The AI-Powered Decade Ahead — Mike Walsh — Futurist Keynote Speaker

[00:00:00] The challenge I want you to consider is this: [00:00:08] Do we lack ambition? [00:00:11] Are we so focused on consumer applications [00:00:15] that we miss the opportunity to really seize artificial intelligence, [00:00:21] algorithms, and automation, and use it to transform our world? [00:00:27] You see, I've got some bad news for you. [00:00:32] ChatGPT is not AI, nor is Claude, nor is Alexa or Siri. [00:00:40] Yes, they're a kind of AI, [00:00:42] but they're a very small part of what's really possible. [00:00:47] You see, while there are large language models, [00:00:50] very soon there will also be large behavioral models, [00:00:54] large action models, and eventually large world models. [00:00:59] Because the only kinds of models that will change the world [00:01:04] are models that understand the world. [00:01:07] And to get to that, we not only need data and computational power, [00:01:13] we need people like you who actually understand the physical elements [00:01:19] that create the world we live in. [00:01:22] So if you actually see it like that, [00:01:25] you start to realize that we're not asking big enough questions. [00:01:29] I mean, forget about building a slightly smarter chatbot. [00:01:32] What would it take to actually create a world-scale, full-stack, AI-powered, autonomous civilization? [00:01:40] Sounds like science fiction. [00:01:42] But you know, there are places in the world that are a little bit closer to that [00:01:45] than you might think. [00:01:46] I mean, recently I was in South Korea. [00:01:49] They've built a fully automated port now in Busan, [00:01:52] where you've actually literally got robots unloading these ships, [00:01:56] even before they come into port. [00:01:58] Robots are everywhere, in fact. [00:02:00] They're in restaurants, serving you food. [00:02:02] They're actually behind the scenes, in kitchens, cutting up steaks, [00:02:07] or in distribution centers, unloading boxes. [00:02:11] South Korea is the most automated country on the planet. [00:02:14] In fact, per capita, there are more robots there than anywhere else, [00:02:19] even in China or the United States. [00:02:21] One out of ten of all workers today in South Korea are already robots. [00:02:28] Here's the weird story. [00:02:29] No one's complaining about them taking their jobs. [00:02:32] It's got the lowest fertility rate in the world. [00:02:35] The only thing people are arguing about is Chinese robots [00:02:39] taking South Korean robots' jobs. [00:02:42] Welcome to the future. [00:02:44] Why is so much money now going into humanoid robotics? [00:02:48] It's for a very simple reason. [00:02:51] We're running out of data. [00:02:53] Most of the data that these large language models are trained on [00:02:56] comes from things like Reddit or Twitter or romance novels. [00:03:00] But once you've trained these models on all the text that's available for free, [00:03:04] you start to need things that exist in the real world. [00:03:08] To be impactful, AI is going to have to evolve from these language models [00:03:13] to models that understand the world around us. [00:03:16] And this is really the next big shift. [00:03:19] So if you start to now think about this kind of timeframe of the next decade, [00:03:24] if you really want to build a civilization that's powered by AI, [00:03:28] you start to realize there are three very big implications, [00:03:33] three very big problems that I think engineers in particular are well equipped to solve. [00:03:39] And there are problems around energy, work and intelligence. [00:03:44] So let's think about energy. [00:03:48] I think we are vastly underestimating how much energy we need [00:03:51] to power the next generation of applications and infrastructure and industry [00:03:57] that we're going to need to create the future. [00:03:59] I mean, do you realize that already today for every five queries you type into something like ChatGPT, [00:04:06] it takes one bottle of water to cool a data center. [00:04:10] Now, what do you do with this information? [00:04:12] I mean, do you kind of cut back and say, [00:04:15] well, I'm only going to use three queries today. [00:04:17] That'll be half a bottle of water. [00:04:19] Of course not. That's ridiculous. [00:04:21] But it's not as ridiculous as people saying we should stop flying to cut back on carbon. [00:04:26] We should start riding more bicycles. [00:04:28] We should go back to a medieval lifestyle. [00:04:31] Absolutely not. [00:04:33] I think we actually need to change the way we think about energy altogether. [00:04:37] How do we have a more abundant view about energy? [00:04:41] We're going to need lots of power. [00:04:43] There's a reason why the big tech companies today [00:04:45] are actually doing something that was previously thought unthinkable, [00:04:48] which is that they're now considering nuclear reactors. [00:04:51] Last year, Amazon bought a nuclear reactor from Talon. [00:04:54] They're co-locating a data center next to it. [00:04:57] Google announced they're now investing into modular small-scale reactors. [00:05:02] Microsoft is bringing back Three Mile Island. [00:05:05] I mean, ironically, Microsoft was formed just a few years [00:05:09] before the notorious accident that shut that plant down to start with. [00:05:15] The reason why, weirdly in the future, environmentalists will probably even see [00:05:20] that nuclear energy is green energy. [00:05:23] Because there's just nothing else that's going to be able to get us to the future [00:05:27] without new sources of power. [00:05:30] There was a famous Russian scientist who talked about a civilizational scale. [00:05:36] And what he said was that eventually we will classify alien civilizations, [00:05:41] whether or not they've managed to capture all of the energy available to them. [00:05:45] A Type 1 civilization would capture all the energy inside their own planet. [00:05:49] Now, arguably, we're only about 0.6. [00:05:52] A Type 2 civilization would capture all the energy of their own star, [00:05:56] maybe by putting a giant kind of Dyson sphere around it. [00:06:00] But a Type 3 civilization would harness all the energy in the entire solar system, [00:06:05] including even black holes. [00:06:06] Basically, if we're going to level up our civilization, [00:06:10] we need to exchange our energy scarcity mindset to one of radical abundance. [00:06:16] The second big problem, though, is that of work. [00:06:20] What does it really take to transform our productivity? [00:06:25] What kinds of organizations are we going to see in the future? [00:06:28] I believe that we are on the brink of a fifth industrial revolution. [00:06:34] I know what you're thinking. [00:06:36] How do we get to that? [00:06:37] I mean, when did industrial revolutions become like Fast and the Furious movies, right? [00:06:40] I mean, we're up to 10 of those. [00:06:42] How do we get to five of these revolutions? [00:06:44] Well, I mean, stay with me. [00:06:46] We've had five, if you think about it. [00:06:48] We had steam, then electricity, computation in the 1950s. [00:06:52] Arguably, you could roll the internet, mobile internet, internet things [00:06:56] into basically a whole bunch of stuff that happened in the early 2000s. [00:07:00] But around 2020, the massive investments during COVID [00:07:04] that went into digital transformation, [00:07:06] and of course, the recent AI advances, [00:07:09] massive data sets, more compute, [00:07:11] have set the stage for a fifth industrial revolution. [00:07:15] One that, like electricity, is going to transform industry as we know it. [00:07:20] But to understand why this is important, [00:07:22] you have to look at what happened with electricity. [00:07:25] I mean, electricity was arguably invented in 1831, [00:07:28] when Michael Faraday introduced the electric dynamo. [00:07:31] But it took some time before anyone figured out [00:07:35] what that really meant to change the way we work. [00:07:39] In fact, arguably about 82 years before Henry Ford came along and said, [00:07:44] "Hey, so this electricity idea, I know this is wild, [00:07:47] but what if we don't just power factories differently? [00:07:51] What if we design them differently too?" [00:07:54] In the early 1900s, they would go to a big factory, [00:07:57] they would take the steam engine out, [00:08:00] they'd put an electric engine in, [00:08:02] but they'd leave the line shaft or the heavy industrial machinery [00:08:05] that had been around there since the early industrial revolution. [00:08:09] It was only when Henry Ford realized that with power, [00:08:12] we could now have more distributed, decentralized, agile buildings. [00:08:17] You could have different people doing different kinds of jobs. [00:08:20] That's exactly the kind of thinking we need today. [00:08:23] What is the new ways that we can work? [00:08:26] What are the new ways we can make decisions? [00:08:29] What are the new ways we can think about things like quality [00:08:31] when we can now powered by data and AI? [00:08:35] And only by thinking like that, [00:08:37] do we start to realize that there are very different types of companies [00:08:40] that are going to emerge. [00:08:41] You look at someone like Elon Musk, [00:08:43] people think he's insane, right? [00:08:45] I mean, who else is trying to run six companies simultaneously? [00:08:49] He might be insane. [00:08:52] But he does know something that many people have forgotten, [00:08:55] which is people from the East look at this and they go, [00:08:58] that's not weird. [00:08:59] We've seen this before. [00:09:01] It's a kairatsu or a chai bowl. [00:09:04] If you're really going to solve complex engineering, [00:09:07] infrastructure, industrial scale problems, [00:09:11] you can't do it with a consumer application company [00:09:14] that runs from ad dollars. [00:09:17] You need deep vertical integration, [00:09:20] deep relationships with government, [00:09:23] long time scales when thinking about capital expenditure. [00:09:27] You need to think differently. [00:09:29] And, you know, it kind of is something we have to realize [00:09:34] that the world's biggest seller of electric vehicles isn't Tesla. [00:09:38] It's BYD from China. [00:09:40] Where did BYD start? [00:09:42] They made batteries. [00:09:44] Because it turns out that batteries are the most complex [00:09:47] and important part of the entire EV system. [00:09:53] And one of the things that's going to change as we go forward, [00:09:56] when we think about globalization, [00:09:57] this new phase of globalization isn't just about smart supply chains [00:10:02] that use data to be more efficient. [00:10:05] It's about geopolitics. [00:10:07] And that's where things get difficult. [00:10:09] You know, many of you would know this. [00:10:12] This is the SR-71, [00:10:13] one of the most sophisticated spy planes ever built. [00:10:16] It was actually designed without a computer, [00:10:18] using pen and paper in the 1950s. [00:10:22] When they were building it, [00:10:23] one of the first planes built out of titanium, [00:10:26] America didn't have enough titanium. [00:10:28] It turns out the people who had the most titanium [00:10:30] were the people that were trying to spy on, the Russians. [00:10:33] So the CIA set up all these companies, [00:10:36] and they pretended they needed the titanium to build pizza ovens. [00:10:41] So that's how the Cold War started. [00:10:44] One country trying to build pizza ovens and telling the other. [00:10:48] But as we move into this next phase of globalization, [00:10:51] that's just not going to work anymore. [00:10:53] We don't need geopolitical competition. [00:10:56] We need geopolitical collaboration. [00:10:59] It's only about working together, [00:11:01] talking together like we are here today, [00:11:03] that we're going to build next generation companies for the future. [00:11:06] Because I believe the future valuation gap [00:11:10] is going to disappear between the idea of the technology company [00:11:13] and the traditional company. [00:11:15] And it's going to be replaced by companies that are AI leveraged [00:11:19] and those that are not. [00:11:21] You see, in every market, in every category, in every sector, [00:11:26] there's going to be winners and losers. [00:11:28] There's going to be some companies, some of you are here today, [00:11:32] who've figured out that by leveraging technology and AI, like Henry Ford, [00:11:37] to work in a new way, [00:11:39] they will completely invent new operating systems, [00:11:42] new operating models, [00:11:44] and they will change the markets that they work in. [00:11:47] And that brings me to the final implication. [00:11:50] If we really want to build an AI-powered civilization, [00:11:53] which is we need to start to think about intelligence [00:11:56] and talent and decision-making in new ways. [00:12:00] Even at the very basic level, [00:12:02] you need new people that work in very different ways. [00:12:05] I mean, think about it. [00:12:06] There's a difference between the way people in the past [00:12:10] used data and made decisions in the way that people do today. [00:12:14] Now, many of you are familiar with Excel, I'm sure, right? [00:12:18] You're probably even very proud of how good you are at it. [00:12:21] You might be one of those annoying people that have mastered pivot tables, [00:12:24] and you love showing people how good you are at using that tool. [00:12:28] That is until you meet some new intern [00:12:31] who tells you that you never need to use Excel again, [00:12:34] because they've figured out that you can upload Excel [00:12:37] into a generative AI platform like OpenAI, [00:12:40] and you just ask the AI to do the work for you. [00:12:43] And in fact, analyze the data, generate beautiful pictures, [00:12:47] and turn it into a presentation. [00:12:49] But there's another set of people that are even smarter than those. [00:12:52] They use tools like Notebook.lm, right? [00:12:55] They actually just upload the sources to that, [00:12:58] and then they actually talk to it as an AI that's grounded [00:13:03] in that particular data set. [00:13:05] So now they're having conversation with, like, [00:13:07] the world's smartest analyst about that data set. [00:13:10] What I'm trying to say is every new tool that's introduced [00:13:14] is going to lead to a generation of people [00:13:17] that have learned how to think much more cleverly [00:13:20] and leverage their own time more effectively. [00:13:23] So we have to be those people. [00:13:25] We have to let go of the past. [00:13:27] Everything that's made us successful [00:13:29] is exactly what could kill us in the future [00:13:32] if we don't embrace new ways of thinking, [00:13:35] new ways of changing, and new ways of leading. [00:13:39] Ultimately, even something like intelligence [00:13:42] is not going to be solved as simply as people think. [00:13:45] There are many people in the AI community [00:13:47] who think that human intelligence is just a scaling problem. [00:13:51] If you throw enough data and compute, [00:13:54] you'll eventually create a machine that's as intelligent [00:13:57] and as capable and as creative as we are. [00:14:01] I'm sorry, I just don't believe that's true. [00:14:03] What I do believe is true is that there are people, [00:14:07] many of us here today are in that category, [00:14:10] who will leverage these technologies [00:14:12] to augment their own capabilities [00:14:14] far beyond anyone thought was possible. [00:14:18] So when I think about the future of intelligence, [00:14:21] I think it's going to be measured, [00:14:23] not by knowledge or technical mastery, [00:14:26] but the ability to curate capabilities [00:14:29] and navigate complex choices even in the midst of high ambiguity. [00:14:37] I think you're going to be those people.