AI News : OpenAIs SECRET Chip Project, SearchGpt, AI Race Heats Up, Stunning Humanoids

So AI news is on fire. So let me get into some of these stories. One of the stories just coming out today is the fact that OpenAI is planning to build its first AI chip in 2026. It says that OpenAI is reportedly working with TSMC and Broadcom to build an in house AI chip and beginning to use AMD chips alongside Nvidia's to train its AI. Reuters reports that OpenAI has, at least for now, abandoned plans to establish a network of factories for chip manufacturing and instead the company will focus on in house chip design. And OpenAI for months has been working with Broadcom to create an AI chip for running models which could arrive as 2026, reports Reuters. Now, I think this news is bigger than people think because this might just seem like some boring tech news about the back rooms of the chip design, but let's actually take a look at things here. It says that they're planning on getting this chip deployed for 2026 with dates in mind. I always understand that dates are important with AI timelines. Many people have speculated, and I think it's now common consensus that one of the bottlenecks to AI development and AI capabilities is of course, the compute capabilities. Now Nvidia's chips are great for running AI, but what happens when we get specialized AI chips that are completely designed from the ground up that are built to train OpenAI's models? Now OpenAI has been working on this with Broadcom, which is a company that designs these AI chips for months. And if this is true, this could be part of those wider predictions where you see people predicting, predict things like, okay, AI is going to get crazier because the timeline is going to increase and we have these exponentials. Like when we think about all of that data and all of the different algorithmic improvements, the improvements in architecture, imagine how good these specialized chips are going to be that could arrive as soon as 2026. I mean, that is going to be pretty crazy. And this is all because the company basically relied almost entirely on Nvidia GPUs for training. But of course, the craziest thing about OpenAI, and this is why there are actually many product delays, is that you can see that Nvidia's next iteration of their Frontier chips are completely sold out for the next 12 months. Guys, imagine you're trying to build a company and the thing that you need to build your company and the thing that you need to develop your product is, is sold out completely for the next 12 months until the end of 2025. And you're competing with customers like Meta, Microsoft, Google, Amazon and Oracle. Okay, the chip demand is absolutely incredible because these companies do not want to be behind at all when it comes to training AI models and providing inference for these AI models. Now I think this is just the of course, ASI play, the AGI play, which is where these companies are not just piling into the AI hype. These companies are thinking, you know what, AGI is going to be real, ASI is going to be real. And the companies that are going to be the top companies are going to be the ones that have enough compute. And I can't remember who talks about it, but someone said there's going to be a divide between the compute rich and the compute poor. But if OpenAI can actually start developing their own custom chips with Broadcom to build an in house AI chip, I think we're going to see some remarkable level products. It was also recently that we literally got the statement from OpenAI themselves about their next frontier iteration of models. I'm not entirely sure if it was GPT5 or whatever it is that they are working on over on at OpenAI, but they did say that what they are working on is constantly blowing their mind. Now that brings me to today's sponsor of the video I'm going to be showing you guys. The quickest way to get started with an online business is using the insane power of AI. Introducing build you'd store AI. This is an amazing AI tool that lets you create fully functional online stores with products ready to sell inside in just minutes, completely free for the first few days. No more excuses, no more complicated setups, just pure profit potential. So just look at how simple this is. You sign up, you then choose a niche that interests you, anything from pet supplies to yoga gear and build your store. AI handles the rest. It's that easy. Then all you need to do is connect your store to Shopify. And the good news about this is that you can actually get started with Shopify for just $1 for the first 30 days. Then integrate Autods for automated order fulfillment, which essentially is just a fancy term of saying you don't need to lift a finger when someone buys from your store. Autods handles the entire process from sourcing to to shipping. Talk about lazy profits. And just like that, you have a professional online store ready to go with products inside. Check out this example. All built with the power of AI. So what are you waiting for? Click the link in the description below to get your free AI store builder today. This is your chance to create A thriving online store without breaking the bank. This was something that I covered in another video where the OpenAI CFO Sarah Fryer said that AGI is closer than most think. And if AGI is what most people to believe is going to be in 2029, that would bring the estimate down to probably around 2027. Because I would say that most people have the consensus that this technology is going to be here by the end of the decade. And that would mean that a lot of people are going to be in for a shock when we do get that monumental technology. And the craziest thing about this clip was the fact that she said it would blow your mind to see what's coming. And I remember when I made this video, a lot of people would think that it's clickbait, but as you can see, she literally said the stuff is blowing her mind. Now I made a video on this. But basically, guys, when you think about the kind of technologies we've seen, if you went back two years and showed someone GPT4O advanced voice mode, their head would have exploded. If you showed them sora, their heads also would have just like exploded as well. So I can't imagine what they are working on if it's going to blow our minds too.

If Sam were sitting on the seat, he would tell you AGI is closer than most think. But I would.

He would.

What would you say?

I would agree based on what I'm seeing, like one of the best meetings I get to go to once in a while is the research meeting. And it would blow your mind to see what's already coming and what. As we have learned how to take Reasoning models like O1 preview on top of GPT models and the interplay between those, you're now really starting to see some incredible outcomes, PhD level outcomes where you have in your pocket human intelligence, that is PhD level in physics and biology and chemistry, in English literature, like whatever the job is you need to do. If you're a healthcare researcher, if you're a banker, if you're in education, the tool that you are now carrying the power there.

So I don't know about you guys, but I can completely believe them when they say that this would blow your mind. Because I mean, A1 blew my mind, Advanced voice mode blew my mind. And I don't need to get into all of those crazy things, but when we think about the kinds of things going on here, OpenAI are clearly ahead. Now, OpenAI might be ahead of everyone else in the US but. But some people are starting to question whether or not this will be the case forever, you have to understand that right now there is a global war going on in terms of these countries and who has the most power. It's an unspoken war, but one that does exist. And the National Security Advisor, Jake Sullivan said that the US needs to build tens or even hundreds of gigawatts of energy of infrastructure to to power AI data centers or risk falling behind. But AI is basically resetting the boundaries for what companies are going to be able to do. And what that means is that this enables a level playing field for anyone to take the cake. Right now nobody has AGI and nobody has asi. It's basically racing like the Manhattan Project to first develop nukes. And if another country that is more powerful decides to accelerate faster than the US does, then we could be at the perils of that technology. And I know it sounds like science fiction, but when you look at the amount of technology that we've seen so far, this is one of my major concerns. How does the US ensure that it not only manages to advance faster than other countries, but not so fast that it doesn't have catastrophic effects, but for those within the country.

Our competitors also know how important AI leadership is in today's age of geopolitical competition. And they are investing huge resources to seize it for themselves. So we have to start upping our game and that starts with people. America has to continue to be a magnet for global scientific and tech talent. Next is hardware and power. Developing advanced AI systems requires large volumes of advanced chips. And keeping those AI systems high coming requires large amounts of power on power. The Memorandum recognizes the importance of designing, permitting and constructing clean energy generation facilities that can serve AI data centers. So that the companies building world leading AI infrastructure build as much as possible here in the United States in a way that is consistent with our climate goals. One thing is for certain. If we don't rapidly build out this infrastructure in the next few years, adding tens or even hundreds of gigawatts of clean power to the grid, we will risk falling behind.

So I'd love to know your thoughts about the dynamics between the United States and many other nation states. I think that the most obvious race that is going on right now is a race between US and China. It will be interesting to see how these companies manage to navigate a tight ecosystem that manages to let the government understand what's going on. Because one of the things that I realized that when looking at even the Senate hearing was that a lot of these top AI labs actually have a Lot more talent than these governments. I think that when we actually look at the kind of talent in these companies, it's quite clear that I think the government might actually be behind on this. And they actually spoke about this in another video that I did. And in this video I basically spoke about how the ex OpenAI employees are basically speaking about how this technology is going to be absolutely incredible. And this was a Senate hearing where you had these people from the government talking about how they're going to potentially regulate AI and ensure the safety of AI while ensuring that it accelerates. Now, I know a lot of people were basically stating that we need to get a lot more people on board, and I think that is certainly true. It's clear that there is a clear disconnect between what is going on privately inside those companies and, and what's being understood publicly in terms of the governments and the outside world. Now, the reason I actually covered this one, now the reason I actually covered this one because the person you're looking at is Helen Toner, who actually served on the board of OpenAI when the much drama was going on. I couldn't even speak properly there, but there was just so much drama going on at OpenAI that Helen Toner was someone that was actually on the board at the time. So she has the perfect insight with. With as to how far along they were at that time and the kinds of things that were going on. This interview also takes a look at what this individual said, which is of course the man himself, Mr. William Saunders, a research scientist at OpenAI, someone that works on governance and policy at OpenAI. And he basically said that, look, I'm not going to lie, I was working on OpenAI and they are making rapid progress towards building AGI is absolutely incredible. And you guys need to do a better check of ensuring that these companies are put in line.

AGI, they are raising billions of dollars towards this goal. OpenAI's charter defines AGI as highly autonomous systems that outperform humans at most economically valuable work. That this means AI systems that could act on their own over long periods of time and do most jobs that humans can do. AI companies are making rapid progress towards building AGI. A few days before this hearing, OpenAI announced a new system, GPT01, that passed significant milestones, including one that was personally significant for me. When I was in high school, I spent years training for a prestigious international computer science competition. OpenAI's new system leaps from failing to qualify to winning a gold medal. Doing better than me in an area relevant to My own job. There are still significant gaps to close, but I believe it is plausible that an AGI system could be built in as little as three years. AGI would cause significant changes to society, including radical changes to the economy and employment. AGI could also cause catastrophic harm via systems autonomously conducting cyber attacks or assisting in the creation of novel biological weapons. OpenAI's new AI system is the first first system to show steps towards biological weapons risk as it is capable of helping experts in planning to reproduce a known biological threat. Without rigorous testing, developers might miss this kind of dangerous capability. While OpenAI has pioneered aspects of this testing, they've also repeatedly prioritized speed of deployment over rigor. I believe there is a real risk they will miss important dangerous capabilities in future AI systems. AGI will also be a valuable target for theft, including by foreign adversaries of the United States. While OpenAI publicly claims to take security seriously, their internal security was not prioritized. When I was at OpenAI, there were long periods of time where there were vulnerabilities that would have allowed me or hundreds of other employees at the company to bypass access controls and steal the company's most advanced AI systems, including GPT4. No one knows how to ensure that AGI systems will be safe and controlled. Current AI systems are trained by human supervisors, giving them a reward when they appear to be doing the right thing. We will need new approaches when handling systems that can find novel ways to manipulate their supervisors or hide misbehavior until deployed. The super alignment team at OpenAI was tasked with developing these approaches. But ultimately we had to figure out as we went along a terrifying process aspect when catastrophic harm is possible. Today, that team no longer exists. Its leaders and many key researchers resigned after struggling to get the resources they needed to be successful. OpenAI will say that they are improving. I and other employees who resigned doubt they will be ready in time. This is true not just with OpenAI. The incentives to prioritize rapid deployment apply to the entire industry. This is why policy response is needed now.

Remember how a minute ago I was basically stating that, look, AGI will be created by one of these countries. But James Cameron, someone that has directed some incredible movies, quite like the scary Terminator that depicts AI evolving into a super intelligent AI that is unstoppable and kills us. He actually talks about the fact that AGI won't be created by the government. And I do agree with this. I don't think that the government is going to be able to have the talent to be able to develop AGI quite like it did with the Manhattan Project in the previous race for technology.

One of the tech giants currently funding this multi billion dollar research. So then you'll be living in a world that you didn't agree to, didn't vote for that you are co inhabiting with a super intelligent alien species that answers to the goals and rules of a corporation. An entity which has access to the comms, beliefs, everything you ever said and the whereabouts of every person in the country via your personal data surveillance. Capitalism can toggle pretty quickly into digital totalitarianism at best. These tech giants become the self appointed arbiters of human good, which is the fox guarding the henhouse. They would never ever think of using that power against us and strip mining us for our last drop of cash. That's a scarier scenario than what I presented in the Terminator 40 years ago, if for no other reason than it's no longer science fiction.

It's happened now in more OpenAI news. Apparently OpenAI was going to do a wide release of Search GPT last week, but pulled it back for one reason or another, not wanting to clash with another anthropic release. But apparently Jimmy Apple says that this must be very soon though. Now of course, like I said before, if you've seen my past videos, you know just how great Jimmy Apples is at predicting OpenAI releases. So I think that either tomorrow on the Thursday, or either next week, Tuesday or Thursday, it's quite likely that SearchGPT, which is this tool right here, is going to essentially be widely released. So if you've someone that has, you know, if you've signed up to the wait list and you thought, oh man, I didn't get to, you know, use this technology. Of course I would say firstly that perplexity is probably better in many aspects, but if you want to use this kind of technology, it's quite likely that within the coming weeks this is going to be broadly available. If you aren't familiar with any forms of this technology, it's basically just like an advanced function for Google Search where you use an LLM to filter through the results and it summarizes absolutely everything that happens rather quickly and gives you responses that are really quick and efficient. So this is something that you might want to add to your tech stack if you're looking at becoming a more efficient individual. And of course I will cover this if it does get released. And in robotics news, one of the craziest stories that I genuinely saw was from this company called Engine AI. I still cannot believe that this is real. And I made a video on this, right? And if you check the comments on that video, you will literally see every person commenting. The fact that this is cgi, which is essentially computer generated imagery, now I can understand why they would think that, because this looks far too realistic for it to be not cgi. However, this technology is actually real. And the way that the, you know, robot from this Chinese company manages to move, we can clearly see that robotics is now moving towards the uncanny valley. This robot is becoming more energy efficient, which allows it to move in a way that is much more remarkable to standard individuals that simply watch this content online. If you haven't been paying attention to the robotic space at all, this kind of thing is going to make you completely jump out of your skin. But as someone who's been eagle eyed on these kinds of videos, this even I'm not going to be, I'm not going to lie, it does surprise me. It does surprise me, but it doesn't surprise me to the extent that it surprises other people, because I knew that eventually that this solution was coming. And of course, I've been looking at Boston Dynamics videos for quite a while and they have super realistic movement. So for those of you who are saying that, look, humanoid robots are always looking like they're about to squat, and you should understand that that era of robotics, where that is a reality, is shortly going to come to an end. And when you start to see robots walking around in factories in places like this in the future, it's going to become pretty incredible to see. One of the craziest things about this humanoid robot that I personally found was that this humanoid robot company was only founded one year ago. So in one year they've managed to go from nothing to inception to the actual product working that looks this incredible. So what do you think happens 10 years from now when this company manages to advance their capabilities even further? I think the entire ecosystem and space is going to look completely remarkable. Now, it was also today that GitHub actually managed to advance in terms of the model offerings that it has. Thomas Dunk says that GitHub has always been an open developer platform and today we are delivering multimodal choice for GitHub Copilot with Anthropics AI, Claude 3.5 Sonnet and Google Cloud's Gemini 1.5 Pro and OpenAI's O1 preview. So this is something that is completely remarkable because previously we didn't have these models available. So the coding ecosystem is going to change once again. And as many people at the Time were actually using Cursor because it had Claude 3.5 sonnet. And now you can see that GitHub Copilot have finally decided to lift off the reins and introduce all of the other models. So I do wonder what happens to the ecosystem now. Is Cursor going to be as useful as it was? I mean, the natural language interface is really nice and I can understand completely why, why people do use that, but it will be interesting to see how this manages to shake things up a bit. Now, for those of you that think that currently we are moving way too fast and there is just simply too much AI hype, Yann Lecun predicts animal level AI in five to seven years and human level AI within a decade. But it could take longer because there is a probability for things to be much harder than we think. Now, Yang Lecun has long held the stance that the current paradigm of using LLMs for intelligence is completely flawed and that this is not how you get to AGI at all. And I would argue that based on some of the research that I am seeing that I would agree with this statement. But take a look at this video before I dive into the one of the papers that came out recently that has a lot of people spooked the.

Ideas that I talked about earlier. Maybe within three years or something like this, if things work out, we're going to build a good path towards machines that can learn from the real world a bit like animals within five, six years, maybe seven. And then perhaps in the past to human level AI systems within the decade. But there is a long tail on the distribution, which means that it could take much longer because we could be hitting obstacles that we're not far seeing. And it's almost always the case in AI that a new technique appear, a new set of a new paradigm, new set of method, and the community says that's it, now we're going to get to human level AI, we just need to scale it up. That's what people are thinking about LLMs and it's wrong. That's why people were thinking about previous generation of AI techniques and that was wrong. So maybe I'm wrong too. So there is a probability for things to be much harder than we think.

Now many people are not even paying attention to Yann Lacan's statements. But you have to remember that this is someone that is one of the most widely accomplished AI engineers slash researchers, slash scientists. He literally won the Turing Award for conceptual and engineering breakthroughs that have made deep neural networks a critical component of computing. So before you think that Yann Lecun is just, you know, not in the space that much, you have to understand that his accomplishments are absolutely incredible by any measure of standard. So I think what we need to understand here is that whilst yes, these models are going to go ahead, we also have to understand the realism. If you remember, it was only a few weeks ago, days ago, where I covered an AI paper from Apple and in that paper it basically said that current LLMs are not capable of genuine logical reasoning. And the research that they saw that says based on these results they basically aim to replicate the reasoning steps observed in their training data. And this was pretty crazy because basically what this paper showed was the fact that if you have a standard question and if you change some of the variables that are non numeric, well, you know, some of them are numeric and some of them are non numeric. But the biggest thing is that if you change things like the name, like some of the numbers, not to like insane values, but if you change some of the relations between the characters but you input that same question again, there is a wide drop on the models, which is pretty surprising because that kind of thing should not be happening. If the models truly understand the question, then changing the name in these tests shouldn't result in such a dramatic drop. And this was something that was widely documented in the video I'd done where I said that LLMs can't reason as Apple drops this new bombshell because this paper was done by Apple Research, they do have an AI division that is really advanced in terms of their AI research. And now for someone else that also isn't buying into the AI hype, Linus Torvald, the creator of the Linux operating system, he actually says that yes, AI will change the world, but, but currently it's 90% marketing and 10% reality and it will take another five years for it to become clear what AI is useful for.

I think it is going to change the world and at the same time I hate the hype cycle so much that I really don't want to go there. So my approach to AI right now is I will basically ignore it because I think the whole whole tech industry around AI is in a very bad position and it's 90% marketing and 10% reality and, and in five years things will change and, and at that point we'll see what of the AI is getting used every day for real workloads instead of just chatgpt makes great like demonstrations and it's obviously being used especially for I mean in many, many areas, but I mean, graphics design, things like that. But I really hate the, the.