The AI Sandbox - CES, Nvidia News, Cool AI Tools and More!

We're going live. We're going live. Welcome to the AI Sandbox. We're going live. Welcome to the AI Sandbox. We're going live. Welcome to the AI Sandbox. We'Re going live. Welcome to the AI sandbox inbox. We're going live. Hosted by Matt Wall. We're. Hosted by Mad Wolf. It never gets old. I love that little animation. Anyway, hey, thanks for joining me live. This. This is the, I don't know, third or fourth time I've done these weekly live streams. If this is your first time joining me on one of these live streams, I literally come into these with no agenda, like, no talking points. I just kind of let the chat in, the comments steer where we go on this show. And we just have fun and chat, talk about the latest things going on in the world of AI. We'll talk about news, we'll test some tools. I'll share my thoughts on ces. So if you have questions or things you want to talk about or tools you want to test or, you know, news you want to chat about or anything like that, throw them into the comments. This is the first time I'm actually testing going live on X as well. I've been going live on YouTube. I've been going live on Twitch. I've now added X into the mix, so hopefully. Actually, as of right now, I think I've got more viewers watching on X than any of the other platforms. I've got Dylan in the chat. He goes by DW in the chat. So if you see Dylan, he's the moderator, he's actually watching the chats and comments for me and then pulling them up for me in a Slack channel so I can sort of stay focused and focus on the. The conversation and what I'm looking at on screen. So if. If Dylan says something to you or tells you to be cool or to stop doing something, listen to him because he does have the power to boot you out. But let's just go ahead and get right into it. I'm already getting some messages over here from Dylan. I'm going to actually switch so that my. My screen is up as well. I did just get back from ces was. It was pretty fun. CES is really, really exhausting. It's basically four days out in Las Vegas and it's spread across, like, 10 different hotels in Vegas. So in the same day, I'm trying to be at, like, an event going on over in Mandalay Bay. And then like an hour later, I'm trying to be at the convention center. Which is literally the opposite side the Strip. And then later that evening there's an event at the Bellagio which is like all the way back at the opposite side of the Strip again. And so it just, it gets super exhausting. And there's 140,000 people that go out to Las Vegas for these things. And yeah, it's, it's fun, but it's exhausting. And I, I got back on Friday, obviously today's Monday and I'm still feeling exhausted from Cesar. So hopefully, hopefully I've got enough energy to, to hang out and go through all your questions and have some fun on this stream today. But I'm gonna go ahead and check in with Dylan here. He's, he's gonna copy and paste messages over to me in Slack so that I can kind of stay focused. So he says. Phil B. Hey Matt, big fan. Why haven't you told the world about vs AI yet? They are presenting World Economic Forum next week. There will be a disruptor in the AI space versus AI. It's actually not one that I know a lot about. I might have come across it in the past. Let's see, what is Versus AI? So imagine a smarter world versus the cognitive computing company building next generation intelligent software systems modeled after the wisdom and genius of nature. Yeah, this is one I haven't really dug into yet. So. Hey Phil, if you want to tell me a little bit more about it, what excites you about this versus AI? We can, we can kind of dig in on this. Akshit Gar and we talk a little bit about large concept models. Apex, tell us more about the Nvidia supercomputer please. When will it be released? I believe they said that one was going to be released in May. Um, but I'm not a hundred percent sure. I mean they, they didn't give me access to one. I don't even believe they're in production yet. I think when Jensen was on stage and he was showing them off, he was saying that, you know, there's only like literally a few units of it built so far. But my understanding on those is that they're coming out in May. Let me actually just double check that. I think I have like a news article that confirms that here. Where is it? Project Digits. So I think this article actually mentions when it's coming out down here somewhere. Maybe, maybe I'm wrong. Oh yeah. So Project Digits machines which run Nvidia's Linux based DGX OS will be available starting in May from top partners for $3,000. The company said I did see one in person, actually have some. I think. Do I have some B roll of that? I think I did actually shoot a little bit of B roll, but I didn't actually finish naming all my B roll, so I don't know which one is. Let me see if I can find it real quick. Yeah, actually, I don't want to pull it up because I don't know what. My B roll is not labeled properly yet, unfortunately. But I actually showed off some video footage of it in my latest AI news video. So you can check that out. You can see, like, sort of what ports it has on it. It's got, like, an Ethernet port on it. It's got, like, three USB C ports on it, and it's got an hdmi. You could, I think, directly plug it into a monitor and just use it as a standalone computer. But it's designed to be. It's designed to be like a cloud computer on your desk. It's designed for you to basically wire in from another computer, let the inference and any sort of training or things like that run on the device, and keep your main computer freed up. That was sort of the. That was sort of the main idea. All right, so what was the coolest thing you saw at ces? So I did want to talk a little bit about ces. That was like, again, I don't really come into these with an agenda of, like, a lot of things that I want to talk about. I love just going wherever the conversation wants to go. But, you know, the. The one thing that is obviously on my mind was ces, because I was there last week. And I think my initial thoughts on it were I was kind of underwhelmed this year. I felt like there wasn't a giant leap between what I saw at CES in 2024 and what I saw at CES in 2025. I think the Nvidia news was some of the biggest news. We'll get into that. There is a little bit of controversy around the Nvidia news. I put out my news video on Friday and was talking about the various Nvidia news that they announced, and a lot of people were freaking out because, you know, Jensen said some things on stage about the Nvidia video card, and a lot of people are sort of refuting his claims, and I do think they actually have a case for why they're refuting his claims. But at the same time, I also did use these devices in person. I actually got to go to a private Nvidia event most of what they talked about there is still under embargo, so I'm not actually allowed to talk about it until, like, later this week. But one thing I can talk about was they did give us actual demos of these 50 90s running. And pretty much most of these demos, they would have like a computer running a 5090 and a computer running a 4090 side by side. So we can compare the difference. And like, the difference felt pretty mind blowing in person. So, you know, of everybody that's sort of complaining and sort of getting all up in arms about like, oh, it's not even that big of an upgrade. It's not that, you know, the, the stats on these things aren't that big of a difference. I mean, the reality is when you see them side by side, it feels like a big difference. And I did see them side by side, but other than the Nvidia stuff, I was again sort of underwhelmed. Like last year, the big thing that was at CES that everybody was talking about was like, these transparent TVs. This year it was kind of more of a lot of transparent TVs. I saw a lot more of those. I do have some B roll that I can pull up on those. Like, the LG had some really cool. Here, let me, let me find it real quick. LG had these, like, where is it? I think it was lg. Where is it? Sorry, there's a lot of silence on my streams as I look up stuff as I'm talking. Let me find this, though. There was some really, really cool, like, wireless clear TVs that I thought were pretty, pretty impressive looking. But I have so much B roll that it's, like, hard for me to find exactly what I'm looking for because I shot like 10 hours of B roll while I was there. Where is it? I need to do a better job of titling this stuff so I can find it easier on these streams. See? Oh, here we go. This is what I was looking for. We don't need the audio here. So this was like at LG's booth. These are all wireless clear TVs. Last year they had the clear TVs, but they weren't wireless. This year, it seemed like wireless TVs were much more of a thing. A lot of people were showing off wireless TVs, but you can actually see, like, these things will start to move here in a second. But they're all clear. You can see directly through them. They're clear, they're wireless and. All right, come on, show it. Move yeah, see there you can see like right through them as they all, like, move. And you could get a little better glimpse at what these things look like. They just look like clear windows, but they're actually transparent wireless TVs. That was one of the cooler things that I saw at CES were these like, these TVs here. But again, like, sort of underwhelming. There was a lot of EV stuff there. Again, um, oh, the other thing that was really cool. We've also, we've seen sort of videos of this. It was really cool to see it in person where these like, xpeng, like drones. Let's see. So, like, these like, personal flying drones that are like little two seater drones that the. Here's another, like, shot of the drone from the front here. By the way, these. These videos are all shot in log. They're not color graded yet. So I apologize for that. But this is the best I got at the moment. So I've got a few angles of this, like, drone here. And then the cool thing about this is that it actually like wheezes into this like, van thing here. And I'm gonna walk around on this shot here in a minute. And you can see, you could see like, this is the drone right here inside of the back of this, like, truck. I think I actually have some better shots of it here. So here's like a better shot of it with less people around. But that was actually really, really cool to see in person. I've seen videos of it online, but to see it in person is pretty surreal looking. And these are actually in production. People are already paying and ordering these and that apparently you're gonna ship and people are gonna be getting them. Yeah, I see Divine within says head chopper 2000. That was literally like my first thought when I was looking at these is it feels like, you know, I'm six foot three and it feels like those blades are like right at head level. So that. That really, like, scares the heck out of me. But yeah, that was one of the cooler things that I saw at CES as well. But yeah, it was. This year felt more overwhelming to me than last year. I don't know why. It felt like more people. It felt more crowded. There was construction going on at the convention center, which made things feel even more hectic. It's like really, really difficult to get really good B roll during CES because there's so many people. Like, I would be sitting there like filming something and somebody would just walk and stand right in front of my camera, like ignoring the fact that I was filming, which was sort of frustrating. But yeah, those were some of the cool things I saw. They, they did show off more stuff with like AI in it, but I felt like most of the stuff that now has AI in it felt kind of gimmicky. So I thought one thing that might be kind of cool to do on this stream, and I actually haven't even looked through this yet, but like the CES website put up their awards of like, you know, the most innovative and coolest things that came out of ces. So I thought it might be kind of cool to like skim this and show some of the stuff off that came out of ces. But before I do, let me just check in with Dylan real quick and see if there's anything coming across that I need to respond to here. So, AI music, which AI technology or breakthrough do you believe will have the greatest impact in the next five years? Let's see. I'm gonna come back, I'm reading some of these questions out loud, but I'm gonna come, I'm gonna circle back around to them. I just don't wanna. If you've watched my past live streams, you know, I have a tendency to sort of jump around from thing to thing to thing to thing a lot. So I'll like, talk about one thing and then I'll go to another thing for like three minutes and then I'll jump back to the first thing I was talking about and then I'll jump to a new thing and then I' back to the second thing and it's, it gets really hectic. So I'm, I'm trying to do my best in my brain to, to sort of keep things a little tighter and a little more organized and stick on one thing for a little bit longer. Because that's something my ADD brain has been bad at, is just kind of focusing in. I tend to, you know, when I'm working on a normal day to day basis, I tend to jump around a lot. I tend to like look at some news and then jump over to a tool and then jump on X and see something buddy talking about something. And then I go and click on the link and then I read the article and then it links to a YouTube video and I'm watching the. And so I'm just like, that's like what my daily life looks like. I am just jumping around all over the place, watching videos, reading articles, jumping in X scrolling Instagram of my like AI feed over there and I'm, I'm trying to rein it in while I'm doing these live streams. But if you've watched my live streams in the past, you know I'm not very good at reining that in. All right, Brian Mooney, pick up any cool or interesting swag from the conference? No, this was actually the first year I specifically opted not to grab a bunch of swag. Last year I came home with like two bags full of stuff at CES and most of it ended up in either in the garbage can or in a drawer and I've never looked at it again. So this year I specifically decided I'm just gonna grab footage and I'm not gonna like take any of the swag or stuff home. I don't tend to wear like a lot of brands on my live streams, so I don't really care about getting the shirts with other people's company logos on it and things like that. That doesn't really interest me that much. The one thing that I do tend to pick up if, if companies offer them is water bottles. Actually you can't see it, but I've got like a giant collection of water bottles on my bookshelf to the right. I've got 1, 2, 3, 4 Google water bottles, an Insta360 water bottle, two Nvidia water bottles, Microsoft HubSpot water bottle. For whatever reason, I like grabbing the water bottles, but I didn't actually see many this year, so I didn't grab them. So Jason Viers, have you played with any multi level agent frameworks like Relevance AI? I'd like to learn more. We'll. We'll circle back around guy t1llm. Which smart glasses are better met? Smart glasses are better Meta Ray Bans or the secret glasses from Google you had in London. We could circle back around. But the secret Google Astra glasses, they're not much of a secret anymore. They've been announced. But those were definitely quite a bit better. Just because they had the heads up display. They pretty much did everything the Google metaglasses did, but they also had a heads up display in them. So like they had that one extra thing going for it. All right, let me get back to CES here. So this is. These are all the things that won the awards at ces. A lot of it I didn't see like there's there. I think if you're there for four days, it's impossible to see everything at ces you kind of have to pick and choose which areas you want to go to. So there's so much that I actually didn't see, I did watch the Samsung Keynote. The Samsung Keynote was interesting because it sounds like what they're trying to do is sort of blend the Internet of things with AI. So everything is sort of communicating and talking with each other. That concept was really, really interesting to me. So basically, for example, let's say you have like their Samsung Smart ring. They have. They're building technology so that, like, you have a smart thermostat. And if you're sleeping at night and your ring notices that you're getting really, really hot, it will interact with your AC and it will crank up your AC for you because it's noticing you're getting hot at night. You know, they also have like, smart window covering mechanics that will, you know, open and close your blinds for you, things like that. So, like, at night it could notice or you can tell it you want to wake up at a certain time and have it sync up with your alarm clock. And when your alarm clock goes off at a certain time, it automatically opens your blinds and lets natural light in. They have different things with, like, the laundry machines that interact with different appliances in your house. And I don't even remember all of the, like, weird connections they were making, but they're basically talking about AI that sort of preempts your needs instead of AI that waits for you to prompt what you want. So like I mentioned with the ring, it notices you're hot and should need to cool down. Well, it will preempt that and turn on the ac. They showed off a Samsung Smart fridge that integrates with Instacart and it looks inside of your fridge and when it notices you're out of eggs or out of milk, it just automatically orders it from Instacart for you. And then the same day that stuff will arrive. So it will like, preempt your need for that stuff. That, that kind of thing was really interesting. But if you watched my previous livestream where I used an AI agent to order toilet paper, you can see how that kind of stuff could go wrong. It racked up $700 worth of toilet paper when I asked it to go buy me a roll of to. So some of that stuff kind of I don't know how it's going to work out yet. Like, if things are hiding behind other things, is it going to be able to detect that you have it? Things like that I'm not totally sure about yet. So, yeah, we'll find out how all that plays out. I was most impressed with their AI were actually from some of the bigger companies. My head I can't think of any new companies. Let me. Let me actually look through some of my notes here. I saw a company that was selling AI stethoscopes. There was a company that was doing, like, AI Optimized water. I mean, a lot of it felt really gimmicky to me. Like, they're putting AI in it because AI is the buzzword, as opposed to, like, this thing is actually way improved with AI. It just kind of felt like, oh, it now has AI, but it's actually not that use, like, the AI doesn't make it that much more useful. I'm just kind of looking through my notes here. There was a smoker called the Brisket Barbecue that. This is a barbecue that has AI in it. That sort of like the crappy framed shot. Sorry about that. So, yeah, this is like an AI barbecue smoker that kind of keeps track of the meat temperature and lets you know when your. Your meat is ready. Do I think that's something that's, like, game changing, world changing? Not particularly. Yeah, just. I don't know. There wasn't a lot of. I. I feel like there was less AI this year than there was last year. Last year, I felt like last year, I felt like every single exhibitor at CES kind of had like a Now with AI banner on their. On their products. This year, it was almost like, just implied. Everything has AI. Like, they stopped putting it in the signs. But, like, when you would go and talk to these companies at these various booths, they would go, oh, yeah, yeah, we have AI in our stuff now, too. But it wasn't like, the thing that they were announcing. I'm also curious if anybody in the comments was at ces, because I'd love to hear your thoughts and, like, what some of your takeaways or some of the cooler things with video of it. HP had this, like, AI. This is actually more of a software thing. It's not really a hardware thing. I thought at first it was like, this printer was the special AI, but what this was was it's an AI that lives between when you press print and then when it goes to your printer. And what I liked about this is, like, let's say you're on, like, a website, you're on, like, a TechCrunch page, and you want to print the article from TechCrunch. It actually used AI to automatically remove any ads from the article. It removed the header, it removed any sort of menus, and it would literally just print the article and it would, like, sort of look at whatever you're trying to print. Use AI to clean up what you're trying to print and then print that for you. I thought that was something that would be really useful. If I'm honest. I don't really print a whole lot myself, but I could. I could see how this would be really useful for people who do print. You can actually see on my little screenshot here, like, this is an article that has, like, all these ads on the sidebar, and it has, like, the profile pic of the writer and things like that on it. Well, what this AI would do is it would clean all of that up and then just make it look like you were printing a Word doc that was nice and clean and that sort of thing. I thought that was pretty cool. I mean, you can obviously do that by copying and pasting it into a Word doc or a Google Doc and then printing it, but it just kind of saves you that extra step that was. That looked to be pretty handy to me. Yeah, let me. Let me scroll through here and. And this is, like, sort of my way of refreshing my memory of some of the cool stuff that was there. What else do we got? There was. Oh, the smart irrigation system. That was actually kind of interesting. It was an irrigation system that makes sure you don't waste any water. It would, like, figure out the optimal spray distance so that you're not, like, spraying your driveway as well. It would only spray what you were trying to get coverage of. That was pretty smart. I liked that a lot. What else? What else? Again, if you were at ces, shoot me some comments about what you saw and what you thought was really cool over at CES as well. Would love to hear it. It'll also help refresh my memory. What else? What else? A lot of this stuff I didn't even see again. It's so, so massive. I felt like there was less robots this year, too. I saw a lot of videos online of people posting, like, various humanoid robots, and I did see a few, but there was way more last year. Last year it seemed like there was tons of boost that had humanoid robots. This year I only saw, like, I don't know, maybe 3 or 4 boost. All right, I'm looking through the comments here that, that Dylan sent me. I'm specifically looking for stuff around CES so I can try to keep myself focused for a minute here. Oh, that's awesome, Mr. Wizard. I got a job as a Lang graph engineer in part thanks to this channel. I love to hear that. Any demonstration of AI based video games or was it just What Nvidia showed with dlss. Interpolated frames are fine and all, but I want to see more substantial graphics generated by AI for a game. I did actually get some demos of games. Let me see. You're not going to be able to tell from my footage how well the games look though. Unfortunately, it was something that you're going to have to take my word for that. When I saw it in person, it looked really good. So like these guys here, they demoed a game being played on the 5090 and I mean, here, I'll pause it real quick. You could see in this game, like all of the reflections of the ceiling on the game and then they actually showed it turned. They turned off one setting and then turned it back on. I think that was the DLSS setting that they turned on and off, but you can see the reflection. They turn the setting off and they turn it back on again. Not the greatest footage because we're all standing around looking at it, but it looked really, really, really impressive in person. Obviously running this through a stream using log footage that's not even color graded, you're not going to get the true effect of what this looked like. But it was really, really impressive looking in person. Like, everybody that I was standing around watching this with was blown away by how good the graphics looked when we were actually playing it. Let's see. Yeah, let's, you know what, let's talk about Nvidia for a second because that was some of the, the biggest, like, negative commentary I got about the latest AI news video I put out. I spent the first few minutes talking about Nvidia and all of their announcements. You know, I was really excited about the announcements because I saw it in person and was actually pretty impressed by what I was seeing and how good everything looked. But a lot of the comments were like, oh, Jensen was lying. And he's, you know, claiming that it's getting 240 frames per second, but it's really not because a lot of those frames are generated frames. And I do think that some of those criticisms are, are accurate. I do, I do think that when Nvidia was giving their keynote, there was some of that stuff that was stretched a little bit. You know, when you are playing these games at 240 frames per second, it's not like a real 240 frames per second. It might be like, you know, 120 frames per second, but every other frame is generated by AI and then it uses frame interpolation. And when you introduce frame interpolation into like videos, it creates a slight, slight latency. So, you know, there is that. The other big criticism people were giving is that they didn't really upgrade the RAM much, the 5070. And the 5070, I believe, has 12 gigabytes of RAM. And I think the TI might have 16 gigabytes, I believe. Let me just convert. Confirm that real quick. RTX 50 series. I just want to double check what the RAM was on each of these. That was the other thing a lot of people were complaining about is like, RAM has gotten so inexpensive now. Why didn't they crank up the RAM on all of them? Yeah, so our 5070 actually has 12 gigabytes of RAM. The TI has 16 gigabytes of RAM. The 5080 only has 16 gigabytes of RAM as well. And then the 5090 has 32 gigabytes of RAM, you know. And so a lot of people were kind of going, why didn't they. Why didn't they crank up the RAM a little bit more? You can see the 4090 is a pretty good jump from the 5090 is a pretty good jump from the 4090. So you get 32 gigabytes versus 24 gigabytes. But, you know, other than that, they didn't really go too much further on the ram. Right. 40. Let's see. 4070 here actually has. Yeah, they kept the RAM the same. I mean, the new one's GDDR7 RAM. The old one was GDDR6.x. But for the most part, they didn't really make big boosts in the amount of RAM they're giving Everybody in the GPUs. I know that was a big complaint that a lot of people had as well, and I think, I think that's a valid complaint. I mean, I. I can't disagree with that. Let's see, Dragonfly theory. How would one go about getting into ces? I looked up tickets and it said only industry professionals. I mean, I. So if you're a content creator, if you have, like they actually have. Here, let me, let me show you. I can actually give you exact information about how to get in. I think the easiest way to get in is to be, you know, part of the media, in my opinion. But there is, like, sort of minimum follower requirements that you have to have. So let me see if I can find the answer to that question. Don't remember the exact numbers. There was a page. I don't remember where it was exactly, but there was a page that shows the guidelines to actually Get a media pass. I think you have to have, like, 50,000 followers or something, but I don't remember exactly. The other thing that's really cool, too, is if you want to. If you want B roll from CES, there's actually a website called cesbrolle.com where they actually have professionals going around and taking B roll of, like, most of the booths. So if you're making a video and you weren't at CES and you want to, you know, share what happened at CES on a video without actually being there, they literally give you all the B roll from the event. I take my own B roll because I, you know, I want shots that sort of prove that I was there. But you don't even have to be there to get B. You can just wander around and look at everything and then grab the B roll they create for you later if you want registration info. Okay, so who can register as media? All right, here you go. Digital content creators. So if you're on YouTube, you have to have 40,000 or more monthly views. So they don't even care about subscribers. They just want to see that you have 40,000 or more monthly views. If you're on TikTok, they want to see 50,000 or more monthly views. You're on Instagram, they want to see followers. So 25,000 or more followers. Twitter, 25,000 or more followers. If you're on Twitch, 40,000 or more views, 25,000 or more LinkedIn followers. Or if you have a podcast, 10,000 or more monthly podcast streams. You can also go in as a freelancer if you have articles published, three articles published within the last nine months. Letter from an editor or producer of the station you work for. So, I mean, like, if you look at the media page, this is probably the easiest way to get into ces is look at the media page and then figure out how to, like, engineer this. You know, maybe start a blog and make sure that you're writing at least three articles a month on the blog about. About tech and media, or start a podcast and, you know, 10,000 or more monthly podcast streams. I don't really know if that's a hard number to get to or not. It probably is. But, you know, I. I actually hit a few of this criteria here. As far as, like, going as part of the tech industry, I don't actually know what the requirements are because I've always gone as part of the media. And then I think. I think if you're part of the tech industry, it costs like, 150 bucks or something like that for a pass. All right, what else you guys asking over here? Is there going to be a dedicated CES recap? There might be. I'm. I'm not sure yet. Again, my plan was to do this live stream and talk about CES on this live stream because I, again, I didn't really feel like there was a giant leap from this year to or from last year to this year with a lot of the tech. So I'm kind of still debating whether I'm going to make a full dedicated CES video. Dylan, who's my producer and also helping moderate this, him and I are going to be chatting about that and seeing if it makes sense to actually put together a sort of deep dive on the coolest stuff that I saw at Cesar. I'd like to. I. I would say I probably will. It just won't. It's just taking some time because I did shoot a lot more B roll than last year, so it's, it's going to take a little time to like sift through the footage. I don't have. I didn't have like an on site team like a lot of other creators do that. They were feeding them the B roll and then an editor was in the hotel room editing the clips together that same day to put them out. I don't, I don't have those kinds of resources at my disposal while I'm at the event. So if I do, it probably won't be until like next week or something because I do have to sift through it all and, and figure out what footage is worth showing. Let's see what AI glasses advise us by for daily work. I mean, right now the only AI glasses that I personally use frequently myself are the Meta ray bands. And I barely ever actually use the AI features in them. I mostly just use the camera and the speakers that are on them. But I do think the Aster glasses, once those are available from Google, those are going to be really cool. I also hear rumors that sometime in 2025 the Meta Glasses are going to add like a little heads up display inside of them so you can actually see, you know, translation and subtitles on them and get turn by turn directions if you're using your glasses to go somewhere. So the Meta Ray Bans or the Astro glasses, once they add the heads up display inside of them are going to be really, really helpful, I think. Let's see. Turundus AI. The Withings mirror was intriguing. Do you think combining smartwatches, blood pressure monitors and body scans into one system is the future. Who will lead this? Greetings from Estonia. Yeah, I did think that was pretty cool as well. Let's see. I actually do have a little bit of footage of that one as well. So this is the Omnia mirror here, basically. Let me pause this video. It's this mirror that you stand in front of, and it's got, like, it syncs up to your smartwatch. So it will show, like, your, you know, your daily steps and your daily activity and your heart rate and all of that kind of stuff, and it's all in your mirror. It also will do things like give you positive motivational messages when you steer it. Stare into the mirror in the morning and things like that. I definitely thought that was really cool. Is it the future? I mean, probably, but I think, like, costs need to come down on a lot of the stuff that's in it before. Like, everybody's putting them in their house. You know, you. You would probably want this, like, in your bathroom, right? So, like, when you're getting ready in the morning, it's showing you all the data, like, as you're getting ready. I don't know if I want, like, a separate mirror that I'm going to go use. Like, I kind of want it, like, built into my bathroom. But, like, everybody's already got where they live, and so they would need to, like, replace the mirror they have already. So, I mean, I could see this being the future, but I think it's going to be a fairly long transition before, like, a lot of people adopt this kind of tech. Let's see. Frog. Thank you so much for the super chat. I appreciate it. What are your favorite five movies ever? Oh, that's interesting. Very, very random. I don't know. I haven't really thought about it. So I love almost everything Christopher Nolan does. Batman Begins is probably, like, my favorite movie. I've seen that movie so many times. This is very cliche, but trying to blank on the name of the movie. Dang it. It's like a classic. The Shawshank Redemption. That's like, every dude's favorite movie. I love Batman Begins. I love Shawshank Redemption. I don't know. I'm struggling. I watch. I've watched a lot of movies over the years. Those are the two that come to mind is probably, like, my top favorite movies. I don't know. I'll have to think about that and get back to you. Ali Edian says last year CES was better than this year. Ces. I agree. It was this year I. I was Honestly, fairly underwhelmed, and I think it was because I was there last year, and a lot of the tech felt fairly similar. It didn't feel like a giant leap from what we saw last year, where last year felt like a fairly large leap from, you know, previous years. One thing that I thought was cool, I don't think this really has anything to do with AI, but I thought this was really cool, was this robot here that cleans your windows for you. It's like a rope, Roomba, but for your windows. And I want one of those for my house. I feel like this might be, like, kind of the future of, like, cleaning hotel windows. You know, you see the dudes on the. The scaffolding that they pull up with the ropes, and they're all cleaning the windows. I think, you know, a lot of these types of machines, you know, bigger versions of them, are probably the future of that. But it's really cool to see, like, you know, home versions of this kind of thing. Let's see. All right, I'm going to go back through some of the questions. We can. We can circle back around to CES in a bit. But, yeah, those are. Those are pretty much my thoughts on it. I did want to share a few of the cool things that I saw. I think my favorite part about ces, if I'm being honest, is just like meeting and talking to other people at ces. I got to hang out with Kevin, who runs the Futurepedia website and YouTube channel. That was really cool. I got to hang out with Dylan. Curious who. That's another AI YouTube channel that I recommend checking out. I got to hang out with Matt Berman. He was out there. Unfortunately, he lives in la, so his area was. The fires are obviously going on in LA right now, so that was a lot of concern. And I think he had to leave early due to that. So that was. That was unfortunate, but it was really great to hang out and spend some time with him. I got to hang out with Saj, who runs the Skill Leap AI channel. I got to hang out with Jacqueline Dallas, who runs. I think her channel is called Nothing But Tech. So a lot of really cool other creators there that I got to spend some time with. That's probably my favorite part about CES is just so many other creators are out there. Linus Sebastian from Linus Tech Tips was there. Didn't get to hang out with him, but I saw him there filming. Same with Ijustine. She was, you know, running around filming a bunch of stuff, so saw her. Her process behind the scenes a little bit, which was cool, but just like, seeing and hanging out with other creators is probably my favorite part about the event, even more so than a lot of the tech. So I absolutely loved that aspect of it. But, yeah, again, it was a little bit. A little bit underwhelming and a lot more AI stuff that I feel like doesn't need AI. You think that's probably weird coming from me because I'm like, the AI guy, I guess, but I. I feel like a lot of companies are putting AI in stuff that doesn't need AI. All right, all right, back to Dylan sending me a lot of questions you guys are submitting, so I know I'm skipping over a lot of stuff, and I want to. I want to circle back around here. Damn. We got over a thousand people joining us right now over on X. Almost 500 people on YouTube. Pretty sweet. Thanks, everybody, for. For hanging out and just listening to me ramble. Because this is, like, I've said this on almost every stream, but I know there's new people tuning into this dream because I'm also sharing it on X this time as well. But, like, what I'm doing right now is literally what I do every single day. I sit down at my computer and I just. I'm just all over the place. I'm looking at the latest news. I'm looking at tools. I'm looking opening up things and testing them out. I'm watching YouTube videos, I'm scrolling X and I'm just sort of jumping all over the place and just keeping my finger on the pulse. Just, what's going on over here? What are they talking about? What's going on here? What's the news? Where's. Where. Where are people's interests lying right now? And I decided to do these live streams because I figured this is what I'm doing anyway. Might as well just, like, turn on the stream and essentially narrate what I do on a daily basis anyway. So that's really the whole idea behind the stream is, like, this is literally the things I'd be doing on a Monday morning. Anyway, I'm just gonna stream it and talk through what I'm looking at. And then obviously, I'm going a little bit deeper because you guys are asking questions and, you know, so it gives me the opportunity to share more thoughts and keep my finger on the pulse of what interests you. All right, so let me. I'm gonna skim back through the questions here. I'm. I'm very add. And I'm sorry about that. So, yeah, we were looking. So I'm gonna move away from ces. You know, I usually go for about three hours on these streams. So if we're still hanging around in a couple hours and you guys want to circle back to ces, we. We can do that in a little bit. But there are some other things that I think are interesting. There's this new thing that came out from the White House, ensuring U.S. security and economic strength in the age of artificial intelligence. I have not read this yet. I opened up the tab before going on the stream. I don't know what this is all about yet. I know that Nvidia put out a statement that they're not happy about this. You know, this thing that the White House put out. So we can probably talk about that here in a minute. Adobe putting out a new tool that can edit 10,000 images with one click. So it can do things like bulk, automatically remove, replace, or extend image backgrounds, but do it in, like, giant batches. So that's something interesting. And I'm not sure. I don't think we have access to this yet. It says Adobe says there will be a fee to use these new tools based on consumption. So just based on the way this is written, I don't believe this is available yet. Oh, yeah. Okay. So digital avatars created using text descriptions and voice recordings will also be available in beta this month. Yeah, so this bulk feature looks interesting, but I don't believe we have access. And if you are trying to do 10,000 images at once, it sounds like it might be fairly expensive. That's something that's kind of interesting. I don't think we need to go too much deeper into that. Here's an article I came across that sort of goes into that whole frame rate issue where it's sort of adding extra frames in between using generative AI. So I was literally reading about this right before we jumped on the stream, basically saying, let's see. So Jensen was basically saying that DLSS4 does not use frame interpolation, but I guess it apparently does. Okay, so DLSS4's multi frame generation uses a technique called frame interpolation. This is the same technique that we use in DLSS3 and the same technique you'll find in other frame generation tools. Frame interpolation works like this. Your graphics card renders two frames, and then an algorithm steps in to calculate the difference between those frames. Then it generates a frame to go in between. So, you know, if you use. If you've ever used like Runway AI's frame interpolation feature. You give it one image and then you give it a second image and then it sort of morphs like an animation between the two images. That's what this is doing. It's. But it's doing it at a super rapid rate, like it's happening, you know, multiple times per second where it knows one image from the game and it knows the next image from the game. But then it adds AI generated frames in between each of those IM to make it seem like a much, much higher frame rate. And a lot of people aren't happy with that because this introduces some additional latency which could potentially slow down the game a little bit. So there's a whole article about that. I was just reading up on it. Somebody keeps asking about versus Let me, let me see what they said about it. I know I asked for more details. Versus has so much going on. Genius program and the spatial web protocol, cheaper, faster, explainable, could be a game changer. You need to do a deep dive. Okay, I, this is something I will look deeper into. It looks, you know, I'm looking at this website. There's a lot of information here. So, you know, if you think this is something that's really, really worth looking into more, I will look into it more. I don't feel like their homepage really does a great job of telling me what this is all about. They've also got some dead images on their website. So yeah, I'll, I'll, I'll dig into verses. I'm not going to go too deep into it right now because then you'll just see me in silence reading the screen and I don't want to do that on a live stream. But that is something I will look into more based on your suggestion. Let's see what else. Allie Klein. Matt, you are the best. Thank you. I appreciate that. I'm just having fun with AI and you know, trying to share the fun a little bit. Paul Harding Ask Chat GPT yeah, I could do that. What, what I'd actually probably prefer to do is ask Perplexity because then it will actually look at the website for me. Actually what was versus AI? So here's what I'll do real quick. What is versus AI and what is it used for? What's novel about it? Why should I care? This is the type of stuff I love Perplexity for because I can give it the URL and you know it's going to actually go and like read the URL here. All right. Versus AI is a cognitive computing company developing next Generation intelligent software systems inspired by natural intelligence and neuroscience. The company's approach centers on two main innovations. Core technology Active inference AI unlike traditional AI systems that rely on large language models versus uses an approach called active inference that updates beliefs in real time based on incoming data. Interesting. Requires significantly less training data and computing power. Makes decisions that are more interpretable and auditable by design Genius platform. Their flagship products suite includes a distributed intelligence operating system that enables developers to build AI applications that can reason, plan and learn real time data processing capabilities that allow AI agents to continuously update their understanding of the world and integration with Internet of things devices, robotics and other systems for practical real world applications. So that's real time. I like this. Allow the AI agents to continuously update their understanding of the world. That's something we're kind of missing in most of our large language models now, right? They're all sort of trained up to a certain date. Yes. They can access the web like if we're using Perplexity and it can pull in more up to date information. But it sounds like this kind of constantly updates the actual underlying training data. So maybe like when future people start using this, it knows based on information that other people asked it. I'm just speculating here. I haven't. This is obviously my first time reading about this here. Efficiency and cost the system requires substantially less data and computing power compared to traditional AI approaches, making it more cost effective to develop and operate real world applications. Smart City management in Abu Dhabi for taxi fleet optimization Supply chain and warehouse management through their Wayfinder application. Autonomous drone navigation in partnership with the European Union. Why it matters Provide a more sustainable and efficient approach to artificial intelligence. Enable the development of AI systems that can better understand and interact with the physical world. Create more transparent and trustworthy AI systems that can explain their decision making process and potentially offer a path toward more advanced forms of AI while using fewer resources. Interesting. So is this different than like large language models? Because I think, I think it like to get, to get to like AGI or asi. I feel like we need a model that constantly learns and like updates from new information as it sort of gets that new information fed into it. I know Sam Altman has recently said they know the path to AGI, but I also, I think once they claim they have AGI there's going to be a lot of people out there there that are claiming oh we, you know that's not AGI, that's just Sam's definition of AGI. Like I think there's going to be some contention around that and I think to really have like a true AGI and eventually get to a true asi. AGI being artificial general intelligence, ASI being artificial super intelligence. I think you need a model that constantly learns and updates its learning as it gets fed new information. So as people have different conversations with it, it's constantly learning based on those conversations. And then you know, everybody that's using that model beyond that is getting the newest, most up to date model based on conversations it had previously, if that makes sense. I feel like to really get to an AGI or asi, I feel like it needs to be that sort of constantly learning model. Let's see how, let me ask, how does this differ from LLMs? All right, just reviewing some questions here while I'm waiting. I really appreciate, you know, Dylan keeping a lot of these questions organized for me because there's, there's a lot coming in right now. We've got, let's see, almost 1600 people live right now. Let's see answered that one. Have you played with any multi level agent frameworks like relevance AI? I actually have not played with relevance AI. I know that Agents was a big talk Relevance AI. I know agents were a big talk during CES and Jensen's keynote. You know, they have what they call NIMs, which are Nvidia microservices I believe. And these NIMs are like little AI tools, right? There's like a text to speech nim and a video processing nim and a large language model NIM and you know, text voice and voice to text and all of these various like AI like apps and services that we've seen. There's like a NIM for each one. But then they introduced what they called blueprints. And these blueprints are when you start linking up all of these NIMs and they're calling those blueprints basically agents. So basically think of a blueprint like let me try to think of an example of like a blueprint. Think of like a video game. And in the video game you want to create NPCs non playable characters where they can actually have commerce conversations with you when you walk up and chat with them. Well, like a blueprint for that that game developers can integrate directly into their game might be a blueprint where it takes a, you know, a voice to text input. And it, there's a NIM for voice to text. And so when you speak to that NPC in the game, it's listening to it, transcribing it into text and then maybe there's a NIM that's like a large language model Nim and the large language model NIM then takes the text that you just gave it, interprets it, gives it an answer, and then there's a text to speech NIM that then basically gives the character a voice and then they start speaking back to you. And so you're using basically three NIMs, voice to text NIM, large language model NIM and then a text back to voice Nim and then all three of those are linked together as a blueprint. And now you can integrate that into a video game. And once one person creates that BluePR, that blueprint is then shareable. And now all game developers can have access to that blueprint and create these AI enabled NPCs in their games. Because the blueprint's been created once and now everybody has access to it. And that's kind of what they were calling agents. You know, everybody kind of have some like different definitions of agents and what they expect agents to do. But Nvidia was kind of using the term blueprints and agents interchangeably. All right, so the leading AI workforce platform that delivers human quality work. All right, we'll take a look at this in a second. Let me jump back to perplexity here. I want to know how this tool we were looking at here differs from LLMs. So this is versus AI. So active inference continuously updates its understanding through real world interactions and sensory input. Yeah, so that's to me what seems like this, what makes this versus interesting that it's got the continuously updating understanding. LLMs remain static after training, unable to learn from new experiences or adapt to changing situations. So yeah, that's, that seems like probably the biggest difference that makes this versus thing stand out. Real world applications. Active Inference excels as an operator rather than a maker, meaning it can run complex systems like hospitals, airports and cities. Make real time decisions based on current data. Continually update its real world, continually update its world model. Limitations of LLMs cannot perceive or reason about real world situations as they unfold. Lackability to take physical actions, generate content based only on its training data, and often hallucinate or fabricate responses. Does that mean these new models won't hallucinate? Let's see. So, technical advantages, resource efficiency, explainability. The system provides transparent decision making processes and report on its own processing, making it fully auditable. That's interesting because right now, obviously a lot of the large language models like chat, GPT and you know, Claude and things like that, that's sort of a black box. We don't actually understand how it came to the conclusion that it came to. Like we know there's a vector database behind it where it's sort of connecting the dots between vectors as it gives the output, but we can't see that process happen. We can't actually see how it decided which sort of vector to jump to next to give us the information. In contrast, LLM decisions are often described as probabilistic guesses that are difficult to interpret. So contextual understanding, active inference builds a dynamic world model that continuously evolves based on real time interactions, while LLMs are limited to their initial training data and cannot understand their update of the world. Okay, yeah, so that to me, that's really interesting. This versus that's, that's definitely one that looks interesting to deep dive even further in. I think we're just kind of looking at it from a surface level. But the big takeaway here is something like this looks like it can do things similar to a large language model, but is constantly updating as new information is sort of getting perceived by this AI model. That's super, super cool. Thank you to who brought that one up, Phil B. Thanks Phil B. For putting this one on my radar. So coming back to the music by Mr. Spicos, which AI technology or breakthrough do you believe will have the greatest impact in the next five to 10 years? You know, I think advancing large language models and tools like what we're looking at on this versus here are probably going to have the biggest impact because as we hit things like aggressive.