Seth Williams: I want you to imagine something with me. Imagine you own a company with 20 employees. Tomorrow morning, somebody walks into your office and says, I've got a new employee for you. This person can write better than almost anyone on your staff. They can analyze spreadsheets. They can answer customer questions. They can write computer code. They can research almost any topic. They can read a 200-page contract in a few seconds. They can make videos and images. They can speak almost every major language. They don't need health insurance. They don't take vacations and they don't sleep. And they'll work for you 24 hours a day for a few hundred dollars a month.

Would you hire that person? Of course you would. But here's where things get a little uncomfortable. What happens to the other 20 people on your team? And now imagine that this isn't happening at one company. It's happening at almost every company in America. Because that is essentially the experiment we're beginning right now with AI. And I don't think we've really wrapped our heads around where this could go. I want to be very careful about something in this episode. I'm not going to tell you that AI is definitely going to take everyone's jobs. I'm also not going to tell you that everything is going to be fine because people panicked about the steam engine and the internet too. I don't think anybody actually knows. But we finally have enough real world evidence that we can start separating what is actually happening from all the hype.

And some of the evidence is incredibly encouraging. Some of it's also pretty disturbing. And if you're a business owner or an investor or a parent or really anybody who expects to be alive for the next 20 years, I think this is one of the most important economic questions we should be thinking about. Because the big question isn't, is AI going to get better? It obviously is. The question is, what happens when intelligence itself becomes cheap?

There is a very comforting argument people make whenever somebody worries about AI taking jobs. They'll say something like, people said the same thing about every new technology. And they're right. When mechanized farming came along, people worried about agricultural jobs. When factories became automated, people worried about manufacturing jobs. When computers came along, people worried about office jobs. When the internet came along, people worried about all kinds of businesses. And somehow, we kept finding new things for people to do. In 1900, about 40% of the American workforce worked in agriculture.

Today, it's around 1%. And yet America doesn't have 39% unemployment. Those workers moved into factories and offices and healthcare, technology, services, and thousands of occupations that barely existed before. So there is a very strong historical argument that technology doesn't eliminate work. And that could absolutely happen again. The Bureau of Labor Statistics still projects total U.S. employment to grow over the next decade. Its current projection is that employment will increase by about 5.2 million jobs between 2024 and 2034. And interestingly, some jobs that seem extremely exposed to AI are still expected to grow. Software developers are one of the most obvious examples. AI can already write code, right? And yet the Bureau of Labor Statistics projects software developer employment to grow almost 16% from 2024 to 2034, adding more than 267,000 jobs.

So clearly this simple equation of AI can do a task, therefore the person doing that task becomes unemployed, that just doesn't necessarily work. Sometimes technology makes something cheaper, which causes people to consume more of it. When computers made it easier to create spreadsheets, we didn't stop needing accountants. When software made graphic design easier, suddenly millions of small businesses started creating graphics that never would have hired a professional designer in the first place. So it may be that AI is going to make intelligence cheaper and the world will simply consume vastly more intelligence.

That's the optimistic case. And there's real evidence for it, but there's also another possibility because this technology is very different from anything we've ever seen before. So why might AI be different? Most previous machines replaced some specific physical ability, like a tractor replaced human muscle, and a calculator replaced arithmetic. A washing machine replaced hours of manual washing. An excavator replaced people digging holes with shovels. But AI is going after something different. It's going after cognitive work. Writing, reasoning, coding, research, analysis, planning, design, customer service, translation, legal work, accounting, marketing, and eventually, as AI gets connected to robots, potentially physical work too. And that's what makes this really unusual. The machine isn't replacing one task.

It's gradually becoming capable of learning almost any task that can be represented digitally. There's one organization doing some pretty fascinating work on this. It's called METR, M-E-T-R. And instead of asking whether an AI can pass some standardized test, they ask a much more practical question. How long of a real-world task can AI successfully complete on its own? Maybe an AI can handle a 5-minute task, and then a 20-minute task, and then an hour-long task, and then a 4-hour-long project, and then something that would take a human an entire day to do. Well, METR's research has found that the length of software tasks frontier AI systems can successfully complete has been increasing exponentially.

And that's a really important distinction because we're not just talking about AI getting slightly better at answering trivia questions. We're talking about AI becoming capable of staying on task, making decisions, correcting mistakes, and finishing larger and longer projects. And if that trend continues, and I'm sure it will, we're going to eventually cross some really important thresholds. At first, AI just helps an employee, but then AI becomes capable of doing most of the employee's individual tasks. And then AI becomes capable of doing the employee's entire job. And eventually, AI may become capable of running processes that currently require entire teams. We don't know when those thresholds will happen, but we can already see what happens when people are given today's AI capabilities.

And that's where this starts to get really interesting. So the productivity numbers are already kind of crazy. Researchers have now done several studies where they give one group of workers AI and another group doesn't get it. And then they compare what happens. And one famous study looked at more than 5,000 customer support agents. Some of the workers got access to an AI assistant that helped them answer customer questions. And the result was that productivity increased by about 14%, which is pretty meaningful. But the part that was really interesting is that the least experienced workers improved the most. In some cases, AI seemed to be transferring the knowledge of the best workers to everybody else.

And just think about what that means. Normally, somebody becomes really good at their job because they spend years accumulating experience. They make mistakes, you know, they learn shortcuts and they encounter weird situations and they figure out what works. And that kind of real-world knowledge and experience has been extremely valuable because it lives inside of experienced people. Well, now, AI can potentially capture that knowledge and hand it to a brand new employee on day one. And other experiments have found even bigger effects. Researchers studying professional writing tasks found people using generative AI finished their work faster and produced better results. There's another experiment involving highly skilled consultants who found that workers using GPT-4 performed dramatically better on tasks that fell within the AI's capabilities.

MIT researchers have described productivity improvements approaching 40% in some kinds of professional work. So, imagine what that could do to a company. What could that do for your company? Suppose you've got 10 people producing 100 units of work, and then AI comes in and makes each of them 30% more productive. Now those same 10 people can produce 130 units. And that's fantastic if your business needs 130 units, but what if it only needs 100?

Now you only need 8 people instead of 10 people, right? And this is where productivity and employment start pulling in opposite directions. For the company, this is wonderful. For the customer, it could be wonderful for them too. Things become cheaper, service becomes faster, companies become more profitable, the economy becomes more productive. But for the two people the company no longer needs, this technological miracle doesn't feel quite as miraculous, does it? And until recently, we didn't have much evidence that this was actually happening. But now we do. So in August 2026, researchers at Stanford's Digital Economy Lab updated a study with a pretty ominous title: Canaries in the Coal Mine. They analyzed payroll data covering millions of American workers, and they were specifically looking for employment changes after generative AI became widely available. And they found something very interesting. The labor market overall was not collapsing. Experienced workers in AI-exposed occupations were generally doing okay. But the younger workers, they were in a different story. Workers between 22 and 25 years old in some of the occupations most exposed to generative AI were experiencing some pretty substantial relative declines in employment. Earlier versions of the research put that decline around 16% relative to the less exposed jobs.

And the effect was strongest when AI could automate the work rather than simply help somebody do it. And that distinction is incredibly important because there are basically two different ways AI can enter your job. It can say, here, I'll help you do your job better. Or it can say, thanks, I'll take it from here. And those are very different futures. And the Stanford researchers found evidence that workers are doing much better when AI complements their work rather than when AI can directly automate it. And this also creates a weird problem for young people. I mean, think about how careers normally work. You graduate from school and you don't know very much, but a company hires you anyway, and they give you all the boring stuff. So you have to research things and prepare documents and answer basic questions and make simple spreadsheets.

You might write some basic code. You sit in a bunch of meetings and you screw things up. And gradually over five or 10 years, you somehow become the experienced person. But what happens when AI is really good at the boring entry-level stuff? I mean, why would a company hire five junior analysts if one senior analyst with AI can do the same work? Why hire an entry-level copywriter if the marketing director can generate 50 versions of an ad in 30 seconds? We could end up removing the bottom rung of the career ladder. And if you remove the bottom rung, then how does anybody actually reach the top? This is becoming serious enough that the World Economic Forum published an entire report in 2026 about AI and the future of entry-level work. Again, this does not mean we're experiencing mass unemployment. We're not.

There still isn't strong evidence that AI is causing major economy-wide job losses. And it's important to note that both things can be true at once. AI hasn't destroyed the labor market, and the first cracks may already be appearing. So let's imagine AI keeps improving. What jobs are the safest? People used to assume technology would replace blue-collar workers first, and then white-collar professionals would be safe. And that prediction is looking pretty shaky. It's much easier to put AI inside Microsoft Word than it is to put AI inside a robot that can crawl underneath your kitchen sink and replace a broken pipe. Because the physical world is messy. A plumber has to drive to your house and figure out where the leak is and move your junk out of the way and then find the shutoff, unscrew some rusty fitting installed in 1978, realize the previous homeowner did something completely insane, and then drive to Home Depot, come back and somehow make everything work.

That is incredibly difficult to automate. Compare that with read these 40 documents and summarize them. AI can already do that. So we could end up with a strange inversion. Some of the jobs we've spent decades telling kids they needed to escape may become extremely valuable. Electricians and plumbers and HVAC technicians, construction workers, mechanics, nurses, physical therapists, people working with machines, people working face-to-face with humans. Meanwhile, some jobs that required expensive college degrees may experience much more pressure. Paralegals and junior accountants and copywriters and analysts and programmers.

Certain financial jobs, certain legal jobs, certain marketing jobs. And I'm deliberately saying certain here because I don't think these professions disappear. The more likely outcome is that the job changes. The lawyer doesn't disappear. But maybe one lawyer with AI does work that used to require three lawyers and two paralegals. The real estate agent doesn't disappear. but maybe the agent who knows how to use AI can serve three times as many clients. The programmer doesn't disappear, but one really good programmer manages a team of AI coding agents. The accountant doesn't disappear, but the accountant stops spending six hours manually categorizing transactions and spends more time interpreting the numbers and actually advising the client. And that brings us to what I think may be the most important concept in this entire episode.

And that is the person who uses AI may replace the person who doesn't. I think a lot of people are asking the wrong question. They're asking, will AI replace me? Maybe, but before that happens, I think something much more immediate is likely. People using AI will replace people who aren't using AI. Imagine two employees, right? Let's say both of them earn $80,000 a year. Both have roughly the same experience, but one learns how to use AI really well. Well, that person can now research faster and write faster, analyze information faster and respond to customers faster, create presentations, automate repetitive work. The list goes on and on as we know. And let's just say conservatively, they're producing twice as much useful work as the other employee.

Well, if the company has to cut one position, which one do you think is going to stay? This is why I don't think the right response is to avoid AI because you're afraid of it. That's probably the riskiest thing you can do. You want to be standing on the side of technology that is increasing productivity. And the data suggests companies are figuring this out pretty quickly. According to Census Bureau data from late 2025 through May 2026, somewhere around 17 to 20% of American businesses reported using AI in their operations. And among the companies with at least 250 employees, usage was around 37%. Another census study found that companies are currently using generative AI most heavily for things like writing, document analysis, and information search.

And most companies still aren't using it everywhere, not by a long shot. In fact, 65% of firms using AI limit it to maybe three or fewer tasks. And that tells me something really important. We are still very early in this process. AI is not fully integrated into most businesses yet. It's barely even gotten started, which means most of the economic effects we're talking about haven't happened yet.

So what happens to entrepreneurs exactly? Let's flip this around because everything I've said so far might sound kind of scary if you're an employee. But if you're an entrepreneur, there's another side of this that is unbelievably exciting. The cost of starting and growing a company may collapse. In fact, I've already seen the cost going way down in my businesses. Think about all the things a small business owner traditionally needs help with. Graphic design and copywriting, customer support, research, accounting and programming and video editing, website development, sales, marketing, legal documents, data analysis, administrative work. You couldn't possibly be good at all those things. So you either have to hire employees, hire freelancers, or just not do them.

And AI is starting to give one person access to all of these capabilities. Not perfectly, but increasingly well. I already see this in my own work. There are things I can do today with AI that I absolutely would have had to hire somebody just a few years ago. And I don't think we've begun to see what happens when AI agents mature. Instead of asking AI to write one email, you give it a goal. Say, find every customer who hasn't responded in 30 days. Figure out where each conversation left off, draft an appropriate follow-up, send the easy ones and flag anything unusual for me. And that's not a tool anymore. That's starting to look like an employee. And by the way, if you're using a CRM like Stride CRM, everything I've just told you is absolutely real. And I'm able to do that right now.

If I hook up Claude Code to the Stride CRM API, I can hook it up to my Gmail inbox, my Google Analytics, pretty much all the good software I have, assuming it has an API key and many of them do now. And I can ask it to tell me insights that I just would have been completely blind to before. And I can even tell it to do things on my behalf without ever even logging into the software. It's unbelievable. Now, imagine giving a solo entrepreneur 20 of those things. One handles customer service. One manages advertising. One watches the books. One researches competitors. One manages the website. Another one follows up with leads. Another one creates content. One handles scheduling and one analyzes sales calls.

Suddenly, a one-person company can have capabilities that once required 20 people. I think we could see an explosion of extremely small companies generating surprisingly large amounts of revenue. We already saw something similar with the internet. The internet gave small businesses access to global distribution. AI gives them access to cheap intelligence. You put those things together and entrepreneurship starts looking very different and very attractive. This is where things can get a little bit complicated. Suppose AI makes the economy twice as productive. Well, that's fantastic. We can produce twice as much stuff, but who gets the extra money? Imagine a factory with 100 workers. Let's say it produces $10 million worth of products. And then the owner installs AI and automation and actually learns how to use it.

Well, now 20 workers plus the machines produce $20 million worth of products. Society is richer, right? But 80 people lost their jobs, and the owner is making a fortune. This is why economists are paying so much attention to the difference between labor and capital. Labor means you make money because of work.

Capital means you make money because you own something. Maybe you own a business, or stocks, or real estate, or machines, or intellectual property, or AI systems, or whatever it might be. And several IMF researchers have warned that AI could increase wealth inequality because the owners of the technology and other productive assets receive a disproportionate share of those gains. And one IMF model actually found something really interesting. AI could reduce some forms of wage inequality because it may threaten high-income cognitive jobs. But at the same time, it could substantially increase wealth inequality. Why? Well, because the people who own the productive assets capture more of the upside. And that distinction is enormous. I mean, imagine a world where intelligence becomes incredibly cheap.

The ability to write a marketing plan isn't scarce anymore. The ability to create software isn't nearly as scarce. The ability to analyze a contract isn't as scarce. And neither is the ability to create an image. So what is still scarce? Things like land, energy, housing, natural resources, infrastructure, prime locations, brands that people trust and ownership.

And this is where I think this conversation gets particularly interesting for investors. Because if intelligence becomes cheap, then physical scarcity matters more. Because AI exists in the digital world, but AI runs on physical stuff. It needs chips, right? Those chips go inside servers. And then those servers go inside data centers. And those data centers sit on land. They need electricity, water, fiber connections, roads, transformers, and they need enormous amounts of physical infrastructure. And the scale of this build-out is becoming pretty staggering. The Federal Reserve has actually started tracking AI-related investment because spending on software, computers, and related infrastructure has become large enough to contribute meaningfully to economic growth.

Data centers already consumed around 4.4% of U.S. electricity in 2023. And some projections suggest that American data center electricity demand could rise dramatically by 2028. And this is creating something that I find fascinating as a real estate investor. One of the most advanced technologies humanity has ever invented is creating enormous demand for one of the oldest assets in human history. And that is not just any land, but land with power, near transmission infrastructure, near fiber, in places where governments will allow data centers with water access near growing population centers.

So AI may live in the cloud, but the cloud has a street address. And make no mistake, there are already rural communities around America dealing with proposed massive data centers because AI companies need physical places to put all this computing power, and this creates winners and losers. A farmer may suddenly discover that a piece of ground worth $10,000 an acre is strategically important to a data center developer. A town may receive millions in new tax revenue and an electric utility may suddenly need billions of dollars of infrastructure. Nearby residents may see power costs or water concerns become political issues and investors are going to start looking at real estate a lot differently. For the last hundred years, one of the most important questions about land was, where are the people going?

But in the AI economy, another question may become increasingly important. Where is the power? So what happens to housing? This gets even more speculative, but I think it's worth thinking about. If AI dramatically changes where people work, it could change where people live too. Remote work has already started spreading jobs from offices and AI could push that even further. I mean, suppose a company that once employed a thousand people in downtown Chicago eventually employs 300 people plus AI systems. Well, that's 700 fewer workers who need to commute to that office. And if you multiply that across thousands of companies, what happens to office buildings? What happens to downtown restaurants? What about parking garages? What happens to apartment demand near major employment centers?

Now, on the other hand, suppose AI creates enormous wealth for a relatively small group of entrepreneurs, investors, and highly productive workers. Where do those people want to live? Well, we may see housing demand become increasingly concentrated in desirable places, right? Beautiful communities, safe communities, places with great schools and great outdoor recreation and low taxes, places where wealthy people simply want to be. In other words, if work becomes less tied to geography, lifestyle may matter more in determining real estate value. Now, that's not a prediction, but it's something I would absolutely be thinking about if I were buying property with a 20-year horizon.

So I'm a parent, and I know there's other parents that listen to this. What should we be telling our kids? This might be the part of this whole conversation that I struggle with the most, because for generations, parents gave their kids some version of the same advice. Work hard in school, get good grades, go to college, learn a valuable skill, get a good job, work hard, save money, and you'll probably be okay. But what skills should a 12-year-old learn today? Should they spend time learning how to code? AI can code. Do they need to learn how to write? Well, AI can write. Graphic design? Well, AI can create images. What about accounting? Well, AI can analyze financial statements. Law? AI can read legal documents.

Languages? Well, AI can translate almost instantly. I'm not saying kids shouldn't learn those things. I actually think the opposite, but I think the purpose of education may need to change. Memorizing information becomes less valuable when every human has access to a machine that knows vastly more information than they could ever memorize. The valuable skill becomes knowing what questions to ask and knowing when the answer is wrong. Knowing how to think, how to communicate, and how to persuade, how to work with people, how to make decisions when there isn't an obvious right answer, knowing how to build things, knowing how to take responsibility, knowing how to earn trust. Curiosity becomes incredibly valuable. Judgment becomes very, very important. Initiative becomes critical. And perhaps most importantly, agency becomes valuable.

Now, when I say agency, what am I talking about? I mean the ability to look at a problem and say, I'll figure this out. Not, nobody taught me how, or that's not my job, or tell me exactly what steps to follow. But instead, I mean, I don't know how to do this yet, but I will find a way. Don't get me wrong, agency has always been valuable, but now it's going to become even more so. AI may actually amplify the difference between people with high agency and everybody else. Give two people the same AI. One asks it to write funny birthday poems. The other one uses it to learn a new industry, build software, research a market, automate a business, and launch a company. Same technology with completely different outcomes. And I think teaching kids how to use intelligence may become more important than teaching them how to store information.

The U.S. Department of Education is already acknowledging that AI is going to play a role in education. Schools are experimenting with it. Universities are wrestling with it. And I don't think the answer is going to be banning it. That would be like schools in 1998 saying, yeah, we don't want students using the internet because they might find the answer. At some point, the tool becomes part of the world. Education has to adapt to the world students are actually going to live in. So let's talk about this scenario that nobody knows how to handle. What if AI becomes so capable that there simply aren't enough economically useful things for humans to do?

Historically, every time technology has eliminated jobs, new jobs appeared. But that historical pattern isn't a law of physics. It's possible AI could create millions of jobs we can't currently imagine. And that's happened before. Nobody in 1980 was training to become a YouTuber. Nobody in 1990 planned to become an app developer. And nobody in 2000 was trying to become a social media manager, right? Entire industries appeared almost overnight. AI could do the same thing, but there's a logical problem. Suppose AI becomes better than humans at almost every cognitive task. If we invent a new job tomorrow, why couldn't AI just learn that job too? That's what makes this technology different. The traditional response to automation has always been, humans will have higher value work.

Okay, what happens when the machine can do the higher value work too? I don't know. And nobody does. Maybe human labor becomes valuable because humans want other humans involved. That's possible. People still pay to watch humans play chess, even though computers are better. We still pay humans to play football, even though machines could throw a ball a lot farther. People go to concerts, even though recordings exist. We buy handmade furniture, even though factories can make tables cheaper. So human involvement itself can have value. And maybe we end up with an economy where authenticity becomes a luxury product. Made by a human, written by a human, taught by a human, performed by a human, advised by a human. Those labels could someday mean something. Or maybe productivity becomes so enormous that society changes its relationship with work entirely.

That's where ideas like universal basic income enter the conversation. If machines produce most of the economic value, governments may eventually decide that some portion of that wealth needs to be distributed to people who aren't participating in traditional employment. And that opens up enormous political, economic, and philosophical questions that I'm not even going to try to solve here because there's an even deeper question. If people don't have to work, what do they do? Because work is about more than money. And this is the part economists sometimes miss.

A job isn't just a paycheck. Work gives people structure. It gives them responsibility and goals and social interaction. It gives them a feeling that somebody needs them. And it gives them a reason to get out of bed in the morning. Think about the first question we often ask somebody when we meet them. What do you do? And we know exactly what that means. We don't mean, what do you do on a Saturday afternoon? We mean, what is your job? Because like it or not, in our society today, our work has become part of our identity. So imagine a world where millions of people become economically unnecessary. Even if you could somehow guarantee everybody enough money to live comfortably. Well, you've still only solved one part of the problem. because people need purpose. They need a challenge. They need community.

They need something difficult to accomplish. So maybe we spend more time raising children or creating art or building things or volunteering or starting businesses or playing sports, exploring, taking care of people, learning. Maybe removing the requirement to spend 40 hours a week earning money becomes one of the greatest improvements in human history. Or maybe millions of people struggle with the feeling that society doesn't need them anymore. Again, I don't know, but I think this is a much deeper question than whether AI will take somebody's accounting job. It's asking, what is a human being for in a world where intelligence is abundant?

So what do we actually do about all this? Well, after researching all this stuff, I've come away not feeling completely terrified but also not completely reassured. The truth is, the evidence doesn't support the idea that AI is currently destroying employment across the country. It's not. But I also don't buy the argument that we can ignore this because previous technologies created new jobs. AI is moving into cognitive work incredibly quickly and there are now measurable warning signs in some of the youngest and most AI exposed workers. So what do we actually do with this information, right? If I were trying to prepare myself for the next 10 or 20 years.

I would probably focus on five things. I would become extremely good at using AI. All right, we don't want to compete against the machine. We want to learn to operate it. And the person who knows how to multiply their abilities with AI has a much better chance than the person pretending it doesn't exist. And there are all kinds of great opportunities to do this right now. You could do this self-taught route through just watching YouTube videos. I know I've done a lot of that myself. Or you could even go with a paid guide, someone like Callan Faulkner, who has now been teaching this for years. She's got an upcoming cohort where she's teaching the AI Employee Bootcamp. This is happening on September 29, October 1, and October 6. And for the price she is charging for this, there is incredible value here.

We actually have an affiliate link if you want to sign up for her course and support the REtipster community. Just go to retipster.com forward slash A2A. That's the letter A, the number two, and the letter A. And you can also sign up for her higher ticket Automate to Accelerate eight-week guided course. And you can check that out at retipster.com forward slash A to A. I've been through it twice now. She does an amazing job. I've learned a ton of stuff from the insights I've gained just going through her material. So that's totally worth checking out. And by the way, we'll have clickable links to both of those things in the show notes for this episode. If you want to click through our affiliate link and get yourself some incredibly useful education that everybody needs right now in this emerging AI age we live in. So first, become extremely good at using AI, however you want to do that.

Second, develop skills that AI struggles with. So we're talking about relationships, trust, leadership, judgment, negotiation, physical skills, human connection, taste, and decision-making, and the ability to operate in messy real-world situations. If it's something that AI can't do well or inherently can never do, that's something you want to be really good at as well. And third...

Own productive assets. This one is especially important to me. If AI shifts more economic value from labor toward capital, you don't want your entire financial future dependent on selling your time. Own businesses, own stocks, own real estate, own intellectual property, own tangible assets that produce value. And you don't need to be wealthy to start doing this. Ownership becomes increasingly important in an economy where human labor may become less scarce. And fourth, keep your fixed costs under control because nobody knows exactly how disruptive this transition could be.

Technological revolutions create incredible opportunities, but they also create volatility. The person with enormous debt and no financial flexibility has a lot fewer options, but the person with savings and low expenses and liquid capital can take advantage of disruptions. And fifth, stay curious. This technology is changing way too quickly for anyone to just learn it once and be finished. Whatever AI looks like today is probably going to seem primitive five years from now. The winners won't necessarily be the people who understand the technology best right now. They're going to be the people who keep learning. So as we wrap this one up, as I say all these things, I have to admit, there is an irony in all of this that I can't stop thinking about. For thousands of years, intelligence was one of the scarcest resources on earth.

If you needed expert legal advice, you needed a lawyer. If you needed financial analysis, you needed an accountant. If you needed to understand something complicated, then you needed to find someone who understood it and so on. We built an enormous part of our economy around the scarcity of human knowledge and human intelligence.

And now we're building machines that make intelligence abundant, maybe extremely abundant. And whenever something that used to be scarce becomes abundant, the economy reorganizes itself around whatever is still scarce. That may be human trust. It may be relationships. It could be authenticity or it might be energy. It could be physical infrastructure or land. We don't know yet. But I think that's the question worth paying attention to. Not, is AI gonna take my job? The bigger question is, what becomes valuable when intelligence is no longer scarce? Because if we can figure that out, we get a much clearer picture of where the opportunities are going to be. And if you're an investor or a business owner or a parent trying to prepare your kids for the world they're going to inherit, that's probably a much more useful question to be asking.