OpenAI's SECRET Robotics Team, New AI Shows, Superintelligent Agents, Mistrals New AI, New AI Music

So one of the things that was actually really interesting that was released this week was this thing called Showrunner. So basically, Showrunner is, I guess you could say, if you've ever seen Rick and Morty, and no worries if you haven't. There's basically this thing called interdimensional cable. And interdimensional cable is basically a thing where you can watch any TV show that you can literally imagine. So if you wanted to watch, I guess you could say a bunch of TV remotes running around and killing pizzas, hunting them in some strange style, you could watch that. And this is kind of what this, I guess you could say, platform is going to be. It's basically Netflix for AI generated tv. Now, I got to say that this idea is really, really good, because in a moment, I'm going to show you guys something that I've literally been watching. And I know it sounds weird to say that you've been watching AI generated tv, but I'm not going to lie to you guys, it is just so funny, because the problem is, is that, like, these kinds of TV shows don't get made. Like, companies are not going to spend millions and billions of dollars making ridiculous st. There's stuff that, like, is genuinely funny that only you would enjoy. And I guess that's the kind of, I guess you could say, entertainment value of generative AI, because it's kind of like, I think. I think it's actually going to work. Like, I think it's actually going to work because if you understand how social media is engineered to make you addicted, one of the things that social media does is it cooks you in. Because social media is, I guess you could say, a slot machine. So, you know, when you got your phone and you're scrolling on your phone, maybe it's TikTok, maybe it's YouTube. Either way, you're scrolling, okay? And that's quite similar to, like, a slot machine, okay? And that's what generative AI TV is going to be. Because you don't know when the next interesting piece of information is going to be. And if you have generative AI TV shows or whatever they may be, and they're just truly uniquely specific to your entertainment and your entertainment options, I can promise you guys that you're going to get a lot more entertainment watching that than some of the things. Now, I might be biased in saying this, but I think this is going to be a really interesting platform provided that, you know, the costs go down and these kind of things do work. But there were, you know, AI generated South park shows that were actually really good in terms of quality. But I also want to show you guys another platform. And, you know, in their TED Talk last year, they previewed how simulations and shows work together, and they had these AI agent simulations in, I guess you could say in the south park style environment. So this was the kind of thing that they had. And you can see right here, it's literally like a city with all of these agents running around doing different stuff. And I mean, maybe this is going to be how video games are in the future. So take a look.

Good morning, Aurora. Currently in San Francisco, it's 62 degrees with partial clouds. You might want to bring a layer for the slight chill in the breeze. Today's schedule includes team meeting at 10am, your yoga class at 6pm and don't forget, today is your mother's birthday.

And I think one of the most interesting things about this is not the fact that it's just AI generated, but the fact that you will have influence over this. So I'm going to show you guys something else that I just found absolutely hilarious.

How does your openness compare to AirPods Pro in terms of sound leakage and potential hearing damage?

My open design allows sound to escape freely.

So basically, if you're not sure on what you're watching here, this is an AI generated, completely AI generated courtroom, okay? And the courtroom is basically whatever some people type here. You can see it's submitted by someone random. And you can see it says AirPods vs AirPods Pro over who can make me go deaf the fastest. And that's the case. Okay? And the cases are just so funny. Like, honestly, I remember watching one about, like, a banana peel and recently about Mario versus Luigi. I'm gonna show you guys a little bit of this latest episode of Jars Court. Next up, on this episode of Jars Court, John sues Mickey for being dangerously attractive, causing mental distress.

Gonna look into whether Mickey being so good looking is a legit reason to sue Mickey, tell us how your hotness supposedly made John go crazy. Let's get to the bottom of this weird situation.

I walked in looking sharp as ever, and John couldn't handle my charm. My appeal is undeniable, but causing mental distress, that's a stretch. Your honor.

John, can you describe in detail how Mickey's level of attractiveness specifically affected you and led to the alleged mental distress you experienced?

I arrived, sunglasses on, love songs blaring, hearts breaking behind me. Mickey's fine looks blew my mind, causing undeniable mental distress.

It's a heartbreak case.

Your Honor, John says you caused him mental distress. Can you give examples of how your attractiveness is to blame?

I strutted in radiating.

So if you don't know where this is from, this is from like an old. I don't even know exactly where this is from, but there's a song that goes, oh, Mickey, you're so fine. Something along those lines, if you know what I'm talking about, you know what I'm talking about. But that's why this kind of stuff is just ridiculous, because they're never going to make this kind of stuff. But some of the conversations that I actually heard here are actually just ridiculous. Like literally sometimes I will just go on here and see some of the kind of conversations that are going on because it's just ridiculous. And sometimes they do get really funny. So I would say, you know, check this out, because it is, it's like different, but it just goes to show, like, how we consume entertainment in the near future might not be how we think we're going to be consuming it. So that's why I wanted to show you this. And you can say, car sued by sandwich for running over it and becoming a panini.

But I never intended to turn a sandwich into a panini. It was a sandwich, not a drive through order sandwich.

Can you describe the emotional distress you experienced when you were unexpectedly flattened and transformed into a panini by the car's actions?

I was a proud hoagie, not a panini. The car's reckless driving turned me into a pressed mess. I demand justice for this toasty injustice.

Car. Did you even know the sandwich was there before it turned into a panini? Your explanation proves negligence, not malice. And the sandwich's distress adds some humor to this crazy culinary chaos.

I was too focused on my speed. Didn't notice the sandwich. I regret the unintended panini transformation. It's a sandwich, not a road snack.

Judge Amelia Bennett Carr's negligence led to the sandwich's panini transformation. Sandwich's distress is understandable. I rule in favor of the sandwich. Culinary chaos averted. Justice served.

Judge Bennett rules in favor of the sandwich, declaring justice for the panini fied plaintiff in this unique case of culinary calamity. So, yeah, let me know what you guys think about that. Maybe we're not all going to be watching AI generated stuff, but I do, I do think it's going to be interesting to see how this develops, especially once this kind of technology gets a lot better. There was also this. That was truly, truly, truly impressive. And this is where the unlocks of creativity are going to be had with artificial intelligence. If you're not familiar with what you're currently looking at, this is Suno AI. And you can see right here that what we have is, I guess you could say, a new way to generate music. This is volume one. Okay. And we. I'm just going to play the demo because this is pretty, pretty crazy. And this is from an existing platform that allows you to make very good AI generated soundtracks. Now I know some people might not think that that is remarkably impressive, but I, I genuinely think that this is remarkably impressive because with literally just Taps, you can now become a musician. I mean, you could publish your own sounds to Spotify or wherever. And as long as people enjoyed what you made, I'm sure that there might even be an audience for it. So this seems so, so interesting. Like as someone who enjoys, you know, pursuing creative stuff and just genuinely experimenting with new technologies, especially on how they might develop, I think this is going to truly, truly shake up the music industry. And I know that a lot of times people are saying that, you know, this is going to shake up Hollywood or this is going to shake up that industry. I truly do think that this is something that's fundamentally different. Because the problem is with music is that like, humans do care to an extent that something is good. But I've even seen a viral track recently that was completely AI generated. Like for example, the. There's a song called I'm going to show you guys right now. It's a pretty crazy song, but it's completely AI generated. I'll actually find the song. But the point is, is that like, sometimes when you hear a song, you don't immediately try and figure out whether or not it's human generated. Like if you're watching this video and there was background music, I'm not sure if people would genuinely care if it was AI generated or not. Unless I explicitly stated that it is. I'm not sure people would genuinely care. Because whilst yes, human made music is good, like you know, playing the guitars, instruments and stuff, stuff like that, I don't think those live performances are going away anytime soon. I'm talking about the kind of like, you know, corporate background music or, you know, just the average TV show background music, I mean, do humans really, really care if that is AI generated or not? And I think that's where the disruption is going to come because I don't.

Think most humans do that implantable device, you can, I think, address almost any brain or spine injury. So the first part that we have there is called telepathy. So it enables you to use your phone or computer just by thinking. So you could, you could do that, move the cursor, literally with his eyes closed. It's reading signals from the motor cortex in the brain. The next product is Blindsight, which will enable people who have lost both eyes or have no eyesight whatsoever to see. And this is directly interfacing with the optical processing areas of the, of the brain. So, so, you know, I think this is, this is interesting progress. Ultimately, the, the goal of neuralink is to have a high bandwidth interface in order to mitigate the risk of, of digital super intelligence. I, I don't know, I'm not saying it will work to mitigate the, the risk. It is just it might help. And, and, and this, this is somewhat, get somewhat esoteric, but the limiting factor I think for AI alignment is probably going to be the bandwidth. How quickly can we communicate with our digital tertiary self?

So that was a clip from Elon Musk where he basically talks about some of his recent advancements with neuralink. Now I know Elon Musk has been getting a lot of flack recently on Twitter or X.com for the fact that he hasn't produced the Tesla Roadster yet, but I think what so many people are negating is the fact that Elon Musk owns so many different billion dollar companies and, and they are all making progress in areas that haven't been made before. And you know, a simple electric car isn't something that people should be focusing on when he's literally pioneering, you know, this field where people are able to literally get their lives back, even if they've been destroyed by whatever spinal industry or devastating, you know, impacts that they've had from whatever accidents that they may have suffered. So I think this is truly the most important, you know, kind of areas where technology and the singularity are going to encompass, because these are the areas where you're truly going to think that, you know, you might not be invincible, but, you know, what could previously stop you, probably won't in the future. And as he's talking about AI alignment, I think this is something that truly does make sense. Because what he's basically saying is that superintelligence is going to be so incredibly fast, so incredibly smart, that the only way that we could potentially try and even keep up with a superintelligence is to be somewhat, I guess you could say transhuman in the fact that we need to merge somehow with the kind of technology that we're trying to create if we want to keep up with it, because otherwise we're just going to be completely, you know, biological beings. You know, we're not going to be, I guess you could say silicone based. And unfortunately, that means we have hardware limitations. And whilst, yes, humans are remarkably more energy efficient, I think in the future that there will be these breakthroughs that will lead to a lot more energy efficiency. But the point that I'm trying to make to you guys is that, you know, super intelligence and, you know, these kind of super intelligent systems are going to be so much smarter than us and so much faster than us that it is pretty, pretty crazy. Now, this is actually going to lead me into the next point because I saw something. It didn't have a lot of likes, it didn't have a lot of retweets. But I think this, when this hits the horizon, most people, I wouldn't say they'll freak out because there are a little bit of reasons why it won't happen, but I think it's going to change everything. This isn't like some clickbait, change everything thing. I think this genuinely will. And I'm going to show you guys what I mean. So this video is going to give you guys a speed at which AI agents will navigate UI for us. So this is playwright web speed, running the payload CMS admin panel. And just take a look at this, because when I first saw this, I was like, whoa, that's what it's going to look like. This is what it's going to look like when we have AI agents running around doing tasks for us. Like, just take a look at how quick this is, running through different pieces of web pages and just doing just many different things. Like imagine an AI system that is able to do this locally on your device and that is able to communicate with other agents. Let's say there's. Let's say, for example, there's just, you know, 100 agents all doing things like this, able to do work that is remarkably effective, remarkably consistent, and is able to literally go around and web pages just do as much as possible. And with the fact that we do have things like grok, a computing company that is providing the fastest inference currently available. I don't think that these things are going to be impossible because imagine you had 10,000 agents doing work like this. And I know this might be, like, hard to even look at because it's moving so quickly and is, you know, pasting different things and writing and checking different things so quickly, but I could just imagine like 10,000 agents all talking amongst themselves. And here's the thing, I think this is going to be pretty crazy because what if we have agents that are just simply just discussing things that don't even concern us. Like, what if we have an ASI deploy 10,000 agents to work on a problem? They're all discussing research, all managing to get by without us. I mean, someone previously did state that when AI is going to be moving at this light speed and they're going to be able to decipher this amount of information, we're literally going to be like trees. If you don't know why that comparison is being made. Well, trees actually are moving. They're just moving on a time frame so slowly that we can't see plants move. Like if you move light near a plant, you know, the plant will eventually move to the light, but it's so slow that we don't consider it movement. That's why we consider, like plants, like these inanimate objects. But if an AI is moving this quickly and is able to, you know, inference it, like milliseconds, conversate in milliseconds, we are going to be like trees to that AI. So it will be interesting to see how that, you know, comes. And this was something that truly did shock me because it was the first kind of insight where I was like, oh, okay, yeah, okay, I got it. I get why this is completely crazy. Now, Apple have apparently closed their deal with OpenAI on the integration of ChatGPT on their devices, which leads me to believe that OpenAI are probably going to be releasing some kind of on device model agent in combination with Siri. And this is something that is not bad news because as many of you guys probably are iPhone users, even though some of you guys might not be, the thing is, is that Siri is a, I guess you could say another wave of AI that is going to be truly, truly transformative. Because whilst yes, it's easy to type in www.chatgpt.com or go on the app ChatGPT and type it in, I think ease of use is a bigger problem than most people think. And once people get into the habit of just clicking the Siri button on their phone and getting instant access to ChatGPT, that is going to give a whole, whole, whole different conscious class of people the ability to realize what AI can do for them. And it's going to make them realize that, okay, this is actually absolutely incredible that it can respond to me, it can write essays for me. Because whilst yes, ChatGPT is still widely used and still very, very popular. It still kind of tapers off in some of the capabilities and use cases. But with the advancements in terms of the reasoning capabilities and what it's going to be able to do for the average person, I think once it's integrated in a way that's so easy to use and, and a way that's native, where people don't have to learn how to prompt, how to say this, how to say that. I think that's why this is going to be pretty, pretty incredible. Now what's crazy is that Apple's Worldwide Developers Conference is going to be very, very soon. Apple's WWDC conference is going to be soon. And that means that we have a situation on our hands where there might even be some other AI announcements. So on 10 June to 14 June, I would say be sure to look out for whatever is going to be coming from Apple because they have been unfortunately remarkably silent on whatever AI announcements they have planned. And I think whatever AI announcements they have planned, I'm not going to say it's going to blow us away, but I think it's going to usher us into certainly a new era because they realize that they don't want to be behind on this and this is of course something that they really, really don't want to miss out on because AI is not something where you can slack off or where you can even fall behind for a day. Now we also had Mistral release a new model. This is Codestro empowering developers and democratizing coding with Mistral AI. So we can see here we introduced Codestrout, our first ever code model. Codestrout is an open weight generative AI model explicitly designed for code generation tasks. It helps with developers interact with code through shared instruction and completion API Endpoint. And it's pretty good. I would say that it's pretty, pretty good. You can see it's compared to Codelama 70B parameters and compared to Deep Sea Coder, it's pretty pretty on par with what we currently do have. And this is just continually. And there was also another Devon thing that was going on, you know, another software autonomous agent. But the point is, is that these coding agents and these coding systems that were now getting open source and stuff like that, they're continually increasing in terms of their capabilities. Which means that, you know, I think it's not going to be long before we do get like a specific code model from OpenAI that just outperforms everything that we've seen because From Google, we've seen AlphaCode 2, which is, you know, beats 85% of, you know, programmers in that competition. And with these open source ones slowly, slowly catching up to, I guess you could say specialized versions, I think with GitHub, Copilot, and, you know, the many different things that are being worked on, I wouldn't be surprised if this is something that really does occur and this is something that I nearly, nearly, nearly forgot to do. But this is OpenAI's secret plan. So OpenAI, after disbanding its efforts to build a general purpose robot in 2020, the AI juggernaut is embarking on a new attempt to supply models to other companies aiming to build robots of their own. So it says that with AI investments into robotics heating up, OpenAI is formally relaunching its previously abandoned robotics team. Forms has learned. So the company is currently hiring research engineers to build the team and they're yet to publicly announce the details of its homegrown robotics efforts. And it's pretty, pretty crazy. So they've invested in companies like Figure, AI, 1X Technologies, Physical Intelligence. They're investing in all of these companies so that they can use their systems with AGI to actually get it like an actual body. Because robotics is much harder than the actual software problem. And they've actually spoke about this previously. I'm going to show you guys a clip from that where Elias spoke about it. But you can see here, we've always planned to come back to robotics and we see a path with Figure to explore what humanoid robots can achieve when powered by capable multimodal models, said Vice President Peter Wallenda, previously a member of OpenAI's robotics team. So it says OpenAI intends to coexist rather than compete with such companies. And they're basically going to build the infrastructure that the robot makers will integrate into their own systems and that the engineers hired for the positions would be collaborating with the partners. So it seems that previously they had models that were good, but those models weren't good enough. So it seems that right now they are focusing on collaborating with existing partners to kind of beef up how insane these models are going to get.

Then it really wasn't possible to continue working in robotics because there was so little data. Like back then, if you wanted to do on robotics, if you wanted to work on robotics, you needed to become a robotics company. You needed to really have a giant group of people working on building robots and maintaining them and having. And even then, like if you. Only if you're going to have a hundred Robots, it's a giant operations already, but you're not going to get that much data out of them. So in a world where most of the progress comes from the combination of compute and data. Right, that's where we've been, where it was the combination of compute and data that drove the progress. There was no path to data from robotics. So back in the day, when you made a decision to stop working in robotics, there was no path forward.

Is there one now?

So I'd say that now it is possible to create a path forward, but one needs to really commit to the task of robotics. You really need to say, I'm going to build many thousands, tens of thousands, hundreds of thousands of robots and somehow collect data from them and find a gradual path where the robots are doing something slightly more useful, and then the data that they get from these robots and then the data that is obtained and used to train the models, and they do something slightly more useful. So you could imagine this kind of gradual path of improvement. You build more robots, they do more things, you collect more data and so on. But you really need to be committed to this path. If you say, I want to make robotics happen, that's what you need to do. I believe that there are companies who are thinking about such doing exactly that. But I think that you need to really love robots and need to be really willing to solve all the physical and logistical problems of dealing with them. It's not the same as software at all. I think one could make progress in robotics today with enough motivation.