's more important to digitize the everyday workflow than try to modernize some old database that might not even be representative of how business is done or the hacks our workforce use in everyday life. The next piece is don't make this too hard. Don't search for some unique solution or overpay. Because the R and D and commercial sector of the AI world is has held a belief that it's a due right to publicly develop AI and provide it out to everyone. Which is correct for the most part. Remember the duality thing. So it's likely your solution or your problem already exists on some GitHub repository out there as we speak. You just need people to bring to bear to understand what it is. And that's where things matter. You should be developing people, employing a skilled, educated, solution oriented workforce. AI people are solution oriented. That's the purpose of it. So let me give you an example. The federal government has a long history of recognizing the critical and significant value of human foreign language aptitude and proficiency. That is, can you speak Mandarin or Farsi or Arabic or any other number of languages that we hold value to businesses? Think about this in international operations then in that case we want to cultivate that. So in the federal government, we're working to implement a computer language initiative where you treat aptitude and proficiency in programming languages equal and commensurate to human foreign language. Right. This is just but one example. But if you're in a company that doesn't have an opportunity to contemporarily learn for your workforce, you're just wrong. Just stop now. Education is essential, especially at the start to this journey. And why is this important, especially as American companies? Here's the reality. This year we're going to graduate 50,000 people with STEM like degrees for 500,000 open jobs. But we live in a world where the barriers education again weren't lower. So I don't care if you have a college degree. Have you done it? Can you do it? Do you have a passion? I believe that our private sector and the public sector have an obligation as a nation to drive down the gap on this topic. Because in only a few short years, insert Dr. Evil reference here, there will be a million open jobs. Only going to that. And then lastly, the ethical, moral and legal use of artificial intelligence. We need to start setting standards. And by the way, this is going to cost money. This is going to take time. This will take away from the bottom line. But I believe that in 2020 and moving forward for companies, maybe it's not reflected in the quarterly report, but you can do better. I make more money by doing good. And that's hard, right? It's not always in the numbers, it's not always in the table, but I think it's really important. So ultimately, data matters. Focus on your internal development. Education is a lifetime sport. Have the right people, be relatively informed, much like you. And I know not to stick a fork in an outlet or tin foil into a microwave and maybe stop chasing the next big AI thing. Dive in. Understand it. You'd be amazed how often people get focused on building the next bigger and better thing when you could have taken on what is seemingly a toy problem and entered into this really nice virtuous cycle of making your business better.

Well, let me ask you this. In your opinion, for a large organization, now I know you are currently in a large organization, one of the largest on the planet. You know, they don't get much bigger than that. It seems like you have choices on how you design and field new initiatives. You can either centralize or decentralize. For example, in the Air Force, it's a good idea. I think that design of aircraft and weapon systems is centralized for a lot of reasons. What about AI? Corporations, I think face similar decisions. You can centralize, you know, bring a team right up under the CEO and put all AI initiatives under there and all Funding must go through the and you follow our way and then you probably feel it faster. But you may need to decentralize because that's where you get all of the use cases out on the edge and where the innovation frequently occurs. And do you have any thoughts on corporate America and which of those two models they should lean towards?

I think the dreaded answer both. Let me explain. You should centrally identify your proverbial tech stack, that is the specifications, how data communicates, the software you develop on. By the way, they're all good. It all works. Just pick one. Because what you ultimately want to do is you, because of the inherent nature of AI, needs to be developed on user data. That's the whole point. So if it's not pushed out to the field per se or to your workforce, whatever you're building in the boardroom, it's probably worthless to them. Right? So what you want to do is centrally identify the technical specifications and how it has shareability and portability. Right. When it comes to the work that is performed on it. You want to push that out to your workforce on cross functional teams and then say I pin the rows on you to make machine learning in your department, in your organization, but at least it will be machine readable to these standards and then that way we can all share throughout the company.

Cool. Mike, in our remaining time, I want to dive a little deeper into another subject and that is you as a person. We know a bit of your background. You joined the Air Force 2007. One thing I got as I was reading the book was this impression, this deep understanding that you are one of these lifelong learners. You just continually are learning. That's your research is apparent. So I want to start with your foundational story. Why did you join the Air Force to begin with?

I am consistently reminded internally, the internal dialogue is the most important one, how grateful I am to have experienced all this. Because we need to remember 1% of this country served in the military. Nowadays it's not quite what you think. So we say things like thanks for your service. But not everyone knows what that means. And I'll tell you this much, I never met a single person in the military. Besides for a couple retired business partners at law firms, I never met a single person before going to the Air Force Academy. I grew up in a place where it just didn't exist. It's the north. Michigan just doesn't have that. So one thing that I learned growing up, and I think that's worth sharing, is you can't always reopen doors if Your door closes. There's no preordained right to reopen that. So when the Air Force Academy opportunity came along and it was really inspired by playing football, that's how I got there was high school football. I always said to myself, well, I hope I never wonder, what if, what if I went there? What if I did take that? You can always go to another school. And I think that that's something that I've always been inspired by. It's something that I often tell young leaders, never, never wonder what if. It's the worst thing that you can do. So that was the inspiration for the Air Force Academy. And I couldn't be more grateful because of the people I get to meet and the places I get to go. It's pretty great. And then lifelong learning, I'll tell you, it's my grandmother I grew up with. You know, she has always instilled that in me. And I think of my love technology in these topics. I think to books back like Sagan's Cosmos. I think of Hawking's Universe in a nutshell. I think of being about 10 years old and reading Brian Greene's the elegant Universe over and over again. By the way, you know, unfortunately, string theory didn't really pan out. Or maybe more recently, like Ferrari Sapiens, Tolstoy, Huxley, Virginia Woolf, and so many more. And all these books and authors in including kids book like if you give a mouse a cookie or goodnight Moon, they have something in common. They're debating and discussing the topics of consciousness, theories of physics, biology, social realities, technology, all the rest of the things that constitute the human experience. Right. And that's what good literature, that's what good podcasts like this conversations do. It's a candid dialogue. And for me, the concepts of consciousness, experience, social order, biology, the human story truly is brought together. And especially on the topic of artificial intelligence. And many people bring those topics to light, but fewer have to life. And that was the intent.

Awesome. That's great. You mentioned several books there. What are you reading right now? Anything current that you that we ought to know about.

I'm having a hard time reading NonFiction after about 27 months of this book, I'll be honest. Most recent nonfiction that I've read though is Sean Carroll, who I'm so grateful he was my acid test as an endorsement for the book. Incredible quantum physicist, thinker, the whole shot. His most recent book, something deeply hidden by the way. Anyone can understand and talk about quantum mechanics. It's strangely all about the human experience right now, much like AI, everyone can converse about it. It's frankly more important than humanities. And then as a fiction book, I've gone back in time about seven or eight years and I'm reading Neil Gaiman's American Gods at the moment just because I need to get my head out of it. I also read the Three Body Problem a couple weeks ago. Yeah. But it was just a little much.

It gets a little dark, doesn't it?

It gets a little dark. So great book, incredible writing. I was so, you know, when I was reading the Three Body Problem, after you write a book, when you're reading a book, you're actually weirdly reading, but thinking about the author and what they're doing and what they were trying to do and picking up literary devices or wondering to yourself, how could you write this? Right. And you're just like in awe of them. So the one thing about the Three Body Problem is, however, that was translated in the way that just, you know, got the essence of it and the beauty of the language. I was just hung up on that. More so than the story.

You know, something I keyed on there is. And it fits with some of the things you said previously as I was reading that science fiction. It's clear from the first page you're reading science fiction from a different culture than the American perspective. So I, you know, as I was reading it, that was impactful for me that this is the way a different culture is writing science fiction. And I enjoyed it more because of that.

I think it's very meta. To this conversation. On AI, we say bias. The thing I learned most about this writing this book were my own biases. I brought into the conversation itself the way that I wrote. Right. And I think that's something special. Always paying close attention to, you know, you might think that you're righteous or in the right, but look at the other side.

Yeah.

See how maybe your actions might be perceived as perhaps offensive or selfish in nature. You have to look at both sides. So yeah, I love reading science fiction. I love reading fiction and certainly non fiction of how other societies see the world because there is, there are little nuances to it.

Right. And you mentioned science fiction in the book too, and that was. You made some good points there. And science fiction can be very enjoyable. It's great drama. You think about science in the future and it can really expand your thinking and fuel your imagination. But sometimes it actually, I mean, if AI is always the bad guy, maybe it's not so helpful to the conversation.

Maybe it's not so helpful in the way that our society has depicted it. I will say a most recent book from from dear, dear friends Peter Singer and August Cole. They just released Burn in, which is a fair depiction of how the world could pan out and how it could turn out. What I want to do for those science fiction books or for people who discuss artificial general intelligence or the killer robots, existential risk is important. We have to talk about that stuff, but we are lacking a context to understand how we get there. For myself, when there's a fire at the door, I don't really care about the lightning in the distance. So what I wanted to hope to do was journey so that when you're reading science fiction or you're reading about general intelligence or existential risk on a number of topics, you have a better context for how that future could come into existence. So we can make smart choices now, right?

Hey Mike, thanks for all this. I really appreciate it. What a great book. And we'll have links to it for anybody who watches or listens to this this so they can rapidly get to the book. But I wanted to ask, any final comments or thoughts? What's the message on AI today?

No, Bob, I'm so appreciative of you and all these conversations. I think this conversation embodies what you're all about and having difficult conversations and on new topics. I think lastly for our businesses out there and for people listening, my only piece of advice on seeing this world into our existence in the way we want to see is team the techniques of the old with the ideas of the new. Your boardroom or the business meeting every single Monday at 9am or Tuesday, whenever your time is, should probably look a little bit different than it does right now. I mean truly, age doesn't dictate experience in this new domain. You're talking about doing machine learning. You haven't been able to do it very well for the past, you know, anytime before eight years ago, nine years ago. So in fact a young person might be the most experienced person you have to offer on a strategic imperative. And I just think making sure we're all communicating better is what's important now.

Great, important message. Thanks again Mike. Really appreciate it.

Thanks Bob. Appreciate it.

Thanks for listening to this OODA Loop production. For the latest analysis on cybersecurity technology and global global risks, please visit www.oodaloop.com.