Sam Altmans Stunning Prediction Is Deeper Than You Think

So today, Sam Altman dropped another blog post that most people are probably going to overlook. But it's hard to grasp just how important this blog post is and the key things that he said that will guide us in actually predicting where the future is now. Before diving into the blog post, I want to remind you all of another blog post Sam Altman wrote in 2021 titled Moore's Law for Everything. And the reason I'm bringing up this one called Moore's Law for Everything is, is because this blog post, if we take a look back to the time that it was made and take a look at the things that are going on in today's society, it is remarkably accurate in its prediction. One of the things that he actually said that we can take a look at was that he stated that in the next five years, computer programs that can think will read legal documents and give medical advice. And this is essentially what we are seeing right now. Documents and programs and all these kinds of things are actually happening on a day to day basis. And medical advice from systems like OpenAI's 01 and Google's medical models are slowly being integrated into wider society. So when we take a look at the fact that Sam Alan's previous blog post from March 16, 2021, and it's been remarkably accurate so far, we can take a look at this new one titled the Intelligence Age, which was only published just yesterday, and truly understand where the future of AI is headed. Now. I think in order to truly grasp what's going on here, we need to take a look at the first sense. And he said here, okay, and try not to get too excited. But he says in the next couple of decades, we will be able to do things that would have seemed like magic, okay, and magic to our grandparents. And the reason I think that that is so fascinating is because when we truly take a look at the previous iterations of technology, when, when we think about how crazy it is that we have planes that can fly, we have fighter jets that can fly 1000, 2000 miles per hour doing sonic booms. We've got technology like AI currently that can write things completely from scratch. We've got TVs, iPhones, holograms, drones, we've got all these kinds of incredible technologies landing on the moon, Starlink. I mean, the list goes on and on. The Internet and smartphones have literally changed how we communicate and access information. And some people in the early 20th century couldn't have even dreamed or even imagined this kind of technology. And if we think about this, our grandparents are probably amazed by the kinds of technologies that are around right now, which means that when we get older, we're probably going to have our minds blown by the technology that's probably being developed at that time. Now, he says that of course this phenomenon is not new, but it will be newly accelerated. Essentially referring to the fact that every company right now is really trying to invest billions and billions of dollars in into their AI infrastructure in order to get a piece of that AGI pie. It continues on to state that people have become dramatically more capable over time. We can already accomplish things now that our predecessors would have believed to be impossible. We are more capable not because of genetic change, but because we benefit from the infrastructure of society being way smarter and more capable than any one of us. In an important sense, society is itself a form of advanced intelligence. Our grandparents and the generations that came before them built and achieved great things. And of course, all of those things we can build on top of for the next generation. That basically accelerates the next generation, which is why technological progress, it kind of goes like this in a graph. It kind of just like goes like this because each generation sets the stone for the next one. And it's like, I guess you could say exponential progress in a sense. So he says right here that the story of progress will continue and our children will be able to do things that we can't. And it won't happen all at once. But soon we will be able to work with an AI that helps us accomplish much more than we ever could without AI. Eventually we can have a personal AI team full of virtual experts in different areas working together to create almost anything we can imagine. So this is where we get into another prediction surrounding AI agents and what they're going to be able to do for us and how. I might as well not really say prediction. They're essentially creating these products, so I'm pretty sure that they know what's in the pipeline. When you think about it, this isn't a blog post by some futurists. This is a blog post by the guy who's at the head of the Frontier Labs. So when he says a team full of virtual experts in different areas working together to create almost anything we can imagine, this sounds rather accurate as this actually refers to the leaked OpenAI internal document which they are using to track AI progress. If we take a look at this, we can see that it was only recently that we had reasoners, which are the human level problem solving with the Strawberry model and of course the next big thing. And I think The AI still isn't in the mainstream yet because we don't have this. I think that this is going to be the magical moment when we actually have an insane level of individuals that really, really start to get what AI is able to do. And I think once we actually have that broad scale area where we see that, okay, individuals are now starting to realize that AI is out there and it's performing things on our behalf, that's when the acceleration is going to be truly incredible because it's quite like we're doubling, tripling the population of individuals that are working effectively and working non stop. 24 7. Then of course, this is where he says our children will have virtual tutors who can provide personalized instruction in any subject, in any language, at whatever pace they need. And I think this is one of the things that is widely underestimated because not only are students going to be able to learn a lot more, personalized learning is, I think, in many cases a lot more effective because you can learn at your own speed, at your own pace. You've got one on one with someone that can explain concepts to you. But I think the fact that previously it was only certain individuals that could afford private tutors. But now we're going to get to an age where pretty much anyone is going to be able to afford their own private tutor, perhaps just for $20 a month, and they're going to be able to absorb as much knowledge as they want from pretty much any subject. If you haven't already realized, AI can break down complex concepts and explain them to you in a way that is really easy to understand. You can talk to it as many times as you want, and it's not going to get upset, it's not going to get frustrated, it's not going to get tired, it's going to be 100% completely understanding. Now, I was thinking about this quite a lot and I was like, okay, this is actually pretty cool. But we also need to think about the fact that like, this isn't just for children. We're not just going to have personalized tutors just for kids. For those of you guys who are in the working world, in the business world, you guys actually now have tutors that pretty much understand most advanced concepts. And you're going to be able to use those tutors to perhaps advance your career, advance your business and advance your side hustle, provided you have the motivation to engage in lifelong learning. I think previously, of course, you've always had to read a book or engage in certain educational, traditional methods. But I think this kind of learning is going to truly open up a new area for individuals that learn in a different kind of way. And I think this is really, really exciting, especially for those who want to move further on in life. Then of course it continues to state that we can imagine similar ideas for healthcare and the ability to create any kind of software somewhere can imagine. So I think this first sentence right here is probably one of the most important ones. If you didn't know, in the UK alone and I think in America it's around 2, there's 200,000 deaths annually in the US of A. Where there are just medical errors. And you can see that there are more than 200 million medication errors in the NHS per year, which is, you know, researchers have realized, which is pretty insane when you think about it. And we think about like how traditional, you know, healthcare systems are. I think once everyone has access to a personal healthcare agent that's able to, you know, coordinate with their doctors, look at all of their vitals, understand all of their medical history and really look at their health care in a personalized way, in, you know, many different reasoning ways, you know, looking at 10,000 different outcomes for this person, I think it's just going to be absolutely insane the type of medical health care that you get. And I think the craziest thing about this is that this is not even going to be that expensive. I think it's going to be really comprehensive and over time I think it's just going to become remarkably cheap. And even when we look at the most recent models, they've been excelling on healthcare benchmarks. So it's no doubt that this thing of ideas with AI for healthcare, it's truly going to be better. And of course this is one of the more positive things to come from AI. Now we're to going, one of the things that we also have here as well is the ability to create software someone can imagine, like any kind of software someone can imagine and much more. And this is something that people might not realize is slowly happening, but there is a quiet revolution going on in the software development space. And this isn't a video about software developers are dead or gone or this. No, no, no. This is just a small tidbit where I'm referring to the fact that there are companies out there building, you know, these kind of AI agents that are going to be able to build software that is quite personalized. And I think this is going to be one of the largest shakes up the economy has seen because AI is able to Create personalized software that is able to do many internal tasks for you. So I'm wondering, is this going to eat the SaaS economy? I mean, there's many different implications for this because not only does it allow individuals to become more effective because they're able to develop their own pieces of software for personal purposes, but this is just something that's pretty crazy. Now take a look at this repli AI agent, because I was supposed to cover this, but I unfortunately didn't get to. But this is, you know, an AI agent that's able to literally do everything for you from start to finish, all the way up until deployment. And if we think about the fact that this is the early days, how is this kind of technology going to be 10 years from now?

This is actually a prompt from a TED AI talk I gave about 10 months ago about agents. We didn't have the product back then, but now let's test the real product with the same. With the same prompt. Make an app of local landmarks based on my location. The agent presented me with a plan. If this plan is fully editable, I just want to add the ability to search. One of the cool things about the progress pane is that you can learn what it's doing. So we don't want this ever to be sort of obscuring what the agent is doing. It should be fully transparent. This could also be a great way to learn how to build things because they agent is sort of explaining itself as it's doing it. That's the app. Let's see. That's exactly the kind of landmarks I would expect to see in Foster City where we are recording this from.

It's awesome.

And so here you can give it feedback, you can iterate with it, you can give it more features. But I'll just say, yes, it looks great because it works. It's exactly what I asked for. And now the agent will move to the next step. And the next step is adding search. We spent a lot of time building the plumbing for this. A lot of what the agent is doing is actually doing the same kind of actions that you would do in the ide. It's editing the code in a similar way that you would edit it. It's installing dependencies using the same tools that we expose to users. So all the UI that you see in the replit, you know, and love is exposed to the agent. So the agent is sort of a multiplier user in the same same environment. We're about to deploy to the cloud. And when we deploy this app, there's, there's no difference between what we built here. You don't have to configure a server, you don't have to do any of you don't have to configure a database. All of that is configured for you. When I had this idea, I had it when I was traveling and that happened on my phone. And actually I can start the project on my phone. I actually finish the project on my phone as well. Anytime you have an idea, it, it doesn't have to be this fleeting idea. You can just put it into Replit and I'll start making it for you. And then we got this initial prototype, but I can also go and browse the code myself. I can use replit's LLMs, or you can use your favorite LLM. The code is all yours, it's there. And you can continue using the agent. You can switch to Replit AI, you can do whatever you want with it. You have five full authority over the code that the agent just created. And there you have it. Replit Agent created repl, configured the Python environment, installed dependencies, wrote the code all in two minutes. And we got an app running that we were able to deploy to the cloud and now anyone can use it. So this is Replit Agent. It's available in early access and would love your feedback.

That is one of the things I always want to reiterate is that when you see AI and it looks pretty bad, remember how bad Mid journey was? Remember how bad these images were? Remember how bad videos were? I mean, when we take a look at how crazy this kind of, you know, acceleration is, we have to truly understand that, okay, this isn't slowing down now. He says here that with these new abilities we can have shared prosperity to a degree that seems unimaginable in today's future. I mean, in the future and everyone's lives can be better than anyone's life is now. Prosperity alone doesn't make people happy. There are plenty of miserable rich people. But it would meaningfully improve the lives of people around the world. Here is one narrow way to look at human history. After thousands of years of compounding scientific discovery and technological progress, we have figured out how to melt sand, add some impurities, and arrange it with astonishing precision at extraordinarily tiny scale into computer chips, run energy through it, and end up with systems of creating increasingly capable artificial intelligence. Now, this statement here is one of the funniest statements ever, because there is a video from someone on YouTube and I think everyone should watch this to understand how crazy technology is. It's a video that basically shows you how to get, like, a CPU completely from scratch. It's pretty funny. I'm only going to play the first 10 seconds, 20 seconds of the clip. But it's a video that I think is so important, because when we see how technological innovations can push the boundaries of what we even thought is possible, we kind of need to think about where we even came from. And it's quite hard to fathom. So, yeah. Now, here's the next thing. Okay? It says this may turn out to be the most consequential fact about history so far. So if you've been half asleep through this video, please just pay attention to this part, because I would say that this is probably the most important statement, okay? Sam Altman says that it is possible, and I didn't mean to do that, that we will have superintelligence in a few thousand days. Now, of course, he, you know, caveats it by saying, it may take longer, but I'm confident we will get there. This statement is absolutely insane. It is possible we will have super intelligence within a few thousand days. Now, the first implication for this statement that I immediately thought about is I didn't even think about superintelligence. I was like, hold on. That means that AGI is most certainly coming a lot sooner than we think. People and critics have always dismissed AGI as some kind of distant pipe dream, but this statement actually puts AGI in a new light. If they are talking, okay, broadly about superintelligence, which on the AI scale, actually comes a lot further after AGI, it means that AGI is almost certain or probably already achieved, which actually leads me to think that we could get AGI and by early 2026 or early 2027, which is not that far away, which is only two years away, which is pretty insane to think about. And I think this also means that acceleration, like I said before, isn't slowing down. If you remember, a few months ago, we did get a few articles that they were stating that, look, AI is slowing down, deep learning is hitting a wall, yada, yada, yada. And I think I actually made a video stating that, look, AI progress is not slowing down at all. Like, you guys have it wrong. And I tried to explain to you all that just because we're not seeing a thousand new models every single day, it doesn't mean that AI is slowing down. These people are making remarkable discoveries internally, and this is just truly accelerating the entire AI progress. And you have to understand that things are only Going to get faster from here on out. OpenAI kicked off the race to AGI, and it's only a matter of time before the world is inevitably changed. Now, after I thought that, okay, look, AGI is clearly within the next one to two years, because if they are broadly talking about superintelligence is clearly here, we have to think that, okay, that means that they certainly did see something. Okay, and that leads us back to the infamous quote, what did Ilya Sutskeva see? Now, Ilya Sutskova has a new company called, you know, Safe Superintelligence. And I think that that means that this statement now holds a lot more weight. The statement that super intelligence is within reach. Is it like echoes now? It feels a lot more, I guess you could say, pronounced, because Sam Altman is stating that, look, we're only going to have it in a few thousand days, which means that Elias at Skiva probably has identified the path. Whether it be this reinforcement learning with chain of thought or, you know, that Q staffing, I think that showed them that all they need is more compute, and then they will probably get there. There's a million videos on deep dives and stuff, and I'll leave a link down to all of that stuff. But I think it's quite important to understand that right now the bottlenecks aren't really in actually figuring out how the technology works, but just getting the energy and the compute to actually power this kind of stuff, which is quite crazy. And when we think about the superintelligence, I always like to refer to the document Situational Awareness, which is where it says it's strikingly plausible that we'd go from AGI to superintelligence very quickly, perhaps in one year. And I think that this kind of statement is not to be messed around with, because this is, of course, from a former OpenAI employee. And I think this graph where you see that, of course, we have an automated AI researcher. And then over time, things start to exponentially increase with, of course, science and technology, robots, military edge, gdpl, all of these things are just going to completely explode. I think that the future is just. It's actually, you know, or so unpredictable that it's incredible. I mean, it's just, you know, incredible. And the worst thing about this is I remember so many individuals were stating that, oh, this guy, all he did was draw a few straight lines on a graph, and now he's claiming that he knows the future. I mean, the thing is, is that when we actually take a look at who Leopold Aschenbrenner is he's someone that's not some random person. He is someone that literally worked on the front lines for aligning super intelligent AI with the legendary Ilya Sutskeva at OpenAI. So if we aren't listening to people like him on the future of AI, who is it that we're going to be listening to? I mean, I truly don't understand. And the crazy thing about this is that across the consensus in terms of predictions for future developments and inflection points and turning points, It's a consistent 2027-2030 that we get this future world that's going to be drastically different than the one we are in now. Now, what's crazy about this is that he then says, how did we get to the doorstep of the next leap in Prosperity? And in three words, deep learning worked. In 15 words, deep learning worked, Got predictably better with scale, and we dedicated increasing resources to it. So I think this one is really cool because, of course, we can see that he's speaking about, of course, the technology which showed us that, look, all we need is more data, better algorithms, more compute, more scale, and we can do incredible things. He says that's really it. Humanity discovered an algorithm that could truly learn any distribution of data, or really the underlying rules that produce any distribution of data to a shocking degree of precision. The more compute and data available, the better it gets at helping people solve hard problems. And I find that no matter how much time I spend thinking about this, I can never really internalize how consequential this is. Now, for me, the superintelligence thing, it wasn't the craziest thing, this was the craziest thing for me was that he said, although it will happen incrementally, astounding triumphs, fixing the climate, establishing a space colony, and the discovery of all physics eventually will become commonplace. I don't think people truly understand what this statement means. The discovery of all physics will be eventually become commonplace. Guys, do you remember what happened once we discovered a few laws of physics? We were able to, like, fly guys. Like, that's insane. Like, sometimes I still see planes and I'm still like, this thing makes no sense. You know, it would look like magic to people, you know, 200 years ago. Look, like absolute, like, our phones and stuff like that. So once we're able to figure out, you know, a few things, there's only a few, like, bottlenecks. And, like, once we, you know, discover them, like, it just. It just remarkably changes how the future is in terms of the Entire fields that we're going to be in, the entire, you know, the way the world works. I mean, like, if you tried to explain current society to someone 220, 50 years ago, it's like, it's quite hard, you know, to fathom the smartphone, the Internet, the connectedness, the travel. I mean, I don't know, my mind's getting kind of blown by, like, trying to explain all of this, but I think the discovery of all physics, I mean, I can truly understand how it happens. I mean, I think once you have an automated AI researcher that's able to truly, like, get all the context from every single research paper and just research in certain areas 24, 7, and we're able to just duplicate that like a million different times, we're going to be able to condense, you know, years of scientific research down to, you know, pretty much a few years or even a few months once the compute becomes available. So I think it's going to be really incredible to see that kind of progress happening in our lifetime. Now, of course, there is no good without bad. And Sam Altman talks about here, he says, as we've seen with other technologies, there will also be downsides, and we need to start working now to maximize AI's benefits while minimizing its harm. As one example, we expect that this technology can cause significant change in labor markets, good or bad, and in the coming years. But most jobs will change more slowly than people think, and I have no fear that we'll run out of things to do, even if they don't look like real jobs to us Today, people have an innate desire to create and to be useful to each other, and AI will allow us to amplify our own abilities like never before As a society. We will be back in an expanding world, and we can once again focus on playing positive sum games now. I think it's always good to talk about, you know, bad things that are going to happen. Of course, the job market is going to be pretty insane once AI, you know, eats around. Like, I think, you know, it's going to affect, like, 60% of all jobs around the world. And I think it's going to be quite. There's going to be a weird period where people just don't have a job, like, you know, a lot of people, and it's going to be, like, really weird. The economy is definitely going to change. It's something that I talk about in my community, how you can, you know, most certainly take advantage of this transition because it's definitely going to be a transition period. And, you know, there's going to be some winners and some losers, and I think I would a lot rather be on the winner side. And if you're in AI, you're definitely early. Like, you're really early. I think the only time you're going to be late to this place is if you are, you know, only, you know, taking action in the next two to three years. Because once we get agents, the general public is going to truly understand what AI is able to do. So I think that that is, of course, something to discuss, but what's crazy. Okay, And I saw this on Twitter, and once again, I'm not like an AI safety person, but I do think this is kind of interesting, and I think it's worthwhile to be mentioned. Okay. Someone, you know, referenced the fact that that Sam Altman's blog from 2015 is a little bit different to the one from 2024. Sam Altman says that development of superhuman machine intelligence is probably the greatest step to the continued existence of humanity. And then in 2024, he says this technology can cause a significant change in labor markets, good and bad, but most jobs will change more than people think. Okay, so this tweet is basically saying that. Look, before he was tweeting about superintelligence, now he was stating that it was dangerous, and now he's stating that this is a good thing, that his company's there. I think either it's one or two things. I think either one, we have the sense that, you know, maybe they've solved the interpretability problem, which is where they truly understand what's going on in AI. I'm not so sure of that because I think that that's quite hard, and that would require incredible breakthroughs. I don't think that it's possible. I just think that it's really hard, especially with something that's smarter than you. Or I think it's. The second issue, which is probably more reasonable, is that they might just be avoiding this kind of AI safety thing because they are raising money again at a huge valuation. So I think they're trying to avoid anything that could reduce that valuation and scare off investors. And, I mean, if you've seen the recent funding round, it's been absolutely incredible. OpenAI has been flooded with investors, and they've had to put in, like, $250 million minimum. You can see here, it says, investors pile into OpenAI 6 billion funding round in unprecedented bets. So this is really incredible. Like, I can't. Like, I mean, I'm not going to lie, guys, if I had that kind of money, I certainly would be piling it into OpenAI, because they are about to eat, like, you know, just billions and trillions of dollars. So I think it's going to be incredible. So with that being said, let me know what you think about this document. I know I was going to make a video on this yesterday, but unfortunately, my voice just disappeared. So if you guys enjoyed the video, I will see you all in the next one.