Welcome to the Intelligence Revolution

Join Lori Zager, Lisa James, and special guest Warren Jackson as they explore the world of AI. In this episode, they compare the Intelligence Revolution to the Industrial Revolution and discuss how to asses the various AI-related investment opportunities. Dr. Jackson is a physicist who first started studying AI as a principal scientist for PARC ten years ago. Currently, he is a consultant for DARPA.

Have a listen!

Find out more and reach out to us on our blog.

Transcript

AI

Lisa James: [00:00:00] Welcome to Nest Egg Podcast. This is Lisa James and I'm here with my partner Lori Zieger and a special guest. We are a team two X Wealth group at Ingles and Snyder, which is a registered independent investment advisor. So today we are going to talk about ai and I'll let Lori take it away with her favorite quote, evidently.

Uh, 

Lori Zager: a gentleman by the name of Dennis Hassabis. Who is the CEO of DeepMind, which is the artificial intelligence arm of Google, and he also won a Nobel Prize, said that the AI revolution is going to be a hundred times bigger than the industrial revolution, 10 times bigger and maybe 10 times faster. We've gotten lots of questions from our clients about the impact of AI and how to invest in it.

Everybody's excited about using chat, GPT. Which is a [00:01:00] large language model. Others prefer Claude. We like perplexity, whatever. Deep Seek is another name that comes to mind. The fact is that none of these companies are traded in the public market, and so the question to us has been we think AI is going to be big.

This is from our clients. We agree, and so how do we participate and. We have started thinking about it and thinking about when these companies do come public, do we wanna participate and what will they ultimately look like? That is how we got involved in this particular podcast and the blog that's associated with it.

Lisa James: Yes. And you know, one of the biggest question is, um, will these companies really be, uh, the ultimate winners or not? And. One of the things that our guest speaker, who is Warren Jackson, who has a, a deep background in analytics and technology, he's been involved in [00:02:00] robotics, uh, neural networks, medical technologies, got a list as long as you're, as your arm of, uh, specialties and inventions.

So we were. Talking about this technology change with him, and he's really the one who pointed out that there are quite a few parallels and things to be learned by looking at the industrial revolution and comparing it to what we're calling the intelligence revolution. So Warren, I'm gonna turn it over to you to talk a little bit about the Industrial revolution and how you see similarities between that.

And the, you know, the oncoming AI revolution. 

Warren Jackson: Thank you. There's lots of. Transitions in human society where there's major changes, where the way of life and things change. And I was struck with the similarities between the industrial revolution and the AI [00:03:00]revolution, both in terms of what it could inform us about how the AI or intelligence revolution might proceed.

And I think we'll get some useful insight, both in how it'll affect society and also what. Possible investment opportunities there might be. The industrial Revolution caused some of the biggest changes that we've seen in a millennium, and it moved us from an agrarian world to a industrial world and eventually to all the inventions that we take for granted today.

The industrial revolution basically started with the steam engine, and the steam engine was a source of power that. Was much, uh, greater than what we had before, and it drove changes initially in things like mining for pumps and also driving looms LA later on, those changes. Started permeating [00:04:00] other parts of society.

Uh, similarly, I would think that the AI revolution, uh, the steam engines are the lgu large language models, which are the basic engine of this AI or intelligence revolution that is performing the function of the steam engine. It's driving the initial changes. 

Lisa James: Well, one of the things you said that I thought was so interesting, Warren, was that it's not just the invention itself and what that invention does, but there is a a critical feedback.

Loop that, you know, you can think of as kind of like a, a positive spiral of, you know, one thing affecting another that affects another in a positive manner and causes growth. So maybe you could, uh, give some examples about that. 

Warren Jackson: So let's start with the industrial Revolution. The steam engine enabled. More efficient [00:05:00]mining, which in turn led to better steels, more coal for the steam engine themselves, as well as all the uses of steam engines like transportation.

In the case of the intelligence revolution, the large language models can assist in the actual manufacturer and development of the models themselves leading to more efficient. Use of energy or other resources, but it also then can spin off to other uses of the engine, in this case, the large language models.

Lori Zager: And isn't it true that one of the LLMs is actually specializing in doing that? 

Warren Jackson: Yes. There are elements where they're using AI to develop the next generation of ai. And AI is also. Designing the data centers and these chips that are all used in the models themselves. And this is how you get this exponential growth [00:06:00] because the technology is improving and it improves itself.

Lisa James: So, um, one of the things that we thought would be helpful for people in terms of understanding what to expect was to think about the stages of development. And there are. Are real parallels in these stages between the industrial revolution and the intelligence revolution. And maybe Lori, you could talk a little bit about the various stages.

Lori Zager: As Warren says, you know, the, the steam engine replaced horses when power was needed. And the first stage is you kind of slot in the new technology for what's. Existing technology. So today a lot of us are used to seeing or using chat bots when we go on the internet and have a question. So no longer is there a person actually answering the chat.

There's a chat bot that is taking care of things. 

Lisa James: Right. So really what was happening there was [00:07:00] that the new technology just provided a different solution to an existing problem. You needed power. Instead of using horses, you used a steam engine. Wow. That like. Produced a lot more things and allowed you to produce a lot more things, and similarly with ai.

It solves the problem of you need customer for an example would be you need customer service for your retail business. So instead of having to hire a bunch of people that answer phones instead, you know you have a chat bot online, uh, that will answer basic questions for people. I can't say that was so successful though, 'cause I know a lot of people, including myself, get pretty annoyed with it chat.

'cause by the time I have a question, it's because I've already looked at all the available answers that a chat PO can give me. But. That's kind of an example where it can do a little bit now, but if it could start to learn or start to process questions better, it could be existing questions. Yes, it could be better.

Lori Zager: And ultimately, you know, maybe the large language models will be able [00:08:00] to do that and uh, in fact write emails. This is just the beginning stages. Of, uh, using ai and then we talk about the second stage of technology, and that's when you have new processes. 

Warren Jackson: The first stage is just direct replacement, simplistic applications of the new technology.

And the second stage you get whole new. Industries are ways of using things that were not even possible with the horse and buggy, where you have things that you couldn't ultimately, like the internal cut combustion machine and other things that arose from the industrial revolution weren't conceived of, and the uses of freeways oh, just weren't conceived of.

So similarly with AI in the second stage, you'll start seeing uses that take advantage of the unique. Aspects of the new technology, which in this case is the large language models. Instead [00:09:00] of chat bots, just replacing a little element of your business, you can re-architect your whole business where you might have no more consultants or middle management might be replaced by AI or some, something like this.

Where a major change in the structure of your business. That, and this would be the second stage 

Lori Zager: and then what we call the third stage. It creates specialization as we think about these large language models, you could use them for different things and we're already noticing, for instance, when we wanna look up what people think about a company.

I spent, um, a couple hours the other morning. Researching a company using perplexity. It's got a lot of information on it. Um, which is 

Lisa James: a large language model. It's a large 

Lori Zager: language model that just has a lot more information and a lot more information than I found. We tried, I tried chat GPT and it just didn't have it.

So it's obviously got some kind of an arrangement [00:10:00] with these company reports, analyst reports, you know, like Bloomberg, it was amazing what they knew. Insider information. 

Lisa James: Meaning 

Lori Zager: insider actions, insider insiders, not insider trading, not insider trading, no insider actions where their company management was buying or selling stock.

Uh, they knew, uh, the number of analysts that were recommending the stock and, and whether they had buys, how many had sell on the stock. They knew what had happened to earnings estimates. I was able to look at a company next to another company in the same business and look at the margin profiles and could see, um, how the mar margin profiles differed between the two companies.

I could look at their revenue profile and see how the revenues differed between the two companies. It was extremely helpful, and in the past I would've read reports. From many various, they've been very laborious. Yeah, laborious. 'cause I would have to pull out the information that I was looking for. And this [00:11:00] was a much more direct thing.

You know, it reminds me a little bit of the early days of the internet. You know, it used to be that everything was pushed to you and then it got where you could then Google really did this, you could ask it, and so rather than having stuff put upon you, you could actually ask it what you wanted to ask.

Yeah, and 

Lisa James: I, and I have to say, I think there's another parallel with the early days of Google, especially when you think about. What teachers would say to students who use the internet to do research, they would say, be sure you know what source you're using. Like just because it's coming from the internet doesn't make it true.

And it just brings up some interesting questions about people who are using AI or Google for that matter today. We have a lot of background deep knowledge. So if we ask a question and we get garbage in response, we have the ability to say, [00:12:00] oh no, that doesn't really make sense. I think one of the concerns we have is at some point in the future when people don't really have deep individual backgrounds based on their own other research and they're completely dependent on AI for information, are they gonna have the capability to say, oh no, that doesn't make sense.

And this is a real thing because I, I use chat GPT and I ask chat GP a lot of different questions on a lot of different topics, and sometimes it answers questions where it does not know the answer. And I will go back and say. Why did you answer that question that was not right? This is what I was told by an expert and it will come back to me and say, I'm sorry.

You're right. I made a mistake. And then it will go find some other information to verify what I just said. And so there are holes. So while we like this product and it's making developments and leaps and bounds, it's not perfect. And I think its biggest imperfection is it doesn't admit when it doesn't know.

Yeah. You know, it doesn't say lot. A lot of men I know are 

Lori Zager: like that [00:13:00] too,

Lisa James: like asking for directions. Oh no, I know how to get there. I don't have to answer for directions. I over more stories than that. But, um, yeah, so I think, you know, you know, you have to use it judiciously. So we use it, but we also think about, well, does that answer. Sense or how much are we gonna 

Lori Zager: depend on it?

So when I was comparing the companies, I knew that the companies one had much more of an international business than the other. I knew that. I just didn't know the actual numbers. Right. That's what, but 

Lisa James: I'm, that's what I saying. And they pulled the numbers, but they had come back and said the other one had a lot more international business than you would red flag.

Right? Yeah. So we know it's good to research things where you have a base level of knowledge. To be able to discern whether the answers you're getting makes sense. 

Lori Zager: No, and I, and I had, they corroborated stuff I had seen elsewhere, but it was so nice in the way that it put the information it was, it was just really great.

So I can just see how it could be really useful. But I guess the point that I'd finally like to make is AI's gonna be most [00:14:00] successful in things that we haven't even thought about. Uh, now and don't even currently exist. 

Lisa James: Yeah, just like during the Industrial Revolution, people didn't think about cars and airplanes.

When was chitty chitty bang, bang. It's, it's hard to imagine what you can't imagine. So anyway, I think that, um, you know, there's a lot of interesting things to discuss here, and one of them is the consequences. Of having AI be a part of our world. Uh, and maybe I can turn that back to you, Warren, in terms of, um, we can talk a little bit more about sort of what happened in the industrial revolution.

What were the consequences there and, and what do we see happening as a consequence of the evolution of AI and current society? What are. 

Warren Jackson: Well, both eras were marked by, uh, transformations of fundamental nature and they reshaped our lives as well as the, uh, infrastructure that led to their development.[00:15:00]

And in the case of industrial Revolution, we went from manual to machine-based physical activity, and the AI is a mind. Improving tool. It enhances the human mind. And since that's our defining characteristic as a species, this has much more impact. Many of the cognitive and thinking tasks, especially the low level ones, we can sort of offload or magnify our intelligence.

Just like an industrial revolution. We could amplify our physical strength. We can now. Amplify our mental capabilities and all cases. Both these changes have led to disruptions the way we've used to do things, organize our labor, was mostly on the farm and we had lots of manual labor and we had to make the transition to, um, factories and [00:16:00]industrial type work structures and co company structures to go along with that.

We'll have to do the same with the. Intelligence revolution will have much more disruption because first of all, it's our mind enhancing capabilities that are being changed and it's much faster than previous changes. And our society structures have trouble adapting at that speed as well as individuals having to retool or retrain.

Um, we will have trouble keeping up with, with the changes, and this promises a more fundamental question, which is what is the role of humans in all this? With the industrial revolution, we always we're the brains. And so the machines were the muscle 

Lisa James: and they were stupid, 

Warren Jackson: right? And then the machines were, and so we [00:17:00] clearly knew our role, right?

In this case, it's not so clear if the machines start taking our roles, what's remaining for humans, humans, months. What happens 

Lisa James: when the machines are smart? And so 

Warren Jackson: then we get to one of the big questions that everybody has, which is called the alignment problem is. To make sure that the machines have our welfare and our goals as their motivating force and not develop some of their own.

The jury's still out about how I transition will go well. 

Lisa James: So I mean, I think so in, in the initial stages when humans are responsible for creating the AI and what the AI does. You know, it, it, it, you're worried about what are the humans doing in terms of what they're creating. But if they then create AI that creates itself without a human or advances itself without a [00:18:00] human, it becomes a process that we have less control over.

And is that something that worries you? 

Warren Jackson: Oh, yes, very much so. 'cause it's not just. I had thought that, for example, we could always just pull the plug, and that's true. But these AI systems are able to manipulate people so they can manipulate people to stop us from pulling the plug. So we're not just potentially fighting the AI itself, but it's AI plus its human allies.

And so the question is when these machines who are starting to set their, not deliberately set their own goals, they take a task and break it up into sub goals, but some of those sub goals could entail. Eliminating the role of humans or some other path that gets them to the goal, but isn't something 

Lisa James: that we wanted.

Warren Jackson: Yeah. It's not a solution you want for us. Right. It seems like there's, 

Lisa James: there's a movie in the making. [00:19:00]

Lori Zager: Yeah. I just heard a commentator who has been a big supporter of Netflix say that. AI was going to allow average people to make unbelievably good content, and so an average Joe using AI could have content that could be put on YouTube and it might rival what a Disney or a Netflix or someone did.

When Warren was talking, I was thinking about. These people that are asking AI for their psychological problems and when they, when they tell 'em, you know, if AI says, well, that's horrible and you shouldn't allow that, can AI be prosecuted for killing somebody or hurting somebody? I mean, you can just, you can take it really far.

Warren Jackson: Humans are not used to encountering an intelligence that's roughly on our level. And more importantly, we are not used to an intelligence [00:20:00] changing as rapidly as. This intelligence is changing. So if we talk and a year later we talk again, I have a expectation of roughly where you'll be. AI has changed so rapidly you can ridicule it or say it's lacking in some capability.

And eight months later it can be from going subpar performance to superior performance. And that is something we're not used to. And so it's very difficult to make statements about. Relative merits of things because of the landscape is changing so rapidly, we've never seen intelligence improve as fast as it's 

Lori Zager: a thing to think about.

With that in mind, and I've compared this not in this particular blog, but in talking to clients about it, the Levi's. Jeans and the gold rush. You know, Levi's made a lot of money and not everybody that mined for gold [00:21:00] made a lot of money. And so, and particularly in this particular situation where things are changing so quickly, it's possible.

That you might make a lot of money with just the people that supply the, I don't know whether you wanna say tools or, right now it's large language models, but it, the people that like, and I say this all the time, the. The large language models are huge users of energy. So if you can buy the companies that supply the energy that's needed for the large language models to run, it doesn't matter which large language model wins.

Right. You, it's still, they all need energy. 

Lisa James: I know, but I think to Warren's point, it's a different situation now because if you went back to Levi's in the Gold Rush, nothing was gonna change about what the gold miners are doing very fast. But the efficiency of AI could change very, very fast. That's true.

Yeah. That's, and that's your back on [00:22:00] an energy company because of the energy use could go wrong a lot sooner than you expect because AI suddenly needs a third of the energy than it did before. And so, and, and that is one of the conundrums, right? Of investing. 

Warren Jackson: Yeah, we, we saw that with the convolutional neural nets, which started making waves and.

2015, and within three years, the compute requirements dropped by a factor of 3000. 

Lisa James: Sorry. A third was really a bad number. Yeah, right. So, 

Warren Jackson: so the same could happen, and we know, yeah, a human baby can learn these things with a lot less energy, so there is a way to do it. Now, whether AI can mimic that or replicate that symbolically.

I think it's quite likely if we make some changes in how current machines are trained and structured, we are very data inefficient right now. Human babies [00:23:00] need a lot less information to get to a same level of performance and applying some of the developmental stages that humans go through to. Training AI in a more intelligent way can lead to some of these economies of scale or in training, so mm-hmm.

It's not at all clear that this is a permanent state of affairs. Now, we will always want the best and greatest. So if we have energy to throw at it. Even if our machine is a thousand times more efficient, then we'll want it to do a thousand times more. And so we'll probably always need energy. 

Lisa James: You think you're saving, but then the fact that you save energy means that you can just do more with it, right?

Yeah. And you actually end up using the same amount. 

Warren Jackson: I heard the statement Intel giveth Microsoft it away. In other words, as Intel created capability, the hardware got [00:24:00] better. The software people figured out ways to use that capability and so the net result was no motion. 

Lori Zager: Well, if that's analogous to what's going on today, that's today's Exactly, exactly, exactly the same, right.

What NVIDIA gives somebody else is gonna take away, it's not 

Lisa James: really taking away, right. It is just growth. 

Lori Zager: Yes. They both grow. It's just one may grow faster than and the other may grow slower. Yeah. 

Lisa James: Well, I'd like to thank Warren Jackson for helping us out in many ways, not just in doing this podcast, but for, uh, positing some interesting questions and really giving us the idea of comparing, uh, the intelligence revolution to the industrial Revolution.

Who are the major players in the intelligence revolution? 

Lori Zager: Well, it's been said that the semiconductor companies are the brains. Data centers are the body and energy producers provide the lifeblood that collectively enable the large language models and drive today's, uh, AI [00:25:00]transformation. 

Lisa James: So, uh, the semiconductor companies, as we know, design and manufacture the high performance chips.

We've all seen the craziness around, you know, the Chinese being able to use the NVIDIA chips and even national security issues about who can use the most advanced chips that we produce. 

Lori Zager: And specifically those chip are GPUs and specialized AI accelerators. Um, and I'm sure there'll be other types, but that's generally what they're talking about.

Yeah, they're the ones 

Lisa James: that really provide the computational power that you need, uh, to train and run the large language models, which are, you know, huge power consumers. Lori, you wanna tell us a little bit about the data centers? 

Lori Zager: They are kind of the physical and digital infrastructure. They store the data, they facilitate communication and they deliver AI services at scale.

Lisa James: Yes. And then the other very important player, which you know, some people think is sort [00:26:00] of a hidden driver, are the energy producers. Because oil and gas companies, even uranium companies and nuclear power plants are very important because. Someone has to be providing the huge amounts of electricity currently needed.

Lori Zager: And you know, this is important because you're seeing all of these companies that are using LLMs invest a lot of money in making one-on-one transactions with nuclear companies or with gas companies. Mo, they're mostly trying to be. Cleaned. So the Amazons, the Googles, the Metas of the world have kind of side deals with a number of, of energy companies.

Lisa James: You know, ultimately the biggest beneficiaries that we expect to see will be the users of the technology because just like in the Industrial Revolution, the companies that can leverage, you know, the new tool, in this case, AI. [00:27:00] To either improve efficiency or reduce cost, or in the case of AI, make better decisions or create new capabilities, those are the ones who are going to benefit the most.

Lori Zager: So I'm just gonna talk about a couple of different industries and name a few companies that are currently using AI in their business. One of the big areas when we think about it are retail and e-commerce. And AI improves the customer experience. Uh, it gives you personal recommendations. It helps the companies optimize inventory and pricing.

And, um, then it enables real time personalized customer support, both online and in physical stores. And, uh, the companies that come to mind are Amazon and Walmart, but there's also. Logistics and transportation. And again, Amazon comes to mind because they deploy AI powered robots and warehouses, uh, for order picking, sorting and inventory managing that improves the [00:28:00] fundamental speed and accuracy.

Um, it's also used for dynamic route optimization and deliveries. Um, UPS has an AI driven system that optimizes delivery routes and, um, it's saving millions of miles on the road and reducing fuel use and emissions. So manufacturing is, uh, benefiting from ai. Again, optimizing production, improving quality, um, helping companies to get a competitive edge.

Siemens is using AI and machine learning to predict equipment failures, reduce downtime, and implement digital twins that simulate and optimize factory processes in real time. General elective is using AI powered predictive maintenance in jet engine manufacturing, and it analyzes sensor data to detect.

Early wear and prevent costly failures. And Tesla, uh, which you would think of with regard to AI is using AI for demand forecasting. [00:29:00] Product optimization and real time quality monitoring across its manufacturing. And then there's energy and utility companies that have adopted AI. And um, duke Energy is a company that uses AI and machine learning for predictive maintenance, grid optimization and demand forecasting.

It helps prevent outages, reduce costs, and enhances grid resilience. And I guess the last example that I'd like to give is, um, medical and pharmaceutical companies that are using AI to accelerate drug discovery, optimize clinical trials, improve personalized medicine, and enhance operational efficiencies.

Um, AstraZeneca's and example that you could use, they've used. AI to identify new drug targets and optimize clinical trial design, enhancing treatments for chronic diseases. And we have 

Lisa James: even more examples on our blog, on our website at two x [00:30:00] wealth.ingles.net. But the other point that I'd like to make is, you know, when you hear about all the things that these companies are doing, is it the chat GPTs of the world that are gonna be the winners or even the NVIDIAs?

Of the world, or might it be all of these companies that are using AI to their advantage. And we have a really classic example to point that out because Xerox invented the personal computer, but Apple was the company that figured out. Why people needed to have one and designed one that became much more popular.

And so Apple went on to become one of the most successful companies in the world. And while that was happening, Xerox went nowhere. And so that's why we say, you know, it's not necessarily. So straightforward, who the long-term winners are 

Lori Zager: gonna 

Lisa James: be 

Lori Zager: as, uh, we were recording this podcast. There was a study that came out from MIT indicating that 95% of organizations that they studied [00:31:00] got zero return on their AI investment.

And, um, I heard someone, uh, who broke down, who read the study and actually broke down all the different parts of the study. Um, went through it with anyone who wanted to. To listen, there were about 300 institutions, I guess 52 organizations that were surveyed. It was between January and June of 2025. What they talked about is these task specific artificial intelligence.

Um, they basically, AI companies sold. Other companies, these task specific AI tools, and that's where they were getting no return on their investment. What they found out when they surveyed the employees of these organizations, that 90% of the employees were using their own large language models, finding them very effective.

Uh, he calls it the shadow AI economy. It's the exact opposite of what they concluded in the study. So 90% of the workers. We're finding a [00:32:00] huge return on large language models, but companies that have spent a lot of money on task specific AI tools, they didn't get a return. It's really kind of funny. I think the first thing that's interesting is that A, who's reading the studies?

Uh, we just, we, we live in a, as I call it Twitter, I guess it's an X world now, so we just look at headlines. But when you look beyond it, AI tools really do help. It's just a question of which tools you use. So just to conclude. Just as in the industrial revolution, um, there will be upheaval and changes in society.

The parallels between the industrial revolution and the AI revolution may have predicted value for investing today, and the early winners, as Lisa just used in the most recent example of Xerox and Apple. May not be the ultimate beneficiaries in terms of creating economic value. Many of the companies that are the ultimate winners may not even exist today.

We hope this has, uh, helped you think about, uh, artificial [00:33:00] intelligence. And, uh, if you, uh, wanna know more about the science, we probably better put you in touch with Warren and not us. But, uh, when it comes to investing, we'd love to help if we can. 

Lisa James: Yeah, so you can reach us at Lisa at two x wealth.ingles.net, or Lori at two x wealth.ingles.net.

We're happy to get questions and, uh, we love suggestions of topics that people might be interested in hearing about. That's all for now, and see you next time.