#293 Unlocking Humanity in the Age of AI with Faisal Hoque, Founder and CEO of SHADOKA

Hi, I'm Maggie, VP of Curriculum at datacamp. And I know firsthand how overwhelming it can be to keep up with the rapid changes in data and AI. But staying ahead doesn't have to be hard. That's why our curriculum team has curated the best learning content to help you and your team stay ahead, covering everything from data and AI literacy for busy executives, to cutting edge generative AI for developers, and everything in between. Whether you're upskilling yourself or your team, we make learning practical, engaging and flexible. It just works. Power your data and AI transformation journey on datacamp.com.

I kind of look at AI as a kind of a mirror of humanity because we're feeding it and then we're being fed from it. So it's a kind of an interesting dichotomy. I look at AI as like it's in a earlier nascent stage where it's a child and it's getting to adolescent and maybe it'll grow up. If you outsource yourself, you're not going to have any value. So even for selfish reason, use it as an aid, not a replacement.

Welcome to Data Framed. This is Richie. As you know, AI is getting rather powerful. It's already having profound impacts on the world. Today we're going to discuss the positive and negative effects of AI and how to ensure that AI is used to do things that benefit humanity in order to systematize making good choices and doing things well. I'm a huge fan of frameworks and today we're going to learn about two of them. Open is a project management framework for AI and Care is a risk management framework for AI. Our guest is Faisal Hoch, a world leading management thinker and technologist. He's the author of 11 books, most recently Transcend, on which this episode is based. Faisal's also an entrepreneur. He's the founder and managing partner at management consultancy Chedoka and the founder and chair at Management Research Organisation. Next chapter. If you think this isn't keeping Faisal busy enough, he also writes for the Institute for Management Development and for Fast Company and he spent 30 years advising Fortune 500 boards. Before we begin, I should note that the proceeds from the Transcend book are being donated to cancer research. So with that, let's find out whether AI is going to help us transcend our humanity.

Hi Faisal, welcome to the show.

Thanks for having me.

Your book poses a really big question, which is if AI is powerful enough to do anything, then what's the point of humanity? What's its value? So I'M curious, did you get anywhere answering this?

Where we left in the book, and I've had a lot of time to think about it over the last three, four years, is that we really don't know the answer. It will depend on what we end up collectively doing. Because if you really talk about what is real humanity, right? I mean, at a fundamental level, you know, you can say, okay, it's our ability to be free to choose. What do we want to do? What do you want to eat, where do you want to go, how pursue creative pursuit, whatever the case may be, once you start outsourcing all that stuff to something else, then you kind of start losing yourself. So that's really the fundamental thing. And I kind of look at AI as a kind of a mirror of humanity, and because we're feeding it and then we're being fed from it. So it's a kind of an interesting dichotomy. But I think what will happen is whatever we end up doing in next 3, 5, 10, 15 years.

That's a good point. There's a lot of uncertainty around this, and we don't really know what's going to happen. But I do find it interesting, the idea that if we're outsourcing bits of our lives to technology, particularly to AI, then, yeah, we lose a bit of that autonomy there. And so you had a great analogy in your book around. It's a bit like we're parenting AI. Do you want to expand on that?

Sure. Look, I don't know whether you have children, but I have a son, and he is now 22. And I was once upon a time a child, and I had my parents. And as my parents gotten older, I've seen this progression where the parents become the child and the child become the parent. So I look at AI as, like you say, in a earlier nascent stage, where it's a child and it's getting to adolescent and maybe it'll grow up. So what happens is that, you know, when the parent become a child, you basically now start depending on whatever your child tells you to do, right? And they become the parent. So I look at AI kind of from that angle, and my research team includes a lot of philosophy background, so we had a lot of time to banter this idea. So what happens if now we're being told what to eat, when to sleep, how to take care of ourselves, you know, what show to enjoy? Pretty soon it's not letting us to drive because it's too late, all that stuff, because they have taken over. I think that's where we'll end up, by the way. But that's not just me, but a lot of more smarter people who have been working on AI for the last 30 years kind of say the same thing.

Yeah, it sounds very dark, that kind of idea where AI controls us, we lose our sense of sort of self determination. But I suppose, yeah, you're right. I mean, you mentioned the idea that AI might tell you you shouldn't drive at night. That already happens. With insurance you can have the black box in the car that sort of checks what you're doing. And if you speeding, then it's going to charge you more for your insurance.

Yeah, I mean, I mean this is like three years ago, I was driving on Brooklyn Bridge in New York, right? And all of a sudden the car just braked itself. And I said what happened? And it wasn't because there was not enough distance, it happened because it heard some noise and it reacted to that. And I'm saying this is crazy. And then started to veer off on the left hand side. And I don't drive one of those electric autonomous cars. This is a big, you know, it's a normal, it's actually a Jeep and it's like, why are you taking over a completely capable driver? And I know what the distance and all that so you know, you can stop thinking how far it can go, right?

Yeah. Actually. What do you think the end game is there? Like if you take it to an extreme?

This is the conundrum and this is the question, right? So just because we can, should we? Right, so that's really the question. And as you know, in our conversation, I'm sure we'll get deeper into it. We've seen this progression of many types of technology over the history of mankind. And we have been in the pursuit of creating something smart that will help us. Since, you know, as, as we talked about in transcendence in 11th century, 10th century time frame, this is not a new pursuit, but we're finally here, right? So now that we're here, the question is how far do we want to take it and should we take it? Right? So those are the bigger question because just because we can create car that can sense whether. I'm feeling sleepy and I didn't have my espresso, should it just automatically drive me to an espresso bar? No, no, I don't think so. You know, I want to decide that for myself and just because I bought a pair of shoe two years ago, I don't want to be told that it's Time for me to buy a new pair of shoe. You know, I want to decide whether I shoot or shoot them. Right. So that's the dark side, but good side is, you know, imagine like which I've kind of noticed with my mother especially, you know, because she recently passed away. But she was suffering from dementia and whatnot. She had to have constant companion and she was in nursing home. I could easily see a AI companion that could have helped her. She's originally from Bangladesh and she was fluent in English because she lived here. But as time went by she would start in English and then veer off to her mother tongue and then she had a very difficult time to communicate. And you can easily imagine AI assistant taking over and translating it or guiding her and having the conversation. So there's lots of, lots of good stuff along with all these bad stuff that we can scenario. So as you kind of seen from the galley of Transcend, I'm not like picturing a dark picture. I'm. We kind of tried to take a pragmatic approach whereby we say look, we're pragmatically optimistic, but what happens to humanity depends on what we do to yourself or ourselves.

I think this is a very good point that yeah, people have been thinking about intelligent machines for hundreds if not thousands of years and it's just sort of become a lot more urgent in the last few years. And I have to say I really like the idea of having assistance for people with dementia because that's just exhausting to deal with like for months or years at a time for loved ones. So yeah, that seems brilliant. Use case. And in fact another one of the big themes in your book is around AI sort of affecting almost everything. But a lot of the impacts are relatively trivial and there's only going to be a few very important use cases. Do you have a sense of what those use cases are or what those are going to be?

I think the use cases depends on who you are and what you're interests are. Right. So you know, we talk a lot about purpose, but so it depends on what is it that individual purpose is, an organizational purpose is. I probably shared with you that the book proceed goes for cancer research because I have a cancer surviving son and so I see like a drug discovery, optimizing patient care, cutting cost for services, you know, those kind of application. Huge. And then even as a researcher of technology, I constantly find myself using on demand research even for the stuff that I have done over 30 years of my career. It's very easy now having all these manuscript in A large language model or a library of code base and say, okay, pull me this, pull me that and it's instant. Right. Used to be like take days and days to figure out even what I did because I don't remember all the stuff I did or didn't. So there are a lot of use cases, international security use cases where you can maybe come up with predictive models. Not offensive. But from a defensive point of view, there's all kind of great use cases. I don't really find that idea of, okay, can I create a cutesy video and a fake video? Very. I mean to me it's like trivial usage. It's entertainment. We're already gone as a society. Like who likes to be entertained 7 by 24. And that's why you see all these emergence of TikTok and reel and all that. People are addicted to it. And AI is obviously helping that because it's feeding you by your behavioral pattern and algorithm and allowing you to create all kind of entertaining things. I don't find those use cases very meaningful. That's just very personal being. Right. The use cases that I talked about, that could be game changing for various reasons.

I have to say I have actually been quite enjoying all these kind of recent AI videos of cats cooking or whatever. Bit of a trend. But I'm right there with you that actually, yeah, probably using AI to cure cancer. Slightly bigger use case for sure. Yeah. So there's different sort of levels of things. And you're kind of right that there are some huge problems that humanity's been struggling with for decades, centuries, but. And we need all the technological assistance we can get to solve them.

So for sure, I mean, look, I mean even like, you know, we thought things we take for granted, but once you start thinking about it like, you know, the medication that you take, you and I may be exactly the same age group, but the reality is that based on your DNA structure, based on where you are, based on what kind of medical history you have, the same panel may not have the same effect on, but we get prescribed the same exact same thing. You got blood pressure, so you take this. We're not at a level of custom service and custom care yet, and we're getting there. But I think if the AI could help us tremendously to collecting that data, genericizing data and recognizing the pattern and giving you the right solution for the right problem. That's what I'm most excited about. And you can apply that portion of it just about anything, whether that's business model improvement or whether that's cancer cure, whether that's geopolitical relations, you name it, you could probably apply the same that basic fundamentals.

Yeah, I like the idea. Personalized medicine's been sort of hyped for like at least decade, but it's only just sort of starting to make inroads, I think. Yeah, lots of cases for AI making personalization cheap. It seems pretty useful. So I do love a framework and I know your book has two of them in there for like how we can get into the sort of practical details of working with AI. So do you want to tell me about your open framework?

You know, open framework. So we tried to be useful, but we also tried to be very simplistic in the sense that anybody can grasp it. You don't need to be a management scientist or a computer scientist to understand how some of this framework works. One of the things we did try to do is that we call it a framework and then we have embedded a lot of the sub methodology and we don't really call this a methodology because we borrowed or applied lots of different discipline in the framework, meaning one extreme philosophy for mindset, another extreme like things like management science, centric stuff like SWOT analysis and pastel and agile and you know, management type of stuff in the framework. So if you look at open, the idea of open was we have to be open to possibilities. That's where open comes from and a care, which we'll talk about in a second, which is that we have to care to protect humanity. So you have to be responsible and create the governance model. So that's where the open and care, those two word comes from. The open stands for outline, partner, experiment and navigate. And what it really boils down to that you have to outline what's the purpose that you're trying to achieve, which we kind of touched upon in our conversation. What's important to you and who you are, what do you want to do with your life and how do you want to use anything or do anything with these technologies really depends on that kind of thing. So you basically come up with outline auto possibilities. That's the first step. Second step is that then you have to decide who do you want to partner up and what do you want that partner look like? When I'm talking about partner, when I'm just not talking about a vendor or technology platform, should you use ChatGPT or should you use Microsoft or Google? I'm not just talking about that, I'm talking about what do you want your AI as a partner, what do you want it to be doing AI is already moving to agents, and we're talking about agentic AI. Obviously, there's generative AI, there's analytical AI and there's automation AI, a variety of AI. So you have to kind of decide what do you want that AI to do, and you have to define that AI's Persona as your partner. So if you want it to be a research assistant, like that's actually one of my biggest use cases these days, then that's a Persona I define. It's some criteria and it does certain things based on the things that I do. If you want it to be an assistant that does your scheduling and maintain your calendar, you can do that too. So you have to kind of decide what kind of partner you want the AI to do, but also who helps you to make that happen. And then you have to experiment to see what's working and what's not working, and then kind of narrow down what's really meaningful and what's going to give you the biggest return. However you want to define return, the return could be that you want to solve world hunger, or it could be that I want to quadruple my revenue over the next two quarters. Right? So whatever the goal is, you kind of decide based on that thing. Now once you do that, you know, Maybe you had 10 ideas and you have narrowed it down to two or three, and then you experimented with three and said, okay, now I'm going to navigate to implement and measure and track to see whether it's really adding the value or not. So within those four segments, there are many, many steps which, you know, we wouldn't have the time to get into it, but basically it takes you step by step process of how do you do it. And the reason we kind of done that because, you know, unlike any other technology, like, you know, you can look at, okay, like CRM or supply chain or customer service, pick any, any technology. You know, those are kind of, to me, passive technology. It was not an active participant in your thought process and your engagement and whatnot. I mean, your Outlook calendar or your Apple calendar, whatever, calendar is still passive. You input and then it, you know, you look it up, right? So it's not telling you, oh, it's time for you to schedule a doctor's appointment because you haven't had a physical for one year. So it's not doing that. Meaning the last generation of technology, this stuff is an active participant. So thinking about from that perspective requires a different way of thinking and different way of, you know, experimenting and executing. That's why we laid out those four steps.

I do like the fact that it's just sort of these sequential steps and it takes you through from trying to figure out what you want to do, this outlining planning phase right through to navigating how you're actually going to use the thing you mentioned it was inspired by some existing sort of project management frameworks. You mentioned Agile and a few others. Did you have to customize it to take account for any sort of facets of AI? What do you do to tweak it to make your own?

Yeah, absolutely. I mean the very example that I just gave you, that is an active participant versus a passive participant changes everything because in the old days you would come up with a project plan and say, okay, look, I'm going to do this development and I would test it and then design it and roll it up. It's a bit different because it also depends on how sophisticated and complicated you're making your whatever that you're trying to implement. But it could get quite involved and in that scenario it's a bit different. I mean, it's not like thinking about it kind of a sequential process, even though you can argue, well, that is really not sequential. It's, you know, you used to do waterfall, blah, blah, blah, when you can get into all kind of philosophical discussion. But the very nature of this technology is that you have to consider it as a partner. So you have to treat it as a partner and then you have to put guardrail. You know, how much partnering do you really want to do? Do you want to become a child or do you want to be remain as a parent? And it's still kind of under your tutelage. That's really the bigger question.

That certainly is very interesting, the idea it's a partner in some sense rather than just some sort of passive piece of software or passive project. Can we make this a bit more concrete? Have you got any examples of how businesses might make use of the open framework?

Yeah, a lot of this open framework or the framework that I have prescribed in, it's implicate derivation of my previous work. Like I've spent 30 years in organizational and technology transformation and you know, if you look at the old days of digital transformation, I mean, you used to have like, okay, I'm going to do an assessment and see where we are and I'm going to look at what the opportunities are and you're going to come up with a series of innovation portfolio or some portfolio things. I'm going to allocate resource and money and then I'm going to execute it. In that example that I gave you, you're kind of doing very similar things, but the biggest issue is that because a lot is unknown and you're really trying to create a Persona with some of these things that become part of your organization. What I mean by that now you have human resource and you have AI resource, right? So you're allocating tasks not just to human, you're also allocating tasks to this thing. Whether that's an agent, whether that's a behind the scene automator, Whatever the case may be, it's still a resource. So when you are coming up with these exercises of doing it, your thinking pattern is more looking at these opportunities as actual resources versus like, you know, the classic software development model. So that's a huge, huge shift. And you ask me is any companies or organization are using open type of framework. I would say that we are still at an early stage of doing that. There are bits and pieces of these practices and that I have already seen in last three, four years where you know, people are kind of looking at this differently and organizing themselves differently and assessing things differently and allocating resources differently. But I wouldn't say that it's at a maturity level. So since you brought up agile, I mean, you know, you do classic maturity model to see where an organization is. I would say in AI development cycle, most organization is probably zero, maybe one. There's a lot of talk, but you know, unless you're talking about like a JP Morgan trading system or you're talking about a handful of things that Amazon are doing or a couple other people are doing, we're not quite there yet. But the model that I prescribe, it is based on the current practices of many different type of organization, AI or not, but also includes AI models that I think it's going to become more important because it's not just about systemic thinking. It's a thinking system that allows you to constantly adopt and change midstream as we find out what's really working and what's not working.

That's fascinating stuff, particularly the part about how you need to start thinking of these as resources rather than just like pieces of software. So I'm curious, is this going to be resources like so you think of cloud computing resources or is it going to be like people team? We're talking about human resources here.

No, I'm talking about human resources, I'm not talking about system resource because you know, if you go classic network stack, you can say, okay, well here's my infrastructure stack and here's my system stack and here's my application stack. And these are resources, right? And I want to spawn off a resource to execute a particular set of tasks. I mean we've been doing that for last 30, 50 years, various from mainframe to, you know, land to cloud computing. I'm actually talking about considering this as part of your resource that actively does things. So what I mean by that, think about this way, it's already happening in the sense that you call up for your current needs service. There's a chatbot that just answers your phone and, and it's taking your information, digging up your information. That is I just allocated my chatbot to do my first level of service. So that's my level one service. And then you said, okay, if they do the level one service, if it's xyz, then I'm going to pass it to my level two, that could be another chatbot. And the level three is actually human. So I consciously decided that my level one service would be done by Chat one or agent one, which is an AI agent. Level two would be done by another set of AI agent and level three would be done by human resource. Right. So if that is my organization, then I've just structured our organization by Saying My Level 1 service is staffed with this basic chatbots or agents. And then level two is slightly more intelligent who looks at exceptions and doing exception processing and deciding, oh, based on this, I pass on to this. Based on this I pass on to that. So you're consciously allocating work. So it's not just about system design, it's also your organizational design and task allocation.

I guess just taking that to the next level, that is you have to start implementing like human resource procedures for your AI. Then you're going to have to give them like a performance review, certain targets, all that kind of stuff.

It may sound hallucinating and funny if I can use that term, but yeah, because you might have seen in Transcend we talk about coming up with racy charts. What racy chart is, is basically allocating responsibility based on who has the decision power and who has actually doing it and who's leading it and all that stuff. So I do prescribe that if you want to implement agent, then you have to create procedures in the form of racy at highest level in the form of a racy. And then you have to decide, okay, well is it really doing it or not? And in the, in the context of performance, it's not just the, is it really doing the task? It's like how accurately is doing the task? Right, so you don't want to have an AI agent guide you. Oh, you got headache, you got fever, you got a broken bone, that means you need to immediately go to surgery. The AI is not intelligent yet. But this is just a dumb example, if I can use that word, of a simplistic way of looking things. So I highly prescribe that there is human intervention in sudden critical decision. You have to decide what's critical and what's not critical based on the situation. But I don't ever prescribe that we completely take our paddle in a hands off. It's like that car scenario. Do I want my car to really veer off middle of the road because it sensed that my blood pressure has gone up slightly higher than what it was this morning? No, I don't. So you have to decide. But even though we're saying this tongue and cheek way, but that is, we'll have to get to that level of granularity. If we don't, then we're not going to be successful. It will start doing its own thing. Here's the other reality, which is a simple mathematical reality. You as one person or I as one person versus a network of 10,000 of us that has every facet of life. Learning from biology to philosophy to organizational management, system development, whatever the case may be, will always be smarter than us. So if you want to allocate resource and then you also want to give the power of allocation of resource, is going to come up with far better, more efficient model, maybe name it beta, right model. But it will always come up with a better and faster way of doing things than you and I would. So there is the danger.

Yeah, as a human, you're not competing against a single AI. You might be competing against a whole network of 10,000 in terms of performance. So certainly like AI, dramatically more scalable than humans in many.

Yeah, I mean that's what really is. I mean like even my little example of I loaded up by 10 manuscript, I can only think about, oh, did I write this here? Did I write that there? Versus I'm just asking you. You wrote here, you're here. It's like my brain doesn't function as fast. Even in that simple example of retrieving a text from a bunch of PDF manuscript. The combination of OCR and my generative, generative way of interacting with the engine to creating the data is a lot faster than I can do.

Okay, so there's one more thing I wanted to talk to you about. The, the open framework. So it Seems like one of the big themes of this is you need to open yourself to the possibilities of AI. You need to educate yourself. What is it you need to learn around AI?

I think there's actually three factors of learning. One is obviously you have to learn all the technology that are becoming available because it's almost ridiculously impossible because every day there's something new, but you have to keep up with it. Second thing is you have to understand and learn and hone your critical thinking skill because that's how you're going to evaluate what's real and what's not real and how do you really use it for the purpose that you're pursuing. And the third thing you need to learn is the art of possibility of applications. How can you use this for this or how can you use that? That's necessarily not, oh, I have this tool, therefore I can do this. It's a little more imaginative. Can I do this? I would like to do this. Is it possible to do that? And so when I say you have to be open, that's what I mean by open. It's like the age old saying that more curious you are, more learning you can gain. And the more learning you gain, you know how little you know, therefore you just keep pushing yourself and, and that's how you become innovative. It's a kind of the same model, except that here we never faced with keeping up with am I outsourcing my critical thinking? But here we are, right? So the more you tell your chat GPT to craft your email, the more I gotta interview. So let me just, you know, here's the documents to come up with 10 questions. People are doing it or like, you know, like the, you know, I saw a show yesterday morning. Last night I turned on my Netflix and it's already suggesting my next show should be this because I just saw this or do I really need to know that? Or can I decide that myself? So there are various level of openness, if I can use that term. And we can probably spend a whole show around philosophy. But in the book, you know, you saw, we prescribe a lot of western philosophy as well as Eastern philosophy, right? So the notion of the Roman philosophical view of knowing thyself or you have to be mindful and you have to have a beginner's mind from Buddhist philosophy, we kind of tapped into both of those thought process because that is humanity is once you are done with it, then what are you? I mean then you're a mechanical being, right? Because you don't have original thoughts anymore.

I have to say you mentioned Netflix and this is something I'm definitely guilty of. It's like, oh, Netflix will like recommend a show, click on it, start watching it. Over years, the quality of my Netflix recommendations have gone down and down and down because I keep clicking on the thing and now it's just like baking shows and things about people with plastic surgery selling houses. So yeah, just because this is what gets recommended. So yeah, certainly it's a good thing to know that you can ignore what the AI tells you and just think for yourself a little bit.

Again, we end the book by saying that you have to stay devoted to, to what's important and you have to detach yourself from that level of convenience. Again, it comes from the devoted, detached, Eastern philosophical mindset, right? So because if you don't stay devoted to force yourself, it's a discipline issue, right? So because, I mean, human beings are very convenient centric, right? And you know, it's easy. So let's do this, it's easy to do that. So let's see this. And he's easy to order food from the same people. It already has my order. I just click a button and it's just delivering the same food, right? So the convenience factor is what we have to detach ourselves from. And we have to stay devoted to critical thinking and forcing ourselves to say, oh no, I don't want to do that, let me try this. And you know, it's a double edged sword in the sense that human beings are very creative and we reach to this creativity level because we force ourselves to learn something different and new. And it took effort to learn. You can still do that. And you can argue that you can learn faster if you stay disciplined. So the trick is, can you stay disciplined to learning and use this technology to aid you to the learning versus letting it do it for you? Do you really want the AI to write your next poem or do you want to see whether you can write it? It may be shitty, but at least you wrote it. And in the process of writing it, you researched 10 other poetry written by some of the greats. So it's that kind of discipline that's going to take us there, I think.

Sounds like a lot of hard work, I have to say. This display idea, but yeah, incredibly important. And I do like the idea of doing things for yourself sometimes without the use of technology, just particularly as you're learning, just to make sure that it really sinks in. Okay, so let's talk about your other framework, the CARE framework. Talk me through it.

So the Care is, as you know, there's a lot of talk around the responsible uses of AI and putting guardrail and putting governance around AI. So and we call it care because you have to say why do you want to be responsible and why do you want to cover? Or the fundamental note is that the fundamental point is that you want to do that so that you can actually protect humanity. You have to care about what you're doing. So the CARE stands for the first Alphabet stands for considering catastrophic situation. So it's really coming up with a risk portfolio. So the idea is that, okay, you got those 10 ideas and now start looking at, okay, what, what are the catastrophic things can actually happen in the process of doing those ideas. And you have to look at short term, long term and medium term risk to see what those are. And we talk about four P's which you might have seen in the book. It's about the planet, people, the profit and product, its impact on those four things. And considering it from that point of view, just like in open, you let your imagination go wild. What are the possibilities here? You can actually start considering the worst case scenario, which is healthy. I mean, if you talk to the psychologist and especially the organizational psychologist, they'll all tell you it's healthy to think about worst case scenario because that's only way you can become pragmatic. False hope is not way to live. Because if you don't consider any risk, you will dive into situation that you will not get out of. So the first step of creating these things are coming up with this notion of risk portfolio and allocating risk matrix. That's number one. Number two is that based on that risk matrix and the risk criteria, you have to assess the various element of opportunities that you have. So that's the second phase of care assessing. And then the third thing is that you have to put some guardrail or regulation, you have to regulate whatever you're trying to do. In a technical term, what that means is like okay, where do you want to put your set of data? Where do you want to apply, let's say zero trust architecture to protect your network, et cetera, et cetera. So it's both from a management leadership point of view, but also from a technical point of view, you have to put those guardrails. And the last thing is that you call it exit. Exit is really figuring out up front what's the exit. When things really go bad, you have to have some sort of a kill switch because things may actually go bad. How do you shut it down, right? So you Know, for example, you have some sort of a agent running around that's cataloging everybody's Social Security for a particular purpose. And before it knows, you gather everybody's Social Security and then you just push it out of the way. So do you really want that to happen? Or let's say, you know, I mean, I was last year, a year before last year, and I was working with one of my government clients and we're talking about unclassifying old stuff that's no longer relevant. And people have the right to know these things. So can we use a AI engine to kind of look at those documents and put they call doctorates so you can't read the whole thing, but you can get a sense of it. And these are done manually. It's a very manual, intensive process. And that's part of the reason a lot of these things doesn't get actually unclassified. So the question is, where does the data reside and how do we decide that the doctoring of the data is actually accurate? How do we know it's not appropriate hallucinating? I don't know whether you played with any of this model, but whether it's cloud or whether it's chat, GPT or any other model, you know, you put a PDF and then it start, veers off and start telling you stuff that doesn't, it's not true. I mean, it just doesn't exist. Right, because it's tapping into other data sources. So I mean, it's a little better. Like, you know, what I did with my manuscript, you know, it's a closed model and it's my little large language model just for me. But so it doesn't veer off as much as, you know, when you use these publicly available models. So you have to have an exit pattern. That's really the point. And you cannot do these things without having, you know, the bigger the opportunities are, the greater the risk are.

So compared to some other risk frameworks. So the first step is catastrophize. Now my understanding is a lot of risk frameworks, they just like list every single possible thing that can go wrong, whether it's small or large. And then you worry about like how big the problem is later on. Why did you opt for going with the worst case scenarios, these catastrophes?

Because unlike other technology, this technology just like there is enormous amount of opportunity, there's also enormous amount of risk. I mean, it's not just my word. I mean, you know, I mean you can look at like Godfather of AI, but they all, we all acknowledge that the risk of AI is nothing like we have ever seen. And you can get into a whole conversation, okay, well what we talked about, this framework by the way, can be applied for an organization that developing stuff meaning using available models and available infrastructure. It could be that the government agency, it could be individual. Right. So we actually show that in individual government and businesses. But the governments also have huge responsibility thinking this way because you have to really put legislative framework that pushes for responsible innovation and utilizing technology for doing good, but also having the guardrail that allows you to secure national security, secure population, secure people's identity so that the agencies don't become Big Brother and tap into our privacy. Rogue nation doesn't have access to these things, et cetera, et cetera. Right. So that's why the risk factor, I mean depending on what is it that you're trying to do, becomes big or small. So it is utterly important to think about that kind of catastrophic situation versus like other technological thing. Okay, we have system down so people doesn't get to access their, I don't know, bank account for a couple of days. Yeah, that's a risk, but it's not like life and death. You'll still may survive if you don't have access to your bank account for a couple of days. So that's why we kind of looked at it from that point of view. As optimistic as we are, I also want to be as cautious as possible because this is a double edged sword.

I'm curious. So I mean you mentioned that these things can sort of scale right up to existential risks. And I know a lot of serious AI safety researchers do think there is a worry about like a real risk of a sort of powerful AI, like just wiping out humanity. So the probability of this happening is known as like a P doom number. I'm curious, do you have your own PDOOM number? Like what's the chance of things going that wrong?

No, I, I don't, I don't have it. But you know, I mean I'm more, I look at this and the more I play with it, more I apply it, you know, you know what I'm saying? I'm playing with it. I'm not just talking about as a consumer playing with it like a generative AI. I'm talking about enterprise level type of stuff. I have serious concern. And you can also look at short term risk and midterm risk and the longer term risk. Longer term risk is that doom and gloom where it just taken over because we have become child and it no longer needs us to make any decisions. It's making its own decisions, manipulating market, manipulating which territory belongs to who, starting war. I mean you can imagine. And then all of a sudden you can go sci fi model where it says we are doing all this to protect humanity, but in the process of doing it, they're really taking over and making us a secondary entity. But if you look at, let's give you like a simple three to five years of basic fundamental thing that we can all relate. So imagine now 40% of population got replaced, meaning their work got replaced. Because you have AI robots making your hamburger and you have self driving cars, driving the car. So you don't have any taxi drivers anymore. You have drones delivering your delivery stuff. You got drones, AI slash robotics filling up your shelves in the, in the warehouses. So all of a sudden 40% people are unemployed right now. Let's take it to the next level now. Next level of people are basically knowledge workers. So, so you say, okay, I used to need 10 people to do the research work. I need now two, I don't need like 10 trauma nurse, I need only three trauma nurse. Because those AIs are much more tolerant of rambunctious patients and people who are in pain, right? So all of a sudden five years from now or seven years from now, you get 70% global population that no longer are doing what they were doing before. Now some of them will find different jobs and whatnot, but vast majority of people may feel that they get completely marginalized. You look at today's unrest globally, which is human being creating the unrest for human being. You are now got unfairly displaced by these things and there will be massive unrest. So that's a dark picture, but I think that's where it will start. Not AI taking over. It's marginalized people feeling they're no longer being fulfilled. So you saw in the book, we talk about masterless hierarchy, which everybody knows. So if everything is fulfilled, meaning you don't have, and let's assume that somehow we don't need income anymore because it's been provided to you. There is universal pay and all kind of stuff and you don't need to make a living, but there is this innate human need to create and to be productive. Once that's gone, then you have a lot of unhappy people. When you have a lot of unhappy people, that's where a lot of these kind of disaster scenario props up all over the world.

Okay, yeah, certainly mass unemployment, like you've mentioned the idea of like, well Maybe it's like 40% of people unemployed, maybe even higher. That's going to cause big societal problems. Do you think that's likely to happen and what's the next step after that?

I think it's already happening. We keep talking about reskilling and we're talking about we have to prepare for. Part of this legislative framework that I talk about is preparing the workforce for this kind of shift is critical and that can be only driven by the global leaders because it's the awareness and creating that infrastructure that allows you to do that. So if we don't do that, that's why I say what will happen is what we do or don't do. I mean, human history has gone through a lot of ups and downs. So time will tell whether it's upswing or it's downswing.

Yeah. So that sounds like a fairly terrible future and I'm hoping that we manage to prevent it. So can you talk me a bit through about how do you give people the skills to avoid being out of work?

I think we have spent a lot of time talking about stem and we do spend fair amount of time skilling people with stem. I have a counter argument that I think the most critical skill set that we have to provide to maintain this upper hand is we're going to have to start teaching people how to ask the right question and develop a set of critical thinking skill that allows people to take advantage of this technology to be part of the process. Not being marginalized. That takes a conscious effort by individual, but that also takes an institutional effort to do that. Meaning what you're teaching your kids in school, what you're teaching people, people in universities, what is the organizations preparing their resources for the next level of change, etc. Etc. Because if you don't do that, these are like a collective effort which we talked about. Like, you know, if you surrender yourself to all convenience, then you become dollar and dollar and dollar and you're not really honing your skill set so that mentality has to shift. I mean, already we got nonstop research on how much time people spend on social media, useless social media, and how much time people waste on consuming things. That makes no sense. Because we're highly entertainment driven, we have shifted into a highly entertainment driven society. Combine that with outsourcing, even our thinking pattern, that's where the disaster can become very interesting.

A lot of people tend to be worried about this sort of Terminator scenario for like AI catastrophe. But actually the thing we keep coming back to is just AI just taking over our lives. It sounds more like the cartoon movie Folly. Whereas in the future everyone's just like sat in hover chairs or whatever.

And yeah, I mean we're already talking about going to Mars. So you can say, okay, we destroy the planet, so let's just take like a handful of humanity and push them to Mars. But it's still with AI. So when even if you go to Mars, you're still like being controlled and manipulated by AI.

So another sci fi dystopia to worry about. All right, so I just want to talk about the E part of care. This is about exiting and this seems like a pretty radical idea. If AI is going wrong, you need an exit plan. You've got to be able to turn it off somehow. I guess not many sort of risk management strategies talk about like, well, you just need to pull the plug. Do you want me to talk them through? Why did you decide that was necessary?

So I mean there is no like one AI, right? So it's not like there's a AI and there's like a centralized thing and you just go on unplug and then you're done. I mean it's a, it's really in the context of this framework because we talked about from a very pragmatic point of view in the sense of okay, well how does the individual use it, how does the organization use it and how does a government agency use it or apply it or develop it? You have to look at from that micro point of view, not necessarily just macro view. So each person will have to have the responsibility to pull the plug when thing goes wrong. And that's important, important because if we don't, it becomes like a virus. Think about, I mean, like when does a pandemic happen? When two people die, it's not a big deal. Let me go in the mall. And then you infect a thousand people, then the thousand people infect million people. I mean AI is a network of system, the global network of system and global network of intelligence. So you can't really pull a central plug. You have to be cautious of what, what you are individually building and putting it out there and then decide, do we leave it all that on individual responsibility? Well that never works because that, you know, that's why you have laws and regulation and compliance and legislation. So you have to focus on that element of it as well. And you know, you, in the book we talk quite extensively about governance model and who governs and all that stuff. And it's the governing body's job to decide where to pull the plug and when to pull the plug and if they don't, then there has to be a consequence and those consequences has to be legal. So for example, you know, we started talking about Big Seek I think earlier. I mean it's look at, okay, so should we allow them to be in every marketplace, whether that's Apple and wherever and everybody's downloading it and then they're using it and it's sucking your data and now we have all that data in the hands of a different type of mindset that believes in very different things as we do in west as an example. Should we allow it to do or not? I mean that's a legal, that's not just a moral question, that's a legal and immoral question and it's an international geopolitical question. Right. So it's much broader than that. But in the context of this is like everybody needs to have a exit button or switch up point for anything they do. Otherwise you're going to create a massive pandemic which there's no point of return.

Okay, yeah, an AI pandemic just sounds like a new kind of disaster movie as well. So I think the reason I said this idea was controversial was because that was one of the clauses that was proposed in the California AI safety bill was that powerful AI has to have this off switch and that was one of the reasons that the bill got rejected was because a lot of companies didn't want this. Do you have a sense of why companies didn't want to include this idea?

Because there's a conflict of interest. Right. Because we're also in a race of innovation. So the idea is that okay, well are we going to snag them the innovation and if we put this kind of guardrail exit point, my humble opinion is that we need to push the boundary of innovation but we have to be responsible innovator and there has to be consequences for not following those guardrails. And I am of the proponent of doing those kind of things. Now we don't see this effect because it's not like you can relate, you know, if you talk about nuclear assets or nuclear energy because we have seen it, we used it and we demolished cities and you know, with atom bombs and whatnot. So we now are very cautious about who has access and what can we do with it and all that kind of stuff. We are not that with the information or network or computing technology yet because we haven't seen those disaster, we've seen glimpse of it. Like, you know, remember that about a year ago when the airport starts shutting down because a Cloud strike AI engine decided that there was some malware and it started to shut down the system. So we haven't gotten to that point yet. And I hope that we don't have to get to that level to decide where is the pull out switch and what's really the guardrail. Now even if you look at pharmaceutical industry, you know, I mean even there, there are gun rails in the sense that not everybody can set up like a pharma factory and start. I mean that's why we talk about this illegal drugs all day long. I mean even with all those guardrails we still have having all these drug problem all over the place. But even in that industry it's far more regulated than let's say information technology or computing technology. So don't we not want to have at least similar level of God will? So if we're not, then we're not being responsible. And you know, if you look at the US vs Europe vs China or other parts of the world, there's different viewpoint. But I think we have to push the boundary of innovation. But we cannot not be responsible if we are not. Those disaster scenarios that we talked about is bound to happen.

Okay, yeah, I like the idea of like you got to take responsibility for any AI you're creating or you're using. But we've had a lot of talk of possible disasters here. I really would like to finish on a happy note. So I mean since your book's called Transcend, suppose everything goes well with AI. What can AI do to help humanity transcend its limitations?

I think that AI could give you a better quality of life like the scenario we talked about, whether that's a dementia patient or providing a better drug discovery for really bad diseases like cancer, allowing you to be more productive and creative. I find myself being much more creative and productive in terms of my research work than I was even a year ago. Because now I have a vast access that can get me stuffed. So it can be a better assistant, it can assist in many different ways. It can make organization more productive as a result of people that are in the organization can focus on more creative pursuit. Because ultimately, you know, if you going back to Maso's law, I mean once everything is met, what left is fulfillment. That's very human. Right. So we want to be fulfilled by pursuing our gifts and desire to make a difference or whatever. Right. So I really wholeheartedly believe in that. So from all those point of view, you can think about climate, disaster, all kind of stuff, predictive modeling, AI could be an enormous tool. But if we apply it the way, in a responsible way and with the intention of doing good, it's like, you know, nuclear energy is a good example in the sense that, oh, we can create nuclear warhead, or we can create energy source that can actually light up the entire city with very low cost. It's our choice. So it's AI, I think, is a similar choice. The difference is though, however, that it's a very. It has a democratic access. Anybody can access it and anybody can do whatever they want to do with it. And I don't think you can leave it up to just personal responsibility because there will be always some people who want to do bad. So you have to have laws, and the civil society doesn't work without having laws and legislations.

Okay, yeah. So scope for really good things or really bad things. We somehow ended up going from happy things back to chance of wars again. So, yeah, maybe just to finish, like one of the recurring themes seems to be about personal responsibility and also just societal responsibility on a broad scale things like legislation. Do you have any final advice for the audience and their interactions with AI?

I think you can look at social media as a use case. I mean, it has done a lot of good and it has done a lot of bad. I think AI will only enhance that because it's a mirror. So in terms of use it, use it to do something good with it, don't use it to do bad. That sounds so basic and so moralistic, but that's what it is. If we collectively all start, don't have any responsibility for doing anything like identity theft. You name it. You can do all sorts of things, but you can also use it for all kind of things. Be more creative. So try to consciously think, what are you using it for? And then for a very selfish reason, don't outsource yourself. If you outsource yourself, you're not going to have any value. So even for selfish reason, use it as an aid, not a replacement.

I like that. Just be mindful about trying to do good with the technology you're given. Excellent. All right, lots of wisdom there. So thank you so much for your time, Faisal.

Thanks for having me, Ruchi. I enjoyed our conversation.