Advice for working in a post-AGI world

So GPT 5. 3 came out, Opus 4. 6 came out. It's really clear AI is getting to Terminator 2 levels of competence. And naturally, the question in everybody's minds is when's McDonald's releasing that Sichuan sauce again? No, it's obviously what the [ __ ] am I going to do with my life if AI models are better than me at basically anything that I could possibly choose as a career? Well, that's what this video is for. My goal is to give you guys a list of different things that you can do that are likely to be insulated against the AI takeover in the next few years. Your first option is getting really good at something that interfaces with the real world. And I'll run you guys through some examples in a second. But have you ever wondered why AI models are getting really good at writing and art and media and programming, but then they're lagging behind on the more monotonous, boring things like scrubbing my toilet or taking out my trash. Uh, I wish they could do those things, but the reality is there's way more training data available for our 26 character English alphabet than there is sensor and actuator and pressure and heat data available on something that I don't know happens all the time. Like let's say you going into a clinic and getting your blood collected. Like if you think about it, we have billions or trillions of words of written text available on the internet because that was what the internet was made for. It was made to communicate primarily through a textual medium. is how much data do we really have on some advanced robotic prosthesis feeding a needle into an arm? So, there's a couple things there, right? If you're in writing or design or media, there's still some options available for you, which I'll cover in a moment. But, but if you're approaching your career at sort of the start line with no pre-existing skill set, it makes more sense as this upcoming economic transition is occurring to learn a skill that does interface with the physical world that has some sort of actual realworld presence. Most digital white collar work is automatable fairly quickly. I mean in the next 6 to 12 months, Dario Amod, he's the founder of Anthropic, which is the group of people that design the Claude series of models. Um, he thinks that we're going to automate the vast majority of all software engineering. And I mean like software engineering underscores a massive chunk of most people's daily lives at this point, but it's not all of our lives. I mean, you exist in the physical world. You don't exist in the digital one. So, what do you do with your life when you step away from the computer? The various services, pieces of infrastructure, and organizations that manage the experiences you have outside of your computer are quite important and likely to have a fair amount of runway yet. And there are other implications to this as well. Consider nursing for instance. If you guys are a nurse, you're typically dealing with um at risk populations. people that you know might be a little bit frailer. Consider working in a nursing home or some sort of long-term care facility or even at the hospital. And so in addition to there being you know a direct lack of data in this field, there are also a fair amount of regulatory concerns and also safety concerns uh about automating work that deals with people like this. Not to mention just the desire for comfort. You know would you rather get treated or dealt with by some cold robotic machine or you know a nice smiling warm friendly nurse. I would obviously choose the latter at least at this point in time. And so in addition to not having the data existing and having to collect all of that data, which is likely a process that would take, you know, 3 to 5 to maybe 10 years, we also have to have those alternatives, these automated solutions approved by regulatory bodies because we're dealing with people that are more at risk, you know, frailer and so on and so forth. And it's not just health either. I don't just want to make everybody think I'm saying get into nursing. um you know dealing with realworld physical things extends far beyond that. What about the handyman that needs to service a building, some sort of maintenance technician. Think about the diverse set of general skills required by like an electrician who comes in to both, you know, wire your home, but who also needs to understand a fair amount of basic carpentry. And not only that, they have to be able to maneuver and navigate environments that may not necessarily be set up for some sort of robotic apparatus. In general

the real world roles with the most runway are likely to be the one that does not uh have the most specialized skill set, but counterintuitively actually the more shallow and broader skill set. So things like general purpose handymen or maintenance technicians or automotive technicians to be honest, people like electricians like I was talking about, you know, construction workers, things like that. which is hilarious to me just because a few years ago the number one piece of advice to people that wanted to make more money was always like learn to program computers. Uh now there's going to be like construction workers at the site watching a poor little programmer stumble on by and they're going to say pick up a hammer little bro. The second way is to help people and businesses implement AI models. If you guys think about it, with models getting more intelligent, less costly, and more capable in every way, shape, and form, you know, this future extension of our models today are soon to become essentially the highest economic good. So, if you guys can help people and businesses and specifically organizations that are high lever like NOS's, uh I don't know really powerful charities or even government organizations. If you guys can help them implement AI systems, you can get a share of that massive unbounded potential upside. Alternatively, you can also help raise awareness about the direction that this technology is going for people that might not have as much of a background or might not be as acquainted by it. You could help people prepare for a future that, you know, might look very differently, at least economically speaking, than our current one, similar to what I'm doing today. You could create video content all about how to implement systems in businesses where you see opportunities to build things that bridge gaps. You could help people transition from roles that are maybe heavily people based into ones that do interface more with robotic process automation. You could learn how to prompt these AI models in more efficient ways and then share those prompts with other people. These are all things that are going to be very

valuable. Although I should note these are going to be valuable for a much shorter period of time than something like uh real world jobs that interface with the physical world and have presence. I would think two to four years of major runway as opposed to the previous idea of you know 5 10 to 15 before the proliferation of robotics. And this is a major chunk of my own content. I'm thinking about various ways that I can help accelerate the transition to a more efficient economy. One is both the awareness of opportunities available. Another is the awareness of various windows that are closing that I think people should take advantage of. The third is helping people build and then implement systems that are driven by large language models, AI models, automated platforms and so on and so forth into their businesses and high leverage organizations. And then another is building an audience of people that I can effectively be a good consensus mechanism to um as the pace of change continues to increase. The third thing you can do to carve out a chunk of the future is building a brand around positive human qualities that might not necessarily be directly related to our economic productivity. One of the most

interesting phenomena for me over the last 20 years has been the advent of uh video game streamers. If you think about it, video game streaming has like a near zero economic value. Like what is a video game streamer producing for the world? Are they making our roads better? probably not. You know, it's a consumptive media. You're watching somebody play video games and they usually comment on it, but it's not the commenting that is really constructing our damn buildings or improving our GDP. So, the value of video game streaming is divorced from the economic productivity of it. In addition, video game streaming is also a really interesting situation where models and AI and tools and systems have actually already been better than human beings for quite a long period of time. And despite that, it's still thriving. The reason why is because the value that it provides is not like an economic good. The value that it provides to human beings, it's a sense of connectedness. community. Have you guys ever seen like video game streaming chats blow up when the fella or chick does something funny in the game? It's insane. You feel like you're surrounded by thousands of other people. And if you combine that with the fact that, you know, software has already been better than us at a variety of these games, but people still choose to watch human beings play them, I think you might be moving towards a realistic vision of what future careers actually look like. Like check out Magnus Carlson. Magnus Carlson is the best or one of the best chess players on planet Earth. But robots have been better than human beings in chess for I mean I think at least a decade now, probably significantly more. So what that means is people don't watch Magnus Carlson because he's the best at what he does. People watch Magnus Carlson and he's developed this fanatical audience. Because instead of providing the most economic value and objectively being the best at something, what he does is he extols human qualities and virtues while not directly related to the economy we find very admirable. For instance, he's funny. He has a sense of humor. He is courageous. He's brave. He acts quickly and decisively. Sometimes he's daring. He possesses courage. These are qualities that human beings respect and naturally understand in the people around them. We really like these things. And while historically they've been related to one's ability to produce economic value, they're not the end- all beall. Clearly situations like video game streaming with Magnus Carlson and you know millions of other people that do this thing show us that it is not necessarily our ability to make money or whatever that is the most important thing to people. people at the end of the day are just these human qualities which have been hardwired and genetically encoded in us to be admirable. So, if

you can get really good at these qualities and then if you can illustrate them and demonstrate them in a medium that has very high leverage, something like video game streaming or potentially some other media, um, you know, I personally think you have a very long and very healthy runway ahead. And I even think that might be an example of what our future economy might look like. You know, instead of us dealing in dollars and cents since AI models are going to be better at us than both producing these things, planning these things, building infrastructure to improve GDP and build roads and whatnot. Um, you know, I think a potential future for an economic system might be some sort of credit based thing where we extol positive human virtues and then get paid as a result of that. The last point I'll make on this is, and this isn't directly related to what I was just talking about, is I just want you guys to think economically about supply and demand. Over the course of the next 5, 10, or maybe 15 years, we're going to see trillions of agents, maybe quadrillions of agents. And each of these agents will be as economically productive as you or I. And so in a future where you now have a quadrillion agents and then you have just a few billion human beings, which is the scarce resource economically speaking? Obviously, it is the human beings, right? And so what is essentially human about that scarce resource that you can extract and that you can use to sort of predict where the future of an economy

looks like. Anyway, I hope you guys enjoyed that video. If you guys want help implementing and rolling out AI systems to high leverage organizations like I was talking about there, check the description. Consider joining Maker School. It's my daily accountability program where I show people how essentially to build these sorts of systems for businesses and other organizations uh and then improve both your leverage and then accelerate the transition to a more efficient economy. Always a pleasure. Looking forward to chatting with all y'all tomorrow. See youa.