When Data Drives the Revolution ft. Sol Rashidi

0:00Do you know the number one obstacle to financial success? time or lack of it without enough time you can't learn efficiently plan effectively or focus on the right strategies that's why real vision offers you a simple and efficient way to gain expert knowledge using time-saving market tools and leverage the brain power of our community to help you succeed faster get a taste of financial freedom with our offer at realvision.com forward slash free. That's realvision.com forward slash free.

0:42Welcome back to Real Vision. I'm Ash Bennington. We're joined today by Sol Rashidi, C-suite data leader and author of the new bestselling book, Your AI Survival Guide, scraped knees, bruised elbows, and lessons from real-world AI deployment. Saul, welcome to Real Vision. Thank you so much, Ash. It's amazing to be here. It's so great to have you here. And boy, we couldn't have had better news flow leading into this with the Apple intelligence announcement and the kerfuffle around it last night. Gosh, where do we begin? Obviously, huge topic. Let's start at the very beginning. Let's define AI in the context that you see it.

1:23And I think I is kind of like having, it's so funny just to back up that I'm having this conversation with my kids and I'm teaching them a little bit about prompt engineering. And my six-year-old daughter, who's obsessed with axolotls right now, I'm like, you can create any type of axolotl you want. And so we would go into chat GPT, I'd open up the dolly, prompt, and I'd be like, okay, go ahead and type what kind of axolotl you want. And she went through a few exercises and generated images. And I'm like, well, you've got to tell if it's doing a good job or a bad job, just like mommy tells you when you're like getting good grades or not so good grades.

1:58And be specific in detail. So she'll go through a second round and a third round and then a fourth round. And now every Saturday is part of our weekend creative time. She gets 45 minutes just doing a variety of props, creating different axolotl images and colors. And she'll write, make it cuter, but this time add uniform horns, things like that. so the way they explain to them when they ask well ultimately what is ai and i said well it's kind of like a super smart shooter um that can learn to a degree what we humans have learned um it's great at certain things and not so good at other things like it can't necessarily read emotion um it can't process uh emotional intelligence um there's certain cognitive capabilities it can't really quite do but if it's something to be learned that's repetitive in nature or something to be learned that has a finite set of responses or something to be learned that we've learned in school, like grammar and spelling, it can actually learn what we've learned up to a certain point.

2:59And it learns so in zeros and ones, because that's ultimately computer language. So a combination of software, data science, code, hardware, storage, compute, all of that is needed as a necessary foundation. just like you have food, just like you have a root, just like you have mommy and daddy that take you to school and bring you back home. Those are foundational needs so that it can learn and grow as you are learning and growing. And then through our teachings, it learns with time and it gets better with time. So if you ask me what AI was, that's pretty much how I describe it to my kids. It's the ability for a computer to learn what we learn to observe.

3:41It doesn't necessarily pass that point, but right now there are still limitations to it. as a student of Charlie Munger, I love this idea of invert, always invert. What is not AI? Because if you listen to chief marketing officers in Silicon Valley right now, everything under the sun is AI. So what's not AI? Yeah. And may he rest in peace. I was a big Munger fan. Um, and I learned Munger's philosophies and principles to a dear friend of mine about a few years back. So there is a lot of sleep oil. There is a lot of hype. There are things that we are putting the AI logo on it because it sells better.

4:17It gets you in the door and you can charge a framing for it. I unfortunately can't lie about that. And I think for people who aren't in the space or people who don't know what questions to ask, it's hard for them to make the distinction that ultimately makes the difference. So in our world, what's not AI is decision trees that have been scripted. if this then that if this then that if this then that so if x it's y if it's z it's b if it's a choose c like those are specific script that you can code that you can automate that could essentially very finite in nature there's no ability or need to think to come up with ad hoc responses to be able to understand contextual references but the beauty of ai is it can understand context after it hits a certain level of maturity.

5:12So like if I were to example, as an example, type in wet feet. What's not AI is they don't know what wet feet means. What is AI is they would understand that, okay, am I standing in the rain and fundamentally my feet are wet or am I getting wet feet before I'm about to get married? There's context, there's language. It can have a dialogue with you as a result. So like that would be a good one. Or another one is the word gray. okay are we talking about mood or are we talking about the color what is not ai can't tell the difference it's just a word with more letters but if it is ai it can understand context because it learns that there's fundamentally a distinction between rays and i'm in a gray mood versus rays and the color so what was it that compelled you to write this book obviously writing a book is an enormous undertaking what did you feel needed to be communicated what did you think wasn't being said in the marketplace of ideas?

6:06It's interesting. I helped IBM launch Watson in 2011. And it means several leaders and several teams, right? It takes an army to do something like that. And I've said it in the book, and I've said it a few times. I don't think IBM got the credit it deserves for launching something at that magnitude. And it's always hard, right? Being first to market, first to launch, because all arrows are pointing at your back, and the exposure in the limelight is large. So people can be very, very fearful. But when Watson went live, right, it was the first commercial-grade AI application. It was B2B, but it was enterprise-grade, and it could be deployed at organizations of mass scale.

6:44And I've been doing it in 2011. It was me connecting with clients, establishing strategies, use cases, being the project lead on deployments. And so I've been doing deployments since Watson went live up until a few years ago. I should say a few years ago, up until about 11 months ago. and now I'm consulting and advising and helping guys do the deployments. But I've rolled up my sleeves. I've gotten into the weeds. I've gotten into the nitty gritty of things. And it's not things that you read about. There are so many nuances. And I think, you know, everyone assumes if AI is going to work, the technology problem.

7:19But 70 % of the reasons why AI applications or AI deployments don't work, nothing to do with technology. But unless you've done the deployment and unless you've gotten your hands dirty, everything is either rumor hearsay or regurgitation of something that's been read and so when things sort of the hockey stick curve the ai trend started going upwards you know people were encouraging me to write they're like you've been in space longer than you know and i was like well let me just do some research because i'm not a writer i've been so busy doing the work i haven't written about it and every ai book i saw out there was either from a researcher or an academic or someone claiming to be an AI expert, but they had, if you look at the LinkedIn profile, there were no signs of it for like until a year and a half ago.

8:04And I'm like, you know, you've got a bunch of kind of your risks and pocus-pocus out there. I'm like, who's talking about what it actually takes to make it work? And there wasn't. So with the encouragement of a few peers of mine and friends of mine, I was like, all right, I'm going to talk about the real, real, real, like real world, excuse me, real world deployments. and you had to have had a lot of failures to peek out a few successes, if that makes sense. It's kind of like the winery business. You've got to invest a ton of money to make a little. You've got to have gone through a lot of lessons learned, scraped knees, bruised elbows, failures, what sees assumptions that didn't hold true in order to be able to take things out of prototype and POC and into production.

8:46And so I think it's one of the reasons why most things stay in perpetual PFC purgatory. They're just not familiar and no one's ever coached them or advised them of what to watch out for because everyone's trying it for the first time. So I mean, to precisely that point, you've got, I think, something like 30 different applications in production either today or have been in production. What have you learned from that experience? Yeah, there's nearly 40 products or capabilities or applications or solutions, however you define it that I have in production to the state. Two have been retired, but I still have nearly 40.

9:23The first is 70 % of the issues have nothing to do with the tech. The tech part, figuring that out, is actually the easiest. It's dealing with the individuals, the agendas, right? There's just so much to dissect there. But the first is 70 % has nothing to do with the tech. The second is it's oftentimes, we talk about strategy and it's a magnanimous proposition on a deck. But I think what some people fail to acknowledge is your strategy, if you can't execute on your strategy, it's pure hallucination. Meaning delivery excellence and your ability to execute on it is just as important as the pretty powerful that you put together.

10:01And what people sometimes forget with AI is they may not have the fundamentals of infrastructure or data access or talent in place. So why pick a strategy and build this magnanimous profile of what you're going to do when you haven't even invested in fundamentals in the last two, three, five? eight years. And so I always say that bend it, but don't break it. And most people think this like utopic view of where they're going to go without the reality of knowing what they can actually deliver. So I would say that's the second. The third is, you know, a lot of people are going through the process and kicking the tires as well.

10:35What use case should we choose? Well, they're picking a problem to be solved by AI versus just looking at the problem and figuring out how should we solve it so sometimes it's best to choose a hammer and not a sledgehammer to solve the problem um a guy's not the answer for everything and but yet some people just want to shoot one and then i would say that's another layer and then the fourth layer that i could say is you know when it comes to picking the use case a lot of people choose business value as the primary operative to determine whether or not it should be chosen and that's not a good way and the The reason is, whether it's within a business unit or in manufacturing or in supply chain or in picture or in legal, there's a number of business functions and a number of business units.

11:24If they're plagued with a problem or if there's an issue that they want to solve, they all inherently have business value associated with them. Whether it's business cost savings, so it's going to increase our EBITDA, whether it's top-line growth, so this is going to increase our revenue, there's inherent business value in that. AI shouldn't be the tiebreaker. An individual shouldn't be the tiebreaker, understanding which use case should actually go into prototyping. And so a framework that I created at IBM after I had gone through a few deployments is based off of criticality and complexity. You have to also take a look at five factors associated with what makes the use case complicated, and then five factors associated with what makes the use case critical.

12:07Business value is one of the 10 measures. and so it's 10 % it's not 100 % and so that it goes into great detail of how to do the calculations and how to approach it in the book as well so I think like those I would say is there a portion of the 70 % there's others of course Hey everyone we're going to take a quick pause and hear a word from our partners we'll be right back Have you ever wanted to trade Bitcoin but haven't dared try? With Plus500 Futures you can trade crypto without the hassle of opening a wallet With just a few clicks, you can register and start practicing with their free and unlimited demo.

12:44See a trading opportunity? You'll be able to trade it in just two clicks. Feel ready? You can move to real money with as little as$100 once your account is approved. And the great thing is that in addition to crypto, Plus500 gives you access to a wide range of instruments. S &P 500, NASDAQ, gas, and much more. Explore equity indices, energy, metals, Forex, and beyond. With a simple and intuitive platform, you could trade anytime, anywhere. Experience the fast, accessible futures trading you've been waiting for with Plus500. With over 20 years of experience, Plus500 is your gateway to the markets.

13:17Visit us.plus500.com to learn more. Trading in futures involves the risk of loss and is not suitable for everyone. Not all applicants will qualify. Plus500, it's trading with a plus.

13:34so what is it other than adding value on the business side that goes into that calculation what else is most critical in determining whether or not an application for ai will succeed or won't your ability to actually deploy it that's a huge turning back because

13:56if you're wanting to run a marathon and you choose january 1st my goal for this year is to run a marathon you're not going to run the marathon january 2nd or february 1st right it's a muscle that needs to be developed so you're going to have a lot of individuals coming together and trying to align on this stuff and then you're going to bring in internal talent you're going to have to hire external talent most companies spend millions to be able to eat out a few hundred thousand it. That's just the reality of it because it's part of that learning curve. And so if you're going to invest that amount, you have to be very, very poignant and decisive about what it is about this use case that's worth enough, the investment, because we're going to lose more than we're going to gain the first time around.

14:42And so when you take a look at criticality, it's not just about margin savings, productivity, this is capacity, this is a top line growth. The criticality of it Is there an imminent threat in our industry? And we have to do this for relevancy and survivalship. Is the industry consolidating? Are we losing market share? Are we in dire straits that we have to do something desperate that's going to begin changing in our industry, even though technically the industry in and of itself is stable? Are there new regulations that have come through that we are going to be fined or at risk of exposure if we don't do this?

15:16Like these are critical business decisions that you have to go through. And there's a list of the five. And then business value, of course, is one of them. What are we going to get as a result of this? So those things are very critical, but there's four other considerations that we have. Business value is one of them. But the biggest is the ability to actually deploy. Is the data that you need to be able to train the model or to be able to run the solution? Is it in place? Is it of okay quality? And do you have access to it? Or do you have to go and find that one person in IT who has to grant you access to get the data and then send you a file over?

15:51Like, that just won't work. So that's like the first one. The second is the stakeholder, the business executive who's going to support this. Are they a good partner? I can't tell you how many times I fumbled because I had a really bad partner. And so I don't want to fumble again. And so I actually choose my stakeholders very carefully. Do they have a good reputation? Are they going to allot time on their calendar every two weeks to go through and understanding the progress? where we're at and the importance of being able to operationalize this use case once we're past prototyping who you actually partner with and say what are you serving is also very good for us and then there's infrastructure do you even have the basics in place and i'm not talking about like hardware per se but do you have a strong dev ops team so that when you go live with something like this and ai products are live wires they're not like software products software products are something that you essentially put into code, you push it into production, and it lives on forever.

16:51AI products don't live like that. You're only as good as your last piece of data. So it's like this live wire stream that consistently needs to flow through. Do you even have the DevOps mechanism, the CICD pipelines? And I know I'm getting all of it technically here, but it's a basic component of support and maintenance when something's in production. And if you don't have a strong process or a strong team, you won't be able to push it to production. So like there's just elements. So there's 10 key considerations. All right. So when we're talking about pushing to production, all of these underlying back end factors, infrastructure, the mechanism of deployment, ongoing support via DevOps, it certainly seems based on the news flow that we got yesterday out of Cupertino, that these folks out at Apple are going to be deploying this and deploying it at massive scale within the next 12 months or thereabout, not just on phones, but on tablets and on laptops and desktops as well.

17:45It's a tremendously complicated effort, as you point out, in terms of the actual deployment of something at scale like that. When you think about the potential applications of AI essentially on every iPhone in America at a certain point in time and indeed every iPhone in the world, What do you think that world looks like? How is it different from where we are today? What opportunities does it present? And of course, what risks a great deal of kerfuffle, as you might imagine, on Twitter? Some very critical comments from Elon Musk. I believe the spirit of which was essentially if Apple does not figure out a way to segregate data in a way that they can credibly demonstrate to him, he is not going to allow Apple devices in any of his companies.

18:25It's quite a statement. And I know people are going to say, well, that's just Elon being Elon on Twitter. but this is a fear that I think a lot of people have about privacy, about data security. So there's a lot around this. Talk a little bit about how you think about these issues as they're going to continue to move to the forefront of our consciousness. AI has been relatively siloed over the last 12 months that I've been playing with ChatGPT or the Google version. Talk a little bit about what this world is going to look like as we start to wrestle with these issues. Like with any innovation, it's a double-edged sword.

19:06on a day-to-day basis i go through a number of apps and a lot of the stuff i do even though i've automated it even though i use a lot of AI i still have to sometimes go between apps to be able to answer a question or be able to say wait let me let me check on my calendar hold on that's actually in a tab in my notes and then cross connect and then be able to respond via text the day i'm available for a talk or the day of it, I'm available for book signing. Right. So I'm excited by the fact that it's going to increase my professional productivity based on what they claim is coming down the line.

19:37So just to be clear for folks who are following along, essentially what you're saying is we do live still in this very siloed world. AI is a thing out there that you use, but it's not directly interacting with our lives on a day-to-day basis. It's not. And so right now it's still, it's like spot treating a shirt when it just needs to go the wash. You know, at some point in time, it's a futile game, right? You're just going to continuously spot treat. By the way, this is a great metaphor, and I love it because it shows you both the opportunities and the risk, right? When you spot treat a shirt, if you get a little coffee on your shirt and you try to get it out, you know, essentially that you're working on an area that's the size of a quarter.

20:16But when you throw in the wash, you know, you can shrink it, you can do whatever. I mean, this is really the sort of this globalized AI risk where you have the interconnectivity of all the data, all the applications, tremendous opportunity, but also I think as Elon Musk has correctly pointed out, tremendous risk. Agreed. And I know he's a character that people love to love and love to hate. He's not wrong in this one because if I spill my cup of coffee or if I put it on the counter typically, I get seven splashes. I have to now clean all seven. It's laborious. It's like time. It just takes time, time that I don't have.

20:52I get frustrated. But each of those seven is contained. And so if I choose not to clean three of the seven, it's a choice I have. So you have a choice, you have a ton, but you know, on the flip side, it could be more time consuming. It's a bit more laborious in nature when you choose to clean all seven, because it's not integrated, it's not throwing it into the wash. Whereas if you throw it into wash, you're going to clean it all at once, whether you want it to or not, that choice has been removed. right so that's the analogy that i use and so i would love for my apps to be integrated that data to be integrated it will save me a lot of time the flip side of that is and this is why the double-edged sword is it will provide me conveniences but at a cost to my privacy and so everyone has a different spectrum on convenience to invasion of privacy of where if they lie, which is why it's not a clean answer.

21:46It's not black or white. It's shades of gray to be funny. Um, because everyone's on a different spectrum now, specifically about Apple. I think for those of us who were in the industry, at first we laughed when they called AI, Apple intelligence. Now, some of us are kind of pissed off. We're like, that's just really marketing. Like that's what you wanted to do. That was too much in your face. I think the second aspect of it, the company that holds the most data on any individual is Apple. And years since they started, they have done nothing but institute philosophy of our consumers come first.

22:30We respect their pussy. We respect our data. Nothing will be shared and nothing will be cross-referenced. Hey, everyone. We're going to take another quick break and hear a word from our partners, and then we'll be right back.

22:46Right. And by the way, in fairness, it's been relatively easy for Apple to be the good guys in this equation. Why? Because, you know, you spend a ton of money buying their devices. Lots of the services that you use online and SaaS applications, you do stuff in a browser, or you do stuff on your phone, the way that that gets monetized is via your data. Apple has had the luxury for decades now of having these incredibly fat margins on hardware and increasingly on data services that people were willing to pay for. So now you have this point in time where you get precisely to the challenges that you're talking about, where you have data privacy issues and you have this incredibly powerful technology, this very rich set of data that people have been very comfortable giving Apple access to over the years.

23:28And now, I mean, it is just a huge open question. It's not that people feel comfortable. They have no choice. If you want a phone that's as powerful or as beautiful or as elegant or as convenient as an iPhone, you have no choice. What's the alternative? And so by default, you also felt more comfortable because part of their institution principles and values were, we preserve and respect your privacy and we will not show and share your information for any purposes. until we want to create AI products on our own. And now we have to access this stuff. But the goal is they always had it. They never accessed it for gain.

24:07This is the implication where they're showing that they are going to access it for gain. Now, I've got to see some of these products because there is also a lot of hype. They've been talking about Siri for how long and how many times have you said, hey, what's the direction to the nearest post office? And it shows me the direction to Costco. Like completely unrelated. C3 is not the best, right? Their natural language processing technology, the ability to describe speech into what it actually heard. And then, you know, set that in motion with directions. We all know we've gotten frustrated with C3s.

24:37And there's a ton of jokes, especially on Reddit and YouTube about arguments between the individual and Siri. So I would have said, Apple, just don't prove Siri alone. And I think people would have been happy. But they took it like a hundred steps further. And so to do the things that they're saying, I don't know how you can do it. And again, I don't work on this, right? I'm not in the Apple safeguards, but either it's amazing marketing and hype and they're really not integrating things, or if they're going to deliver on what they said, it is a violation of the institutional values and principles.

25:12Historically, they've said, we respect your data. We're not going to use your data for our dating whatsoever. And that's the crux of this right now. So it's a watch what we can see. but a lot of us are pretty busy. We're on edge a little bit because Apple was supposed to be at that one company. Yeah, it's going to be really interesting to see how they navigate these waters with all of the opportunities and all of the threats. Talking of threats, another question that Elon Musk raised last night or a series of questions is about the question of essentially who's going to train the data set and what values is it going to reflect?

25:50Elon Musk retweeting a couple of tweets that I'll let folks go and read on their own. But they're these very provocative questions asking AI, would you prevent a global thermonuclear war if you had to do X, Y, and Z? And there are groups of things that are relatively, I should say, very, I guess, taboo. And the answer is, well, I'll let you evaluate them for yourself. But the question that he's asking fundamentally, it seems to me at least, is this question of what values are going to be reflected in AI when they begin to play an incredibly important role in the way that the day-to-day mechanics of society unfold.