Crazy OpenAI Drama, Secret Claude Command, Thousands Of Humanoid Robots And More..

From looking internally at how AIs now think and respond to us, to even more drama at OpenAI, here are all of the secret stories that you really did miss. Okay, so one of the craziest stories that actually happened this week in AI, and I think this is genuinely absolutely crazy, is that you can actually now see what Claude is thinking. Okay, so if you go ahead and you use this prompt, you say, from now on, use this. I don't know genuinely, I don't know what this symbol is. And instead of this, you essentially are allowed to see what Claude is thinking. You can see here in this special box here, you can see that, you know, I asked this question, I was like, okay, how do I make my post AGI canonics community more valuable? You can see that right here, you can see that it says thinking. Okay, and now I can access Claude's internal thoughts. It says, this request is about improving an online community which doesn't meet the criteria for an artifact. This response will be a list of suggestions which is better presented in the conversation for easy reading and follow up questions. And then you can see it ends by thinking. So basically, if you want to know why this is so crazy, it's because previously what we used to do, like in the early days of GPT4 when we would ask the model a question, what we usually would do was something called chain of thought prompting. So what we would ask the model to do is think step by step and then lay out your reasoning. So this was called chain of thought and this was something that was really, really popular. And you can now see that this is a method that anthropic have baked into the model. And I'm guessing that this is why Claude 3.5 Sonnet is just simply so much more intelligent about its responses. Previously, a lot of people didn't know why the model was so more intelligent. Of course there were a little bit of, I guess you could say, hints within the research paper that they produced. But we do know that the gains that are to be had from this, whilst they are incremental, they are actually really verifiable because we've actually seen that recently when we looked at the benchmarks as well, we saw that this is basically on par or even better than GPT4O. So this is something that is, I guess you could say, kind of surprising because I didn't think this would be baked into the model just yet. Until next, the next iteration of models like GPT5 and of course Claude 3 Opus. But now that this is baked into the model, this is showing me and this is showing us that, look, we have a remarkable amount of growth that's going to be happening with these models because if they literally are starting to implement this thinking area, and if this is actually bringing gains, then this means that these creators of these AI systems, they're not just thinking about how they can scale up the models just by training $100 million models, they're actually scaling up the models via iterative ways to improve the reasoning capabilities of the base model with all of these different things. So it will be interesting to see if Anthropic actually talk about this because in the paper they didn't actually say anything about this. So this is going to be the first time that we're truly seeing something like this. So this is something that I'm excited for and I want to see what kind of other things it's going to be doing because I want to know if it's going to be thinking like this, if it's not going to be thinking like this, and how everything is going to be. Now, in some other AI news, I do want to state that this is pretty incredible. What's going on here is that, you know, Endeavor CEO Avi Emanuel went on not a rant, but he actually spoke about Sam Altman in a very interesting way. He said that Sam Altman is a conman and that Elon Musk gave him a lot of money to create a startup for a nonprofit, only for OpenAI to be turned into a for profit company and we shouldn't trust them to take sufficient caution when creating powerful AI. Now, of course, Elon Musk, I wouldn't say he's got a vendetta against Sam Altman, but there's certainly been some kind of feud between them going on in the past. You know, certainly, certainly some stuff going on there. But essentially what we're seeing here is a really interesting statement. So I'm going to play this for you guys and then I'm going to give you guys my thoughts because I think this is, it's pretty crazy because, you know, Sam Altman never ceases to amaze me with the amount of drama that surrounds Open Air.

And I think he's, he's a con man. You know, I think that, you know, the old shucks thing, I don't know, started out with Elon, gave him a lot of money. Supposed to be nonprofit, now he's making a lot of money. I don't know why I would trust him. I don't, I don't know why we would trust these people not. And they're very smart. And there's probably, you know, you just have to weigh the following for all the stuff they're saying that could be great. And I think it is. Health care, you know, new discoveries, a lot of great stuff. And we have to figure that out. And this is where the regulation has come. I looked at a whole host and thought about a whole host of stuff that's bad. So you're telling me you've done the calculation and the good outweighs the bad? Really?

I gotta be honest, it is a little bit concerning that this is a constant theme iterated by people that have been, you know, I guess you could say, in contact with Sam Altman. This is something that we've seen, you know, people on the board of OpenAI say, people like Helen Toner say that Sam Altman has, you know, deceptive practices, manipulation, lies, just, you know, it's not looking too good for Sam Altman because this keeps happening. And, you know, if one person says something, another person says something, it's like, okay, it's his story versus her story. But if there's a constant stream of people, you know, saying this, at some point you have to start to wonder if there is truth in these statements. So this is going to be something that is, of course, a little bit of cause for concern and creating AGI. This is something that has been a major concern for the public. And this statement right here that we shouldn't trust them to take sufficient caution with creating powerful AI. This isn't the only person that said that. Genuinely, this is not the only person that said that. This is a statement that was also iterated by Daniel Coco Taljo, someone who used to work on the OpenAI's governance team, and said that he's leaving because he doesn't trust OpenAI to act responsibly around the time of AGI when it is achieved. So he actually gave up quite a lot of equity and OpenAI for that. So hats off to him for doing that. Because warning the public about what a company might be doing, I think that's a great public service. And of course, OpenAI's strange governance structure is one that is always going to be brought into question. So it will be interesting to see how we continue to move in regards to that. If OpenAI managed to change their structure completely, or if they just managed to get out of the jam, because Elon Musk did actually recently drop the lawsuit for whatever reason. Maybe he's just thinking, you know what? I just Got to focus on my companies and what we're doing now. In more OpenAI drama, there is news coming out confirmed that someone broke into OpenAI. This is from, you know, this account right here. But it's pretty crazy because I didn't know that OpenAI was hacked in April 2023. Of course, Leopold Aschenbrenner spoke about this and he spoke about how this is basically like a giant security risk. And of course this is, I guess you could say in the future a truly, truly national security kind of movement because it's a threat to the US Security system, the home system, whatever you call it. I know that's not the actual words, but point trying to make here is that, you know, OpenAI weren't public about this breach. But I got to say that being hacked is very different from leaks because we always know that, you know, there are a few leaks and stuff going on around here. But like I said before, I don't think this is going to happen at all in the future because of how serious the technology is that they're working on right now. I'm sure they've got new methods and new ways in order to prevent these leaks. There were screenshots of them hiring certain people to ensure that their data or whatever it is that they're working on is remaining completely top secret. But you can see here that this is pretty, pretty crazy because we have a situation on our hands where if this company continues to get hacked and bad actors are able to get out secret pieces of information regarding superhuman systems, that doesn't bode well for anyone. So it's pretty crazy.

Take a look in April of last year and never disclosed it to the public. First reported by the New York Times and now confirmed by cnbc, details how hackers broke into the startup's internal messaging system and and access secrets. So this raises questions as to how much generative AI companies should be sharing about security incidents and how transparent OpenAI in particular is as its influence grows. The startups executives, they decided not to share news that the hack publicly or share it to law enforcement officials because no information about customers or partners had been stolen. And they didn't consider it a national security threat because the hacker, they say, was a private, private individual with no known ties to a foreign government. But in cybersecurity, that can be difficult to determine and those impacts may not be immediately known. Some experts say that reporting attacks can help identify and mitigate broader threats. In this case, I'm told the hacker gained access to internal communications channels used for discussions among employees. So not the more important systems where OpenAI actually keeps and builds it's technology.

So yeah, pretty crazy stuff going on here because I'm starting to wonder if you know, Leopold Aschen Brenner's claims are true. If you don't know, he's basically someone who used to work at OpenAI on super intelligence safety. And he basically said that, look, if you guys don't fix your safety issues, we're going to have a situation on our hands where these foreign actors, or bad actors, whatever you want to call them, they're going to be coming in, they're going to, they're going to be stealing our secrets, they're going to be using that and they're going to push ahead without regulation and surpass us in the race for AI dominance. And I think that is a very valid threat because AI is something that moves so fast and if you blink you'll miss it or you'll be behind. And only certain companies can be in the race. Like literally only certain companies are going to be able to build only probably like around four or five companies. Meta and a few other companies, to name a few, can only do these billion dollar training runs to get these super powerful systems. So it's important that you do prot. Whatever information or valuable secrets you do have at that time. That brings me onto this section of the video where we have today's sponsor On Demand. A platform that can turbocharge AI innovation and is the fast track to AI product development. This is a cutting edge platform as a service which is actually powered by retrieval augmented generation AI technology. This is a platform that has a lot of amazing things including this very well thought out playground area where you have an array of large language models. Not only that, with these custom model endpoints that you have right here, you also have the ability to create your own model and import your own specific model from Hugging Face. If you do request. You can see that there are a bunch of different models here that you can search for on Hugging Face that you may want to use for various different applications. The ability to create your own model, slash add your own model from the Hugging Face model search. You can see right here, you can search for models that currently exist on Hugging Face. And you may need this as you may want to run different applications such as testing or whatever specific use case that you do want. It also has this amazing plugin marketplace where you have different plugins that can actually have different applications depending on what your target is. For example, if we switch to the top Plugins tab. You can see that there are many plugins here that you can use now. Some of the use cases of this platform are documented here. You can see that one of the things that you can do is hook this up to your smart home plugin, of course, connect it to the cloud and whatever IoT platform you want, like an Arduino, a Raspberry PI. And of course with On Demand, you can then have your own home automation and you can use this to do things in your home. If you have a smart home, if you're not as savvy as that, you can instead use On Demand with a bunch of different other plugins. You could potentially make this if you wanted to and create some kind of latest news chatbot which is able to sync up with all the news plugins, the live tracker plugins, and of course knowledge trained on a custom model. Combined with GPT4 and on demand, we get a comprehensive chatbot that's able to give you up to date information that's rarely relevant due to the retrieval augmented generation capabilities. Overall, the On Demand platform is a very innovative platform that is a step towards what large language models and AI systems are going to be in the future. With a vast array of different plugins and ways to integrate them into your chatbots. It's something that is truly, truly impressive. Now, Korea AI actually did announce their scene transfer and I don't usually do videos on stuff that's like, you know, very artsy and stuff like that because, you know, it's, it's not usually what people are entertained by. But I think this is something that is really useful because it allows you to change the background of your image. Okay, so you can have an image and then you can change the entire background of that image to whatever you want. I think this is really useful. And the reason I think this is so useful is because when you used Photoshop before, the main thing that you always wanted to do was to change the background and be able to composite the image into another image. And this is something that took a lot of time and it was just extraordinarily time consuming. So now that this can be done instantly, I think this is going to be absolutely incredible in terms of how much time this is going to save you. For those of you that, you know, are wanting to step into your creative era, I think this is going to be something that, you know, helps you out quite a lot because this is going to allow you to explore that, you know, bit in a way that you didn't before. Like this was stuff that even, you know, someone who's really experienced in Photoshop, it would take them quite some time to do. You'd have to do all the shading or the lighting. Just trust me, guys, it takes a very long time. Like, there's a YouTuber called Benny Productions, I used to watch his stuff all the time and he goes into the speed process of how to actually do this. And it takes, I wouldn't say forever, but this is a genuine crazy time saver and this is something that I think is definitely going to like. You're probably going to start seeing posts in the future where you see a lot of interesting things. So it will be interesting to see, you know, how all of that does work, especially with the rise of social media and of course, talking about the rise of social media with Generative AI. There was a tweet I wanted to show you guys, but I simply, for the life of me cannot find it. So if I do find it, I'll leave a link in the description and I'll be able to discuss it there. But you can see that humanoid robot Vanguard lined up for, you know, WAIC 2024 is looking absolutely incredible. Guys, this is a conference. I think it's conference slash event. I'm not too sure on the entire details. There is an article that I'm about to show you all, but you can see that, you know, China is not slowing down with humanoid robotics. I made a video in the past three weeks titled Get Ready for China's 2024 AI Robotics Takeover. And I think in the future, if America or other nations aren't smart about what they're doing in robotics, they are going to be left behind by just how much robots China can produce in terms of the efficiency, the scale and the fluidity of the robots, because there are literally so many independent Chinese companies. Like, I could make a video probably every single day about a new robotics demo from China if I wanted to, but I'm not going to do that, obviously. But I'm just saying there are that many, like, trust me, guys, I make videos on a lot of different stuff and a lot of the times I'll see like three different Chinese robots every single week that get released doing impressive demos. So, I mean, if they have like an ecosystem of, you know, continually, you know, looking at robots, what they can improve, the kind of research that they can do, I think they could definitely move very, very quickly up the ranks in terms of the leader, in terms of humanoid robotics. So that will be interesting to see how the next 10 years goes because some people are stating that humanoid robotics is a dead end. Like the video I produced yesterday on the predictions for the future. I disagree, but it will be interesting to see. And you can see here, the Global Times says China's AI's humanoid robots are catching up with global rivals. So Chinese manufactured humanoid robots are catching up with global rivals, company representatives at an industry expo told the Global Times on Thursday. So yeah, it's not a surprise, but there are so many that are literally, you know, just, it's just, it's just crazy. And China's doing some incentives where they want, you know, companies to kind of build out this stuff. So that is also there. It's not just like, you know, they're thinking about this, they are really know, gunning for this and pushing for this. So it will be interesting. As well as this robot stuff, there was also this full size general purpose humanoid robot independently developed by Humanoid robot Shanghai CO. Ltd. To make its debut. You know, it made its debut at WAIC 2024 with a height of 185cm and a weight of 82kg. So this thing is pretty tall. I think that's like six one or six' two. I honestly don't know. But Qinglong Bot has high bionic torso and human like motion control. So this is some impressive stuff for what we're seeing here. Like this is some genuinely impressive stuff for what we're seeing here. Because it was only a couple of years ago that we were starting to think, I wonder what the future is going to look like. And now seeing a lot of these robots and oftentimes some of the stuff we are seeing is autonomous. Of course there is the teleop demo here and there. But I do think that this is really like, you know, when you take a look at how market conditions can affect the results for consumers and the kind of products that we're going to get on the market. Right now we're probably at the most competitive part because I think people like the global consciousness in terms of, okay, well, AI is a thing. Okay, well, humanoid robots can actually be a thing. That part has been ignited in terms of investments, in terms of what people are seeing. And I truly believe that now we're going to probably get even more breakthroughs because companies have financial incentives in order to build the best robots and in order to make these things actually effective. So I wouldn't be surprised if the next five years we do get some crazy humanoid robot demos and we start to get robots that are continually doing work in factories, continually Doing stuff in ways that we never really thought. But, yeah, China, I'm not going to lie to you guys, I do think they probably will be the market leader. But it will be interesting to see the balance between companies like Tesla's Optimus and Chinese factories like humanoid robotics, the other astrobots, the other GR1 robot, the unitary robot. They honestly keep doing some amazing stuff. So it will be interesting to see what kind of things are going to be there and what other kind of products we do manage to get. Because the race is truly on now, and we know that companies and countries are very, very competitive, and they're going to pour millions and billions of dollars into this. So this is probably going to be one of the most exciting times in terms of what we're truly going to be able to see as a consumer, because they're all gunning for the same position, which is, of course, the best humanoid robot that exists. Now, in humanoid robot news as well, we had the Mentibot. So mentibots are mentored by people on their first day to a new place. The owner or coworker needs to show them around. This involves following a person in an unfamiliar environment while avoiding obstacles. Basically, they've decided to add voice commands to robot, and they've added something to where the robot can actually follow you around in novel environments, which is, I think, kind of remarkable considering the fact that this company only recently came out of stealth. They're kind of showing us a very interesting, unique sort of humanoid robot. This is one that is not as well noted as many others are, but I think it still deserves a place in the AI debate community, whatever you want to call it. So it's still worthwhile because it seems rather. Well, because she's giving it commands. She's saying, in fact, let me just let the audio play. But basically right here, she's saying, can you face me? She goes over to here, the robot faces me. Then, of course, she says, can you face me again? And you can see that the robot is managing to follow all of those tasks. So I think, of course, in the future, this is going to need to be a little bit faster, a little bit more fluid. But the progress on humanoid robotics is there. And like I said before, this is definitely accelerating in here. She actually asked the robot to walk backwards, so great stuff on that end. We also have one other interesting perspective here. Google executive Mo Gad says that the mainstream media is hiding the truth of the trajectory of AI and it is on course to end our world as we know it within four years. I'm going to give you guys my thoughts on this because this was debated by so many people on Twitter that I thought that this is something that I want to show you guys, because people are either on the right or the left. Not politically, but on what either side of the argument.

You believe are two paths to that. One path is to lie, but then lying gets you to debate. The other path is to distract, is to give you football and naked women and, you know, music and the story of the scandal between that actor and that football player and, and so on and so forth. And so suddenly we flood you. Well, not we, but the. The media floods you with information, okay, that's so irrelevant that you get distracted from the truth. So you're not even questioning what the most important thing is, okay? And to break through that, someone has to stand up and say, are you even aware? Are you Even aware that ChatGPT today is as intelligent as Einstein? Are you, Are you aware? Okay, Are you aware what the trajectory of that is? Are you aware what's being worked on today in terms of complex mathematics and deep reasoning and, you know, so many forms of intelligence? Are you aware? Okay, because if you're not, you're not acting. If you're not, you don't know the truth because not because someone is lying to you and telling you, oh, no, AI is going to be fine because someone is telling you football is important, right? And the game here is the following. The game here is every human you meet will have passion. They'll say, no, no, I want to influence my future. They don't even know what their future is, okay? The real influential influencers of your future, other than looking inside yourself and making you know the truth, because not because someone is lying to you and telling you, oh, no, AI is going to be fine because someone is telling you football is important, right? And the game here is the following. The game here is every human you meet will have passion. They'll say, oh, no, I want to influence my future. They don't even know what their future is. Okay? The real influential influencers of your future, other than looking inside yourself and making yourself, making yourself the best version you can of you, are a very, very significant economic, geopolitical, climate and technological perfect storm that will end our world as we know it. The world, in four years time, it's not going to be recognizable. What do you think it's going to.

Look like in four years?

I think we're going to suffer significant job losses in an economic, very tough economic environment where lots of the Tyrants in the world will be able to get away with wars. And.

So the reason I included that in this video is because I wanted to talk about how I think people view the current AI debate in terms of these systems. Now, if you don't know, on one side you have people that think we're very close to artificial superintelligence, people like Ilya Sutskova, people that have actually worked on the stuff. And then on the other hand, we have skeptics like Gary Marcus and Yann Lecun that basically say we are so far from AGI. You know, you people have just got carried away with the hype. I'm a bit of both. I think that generative AI is good. I think that it is intelligent, more intelligent than people think. I think that in some cases, in either cases, people just focus on what they want to focus on because humans are biased creatures. Like sometimes when Gary Marcus tweets stuff about AI, he'll often tweet things in studies where they say, like, look, ChatGPT is not even that smart. And the studies that they're referencing, they all use GPT 3.5. So if you're referencing a study that uses GPT 3.5, you clearly have an agenda within what you want to portray in terms of the kind of knowledge you're saying. Basically, if you're doing a study with GPT 3.5, you're basically, you know, you want to make it out to seem like this AI is not as smart as it is right now, which is a little bit concerning. So I think on either side, you know, people have valid points, But I think the things are getting smart and I think they will be super intelligent someday. But I do think it will be a combination of many different techniques, and it probably might not be transformers, But I do think that since the race has heated up so much now, I think that, you know, we're definitely going to get there. And here is a tweet that I want to leave you guys with. It says, so sick of AI, just unplug it already. It's bad enough that the fossil fuel industry and the military industrial complex are hurdling us towards a geological collapse. Now we have one final genius invention of capitalism. A magical machine that eliminates the need for labor by stealing it and it costs to run is the low price of exceeding the world's energy grids. For F's sake. We don't want this technology. We don't need this. Technology doesn't create anything of value. All it does is consume energy and pollute. Everyone hates it. And it will never earn back anyone's investment, let alone the cost of the entire world's energy supply. Just shut it down for good. And this is someone that tweeted this. What's interesting is that this person has 3,000 followers. So it wasn't a large account which tweeted this, which means that there is a large majority of people that do believe that generative AI is a net negative, meaning that the benefits do not outweigh the negatives. And that is kind of interesting. Slash concerning. Well, it's an interesting dynamic, one that I've discussed before because I said, you know, depending on where you are, you can definitely use this to your advantage by, you know, engaging in the fact that you can position your business or your brand as one that is human made and enhancing the fact that you don't use AI, because that's an increasing trend. But I want to know where some of you guys are, because some people like AI and some people don't. And it is clear that there's a growing, growing divide. So let me know what you guys think. If this tweet actually holds any value. And if you do believe that this is true, let me know why. Hopefully you guys enjoyed the video. Hopefully you guys have a great weekend and I will see you all in the next video.