Navigating the AI Era: Understanding, Adapting, Thriving. David Shapiro's Insights

Hello, my friends, and welcome to another exciting episode of Negotiate Anything. This show is produced by the American Negotiation Institute, and with over 5 million downloads and listeners in over 180 countries, listeners just like you have made this the number one negotiation podcast in the world. Hi, my name is Kwame Christian, and I am the founder and CEO of the American Negotiation Institute. Here at AI, we believe that the best things in life are on the other side of difficult conversations. And we are passionate about providing you with the best content that will help you to make your difficult conversations easier, both at work and at home. Lastly, I want to remind you that we offer consulting and conduct trainings both virtually and in person all around the world. Our focus is in three main areas. First, negotiation and conflict resolution. Second, leadership. And lastly, diversity, equity, and inclusion. Check the link in the description below to learn more about how we could work with you and your team. And now, without further ado, let's jump into the interview.

Dave, thanks for joining us today.

Thanks for having me.

Hey, man, it is my pleasure. So how would you get us started? By telling us a little bit about yourself and what you do.

Yeah. So again, thanks for having me. My primary shtick right now is I'm an AI YouTuber. I've been researching AI off and on for about the last 15, 14 to 15 years. Started doing it full time earlier this year. And I focus on everything from fine tuning to AI alignment, everything in between. I also do some consulting. I have about 600 supporters on Patreon, a few dozen of which I talk to on a regular basis to help coach them on deploying AI in their businesses.

This is great. And listeners, I always say it. My friends who come on the podcast, they are very humble. Let me pump you up a little bit, Dave. So Dave's YouTube channel is the best channel out there for artificial intelligence. It's exceptional. And so if you want to learn more about AI, this is the guy to follow, because you've been growing at an astronomical pace over the past three to six months. And it's just great to see you continue to shine because your work is exceptional.

Well, thank you.

Yeah, yeah, my pleasure. And listen, so for this episode, we're going to keep this simple at the beginning, and then we're just going to. We're going to build. And so this is going to start off almost like there's a primer getting everybody started understanding what AI is and then building up to the implications for society in the future. And there's been a lot of extremism on all sides when it comes to this. And it's hard to tell the, you know, separate the wheat from the chaff here. So that's why you have Dave. And, you know, let's, let's go back to the beginning because you said you've been working on this for the last 14 and 14 to 15 years. And I think for a lot of people that would be surprising because they might say, didn't AI come about last year? So let's just start off with the simple question of what is AI? And tell us a little bit of the history about it.

Yeah, so from the biggest perspective, AI is, you might also have heard machine learning. So the idea is that you use computers to learn from data. That's like the simplest, highest level version. But what that learning looks like has changed over time. Historically, most machine learning was about mathematical optimization, linear regression, that sort of stuff, a lot of statistics. With the last decade or so, as computers became a lot more powerful, basically we could throw a lot more math at it. And when you do that, you can have larger mathematical models which we now call neural networks or deep neural networks. And so all the rage right now. Probably most of your listeners have heard about ChatGPT. ChatGPT, at its heart, is just a large language model which is a artificial neural network. And so an artificial neural network, that's a really fancy term for saying that it does a lot of linear algebra really, really fast, and I mean, like billions and billions of calculations every, every conversation that you have with it. So it's at its heart, it's a lot of math and it's predicting your next character.

This is great. First of all, impressive synopsis for something so complex. Now when we think about artificial intelligence, we can say, okay, Dave, I'm following. So what's the big deal? This is taking over the news cycle. What is the big deal? How would you conceptualize that?

Yeah, at its heart, artificial intelligence is a form of automation. And automation itself is not new. We've had electromechanical automation since the beginning of the industrial revolution, 150 years ago. It's also a force amplifier. So a way to think of force amplifiers or multipliers is like the steam engine, right? Steam engines put out a lot more work than humans, a lot more work than oxes and cows, Internal combustion engines. Ditto for all that. So you can think of it as kind of two parts, right? It's that automation engine which allows you to do more cognitive work, faster, cheaper, better. It also is a force amplifier, and that allows one person to do More at one time or do more faster. So from a. From a brass tack standpoint, that's kind of how to look at AI. It's just the next evolution of automation and industrialization.

Okay, this is great. So as a force amplifier, and it's a form of automation, this is great. So we can understand, essentially, we can do the things that we want to do using artificial intelligence with much more efficiency and precision because of these large language models, Right?

Yep. That's the heart of it.

Yeah. And now let's try to paint a vision of the future, because people are starting to say, okay, great, maybe that sounds to them as smarter computers. Computers have been getting smarter and faster and smaller every year. So what's different about this?

So this is what in the industry is called a step change. So in many cases, you have a smooth trend line, Right. Like, cars get X amount more efficient every year. We deploy X percentage more solar every year. So that's a nice smooth growth curve. But what artificial neural networks, especially large language models like ChatGPT, do, that is what. That's a step change, which is there's suddenly a whole bunch of new abilities that were not there before. And so that's like going from an electric typewriter to Microsoft Word. Right. It still fundamentally does the same thing, but it comes with a thousand features that typewriters just are not capable of. And so this is taking that to the next level, at least in terms of language generation and manipulating langu language.

Absolutely. And I think that's one of the things that's so difficult about explaining this technology, because when you think about our human cognitive capacity, we're limited. So, for instance, when you think about numbers, we're limited in our ability to appreciate large numbers. So it goes from 1 to 10 to 100 to 1,000, to a lot. Okay. And that we can't really fathom much beyond a lot. And I think. I think it might have been on your channel or some other channel. They were talking about kind of giving an idea of how rapidly this was advancing. And when you think about the iPhone, it was advancing, what, every year or every few years we're waiting for the new iPhone to come out. It was a big deal. But here with AI, it's like these advancements are happening in a matter of days, not years. Can you kind of give a conceptualize just how rapid this advancement is?

Yeah. So having been keeping my finger on the pulse of this particular technology thread, which started in about 2015, and of course that back then it was a new version, a new model would come out every one or two years. And then a year or two ago, it was about every six to nine months. And now, of course, as you said, it's weekly, where we get new advancements. And not only is the individual companies are churning out results faster, there are now more companies participating. So we're getting this broadening of the field as well as this acceleration. And so we get these compounding returns as results are shared. And the economic incentives are there as well, because we've seen for the first time, companies have gone under or had their market undercut. Biggest one being Google with ChatGPT came out, Google lost. It was. What was it like $100 billion of valuation in a couple months is incredible. So a lot of people see the writing on the wall. And so what we're entering is what's called a race condition, where one company has to race another company in order to just keep up. Right. Because it's keep up or die at this point, at least for some companies. Not all companies are in the same, but that is why we are. That's one of the driving forces of seeing this acceleration in AI, and that's why ChatGPT, they come out with a new ver about every other week now. And then, of course, there's competitors trying to break into the market, but they're way out ahead still.

Absolutely. And I think one of the concepts that can help people to get a better understanding of how significant this technology could be just in the context of the entirety of our species is the concept of the singularity. Can you tell them about that, too?

The singularity was a hypothetical eventuality, end state that was proposed by Ray Kurzweil, a futurist and researcher, where basically these compounding returns that we're seeing, that we're starting to see today, if that trend continues, that acceleration continues, it's like you drop a penny off the Empire State Building. The myth is you drop it, and by the time it gets to the bottom, it's moving so fast that it could shatter a car windshield or kill someone. And so gravity is probably the most familiar exponential acceleration that anyone's familiar with. Right. You drop something from space, it burns up on the way down because of how fast it's going. That is kind of the data equivalent of the singularity, where the idea, and there's lots of different ways to explain it, but the idea is that the compounding returns of more data, more AI, better computers, you get this reinforcement, this virtuous cycle where it keeps going faster and faster and faster. Until it reaches basically infinite growth rate, which obviously there's still going to be constraints of physics. We only have so much electricity, there's only so much water to cool the data centers. There's always going to be some constraints. But the hypothetical limit of computation is much higher than we've even come close to. So the high end is still way out there. But if this rate of acceleration continues, we might get close to those thermodynamic limits sooner rather than later. That's the singularity. And the idea is that once you basically solve all of science, all of math, all of chemistry, we don't even know what's possible. And so the idea of the singularity is that there's the event horizon, which you can't see beyond the event horizon of a singularity. That's the point of the singularity is, okay, something's going to happen. And the estimates vary from 2045 to 2035 or even, I think some people have revised it down to 2028. So five years at the soonest. We're going to get to this point of compounding returns and compounding acceleration that we don't really know how life is going to be like beyond that.

Does your company invest in professional development training? If you believe that your team would benefit from a negotiation workshop, all you need to do is go to our website, fill out the workshop request form, and then we'll set up a time to chat. These workshops are completely customizable, and we've done them all around the country. Negotiation and conflict resolution skills are beneficial across all professions, but they're especially useful in procurement, purchasing, sales, sourcing, and contract management. Our calendar is filling up quickly, and we even have some workshops scheduled for next year. If you think you might want one. I'd suggest reaching out soon so you don't miss out. Check out the link in the description.

To learn more and we will be.

Right back after this.

Yes, and as you say, we don't know what life is going to be like. I am going to ask you to try to explain what life is going to be like, because I think people are understanding. They're saying, okay, this is a big deal. Solving science is a very big statement because we think about just the technological advances from 1900 to 2000, it's astronomical. It's a completely different world. And now to even posit that we're at a point where we could create a technology to essentially end all technologies like this is the platform that just spurs humanity to a place that we can't even Conceptualize. Can you give us an idea of what the potential outcomes could be?

Yeah. So I guess on a sliding scale, on the low end, due to some of those constraints of physics, right. The planet is only so big, there's only so much energy that we can use. On the low end, you're looking at what I call kind of a post labor economy, where because these machines are getting smarter and more efficient, like think about how powerful ChatGPT is. $20 a month. And it, I mean, I don't know about you, but it accelerated everything that I do on YouTube, anything that I want to do. Like I was working on a project with people and I wrote a 32,000 word book in three days with the help of ChatGPT, right? So I'm going 100 times faster than I used to before, right? So on the low end, everyone goes 100 or 1,000 times faster at whatever it is that they're trying to do, at least for intellectual labor. Now, that alone would create new economic paradigms that we're not quite ready for. Because the thing is, humans are very expensive to employ. You need health insurance, you need Benefits, you need HR. But if you can replace 100 people with a ChatGPT subscription, and that's today, that's version one. What is version two, three, four of these things with all their integrations, how many human labors are they going to be able to replace? And then the big question is that economic boost, is that going to translate to other new jobs? It's a big open question and there's a lot of debate about that. So at the very least, we're looking at a major shift in the way that we think about economics. If you look at 200 years ago, 95% of people were farmers. And that's just because of how inefficient it was. Farming was not particularly efficient because we didn't have steam engines and internal combustion engines. You had to do most of it by hand or with the aid of large animals. The industrial revolution gave us that new force multiplier, steam engines, internal combustion engines. This is the next labor force multiplier. So what if 95% of jobs today, you take that same paradigm, 95% of jobs today go away because you don't need it. So what's left? And it's pretty scary to think like, okay, well, only 5% of jobs will remain, and historically speaking, those jobs have always been replaced by new jobs. We don't know if that's going to happen this time. And there's good reason Arguments on both sides. It has always happened up to this point. We'll discover new demands for human labor. It might not happen though, because like I said earlier, humans are expensive. So that's like the low end. We're looking at that level of economic disruption even before the singularity, right? That's coming in the next five to 10 years. On the high end, you have everything possible from people are working on brain computer interfaces. So I don't personally believe this is possible, but some people think like, oh, well, you get those brain computer interfaces sophisticated enough, you could upload your brain to live in a computer for the next, you know, 4 million years or whatever. So. And that sounds like pure science fiction. You know, there's all kinds of other possibilities. You know, you solve faster than light travel, high energy physics, and then we start colonizing the entire galaxy after the singularity. So the truth is probably somewhere between those two extremes of economic disruption and you know, we invent Star wars technology.

Incredible. Great synopsis, because you're absolutely right, because I agree with you. I think the conservative estimation is the fact that we get to that post labor, society, the leisure class, as you describe, and then it creates this really challenging, almost philosophical conundrum where we are really, we almost have no choice but to explore what the meaning of life is because we have nothing else to do, right? And so this disrupts the whole fabric of society. Notice I say disrupts it doesn't mean good or bad. It will significantly change it. And so in order to be able to be ready for these outcomes, we have to be able to have conversations about this. And I think one of the most important conversations is about AI regulation. And I think this is notable because again, this listeners, let me say something. If I ever say anything smart about AI, just know that I'm saying what Dave has already said to me months in the past on YouTube. Okay, don't give me any credit here, but you and your team brought up a really great point where you said this is the only time in history in the United States where corporate interests have come to the government and said, what we've created is so powerful, you need to regulate us. Let's work on how to do that. So let's talk about why meaningful regulation is so important, and then later we can talk about what that actually looks like.

Yeah, absolutely. So earlier I mentioned that race condition, right? So Google is terrified, Microsoft is terrified, a whole bunch of other startups and other tech companies, they're all terrified. They have a lot of inertia, right? They have A lot of market capital that is going to be difficult to dislodge. That being said, we've seen it before. Tech companies can collapse relatively quickly. So that race condition means that they're all incentivized right now to go as fast as possible because it's adapt or die, right? And so the only thing that can slow that down, that race condition down, is a systemic change that affects all of them equally, and that is regulation. So the best model that we have for effective regulation is GDPR in Europe. And so GDPR in Europe, one of the. It's basically data privacy, right? That's a very broad data privacy. And many, many companies took it very, very seriously, going so far as GDPR tells you about things like where data is allowed to be housed, who is allowed to have access to it. And so like if your user, if you're a German citizen, for instance, your data, you have a right one to have that data deleted, but you also have a right for it to stay within German jurisdiction. And that policy was so powerful that people abide by it. And so that's the kind of systemic change that changes the rules of the game, right? And so that is why people like Sam Altman from OpenAI and apologize, Christina Montgomery from IBM, why they're saying, hey, we are advocating for this national blanket regulation policy because that will force everyone to slow down and they'll all stay. They're still incentivized to compete, right? Capitalism isn't going anywhere anytime soon. So that competition, that race, will still be there, but it'll be constrained. It's kind of like, you know, stock car racing, right? You're not allowed to put nitro in your stock car, right? Everyone has the same basic parts in their car, so they're limited in speed. And so that keeps the rules fair, but also safer for everyone. Actually, like that analogy.

I like that too. That was really good. And now again, that makes sense. Why is that so hard.

Hypothetically? It's not hard, right? Because we've done it before, right? We've got the fda, we've got plenty of regulatory bodies all over the world. We've even got international regulatory bodies, right? There's nuclear inspectors that will inspect all the enrichment reactors all over the planet. So we have a model for it. We have succeeded at this problem because the only other time that we've ever created something, there's been two other times that we've created things this dangerous. One, bioweapons and two, nuclear weapons, right? So we have a model for regulating these existential technologies. Now, that being said, one of the hardest problems is actually what's called the Overton Window. That is the range of conversation that you can talk about without getting laughed off the stage, right? And so because of the Senate hearing that we're alluding to, because of ChatGPT, because of Google and Microsoft, that Overton window is shifting. So now people are saying, oh, okay, we're allowed to talk about these existential risks. On the one hand, it's maybe not a full existential risk, but we are at risk for that economic disruption. And on the more extreme end is that existential risk. But now we're allowed to talk about it. Now it's time to evaluate, okay, what are the lessons from history? One of the darkest chapters of human history was a policy called mutually assured destruction. For many people who were alive back then, or alive as I was at the tail end of the Soviet Union and the Cold War with the United States, the policy of mutually assured destruction was about the buildup of nuclear weapons. So the buildup of nuclear weapons was about the idea that no nation had what's called first strike capability. So first strike capability was the ability to completely cripple the other, your enemy, so that they couldn't retaliate. So in this condition, in this. In this race condition, where you are incentivized to have as many nuclear weapons as you could all over the planet pointed at each other so that nobody had first strike capability, then nobody would shoot first because it was guaranteed that everyone would die, right? And so that's like the worst kind of game to be playing. We don't want to be playing that game again with AI, we want to have what's called a win win situation, where it's like, instead of getting everything locked up in this mutually assured destruction, where companies are going as fast as they can because nations are also incentivized to go as fast as they can, there's been political leaders all over the world saying that the nation that wins at AI wins the geopolitical stage for the next century. And that sounds very similar to mutually assured destruction. So we want to avoid that. So it's not that it's. It's complex, but there's a lot of moving parts. There's a lot of incentives, but in principle, one, it's relatively straightforward because the incentives are there. It's a military and economic incentive, but it's complex because you've got a lot of people with antagonistic incentives. So that's where the complexity comes in, because we all want it Right. We want the golden carrot of AI giving us that post singularity lifestyle without the mutually assured destruction component that's threading that needle is the hard part.

That makes a lot of sense. And one of the things that you've talked about is the strategy of national versus international types of regulations. And so there are layers of complexity here because it's going to be tougher to get it regulated and come to some alignment here internally within the United States. But then we. And the only way to really make this work in a meaningful way is to get as many parties as possible, hopefully the entire world on board. And so when you think about the strategy to actually pull that off, what does that look like to you?

Yeah, so the strategy that I advocate is one that is grassroots, decentralized, global. Because one of the chief problems. Let me frame it this way. One of the chief problems of this is what's called a coordination failure. And a coordination failure is we could hypothetically find an optimal solution, but only if everyone works together. But because there's a lack of trust either within nations or between nations, because of that lack of trust, we tend to throw up our walls and not cooperate. But fortunately, we, you and me, everyday people, have a really powerful tool for collaboration. We have the Internet. And so by using this global communication tool, we can actually solve the coordination problem. But then the question is, okay, who does what and why and what's the strategy? Right? And if you distribute these ideas, these plans, and you build consensus on Twitter, on Reddit, on email, on YouTube, we've got these platforms that have global reach. You know, on my YouTube channel, only 6% of my viewers are in America. Right. So everyone else is global. Yeah. Right. So by using the, by using those exponential communication tools that we already have access to, having these conversations, talking with other influencers, talking with researchers, we can build consensus to say, yes, we need an international regulatory body that is modeled on, you know, the nuclear regulators. Yes, we need it. We need a national regulator that's modeled on the FDA or the FCC or whatever. You know, we need a national policy that's modeled on the GDPR but updated for AI. Right? So by building that consensus, like that's, that's how democracy is supposed to work, right? Is once there's a strong enough consensus, the politicians that support it get voted in. The politicians know that the, you know, they, because the politicians, they all look at polls, right? What's most popular, right? What's. What policy is going to get them reelected. It's all about incentives. And so by building Consensus around these things. We can, we can shift the rules of the game game so that those incentives line up and actually incentivize solving the control problem, solving AI and creating breaks for that race condition. But it does require broad enough consensus for the mechanisms of democracy to work like that.

This is fascinating because it makes sense. It makes sense theoretically because if we get enough people saying the same thing, then people would have to respond. In your opinion, is that more of a challenge of essentially winning the messaging battle and just getting more people on board or is there some kind of mechanism that you think would accelerate the process?

Messaging is absolutely critical. First step is getting the message out, having the conversations, having the debates. But messaging alone is not, it's necessary but not sufficient. So that's an engineering term, necessary but not sufficient. If you don't have a foundation, you got nothing. You build a house of cards. So what comes after that is action. And I don't say there are going to be, there are likely to be conflicts of interest, right? There's always going to be some adversary out there that says, actually I don't want to play nice, I'm going to cheat. I'm going to try and stack the deck in my favor because I want the strongest military, so I'm going to cheat. Or I want to have the first quadrillion dollar company, so I'm going to cheat. There's all kinds of, in economics, what's called a perverse incentive. If you can cheat and get away with it, there are rewards for that. So what we have to do is create a system that is so robust, so comprehensive that cheating that breaking the rules is worse than actually following the rules. And so that's what, in game theory that's called a Nash equilibrium. So I don't know if you've ever played the game Monopoly, right? You want to wheel and deal, you get all the properties, you build your hotels. That's your strategy, right? That's everyone's strategy. But the difference is that in a competitive game like that, it's designed for there only to be one winner, right? It's a winner take all strategy. So that is still true globally, Right. Apple dominates the smartphone area. They're neck and neck with Android. Right. And then on the military front, the United States and the European Union, like the Western allies, dominate the in terms of military power. So globally we do have this condition where it's still winner takes all. But how do we create a situation that brings everyone else to the table so that everyone is able to win? Right? Not just People that are ideologically aligned to, you know, using Apple over Android or ideologically aligned to the United States and Western allies and liberalized democracies. So there's a lot of work to do, even above and beyond just getting the message out. But getting the message out is certainly the first step.

Yeah, I love this. It's really helpful to have a clearer roadmap because essentially it's amplifying voices like yours, helping people to see what options are out there and spreading the message. And there's no way to know this for sure, but I'm just interested in your perspective. When you think about the risk of challenges and problems with AI versus the rate at which the United States is regulating, do you think we'd be able to get meaningful regulation in place before something bad happens? Or does something bad happen, need to happen in order for us to make that move?

There have been a few times in history where the perceived danger was there. There's a movie that's about to come out called Oppenheimer right before. And so this. This movie, it stars Cillian Murphy as Oppenheimer. And so the backstory for anyone who doesn't know this is the story of the development of the nuclear bomb. So one of the events that precipitated the Manhattan Project was Albert Einstein wrote a letter to Congress saying. Because basically what he did was he ran the calculation. He said, okay, let me look at a pound of plutonium, look at how much energy it radiates per day, per hour, per year, and calculate out how much energy is total in a pound of uranium and a pound of plutonium. And he realized there is more energy in a pound of this stuff than in millions of barrels of oil. And so he wrote a letter to Congress saying, you need to pay attention to this because there is so much energy locked into this. If you can unleash that energy, it can change the world. And so then the Manhattan Project came in and the rest is history. Right. It was all very, very secret. It's all going to be dramatized in the movie Oppenheimer, which I will be there day one, because it looks real good. And I don't know if they'll have the Albert Einstein aspect in that, but you can actually find there's a copy of the letter online, so we probably are. Hopefully what happened was that Senate hearing with Sam Altman and IBM. Hopefully that was that letter to Congress saying, there is something here, you need to pay attention to it. And so there's been calls online for, like, we need the equivalent of The Manhattan Project, but for AI. So maybe that call has happened, but maybe not. So hopefully we'll see.

This is great, Dave. I could talk to you all day about this. I know this because I've listened to your videos all day. And before you go, just remind the listeners about your YouTube channel, the Patreon, and how they can get in touch with you.

Yeah, yeah, thanks. So, real quick, obviously I've got a YouTube channel. It's growing. Easy enough to find. David Shapiro, Automator. I've been an automation engineer for many years, and AI is just the next evolution of that. I've also got a Patreon, so patreon.com Dave Shapp, which has a private Discord server. So if you want to chat with me, just jump on there. I have limited time, but every now and then a slot will open up if someone wants to have a consultation. I'm really good at figuring, figuring out how to solve business problems with AI. Yeah. And I'm also available on LinkedIn and Twitter. I'm pretty picky about LinkedIn just because sometimes people, they want to sell me something. So I want to make sure that no one's trying to sell me anything before I accept their connection on LinkedIn. But other than that, yeah, I'm happy to connect with people on pretty much any of those platforms. I should be pretty easy to find.

Love it, Dave. Thank you again. This was really, really enlightening.

Yeah, thank you. Thank you so much. It was good talk.

Congratulations. You've just joined an elite club by listening to a full episode. You're now officially on the Negotiate Anything team. So welcome aboard. What most team members do is they subscribe to the podcast because that allows them to automatically get the latest episodes of the show. The best things in life lie on the other side of difficult conversations. Keep learning, keep practicing, and keep getting better. Your relationships will improve, your career will soar, and you'll have the confidence you need to get the most out of these crucial conversations. Again, thank you for joining the team. We're excited to have you, and I.

Will see you in the next episode. I'll catch you later.