The Governments SECRET AGI Project Just Was Revealed...

So as of recently, it is quite likely that the US government is going to be funding a super secret project to unleash superhuman AI. This might be very one of the largest stories in AI because we all know what happened the last time the government decided to do something like this. So I'm going to give you guys a bit of a backstory, but what I'm referring to is the Manhattan Project. Now for those of you who have already seen the movie Oppenheimer the or Oppenheimer a massive wartime endeavor that basically changed all of human history. The US government basically realized that an entirely new weapon was possible and this one was one that could harness the power of atoms themselves. Now not only did they realize this was possible, they knew that Nazi Germany were trying to build it too. Which basically meant that there was a race against time. And currently it seems like there is the same thing happening in AI right now. We have two of the biggest players in AI going head to head to see who can build AGI first. And recently it was revealed that there was the commission's 2024 U.S. recommendations that it was recommended that Congress establish and fund a Manhattan Project like program dedicated to racing and acquiring an artificial general intelligence system for America. So this is the entire document. You can see that this is where they're looking at 10 of 32 recommendations to Congress. The complete list of recommendations appears at the report's conclusions on page 733. And this thing is just absolutely crazy because there's a lot to dive into. So you can see here it says that AGI is generally defined as systems that are as good or as better than human capabilities across all cognitive domains and would usurp the sharpest human minds at every task and among the specific actions the Commission recommends for Congress. And this is where we get into the two of the things that they recommend. And it says provide broad multi year contracting authority to the executive branch and associated funding for leading AI, cloud and data center companies with the stated policy at pace and scale consistent with the goal of US AGI leadership. This first point they're basically stating that they want the government to basically sign long term contracts and provide funding to these tech companies that are leading in AI, cloud computing and data centers. And the idea is to work closely with these companies to speed up the development of advanced AI systems. And they want to make sure that everything happens at a fast enough pace to keep the US ahead in the global AI race. So this is going to be something that I think is probably going to happen anyways when I remember I was looking at the, the recent talk where we had ex OpenAI employees speaking out about AI safety and they were talking to Congress. They were basically talking about the fact that even if you were getting certain people from the government to actually work, look at AI safety, a lot of them are just far behind what is currently at the frontier of these large labs. So they aren't really well versed. And since AI talent is such a free for all at the moment, these companies are literally throwing around million dollar salaries for the top machine learning engineers. What we do have is a situation where talent is pretty scarce. So even if the government does want to do this, I think it's going to be really hard for them to be able to do this. And I might include a clip from the full one and a half hour talk, but they basically said that look, if you're actually going to do this, you need to ensure that you try your best to give everyone the highest salary possible. Because companies like Google, companies like OpenAI, those guys are paying out a million dollars plus in salary galleries, in stock options. And that is something that the US government can't compete with because there's no incentive for someone to get paid 250k a year to work on a secret government project when they could work for the likes of OpenAI, Google and Anthropic working on the frontier of AI for four to five times that amount. And the second bullet point actually focuses on defense, where they state that they need to provide a defense priorities and allocation system which is a DX rating, which is basically where they are ensuring that AI related projects get the highest level of importance and resources protecting them as national security priority. In short, these bullet points are just laying out how the government can work with private companies and ensure that the resources are allocated properly to make sure that this AGI project happens as quickly and as securely as possible. Now I think this is something that is more important than people do think. Like a lot of people don't realize how impactful the AGI technology will be. And I think when you start to realize how impactful it will be, you'll understand exactly why this recommendation is being made. Now some people are looking at this recommendation and they're stating that look, this recommendation doesn't even make sense when we look at what Max Tegmark is saying. He's saying that the new AGI Manhattan Project proposal to Congress is science fraud with a bunch of non scientists pushing false scientific claims that AGI is easily controllable. Losing control to AGI would End US national security. And national security mustn't be compromised by fraudulent signs. Now he goes into an entire document, but I don't want to talk about this just yet. What I really do want to focus on is the fact that this is a really big issue. So if we look exactly at what Reuters are claiming here, Reuters are basically stating that, look, if this does happen, this is going to shift the global balance of power, like once AGI gets created. And Sam Altman's basically stating that, look, going to have this in 2025. And you know, they're basically saying, look, we're going to have super advanced AI systems in 2027. You can see here that we've seen throughout history that countries that they're the first to exploit, periods of rapid technological change can often cause shifts in the global balance of power. So that's someone that's a senior advisor to the company Palantir, which is basically an AI company for US defense. And it's incredible because we do know that if a company or a country is able to get to AGI or ASI first, they're going to have immense power over those who don't. And that was also what Daniel Cocachaljo, who worked at OpenAI said. He said, those who have AGI will wield godlike powers over those who don't. And that was some of that work to OpenAI. And he left because he said OpenAI probably aren't going to be responsible at the time of AGI development. Now also Reuters here, you can see it says that China is racing towards AGI and and it is critical that we take them extremely seriously. Now all of this talk actually brings me back to one of the most important documents for the next decade. And this was of course the document that is basically called Lock down the labs Security for AGI. So basically what this is actually talking about is the fact that AGI is something that is probably akin to nuclear weapons. And that's why I'm bringing it in this video. We were talking about a Manhattan Star like project because if we actually take a look at what we need to do here, we actually need to ensure that if AGI is made, if the US can get to it, we need to ensure that it is held down in a secure base so that the model weights aren't sent off to another competing country that's able to do what they want with them. Sure, the US is going to ensure that they secure the model weights and ensure that they are safe with them, but are other countries going to do the same are they not going to let their AGI run rogue if they manage to get access to the model model with so this PDF was from a former OpenAI employee where he speaks about something that most people did miss. Because even in that document I didn't even look at this for too long, but it was basically where he's stating that currently they're basically handling the secrets for AGI to the CCP on a silver platter and securing AGI and these secrets against the state actor threat will be an immense effort and currently we're not on track. He's basically stating, look, currently if we don't act up and we don't fix our mode, we're going to be in a situation where we manage to leak all of the AGI secrets, all of the Frontier Lab breakthroughs to a competitor country that's going to be able to speed up development because they have less relaxed laws and they're going to be able to race ahead, which means they will be in front after stealing those breakthroughs. You can see here he actually talks about how China engages in widespread industrial espionage. The FBI director stated that the PRC has a hacking operation greater than every major nation combined. And just a couple of months ago the Attorney General announced the arrest of a Chinese national who'd stolen a AI code from Google to take back to him to the PRC in 2023. And that is just the tip of the iceberg. And apparently, and this document was released around seven to eight months ago, he says our failure today will be irreversible soon. In the next 12 to 24 months we will leak key AGI breakthroughs to the CCP. It will be the national security establishment single greatest regret before the decade is out. Apparently the preservation of the free world against the authoritarian state is on the line and a healthy lead will be the necessary buffer. That gives us a margin to get AI safety right too. And the United States currently has an advantage in the AGI race, but we will give up this lead if we don't get serious about security very, very soon. And that's why they're probably talking about a Manhattan style project, because we need to ensure we dedicate enough time, effort and energy to this kind of project. You can see Marc Andresen here says, is this what we see at OpenAI or any other American AI lab? No, in fact what we see is the opposite. The security equivalent of Swiss cheese. The Chinese penetration of these labs would be trivially easily using any number of industrial espionage methods, such as simply bribing the cleaning crew to stick USB dongles into laptops. And my own assumption is that all, all such American AI labs are fully penetrated and that China is getting nightly downloads of all American AI research code right now, basically stating that, look, we are woefully unequipped for the security measures that we need to have in place to ensure that AGI isn't developed by another nation state. And if that does happen, we're going to be in a seriously bad position because it means they're going to have something that we don't have and we will no longer control the free world. And one of the last things that he says here is that if we look out just a bit further, once China begins to truly understand the importance of AGI, we should expect the full force of their espionage efforts to come to bear. Think billions of dollars invested, thousands of employees, and extreme measures dedicated to infiltrating AGI efforts. And what will security for AGI and superintelligence require? In short, this will only be possible with government help. And he's basically stating here that, look, somehow these labs are going to have to collaborate with the government. And with this kind of, you know, recommendation here, it seems like that's exactly what's going to happen. Now, what's crazy about this is that this recommendation was actually put forward by someone who is actually friends with Sam Altman, and someone was actually talking about, if Trump Admin tries to move forward with these proposals, will they end up, you know, being another unpopular cluster mess like the TikTok ban? But of course, he's talking about how this person who put forward this stuff was actually friends with Sam Altman. So it's quite likely this is going to happen and there's probably some government program underway where they're going to be developing AGI without all the kinds of restrictions that we do see currently. When you are deploying these AI systems, you have those safety levels, but I think internally they're definitely going to be rushing ahead, trying out different methods just to get to a completely smart AI as possible. But of course, in a way that is airtight, like probably they're going to have their own servers that are completely disconnected from the Internet when they interact with some. An AI model. And of course, the craziest thing about this is that, remember what I said Max Tegmark was talking about? He basically talks. He basically just said, this race that we're basically entering right now is a suicide race. And the proposed AGI Manhattan Project and the fundamental misunderstanding that underpins it represent an insidious growing threat to US national security. Any system better than humans at general cognition and problem solving would by definition be better than humans at AI research and development and therefore being able to improve itself at a terrifying rate. And he's basically stating here that if we go ahead with this race, we're going to be in a situation where, yes, we're going to get to that AI that develops first, but unfortunately, once we get to that level, we're going to get to a situation where the AI just kills us anyways. Now, I know most people don't believe the fact that AI could kill us or, you know, those kinds of existential risks, but I will leave a link to a video that basically explains everything. And once you watch those videos, you, you understand that it is very, very hard to control an AI because intrinsically they are black boxes, which means that we don't understand why they are the way they are. And even the most known researchers actually say that. Look, it would be more accurate to say that the way we interact with these models is that we grow them rather than we actually build them. Which means that if you're growing something and you don't understand why it is the way it is, growing something that is going to be smarter than you is going to be incredibly difficult. And considering the fact that it was recently that OpenAI disbanded their super align team, I'm really wondering how we are in terms of safety. And one of the craziest things that they talk about here is the fact that if we go ahead with this AGI style Manhattan race, we're going to have no opportunity to solve the unsolved technical problems of control and alignment and every incentive to see decisions and power to the AI itself, which is of course a big problem, because if we rush ahead with developing this AI, we're not really going to be focusing on alignment and how we can actually ensure our complete safety. So this is going to be something that just focuses on how to make the model smart, how to make self proof, how to get to asi. And of course there are a variety of problems with that. And you do have to remember that it was Geoffrey Hinton, who is referred to as the godfather of AI, who said once AI gets smarter than we are, they will take control. And I mean, it does make sense. Humans are the smartest thing on the planet currently. And look at what we've done to everything else that is not smarter than us. So overall, I do think when we do take a look at things, it does seem like we are in a strange race. But I will say this, that if we don't get to AGI and another state gets to it before us, then I think that could be potentially worse. So, I mean, it's a very tough position to be in. We're in between a rock and a hard place. So I think the government is just basically doing the thing that they think they should do, which is of course just get to it and try to figure it out on the way. I know it's going to be pretty difficult to do, but I don't see any other solution. If we do slow down, we know 100% that these other countries aren't going to slow down at all. Which means that unfortunately we're forced to be in this race. Now let me know what you guys think about this. I think this thing is going to be absolutely huge in terms of the future because if you do get to an ASI level system, the kinds of things that are going to become possible are going to be absolutely incredible. So if you guys did enjoy this video, do not forget to leave a like, do not forget to subscribe. I'll see you guys in the next one.