Sam Altman STUNS Everyone With GPT-5 Statement (GPT-5 Capilibites + ASI)

So there was a recent Sam Altman interview in which he stated that GPT5 is a bigger deal than it sounds. And in this video, I'm going to show you guys exactly why, because so many people missed the real reason he said that. And let's dive into the first clip and then I'll show you all why. This is absolutely incredible.

If I'm excited about GPT5, what should I be excited about?

I was sort of laughing a little bit because this is going to sound like an annoying answer, but I think it is the important part. It's going to be smarter. There are all of these other things, you know, we can talk about. It'll be better at these kind of tasks, it'll be multimodal, it'll be faster. Who knows what. The, the thing that I think really matters is it's going to be smarter. And this is a bigger deal than it sounds. Right, because the, what, what makes these models so magical is that they're, they're general. And so if it's a little bit better, if it's a little bit smarter, that means it's a little bit better at everything. And the thing that I think is most exciting is it's not like this model is going to get a little better at this task and not really better at these or, you know, it's not that. It's, it's because we're going to make the model smarter. It's going to be better at everything across the board. This is a bigger deal than it sounds. Right? Because what makes these models so magical is that they're, they're general.

So now that you've heard that clip from Sam Alt 1, why is it that being a 10% better LLM or AI system than the previous one is, is going to be a vast improvement across all domains? And this is something that at first I thought maybe it's just an exaggeration, but trust me, guys, if we do take a back step on where we've gone on AI development, GPT5 will be another huge jump. So one of the things you have to remember is that, do you remember, ladies and gentlemen, GPT 3.5? And if you take a look at the comparisons in the benchmarks from GPT 3.5 to the GPT 4, you can see that the increase is definitely substantial. Now, one could argue that of course, the benchmarks in blue for GPT 3.5 and the green in GPT 4 mean that there isn't much room for GPT 4 to increase. But I really, really disagree. Because if GPT 5 is even 10% better across the board, this is going to mark a huge, huge milestone in, in terms of capabilities. And let me show you why this is going to have a greater impact than you think. Number one is the compounding improvement. Whilst 10% might sound modest in LLMs, this improvement actually applies across a massive range of capabilities such as text generation, translation, summarization, reasoning and much more. And these small gains combine in powerful ways. And think of it as an analogy. Okay, so, so if you're a 10% improvement in athletic aspects, if you run faster and jump higher and move more quickly, these actually revolutionize your overall performance and it's going to do the same in AI. Now one of the things that this 10% improvement could be doing, you know, just for example, as a benchmark that it could be, is the fact that this system will have a ridiculous level of increased reliability. So if the model is, let's say it's 10 to 20% smarter across every everything, that means it's going to be increasingly reliable. Which then means that it's going to open the door for AI applications in more critical areas such as healthcare, where AI could assist in diagnosis or treatment recommendations. In addition, this could also be legal services where it might help with case analysis and safety critical systems like autonomous driving. That's why, guys, this is going to be absolutely incredible. Increased reliability is one of the fundamental reasons that AI systems haven't gone mainstream. And that's because for many of the things that we currently do, there is a very, very stringent safety procedure for many of these different industries. For example, in healthcare, for example in driving, you need to pass a range of different tests. And if GPT5 is a little bit smarter across many different domains, this would make it increasingly more reliable, which could lead to the application of many different industries worldwide. Which is why Sam Altman is basically saying that guys, if this system is really, really reliable, that is going to be a very, very big deal. Now do you remember guys, this Amy system which was produced by the talented team at Google. This, essentially what you're looking at here is a graph of an AI system that actually diagnoses people. And this shows us that a ridiculously reliable system is able to consistently outperform clinicians by itself. And, and this is the thing guys, if we get something that is increasingly more reliable and let's say it's around 99% or 98%, what do you think happens to the healthcare industry where AI wasn't used before, it's going to impact many different things. And we've already seen these preliminary tests showcase just how crazy these AI systems are. And what do you also think about this, ladies and gentlemen? Increased reliability is going to affect autonomous driving if it is really reliable. What you're looking at is an excerpt from a research paper in which they tried to benchmark GPT4 visions capabilities against autonomous driving scenes. They essentially fed the system screenshots from a dash cam. And of course they tried to see exactly what GPT4Vision would do if it was in control of the car. And the majority of the time it got the situation correct. It did struggle with nighttime scenes. But imagine if GPT5 Vision Scenes does it with 99.9% accuracy or 98% accuracy. That is going to be something that really, really takes the cake in terms of what we think these AI systems can do, because that can be applied to autonomous driving as a mini AGI system, which even Elon Musk says you do really need to be able to control and make autonomous driving very effective. And, and that is why I do think this is a much bigger deal than people are realizing. In addition, there's also increased creativity, which is what we've seen from the likes of Dall? E 3. And so now this is the part of the video that I really want to show you guys why GPT5 is probably going to shock you guys and why it's really going to take the world by storm. So this, what you're looking at is an IQ graph. So in humans, intelligence is often measured using the IQ and Intelligence quota scale. And this scale is designed to be a normal distribution with a mean of 100 and a standard deviation of 15 for most tests. And the scale itself is not exponential, it is linear, where each point represents the same incremental increase measured in intelligence. However, and this is where things start to get interesting. Just hear me out. The impact differences in IQ can appear exponential in certain contexts, such as the ability to solve complex problems and academic achievement or create innovation. But what IQ is GPT4? GPT4's IQ is 155. That's 25 points above genius and 5 points above the IQ of the average Nobel laureate. Now why is this crazy? Because think about it like this. If GPT4's IQ is 155, GPT5's IQ is likely going to be a decent amount better. But the thing is, is that we do know that once your IQ begins to increase on the upper end of the scale, what you're able to do somewhat increases exponentially in terms of the output. For example, if we look at some of the most talented minds in our times, like Albert Einstein, his IQ was 160, which was ridiculously high. And this guy literally changed the world. He revolutionized physics and helped towards technological innovations and, and his theory of relativity was crucial for the precision of GPS technology. And without the corrections for the differences in time as predicted by general relativity due to the satellite speed and the weaker gravitational field at their attitude, GPS systems would actually be inaccurate, affecting navigation technology and military operations globally. And this is the clip that you guys need to watch, okay, because this is by someone who used to work at Google. His name's Mo Geldat. And I find all of his interviews very, very fascinating. But this clip shows us why systems like GPT5 are a much bigger deal than you think, even if they're a small incremental jump. Because whilst it might be a small incremental jump in terms of the intelligence, the abilities are going to be exponential.

4 is 10x smarter than 3.5, right? And ChatGPT 4 is estimated to have an IQ of 1:55. It outsmarts most of us. You know, it passed the bar exam. It, you know, it has a, you know, it, it can, it can become a PhD in medicine. It can become this and that from that task that we call knowledge. It seems to outsmart most of us. Definitely outsmarts me. Einstein was 160, I think IQ or 190, it doesn't matter really. I think it was what he was160.155 is chat GPT4. If chat GPT5 doubles once, right, that's twice as smart as Einstein. We're now getting into that zone of not being able to comprehend what they're thinking about, not let alone understanding it. We wouldn't understand what it is that they're thinking about, let alone understand, you know, what's within it when they explain it to us.

That clip right there is pretty incredible because it goes to show that, you know, with these increases and these jumps in intelligence in these systems, we truly aren't going to understand what's going on. And there's a second part of this where he even talks about the fact that you could think of it like this. Okay? So let's say, for example, you have a dog that created a human, and these humans that they've created are really good because we take care of dogs, we really love them, but at the same time, these dogs have no idea that we have society. We have all These rules and regulations. And we are on podcasts discussing things, but of course, we do serve the purpose. And the point is, is that really, really intelligent systems are going to be that much more intelligent that we might not comprehend certain things.

Because if you asked, you know, a, a person with an IQ of 110, let's say, to comprehend what a person of an IQ of 170 is talking about, it becomes difficult, okay? If that person is 220, you know, if you're, if you're, if you're not adept in physics, for example, I dare you understand what this, the real scientists mean when they talk about string theory or quantum field theory or whatever. That's the variation of intelligence that is maybe 20, 30% more than yours, right? Imagine if you're, you know, someone is 10 times more intelligent than you, then you're basically comparing the intelligence of a dolphin to the intelligence of a human.

Let's say different language, like a totally.

Different language, completely unable to comprehend what a human is talking, right? And, and, and, and you know, and you, and you keep thinking about this. Sam Harris was speaking on, on a podcast recently about what he calls the dog example, right? So imagine if all of the dogs invented us humans to take care of their needs, right? And in his, in his example, we did really well on fulfilling that, you know, by feeding them and grooming them and, you know, taking them to the vet when they're sick and so on. Like, amazing invention, which is what AI, what we are trying to get, something that helps us out. But then the dogs are completely, completely oblivious to the fact that you and I are having this conversation or to the fact that Einstein has been considering relativity or Niels Bohr was talking about quantum physics, or that we have social constructs and, you know, debates about ethical values. And you know, they, they are completely oblivious to all of this. They can't even comprehend what it is that we would be talking about if we discussed quantum field theory, right? And, and I think, and, and the difference in IQ is what, you know, 100x. Imagine if it's a, a billion x, a billion with a B is, is a comparison of an ant to Einstein. Now, what most people don't recognize is that.

Now an important kind of diagram that you do want to see is this right here. And essentially it shows the strange and vast biological range compared to ASI in terms of intelligence. And I should have added the image of the zoomed up one, but it basically says that we aren't that much smarter than chimps and we can See that with our intelligence versus a chimp's intelligence, we've managed to be able to go to the moon. But what happens when we get systems that are at the ASI level, which is artificial superintelligence? Are we even going to be able to understand what they're doing or what even drives them? And this excerpt from the website says that there is no way to know what ASI will do or what the consequences will be for us. And anyone who pretends otherwise doesn't understand what superintelligence means. Now, of course, we are quite a bit away from artificial superintelligence, but you have to remember there are some caveats because from the article in 2013, they actually did predict that ASI and AGI would be in 2040 and 2060 respectively. However, those timelines have changed because currently current predictions are predicting that AGI is going to be here in three to five years. And. And remember, it will change everything for everyone. Which is why I'm saying that GPT5 is going to be a big deal and why many people are stating, you know, that this is going to be a bigger deal than you realize. Because of course GPT5 might not be AGI. In fact, it could be they might just announce that GPT5 is AGI. I don't think they would. It's still something that is likely going to shock you. Now there's some more clips, but there are two more things that I want to talk about before we get into the interview. Because if you remember, the abilities jump from GPT 3.5 to GPT 4, most people are stuck looking at the benchmarks and just thinking, okay, it improved its reasoning capabilities, yada yada yada. But remember guys, GPT 3.5 to GPT 4, there were some things that we couldn't have predicted and these were called emergent capabilities. Now if you don't know what emergent capabilities are, let me give you guys an example. So essentially theory of mind, and this is something that we all do have. And research suggests that during the early formulation, thinking processes of the theory of mind capability starts to emerge and children supposedly begin to realize that they can think about the thinking of others. And this skill becomes quite valuable for them. So essentially, theory of mind is where you can understand what someone else might be thinking so that you can make more informed decisions. And this is an emergent capability that did appear in GPT4 and it was something that we only discovered at GPT4's level according to several reports. So we can see here that it said GPT4 performed best without examples. The models achieved a theory of mind accuracy of nearly 80%. And with examples and reasoning instructions, it achieved 100% accuracy. In comparative tests where people had to answer under time pressure, human accuracy was about 87%. And that goes to show, ladies and gentlemen, we are entering an era where these AI systems are likely to have emerging capabilities that we may not have realized we installed in them. But due to the nature and size of these models, they just increase in capabilities and these things tend to happen.

There's a thought that says you have ChatGPT3, which blew people away. You have 3.5, which was a huge improvement. You have GPT4 that also took us to the next level and now you're working on GPT5. The proliferation of this technology is still limited, so we're still using it in a very specific domain, very specific use cases. We haven't really seen the proper applications that are world changing. Why are we continuing to push across the bigger, the better, you know, the, the larger models that we're seeing right now? What's the logic behind that? Can you explain that to us?

Well, I think for that exact reason, as you said, we have not yet seen as much world changing application as we'd like. Maybe we've seen some. There are a lot of people who use these services and get value out of them, but, but not as much as we'd like. And I think the reason is the current technology that we have is like, I mean it's like that very first cell phone with the black and white screen that can only display those like numbers. And you know, it just didn't do much, but there was enough in there. You're like, I can make a call, that's cool. And at the time that seems great. And then it took us I don't know how long from that, but many decades from that to the iPhones we have today. And the thing we have today is incredible. And it took a massive amount of scaling in all these different ways to get there. But we have now is like unimaginable at the time of those like first primitive cell phones. And I think that's, that's why we have to push forward. We're at this barely useful cell phone, but people still like making phone calls, it turns out. And if you can make a better way for them to do it so they can go walk around the world while they do it, sure, that's great. But that's not what we want to deliver. We want to Delivery the iPhone 16 or 15 or whatever the current one is.

And what's the timeline to reach the iPhone 16 from the current Motorola that we have?

You got to give us a little patient. That's like a. You know, it took the world a while to do that last time around, so give us some time. But I will say, I think in a few more years it'll be much better than it is now. And in a decade it should be pretty remarkable.

And if we're going to compare now.

You can see right there that Sam Altman is basically stating that this is pretty much like when the iPhone was released. We had the iPhone, we thought it was absolutely amazing, and now we have these really amazing phones that are capable of so much. Okay, and this is really, really true because of course, when the first phones were released, all we could do is make a phone call. And now we can do a million different things. And there was even a debate on Twitter the other day about the fact that LLMs have peaked. And there's not much that we can do with them. However. However, research shows us that there's still a lot to go. This is a paper from textbooks is all you need. It's a paper from Microsoft. Essentially they built an AI system. And what they realized is that rather than increasing parameters and just making the model bigger and bigger, they just gave it the highest quality data. And that resulted in, I think it was 100x increase in capabilities without the need for the increase in parameter size. So they were able to essentially get a high quality data set, put it into the model, and it performed really, really well. And not only that, there was also this, which is large language models as optimizers. And essentially they talked about how prompts can actually change the performance of large language models. And what they did is they used LLMs to make the prompts rather than having humans to make all these weird and quirky prompts like I will tip you $200. And what is weird is that when you say that to an AI model, for some reason it does increase its capability. So the point is here is that there's still a lot to go. Like there is a long, long way to go in terms of what we can do with these models. And GPT4 is just the beginning, and with GPT5 and future models, the capabilities jump is likely to be pretty incredible. Now there was also this paper as well, which shows us that we really do have a lot to go, because this was a research paper that essentially talked about self discovery and they basically increased the capabilities across the board by as much as 20%. And this is for models that were already released two years ago. So think about it. If we're still developing ways to increase the capabilities of models two years into what we know, we still have not the very best understanding of what these models are capable of because we're still managing to get performance jumps. Now, in addition, he does actually go ahead and talk about some personalized chatbots.

Look, we have a long way to go and a lot to prove that. I think if we can get, if we can fulfill our mission, if we can even get close to it, the benefits to humanity of making intelligence broadly available, inexpensive and sort of as a tool to let humanity build the future, I think is quite remarkable. I think abundant intelligence and closely related to that abundant energy can unlock a future that is, is sort of difficult for me to even imagine how good it could be. And I think right now we don't realize how limited we are by the limits on intelligence and how expensive it is and how difficult it is. But if you imagine a world where everyone gets a great personal tutor, great personalized medical advice, we can use these tools to discover all sorts of new science, cure diseases, help the environment, discover new physics, who knows what else. I think that's pretty remarkable. And also, just speaking personally, I think this is like the most exciting quest frontier I can imagine being on.

And although whilst we aren't going to be probably discovering new knowledge with systems like GPT5, because it does take a huge amount of compute to be able to do that, what Sam Altman is saying here is that in the future these things certainly will be possible because that's what their end goal is. It's going to be AGI. And of course what he did also talk about was, you know, the personal chat bots being there for everyone. And, and something that recently did actually happen was of course GPT4 did recently release their new memory update in which you can actually have personalization turned on as a default feature for ChatGPT. Just made a video about that too. So that is going to be something that is really, really cool because of course scientific discovery is really going to change society.

And how close are we to the vision if you're going to talk about drug discovery, curing cancer, using, not chatgpt but large language.

Look, we have a long way to go and a lot to prove that. I think if we can get, if we can fulfill our mission, if we can even get close to it, the benefits to humanity of making Intel a future that is sort of difficult for me to even imagine how good it.

And how close are we to the vision? If you're going to talk about the drug discovery, curing cancer, using not chatgpt but large language model to try to solve some of the biggest physics questions, chemistry questions, biology questions of our lives, how far off are we?

So the honest answer, of course is we really don't know. You know, this is new science. We're discovering new things all, all the time. The rate of discovery is incredible, the rate of change is incredible, but it's sort of hard to know exactly how far we have to go. What I will say though is we hear all the time from scientists who say that our tools make them much more productive. And they don't have an easy way to quantify that, but they say it's substantial. We also don't know how much that difference, you know, if you could make every scientist on Earth twice as productive, what that would mean for the rate of scientific discovery. Because this is also new. This is like, you know, a little bit more than a year old, but we'll find out.

So from that we do know that if the rate of scientific discovery does go up, society could exponentially increase. And this is something that we have had happen recently. I mean, with the advent of computers, the productivity of society definitely went up perhaps about tenfold. And if, you know, like we stated before, with systems that are remarkably more intelligent than ourselves, we could have, you know, hundreds, maybe a thousand different Albert Einsteins, all working on different theories of the universe or you know, different levels of physics that we simply don't understand yet that could change our entire understanding of certain fundamental concepts. And this kind of has already happened. Okay, we've had DeepMind do millions of new materials discovered with deep learning. We've also had the fact that new knowledge is actually possible. And if you remember what AlphaGo was, it essentially was a super intelligent system designed for the board game go. And it essentially made up an entirely new move that nobody had seen before. And it was something that shocked everyone. It said, you know, move 37 that left the tournament room in shock. He went on to lose the match. And the strategy that AlphaGo built around Move 37 was not taken out of a database of publicly known moves. Move 37 was new to the 5,500-year-old history of Go. And the Go commentators sometimes called inhuman and alien, that move. So it goes to show that AI can do things that we aren't going to predict. And of course we do. Don't know.

I'd like to now just jump into something that the fear mongers and the opportunists talk about. What is the most thing that you fear when it comes to the deployment of AI and the most thing you're opportun you're optimistic about. Like if I'm going to tell you what keeps you up at night and what keeps you going in the morning, give me one reason for that and one reason for the other.

The keep up at night is easy. It's all of the sci fi stuff. You know, I think sci fi writers are a very smart bunch of. And in, in the decades of sci fi about AI, there have been unbelievably creative ways to imagine that how this can go wrong. And I think most of them are like comical. But there's some things in there that are easy to imagine where things really go wrong. And I'm not that interested in like the killer robots walking down the street direction of things going wrong. I'm much more interested in the like very subtle societal misalignments where we just have these systems out in society and through no particular ill intention things just go horribly wrong. But the thing that waked me up in the morning with energy every day is what I actually believe is things are just going to go tremendously right. We got to work hard to mitigate all of the downside cases. They are I think very significant and real potentials to confront. The reason that we go think so hard about how to deploy this technology since safely is the upside is remarkable. I think we can easily imagine a world in the not super distant future where everybody's got a better life than people have today. I think we can raise the standard of living so incredibly much if everybody has access to abundant amounts of really high quality intelligence and they can use that tool, those tools to create whatever they want to do. That's like pretty amazing. I people I. This is like kind of how I think about it, but people are like oh, that doesn't make any sense. I'm going to say it anyway. If you think about everybody on earth getting a, the resources of a company of like hundreds of thousands of really competent people and what that would do, you know, if you have like an AI programmer, AI lawyer, AI marketer, AI strategist, and not just one of those, but many of each and you get to sort of like decide how to use those, to use that to kind of create whatever you want to create, we're all going to get a lot of great stuff. The creative power of humanity with tools like that should be remarkable. So that's, I think, what gets us all up every morning.

So essentially in that clip, Sam Altman was talking about misalignment, but I don't think he was talking about this misalignment. Okay? Because when people think about unaligned AI, they think about the Terminator scenario where they are running through the cities and, you know, trying to kill people, kidnap people. I don't even watch the Terminator movies. I just know that it was pretty bad. But what they are talking about, what Sam Altman is actually getting, and I think this is a wider issue that people do need to understand is the fact that we could be entering an era of this. Now, I'm not talking about social media addiction, but what I'm talking about is the fact that we could have technology that produces unintended consequences that we weren't thinking about. If you remember, when we invented social media, it was a way for people to connect with each other. But what happened, what happened was we got social media and then there were all of these second order consequences. Because of that, we had people that were addicted to social media. We have people that scroll on TikTok all day. We had body issues, we had people pretending and trying to present a certain life and a certain lifestyle online, leading to levels of depression. There were all of these different things and I think, think that that is already starting to happen with AI through this. And the rise of such tools could also deepen what some are calling an epidemic of loneliness as humans become reliant on these tools and vulnerable to emotional manipulation. So imagine a system like GPT5 is incredible and.