MINDBENDING AI ADVANCES DON'T STOP! - Latest AI News & Tools

Welcome back, everybody. Today I'm going to be giving you my complete, honest opinion on all of the AI news that has happened in the past few days. And there has been quite a lot. I am not here to waste your time, so let's jump right into it. First up here, Claude and Chat GPT. Now, both have their own Windows and Mac applications. Obviously, everything, as always, is linked down below if you want to download these. But honestly, there isn't really much to say. It's the same thing as the web browser, except natively in a application on Windows or as I said earlier, Mac. They seem to both be updated pretty well, giving you pretty much all the same features that you would find on the web version of the chatbots. It's a little bit more convenient. You can pull it up very quick. Chat GPT, what would you say to Claude right now? The Chatbot, Hey, Claude, how's it going over there in anthropic land? Sure, we'll send that over. And yeah, you could see they both work pretty flawlessly. They're nice, clean apps. Claude also has support for their projects in here. Custom GPTs are supported in the Chat GPT desktop app. There is one feature, however, specifically on ChatGPT that might interest you guys the most, and that is we actually have advanced voice mode. So if I click this. Hello, ChatGPT. Oh, can you even hear me is the question. Oh, it wants to use my Oculus. This is a known Windows issue. We're gonna make sure it uses the right microphone. So go to your hardware and sound. Go to sound. Default recording device. We're gonna want to set that to this microphone right here, not the Oculus virtual audio device, which is my quest 3. Hey, chat GPT, can you hear me now?

Loud and clear. What's up?

Oh, what's up? How are you doing today? I'm just recording a little bit of a YouTube video, talking about some new updates. What do you think about that?

That sounds awesome. AI updates are always an exciting topic. There's so much happening in the AI world, from new models to groundbreaking research. What's the latest update you're covering?

Well, right now I'm just talking about how you're now available in the Chat GPT desktop app. But also I wanted to test one more thing. Search just came out in Chat GPT, so I wanted to know if you had the ability to do search.

Ah, the ChatGPT desktop app. Nice. As for the search feature, I don't have the ability to perform real time searches on the Internet or access up to the minute information.

My response, well, that's boring, bro. All right, get out of my face.

Got it, bro. Catch you later.

Anyways, yeah guys, that leads us into our next piece of news which is advanced search is back in ChatGPT. As you can see, if I just go to my Chat GPT and then I click this little search the web button. Now we have search and it has access to real time web information. Again, if you remember, I think it was over a year ago Search was initially introduced to ChatGPT and then they took it away after a little bit because it wasn't very good at all. This is way better though. Who is Matt Vid Pro AI? It will quickly search the web and give you an answer that is pretty good YouTuber. It gives a specific sponsorship that I did once platforms have introduced AI voice models named after him. Okay, interesting. And then you can go to the sources on the side here. I am going to say, and I have to be real with you guys. While it is really nice to have real time search back in Chat GPT, I still feel like Perplexity AI is so much better. Even the free version. If you guys have never tried Perplexity AI before, I highly recommend you do because it is easily the best AI search and in a lot of circumstances better than Google. If I say who is Matt Vid Pro AI in here? And this is again just a basic search, not even a pro level search, this is something you get for free. Not only do I immediately get some recent thumbnails and pictures of myself on the side, I like the way it displays the sources a lot better actually in the text right here. So I can see which specific sources relate to whatever part of the text it's talking about. And I think it genuinely gives you a better overview of who I am. For example, obviously tells me I'm a YouTuber. Here's his profile and work. We have channel and content, educational tutorials, updates on latest developments, demonstrations. That's very accurate in comparison to just specializing in artificial intelligence and machine learning content. It still does say tutorials, reviews and discussions, but this is just better with the bullet points. I like that we have channel statistics as well, giving you the subscriber count, how many uploads per week, average video length, content focus, AI voice generation tools, updates on large language models, AI video generators. It's. It's just a lot better, I think. And I'm actually a pretty good test for this because I'm niche, you know, it's not like you're searching for some massive YouTuber. This is something that's not really searched for all that much. I'm going to do another test here. What is the link to his Discord server, which you guys should join by the way. Fantastic AI community that's really up to date. Alright, so Perplexity gives me a link. Matt Vidpro Discord. If I click that. Ooh, this gives the link to the old Discord server, which is not the one that you want. All right, let's see how ChatGPT does. Oh, this is really, really interesting. So OpenAI search did give the correct link to the Discord server. If you click that, this is the correct link, so it gives that immediately. But then it also says you can Join met VidPro AI's Discord server using this link. And this is literally just a link to a live stream that I did, which I mean it does have the link to my Discord server in there, but it's, it's not a link to the Discord server itself. So strange. Looks like both Perplexity AI and Chat GPT web search have their own issues still. I'm going to give Perplexity a quick redemption arc though. That's the old link, I want the new one. Wow. Okay. It tells me some good ways to go find the latest link, which is true, but it won't give me the correct one. But will it work with a Pro search? A Pro search worked okay with Pro search on Perplexity AI. It gives you the correct new link. Okay, pretty cool. Quick TLDR Chat GPT search is very much useful. It's directly in chatgpt. So if you're already doing something in chatgpt and you need information that's real time, it can just go nab that for you real quick and it works decently. However, if you are looking for the most accurate possible AI focused web search, you're going to want to use Perplexity specifically the Pro option. It seems like there is a little bit of head to head competition between the free version of perplexity and ChatGPT web search, at least on the surface. Again, it can depend for a wide variety of searches. So let me know what you think. Get out of it personally in the comments. Let's move on this one right here guys. You need to be prepared for because this is going to blow your mind. This is Oasis. It's the first playable AI generated video game. This company Etched partnered up with DCART AI to build a real time interactive world model that runs 10 times faster than the last Version of this technology we saw, which I think I covered like only a week ago or two weeks ago, something like that. They're going to be open sourcing the model architecture, which is just Chef's kiss, man. Love to see that. The weights and all the research, all open source, really beneficial for the community at large. It's respectable. It is the best way for this technology to move forward and to drive competition from real corporations in this space. There's a demo you can play, but it's actually a little bit flooded by users already, so I don't know if we'll even get to test it out in today's video, but we've got a video and I. Like I said this is gonna blow your mind. Check this out guys. So this is all real gameplay, again, fully AI generated. Obviously they're copying Minecraft here, but we have breaking blocks, we have running around and jumping. Looks like it's running at about 30fps, but there's no Minecraft being played here. This is all AI generated, which is just absolutely staggering. And you can definitely tell in certain circumstances. There's some fuzziness, there's some weirdness, it's very low resolution, but it looks just like freakin Minecraft to like a shocking degree. I played a lot of Minecraft. A lot of you might know. I used to run a Minecraft YouTube channel actually until it was converted to AI. But I mean all of the colors, the background, everything looks really good. The amount of hearts is correct, the amount of slots is correct, and the block breaking animation is pretty much perfect. The amount of particles disappears a little too quickly, but again, that's pretty minor problems. It's still playing Minecraft. Obviously the whole game isn't going to be here when certain things happen. Probably not going to work perfectly. But I assume the basics are there. The very basic level, break blocks, pick up blocks, explore the world, jump punch. I don't know if creatures or animals are in it. Just totally insane though, to think that this is all AI generated. Like you can see some of the grass here is a little bit wonky, but the textures like that dirt texture looks good. The block placing looks good. The jumping in the water was really crazy to me too. Just swimming down, look, this one has like a whole slew of different items too. So I don't know if it starts with that, if they were able to set program that in or what, but I mean it's just absolutely insane. I also thought, wow, this is genuinely a copyright issue. I know that this is just gonna be all Open source and it's a demo. Hopefully the devs of Minecraft Mojang don't really care all that much because I think this is great research. But yeah, I mean, this is basically just copying the game. Minecraft straight up, but using AI technology, it's able to essentially replicate it perfectly. And so much so, I think to a degree it could be considered copyright infringement. I'm no lawyer, but that's crazy. It generates frames based on your keyboard inputs. It can move, jump around, break blocks, build, explore. It runs incredibly fast, obviously, to generate these frames in real time. It's on their upcoming chip, which they call sohu. It's built to handle, quote, unquote, massive video models running 4k for 100 plus billion parameter model. Much better than an H100 apparently. So obviously you still need way more compute to run fake AI Minecraft rather than just playing the real game. But the idea is that it's slowly going to go down and then meet in the middle. And then eventually it's gonna make way more sense to play your games with AI and you'll probably have photorealistic perfect graphics that's probably coming in like, I don't know, 10 years. So cool. Even Elon Musk had to jump in there. Wow, this is happening fast. Can we play Minecraft ourselves? Yes. Yes. Demo. Let's go to the rugged coastline. Waiting for your turn. Begin your journey. Okay. Oh, my God. This is crazy. Dude. Yo. Okay. It looks like garbage for sure, but. And it's running in a little bit of a slow motion, but it's pretty good. Can I double? Oh, my God. I can sprint. It actually works. That's incredible. Inventory. Oh, my God. It's trying its best. It's a little bit slow. It's hard to believe that I'm not really playing Minecraft, though. This is all just AI generated. Can I change my items? Go to 1. Break something. This is obviously gravel we're looking at. Got to give it some time to render that out maybe. Oh, my God. It's having a little bit of a. Bit of a tough time for sure, because it just opened up a whole new cave. I feel like I'm, you know, tripping balls or something here, guys. My word, dude, this is crazy. It's kind of more consistent now. Oh, it's having a hard time figuring out where I am, man. This is so freaking mind blowing. Time left. I only have a couple minutes left, guys. We got to make the most of it. I'm lowering the sound. Continue. There we go. Let's Try going in the water now. I seem to be maybe just a little too short as well in some circumstances. It's working, though. It's. It's so crazy. Look at the distance. This is the beginning, guys. This is the. The first birth, the first AI generated video game you can play in real time. Holy cow. Oh, man. It does feel like you're playing the worst quality YouTube video ever. It definitely needs work, but it's here. Let me into the water. I want to see the water. Oh, okay. I think we made it. I think we made it in the water. Yes. Oh, and I think we're kind of floating because it doesn't know the difference, really, between creative and survival. There's also a free crafting table here and. Oh. Oh. What's going on here? Where are we? It's so. It's like the environment is constantly changing around me. Let me fly. Can I fly in the air? Dang. That's crazy, man. It's pretty consistent about your items, though. Can I drop items? No, not really. I can place them, though. Yeah, that sort of works. It's definitely not very consistent. I think that that's its biggest issue right now is the consistency and obviously the overall quality. But no, man, it works. It works. It's doing a lot of weird things, but it. It works. Yeah. What's in this random chest that's here? We have all kinds of stuff. Give me this item, whatever the hell this is. Oh, man. It's having so much struggle. You could tell it's so rudimentary. Oh. Oh, okay. Oh, man. Yeah. I don't know how they're gonna solve that consistency issue. Don't have much time left, but it's definitely Minecraft being generated. I don't know. I might have to take back what I said about the copyright stuff because this is pretty dang unplayable and unusable, but it's enough to just be like a proof of concept. And the fact that they're open sourcing it is even better. Oh, man. Can I break stone? Okay. Wow. You can definitely break stone way too fast with the dirt. That is not correct. And we're getting freaking items from it too. Nice. Love to see it. I think once less is going on in the world as a whole. Like if you're in a cave, it starts to become more consistent. Look, it just opened up this whole massive cave. It just spawned me in a cave. That's so trippy, man. It's so strange. Wow, that's just crazy, guys. I'm. I'm blown Away. At this point I can't see anything. Something bad has occurred. I think it's. It's like slowly closed me off into a cave where I can't see anything. Where am I? Please help me. Why is my health keep going down as well? That's a little concerning. I'm. But I love stuff like this. I love little demos and stuff we can actually use. Give me the torches. Where am I? Please light up my world. There's my gameplay. I can download it. Oh, there it is guys. Absolutely insane to think about. So yeah, guys, I'll link that down below. You can go test out the demo for yourself. Hopefully you can get in. Alright. Mind blowing. Obviously though, this is like baby steps, baby steps. It's barely playable, it's barely usable. It looks garbage, but they're gonna fix it eventually one day. I mean, think about how AI generated images started out, right? They looked horrible and now they're in a lot of cases indistinguishable, at least at first glance from a real photo that might have been taken or a real piece of artwork. So there you have it. Alright guys, let's jump into image generation now. So if you remember my last AI news video, I talked about a mysterious model that was sort of topping the benchmark charts on various AI image generation blind comparison websites. And we now know what this mysterious model, which was named Redpanda is. It is Recraft V3. It's a quote unquote revolutionary AI model that thinks in design language. It delivers unprecedented quality in text generation, outperforming models from Mid Journey, OpenAI and others. Now they say it's very game changing and it's a good model, I'll tell you that right off the bat it's impressive. It does Photorealism well. It adheres to text very well. Check out this text generation for example. This is a lot of text to put on one screen, especially for an AI image generator, right? And if you use this model on their website, it has very precise control. They have their own canvas mode where you can create all kinds of things with plenty of other tools as well. And it's impressive. I used it. It's also got style input as well, which a lot of people like to see. And they have an API for it as well with vector art and style consistency support. But I've got to be honest with you guys, I am a huge Ideogram fan. So my initial impressions were, alright, let's test this thing against Ideogram for the ability to adhere to prompts. And here was my first test. Simply put, a 3D Pixar animated style lemon character holding a bag of Doritos. He looks upset. On the left hand side here we have the Recraft V3 model, aka Red Panda. And on the left we have Ideogram 2.0. There's a very clear difference here. Not only do we get the real legitimate Doritos logo, like from the company in a nice Doritos bag, spelled correctly and everything, but the Lemon character also looks fantastic in that typical Pixar style Recraft Red Panda. It doesn't even spell Doritos correctly and it's not the real logo. We have some subject confusion going on with the orange slices on the bag itself. And I think that there's just less detail and less professional quality coming out of the Lemon character as a whole. Even members of the community jumped in. We have Michael DeFore here trying it on Flux Dev Local. First try gets the Doritos logo correct as well. The bag is open, so maybe some points lost there, I guess, but not really. And I think the Lemon character also looks pretty great. I think this is very comparable, in fact to Ideograms, even more so than the result we got out of Recraft V3 or Red Panda. I then moved on to do a more detailed, complex test and this is a big prompt. First of all, this prompt is to be in first person view, and then we want a massive mutated Lemon Monster roaring with glowing red eyes. And in one hand he's holding a minivan. The setting is New York City and destruction can be seen throughout New York City where the Lemon Monster has been. In the first person view, we see a lightsaber being held in the right hand and then a cartoon style bomb in the other hand. And it's dramatic and cinematic. This is the result that we got out of Ideogram and it's still not perfect, but it's pretty darn close to the prompt. Mutated lemon monster, glowing red eyes, New York City all there. He's holding the minivan and he looks pretty good, I have to say. There's lots of destruction in the background. And we do have a lightsaber kind of jutting out in like a first person style view. It looks like it could be from a video game maybe. Right. That's sort of what I was going for. We did get the cartoon style bomb, although the Lemon the monster is the one that's holding it. Now here's what we got from Recraft Red Panda and I think this is very Telling we got this big monster, but he doesn't look much like a mutated lemon. Instead, we have another person who's holding a lightsaber. He's not holding a van. He's. It's just not all the way there. And I guess this could be considered a first person view, but it's really not what I was asking for. So you can see there's a pretty big difference in terms of the model's ability to understand your prompt really clearly. And this is really what Ideogram excels at. It's pretty much the reason that I constantly am using Ideogram over any other AI image generator. Still though, that's not to say Red Panda isn't good, because this is stability 3.5 medium. The same prompt generated by yes hand. I don't know if I'm pronouncing that correctly. Anyways, I think that it absolutely screws up here with the first person view and the lightsabers. Like what is going on in this corner here? The Lemon Monster maybe looks a little bit better. Still not holding a minivan, but I think that this is a more acceptable image from Red Panda here in comparison to stability 3.5 medium. However, you'll see Dolly3 did a pretty darn good job as well. Cody Baker posts his result in here. We have a very first person view for sure. Like it's an actual hand holding the lightsaber. The Lemon Monster has a lemon head and he looks terrifying holding the minivan as well. Relatively speaking. We even have the bomb in there. So Dall E3 is pretty impressive. I think a little bit of the smaller fine details though are mushed up more on Dall E in comparison to the other models. But Flux 1.1, man, Flux 1.1 comes out here honestly, competing pretty heavily with Ideogram, almost like side by side. I still think Flux is the king in comparison to Red Panda, at least for prompt coherency. Just based off of these results, and especially Flux being open source, I mean, that's tough to compete with. Look at the way it holds the minivan. We've got the lightsaber in hand and it's just got these glowing red eyes. Menacing. Totally awesome. Here is another Flux Dev result. Continuously getting the lightsaber in there with the Hand Monster. With the minivan. Again, he has a human hand here, which doesn't really match the rest of his body. We've got too many fingers on this side. But it's a Lemon Monster, pretty darn good. Not saying that this new Red Panda model is useless. Not saying that it's bad, I just don't know if it's fair to say that it has next level ability to understand what you want from your prompt. That is definitely not true, at least in comparison to Flux and Ideogram as well as even Dall E3 in some circumstances, as with any image generation model, there is going to be specific areas where it excels. Very far, faux Fur AI brings up that so many models struggle with wet things and soap suds and that is absolutely true. But Recraft V3 is able to get it very well. Plenty of detail here. Looks fantastic. We've got this octopus and I gotta say the coherency with the octopus is pretty impressive. That's a very difficult animal to get correct for image generation models. We've got the tiger in the suds. Looks pretty dang good. This wet lion is also markedly very impressive. You can still tell with some of the finer details. It's like, dang, that's definitely AI generated, but it's a, it's a good test for sure and it shows you that like I said, Recraft V3 is by no means a bad model. It is by no means a useless model. And this is the website where you can actually try this new model out. And it is by no means a bad way to generate with AI. They've got a ton of useful tools in here and I wouldn't blame anyone at all for using this website over something like Ideogram or Midjourney. You have great control over aspect ratio. They have a bunch of different fine tunes like vector art, illustration line and fill for this Recraft model, which is really nice. They've put the work in here onto this website and UI and I think it's very good. You actually have the ability to go ahead and draw in text or bring other images into your prompt. Very comparable to Ideogram Canvas I would say. And maybe Ease even has like a few more features that you don't see with Ideogram and they have a pricing plan that's pretty reasonable I would say. But yeah, those are my thoughts on this new Recraft model. I don't think it's any Kingslayer, but it certainly isn't useless and it's definitely worth knowing about and mentioning. Next up guys, we have a new state of the art text to speech model. It's got zero shot voice cloning, emotional text to speech, long form synthesis, varial speed synthesis. It's bilingual, Chinese and English and it's available on Hugging Face fully open source, by the way, Model waits. I don't think that you can do commercial things with it, but it's still open source, so still promotes growth in the community. It really is quite good. Take a listen.

Would you guys personally like to have a fake fireplace, an electric one in your house? Or would you rather have a real fireplace? Let me know down below. Okay everybody, that's all for today's video and I hope you guys learned a bunch of furniture vocabulary.

I still think 11 labs is better, especially more consistent at pronunciations. However, this is open source, this is free, and it's pretty dang good. Need to show off your latest web app, but don't have the video editing skills. Try screen run today. AI powered zoom, smooth cursors, automatic subtitles and video backgrounds will turn your boring screen recording into a viral TikTok video. It's close. Not as good as 11 Labs, but close. You see, 11 Labs is really, really damn good. Listen to this. Sounds just like me. Let's pronounce some really tough words too. Worcestershire. This word is often mispronounced as Worcestershire. It's actually pronounced in three syllables. Quinoa. This word is not pronounced. Kinoa. Onomatopoeia. This word is long and difficult to pronounce, even though the words it imitates are usually short. It did get onomatopoeia. Incorrect, I think. I don't even know if I pronounced it correct there. But yeah, 11 labs is really good. Yesterday on my Discord server, this new voice generation model was actually cloned and tested on my voice. So here's how the new open source model sounds with my voice. I'm so done with OpenAI and their stupid sore release. Like seriously, why is it taking for keep teasing us with this next big thing, but all we get are vague updates and nothing substantial. It's such a joke. They're just gatekeeping it like they think they're some elite club. I don't get it. What's so special about it that we can't see? Meanwhile, the rest of us are left in the dark and it's beyond frustrating, I swear. They need to stop dragging this out. I'm not missing anything with this nonsense. Just us. Give us the damn product already. You could see it definitely sounds like me, but it skips over some words. It's not as clear and robust, I think, as 11 labs. And also guys, like I said, you gotta join the Discord server because they're already testing the latest and greatest, especially this new Minecraft model. Seems like people have been using it all day on the Discord server. Oh, and of course they're also testing out the lemon and Doritos thing as well. Oh, and what the. What is this? I gotta get off of here. We've also got some quick snippets here. This is CAM I2V. It's a method that can generate videos from images with precise control over camera movements, which is cool to see. First row here is Zoom in, second row is zoom out, third is Orbit left and fourth is Orbit right. And you can see, you know, obviously this isn't great quality generated video, but it's cool to see little projects like this that can do specific camera movement. I think the camera movement here is pretty impressive. Not something we haven't seen before, but I like to see these little projects because if they get open sourced, it can benefit everybody. Apparently this one has code coming soon, but it does have a project page on GitHub if you're interested in checking this one out. And if the code and weights, etc. Are coming soon, stuff like this could be applied to models like Runway gen 3, for example. Larger companies can take this, learn from it, and then implement it. Next up we have a release from Meta called Longview. It's a video LLM that can handle long videos. I don't know if this is the second video model that they put out or if it's the first one still, but either way it's really cool. So I don't care if I'm mentioning it again, and I've already talked about it previously, you can upload videos and it can actually understand the context and what's going on in the video you send, which is cool. Right now we have LLMs that can only understand images, so having ones that can understand video over time is a lot bigger of a deal. People in the video are standing and watching a performance shows. A crowd of people gathered around a performance area, some individuals standing and others sitting on the ground. The performers are playing musical instruments and singing audiences enjoying the performance. And yeah, I mean it's only 10 seconds, but that's still considered long video right now for large language models. Also, all these models are open source. Murf couldn't find a specific license here, but I think we can assume it's a good open source license. Meta is usually pretty damn good about that. We've also got something new from Blockade Labs and they call this Blendbox. This is another interface, another canvas style interface to actually create with generative AI. Although they do have a few things in Here that are different. For example, in this demo, they start off by just generating a basic forest path, and they get that output, and then they generate a woman to be on this forest path. And in Blendbox, apparently they have a model that can generate cut out images of people, for example, or anything. So these images don't have a background. They're basically PNGs, right, that you can drag around the image, so you can put her anywhere you want in the path. And you can see there's an overall output that sort of blends everything together, adds shading, adds detail where necessary to actually make it look like this woman is on a real dirt path. And they sort of go on to add a skyscraper, add some waterfalls. You can see it blends it pretty seamlessly. And you can add all of these fine details overall to get whatever image output you're looking for. It's pretty cool. And then they go through to change the art style and do a few other things. It's. It's cool. Honestly, though, I still don't see this replacing my current workflow. So I did check it out and it looks like you have to pay to actually start to use it. There's no free access whatsoever. If you guys want to see a full video on this, please let me know. I will pay for a plan and we can test it out if you think it's really that interesting and really that cool. Definitely thought it was worth mentioning just to see what you guys think. Now for my final piece of AI news today. Apparently AI has achieved something called sent teleoperation. Rowan Chong does a really, really good job at breaking this down. Photos, videos, and audio have been digitized for decades. True. And now smell is joining the digitization. This company called OSMO has made it happen. They've got a video down here where they essentially take a plum. They take a little sample of the plum. They're able to digitize the plum's smell somehow. I have no idea how they did this, apparently with AI and then they just produce a reprinted liquid. It's just a clear distilled essence of the scent of plum. And they both apparently smell exactly the same. I would love to, you know, experience this in person, but that's pretty crazy. I mean, smell is actually a really important function of the human. It does a lot for memory. It does a lot for immersion. So I think that it could be really impressive to have like a smell sense going on when you play video games or watch movies or something like that. It's something that actually has been tried in the past, I think, at Disney parks in particular. But this is something totally different because they don't even really need the plum. They only need the plum once, and then they can just make a liquid that smells like plum. And apparently they could just do this for anything, I guess. I would love to know what the limitations are. Yeah, it's like Holodeck tech, as Sakaar points out. You just print scents now. Okay. AI technology is always blowing my mind, man. Anyways, if you want to look deeper into it, please read Rowan Chung's little post here. AI just blows my mind every day. There's something new to explore constantly, and that's why I keep making these videos, and I guess that's why you guys are watching them for a nice recap. But I think you guys should definitely check out the Discord server if you want to see the latest and greatest of the AI world posted right when it happens. And if you want to join a community that is really passionate and excited about those latest AI technologies, I'm also usually in there offering my first impressions, my first thoughts, my examples of using the latest AI tech, and hey, it costs you nothing to join it. But thank you so much for watching, as always. I appreciate you guys, and I'll see you in the next video. Goodbye.