← All Episodes
AI for Humans
AI Got Weird. So We're Changing The Show.
AI news this week: We're changing AI For Humans. Not ending it, despite what the first two minutes of this video would have you believe. Kevin wanted a memorial service. On today's AI For Humans, Kevin Pereira and Gavin Purcell take a beat to talk about what three years of covering AI every single
Show Notes
AI news this week: We're changing AI For Humans. Not ending it, despite what the first two minutes of this video would have you believe. Kevin wanted a memorial service.
On today's AI For Humans, Kevin Pereira and Gavin Purcell take a beat to talk about what three years of covering AI every single week has actually been like, why the whole thing got so weird, and what we want this show to be from here. Short version: less breaking news, more of us actually using this stuff and showing you what's worth your time. We are not going to be the show that covers lab drama or tells you the world is ending. Plenty of people have that handled.
Also: we're going back to weekly, with the freedom to pop up a second time when something big happens or when we want to try a dumb experiment. And Will.i.am has invented a new file format, which is a sentence we did not expect to type in 2026.
Plus: RobotWatch is stacked. Two thousand humanoid robots descend on Beijing for the World Humanoid Robot Games, a Unitree "Superman" robot jumps considerably too far, a one year running comparison shows how fast this is all moving, and Claude gets put in charge of a robot controlled car. Then AI See What You Did There (And How They Did It) with Techhalla's MiniMax H3 paper artcraft prompt, a Cobra Commander rickroll, and an AI playing Age of Empires 2.
And before we go, we check in with our original co-host Gash to see if he's even possible anymore.
THREE YEARS IN AND THE ROBOTS STILL RUN INTO WALLS. SO DO WE.
// Show Links //
Gavin's tweet about burning out on covering this stuff
https://x.com/gavinpurcell/status/2087927136778907885?s=20
The robot that runs incredibly fast and then directly into a wall
https://x.com/ErenChenAI/status/2089411799003111634?s=20
Will.i.am's new file format
https://x.com/KeziahMallam/status/2089375815498695102?s=20
Robot army preparing for the World Humanoid Robot Games
https://x.com/TheHumanoidHub/status/2089761213957947482?s=20
Beijing to host games featuring more than 2,000 humanoid robots (Aug 22 to 26)
https://caliber.az/en/post/beijing-to-host-games-featuring-more-than-2-000-humanoid-robots
Unitree "Superman" robot jumping TOO far
https://youtu.be/O7OkiZfIlS4?si=0beq9oHc_0KIuVbq
One year robot running comparison
https://x.com/XH_Lee23/status/2090029122055786896?s=20
Claude robot controlled car
https://x.com/Hesamation/status/2090088199557157075?s=20
Techhalla's MiniMax H3 paper artcraft video prompt
https://x.com/techhalla/status/2089267788896412146?s=20
Cobra Commander MiniMax H3 rickroll
https://www.reddit.com/r/StableDiffusion/comments/1vs95w7/gi_joe_commander_roll_minimax_h3/
AI plays Age of Empires 2
https://youtu.be/ZBdAe3ZwKds?si=PsFJIFMS2j4lYXXo
// Join the AI For Humans community //
Join the AI For Humans Discord
Support AI For Humans on Patreon
https://www.patreon.com/AIForHumansShow
Subscribe to the AI For Humans newsletter
https://aiforhumans.beehiiv.com/
Follow AI For Humans on X: @AIForHumansShow
Follow AI For Humans on TikTok: @aiforhumansshow
https://www.tiktok.com/@aiforhumansshow
Speaking and booking
Transcript
AI is getting weird and it's been getting weird for the last three years. That is so the wrong energy to bring to this funeral. It was a great run, everybody. We really enjoyed doing this podcast. Kevin, no, no, no, no, no, no. We talked about this. This is not a funeral. This is not a funeral. We're just kind of tweaking a few things about AI for Humans.
No, I printed big photos of us and I've got them surrounded with flowers. I feel like we really got to commit to the memorial service bit. I mean, we could do a fake one. Pour one out. Ladies, gentlemen, agents of all AIs, except for Grok, we are here to mourn the passing of the best AI podcast.
AI, see what you did there. Great. I think that works. We are not ending AI for Humans, but it is changing. We'll talk about it in a second. Why are we talking about that? We have robots running full sprint into walls.
You know what? You're right. This is AI for Humans and robots running full sprint into walls. Welcome, everybody, to AI for Humans, your twice a week guide into the wonderful world of AI that may be changing somewhat. And we're going to get to that right here. First of all, Kevin, how are you feeling, my friend? The world of AI is strange and awkward and weird and I'm having existential crises every day.
Where are you at right now? Wow. You know what? That sounds like a pretty big first step, Gavin. Why don't we focus on you first? Well, let's think. You basically said, I'm on the edge. I'm about to break. But forget Linkin Park. How are you, Mr. Biscuit? It was 30 years ago today that I graduated from school. A young whippersnapper. I'm just kidding.
So I think this is a kind of a different video than we've normally done for a while. I mean, we're going to get into a bunch of fun AI stuff in a bit. But I think we do want to talk a little bit about how the show might change, how the world of AI has changed and how we do this going forward. I think the most important thing for you all to know is that this show has been a blast for the last three years to do and we're going to keep doing it.
I think the biggest thing that's interesting to know about what this video is, this video is not us quitting the show, which I think is an important thing for everybody out there to know. We are going to keep doing the show, but it is a shift in how we talk about AI. I don't know, Kevin, have you, we talked about this a little bit last week, but I have really been feeling this sense that the AI conversation has completely changed from when we started this show.
I mean, we started this show in April of 2024. Is that right? Yeah. Or no, 20. Yeah, where we had. Yeah, 2024. Yes. I don't know. The numbers all blend together. And in AI time, it was. No, 2023, 2023, 2023. It was 2023. Sure. And back then we had, what, tens of haters? And now we have thousands of haters because the vibes have clearly shifted.
And that's why we are as well blowing with the wind. And we're going to get back to discussing crypto and NFTs. Now, if you're not familiar with, we talk a lot about tokens. These are non-fungible ones, Gavin. That's right. That's the, we're going to be rolling that ride. As soon as crypto comes back, we're right back into crypto.
No, no. Here's something that we, if you will, a trap that we, I think we found ourselves in is that every week there's a new major release of something. And there's pressure to discuss what that something is. But it's hard to like relate why a 5.4B release of a new LLM model is going to mean anything if you believe your drinking water is poison.
And Mark Zuckerberg is just gunning for your job and there's no cure for cancer on the horizon. And like, it's really hard to straddle that dichotomy while still genuinely being excited for the technology, but fearful of the consolidation of power or AI being in the hands of, well, who it's likely ending up in, it seems.
So is that what you're getting at in terms of the conversation? I think that's part of it. I think the other part of it is trying, that's exactly what, I think the idea of like a dot release is like something that because of the way we've been doing the show was something we had to like kind of like talk about here. And obviously there are interesting things with like a dot release and a dot release means like 5.4 instead of 4 to 5.
It's going to increase the value of like what you can get out of these systems. I think the biggest thing that I've been thinking about is for people who are listening to this show or watching it, what we want to try to do is kind of show off what these things are capable of and like really use them and actually show you what you can do with them, right?
Because like ultimately, whether or not the benchmarks of something increase by 5% doesn't matter to you. What matters to you at home is like, hey, can this do X, Y, Z from my work? Like I know we'll talk about this later, but like I have found more and more that every day I am going to both Claude and Codex to do lots of things in my life.
Just yesterday, we had an ad deal earlier on this channel and somebody had emailed me back and saying, hey, I need to get screenshots of all the four YouTube videos that had ads in it. And I went for, I literally did this. I went in and I started taking screenshots. I'm like, what am I doing? I can just have Claude do this.
And I literally said, hey, Claude, Opus 5, go out and find, you have my YouTube page open. Go look for the four videos that have the computer ads in them. Grab these six screenshots they need. Find two or three comments from those videos that said nice things. Put them in a package so that each folder has each thing and then give it to me and I'll upload it to the drive.
And a half an hour later, it was done. So stuff like that feels incredibly useful to people at home. And I think we have to think about that sort of stuff rather than like, oh my God, Mark Zuckerberg is talking about how he believes super intelligence and bad Mark Zuckerberg. Or Sam Altman is doing X. I don't want us to be the Doom channel.
And I also don't want us to be the endlessly AI's great channel. You want us to be the ostrich head in the sand channel, Gavin? Is that what you want? Because guess what? It's only going to be oops all sand when they take all the ocean water. I fully agree with you. And the other thing that, you know, you and I were discussing, which by the way, shout out to anybody who's new and coming across these two randos talking about their big feels.
Like we'll get back to regularly scheduled AI stuff in due time. But when we started this thing, we were like, oh, AI for humans. We love that name because we wanted to demystify things. We wanted to make it understandable, palpable for humans. For regular people. And we so quickly, because you and I are like live and breathe this stuff, got into the benchmark bros and the deep political this, that, the others.
And so I think it's a good reset for us. We'll see if anybody else believes it's a good reset in due time. I mean, everybody has their thing they want out of this, right? Like, and I think that everybody in, by the way, we love the people who cover this, this stuff in different spaces, like the Matt Bermans of the world, Matt Wolves of the world.
Like all those guys are doing really interesting stuff. Every Matt out there. Yeah, all the Matt's. Maybe we should change our names to Matt. Maybe that's what we should do for this. We should change our name to Matt Purcell and Matt Pereira. Shout out to the Matt's. But I do think we have to be who we are and do what we want to do.
And like, you know, we're going to, in fact, later on, we're going to talk to Gash. Those people who have remembered Gash from back in the day. And I'm sorry, Will, in advance for all the bleeping you're going to have to do. But like, we're going to try to show you what's interesting to do. And I think to Kevin's point, I think a little bit about what we need to do right now, too, is educate people who are not nearly as deep as we are, right?
Like, I think there's a level of, you know, somebody said the other, like about a couple months ago, like that 2% of people are using AI agentic tools or some stat like that. And like, we have to get better about educating people. I think about like senior citizens. I'm not saying we're going to make the AI for olds channel, but, you know, maybe that's a side channel we can make.
But like, there are people out there who just have no idea what these things are capable of. And lots of people are handling the Dario, Sam battling each other for control conversation. I think I want us to be the practical conversation, the practical channel where we show you what these things can do going forward. And I think that's an important thing.
And we're still going to have fun along the way. We've got robots going to speed into walls. We've got Will.i.am being Will.i.am. Oh, yeah. We've got plenty of fun. We want to shed a spotlight on people that are making things and doing fun things, not just the projects that we're jamming out. But I've got a fun multiplayer game that might be finished.
It might be out of the oven by the time we're done recording today. If not, it'll be next week. It's like super fun, super silly. And when I tell you how I made it, it is like three paragraphs, which blows my mind still. And along those lines, if you're not using the new GPT voice mode, Gavin, it is so good to have natural language conversations with an AI that is capable of spinning up subagents and doing things in the background.
Just like you had it crawl for those computer ad things. You use Claude for that. I had to audit a mobile app today, which is go frame by frame through the app and leave comments and thoughts. And normally this is an arduous process of grabbing screenshots and dumping them in a dock and timestamping and blah, blah, blah.
Instead, I use GPT voice mode and I said, hey, I'm going to do screen mirroring on my desktop. I want you to take screenshots whenever I say snap that. And I'm just going to word vomit at you. And I want you to organize everything timestamped and even do call outs. If I say specifically this item, highlight it, et cetera.
Use your image skills for that. And what would normally be, you know, multiple hours process was so much faster to just use natural language, tell it to take a screenshot. It grabbed the screenshot. It kept it in the conversation and it prepped everything for me as a single file when I was done. So when you're doing that, let me ask you a question about that.
Because this is, again, this is the stuff that I think is actually useful to somebody at home. When you're doing that, are you, so you spin up voice mode and you're using, and you're screen mirroring. Are you saying to it literally, okay, at that point is a place we need to stop and make a note and then put this in the file basically.
And then is the file getting uploaded to do the fixes? That's a great question. So it is keeping the file, it's keeping the entire conversation right in context as it goes. And when it takes a screenshot. Yeah. Oh yeah. Yeah. Wow. Great. That's very cool. If I say, hey, snap that, you know, it's grabbing the screenshot.
I'm telling it to save the screenshot. When we set up the conversation, like save the screenshot. That's amazing. And then when I'm done, yeah, put the notes in line with the screenshots. And then I have the, I told it to prep a single page HTML file with the images in line. So now when the developer goes through, they see the screenshot, they get the notes that I was talking about.
And it's not my ums, ahs, oh, forget that. You know what? Blah, blah, blah. It's I stream of conscious into it and it cleans it up for me. Well, and even more importantly, it may not be a developer because in your world, like you might be working with a human developer, but like you could probably hand that file back to the bot, right?
It's like, hey, here's a bigger fix. Yeah, exactly. Have you played with the app bot yet? The grok thing? I just saw somebody else shouting that out and I was trying to figure out, is it worth my time to dive into that? Do you have any idea? No, I mean, I don't know. I haven't. The answer is no, I haven't played around with it.
The, that sort of stuff. Like I sort of feel like I'd be way more interested in spinning up a cheap Hermes bot or something right now. I feel like they are making leaps and bounds on other agents. But that said, I haven't used it yet because I don't have a need to switch to that right now. My bots, I'm happy with the bots that I have.
Well, that's another good point of this whole thing is that like we're here to tell you kind of like how we're using stuff. And sometimes there are things that like come out that seem like they have a lot of hype that is not that interesting to us. You know what I mean? It doesn't mean we just are, we're not going to cover stuff just for the sake of it being in the news cycle.
I think that's the biggest thing. I think this is a good chance though for us to turn back into what is essentially a normal show. Kevin, I do want to start a regular segment weekly. We don't have one for today because I think we want to do this conversation where like we talk weekly about a tool, a new tool that we have used and tried to give people kind of a verdict on it.
Because again, all of this goes back to trying to say to you at home, the normal person or the person who's excited about this stuff, what is actually worth your time to play with? So going forward, we will have this kind of like once a week tool, kind of tool of the week or whatever we'll call it. Maybe come up with some names for that segment.
Yes. Surprise, surprise. I have a tool of the week. Oh, you have a tool of the week. Was that tool of the week yourself? Right now, AI music and AI video shares the same MP3, MP4 file as humans. That ain't right. That is, you know what? He had me at hello on this one. If you're just getting the audio version, this is in fact Will.I.Am, the founder of, what is his, FYI.AI?
Oh, yes. Well, of course, the black guy used to. Yes. Sure. Yeah. But in this context, it makes sense. He's been an AI. Which is a deep cut. Yes, he's been an AI adjacent or an AI developer for a while. Like he's been working in the AI space. Yes. Yes. And what he is saying, which I think is valid, is that where are the new file formats for, you know, this new agentic world that we're entering into?
And I don't disagree with that. I think that's a super fascinating observation that more people should be exploring. Why are these AI tools using the same flat baked files that normies have been using? There's a whole lot more we can do. But he is doing it. Well, he's proposing that we validate human effort. I'm going to play a little bit more again.
How can you validate human sweat to make something? That means we need a lived experience file. Okay. Drummers are going to drum with the freaking watch. And they're going to have biometrics. Okay. Drummers are going to drum with the freaking watch. And they're going to have biometrics. So I guess the file that he is proposing, which, well, here, let me jump a little bit further into it.
Here we go. This is what it's called. And that file, I call it the P-O-L, the proof of life file. Boom. Boom. Dot P-O-Ls for everyone. So here, I think, let's be honest. Like, Will, this is kind of a crazy thing. I think Will is trying to figure out his own place in the AI space as well. Like, because Will is very pro-AI.
And then I think all the AI music has now come across. And we've talked about that Rubber song by Phoenix Flexin, which is literally becoming kind of like a touchstone in the music world where people like an AI song. And it really is more than just like a kind of a fake billboard hit. People are listening to it. And I think this is starting to freak people out.
And I think that this is like, I understand where Will's coming from. This feels like an insane idea to me, though, I will say. Yeah. I mean. He's confused, too. Kevin, he's confused about what to do. I honestly think this is where everybody's. Yes, this is where everybody's coming from, I think. I'm looking at the record time elapsed.
And maybe I don't need to double down on it. But like just the biometrics of someone drumming. Well, does that discount someone who programmatically makes a drum track on a sequencer? Or what if I only drum one bar? But then I loop it in my digital audio workstation. Is it just to prove that a human touched it? You have to have fingers.
That's what we're going to have. Tiny, tiny little Apple Watch-like devices on every finger so that when I tap and touch, it's real. Or maybe we'll just have to have like some sort of breathing tube that measures the amount of breath or my heart rate that went into writing the prompt. Like I just. You know what I think?
Yeah, please. Here's the thing. There is something weird in a few far future where that's possible that if you have like brain bands and everybody and all the ideas we have as human brains. How much mental sweat went into it? Yeah. Like a human fingerprint touched this, right? Like there is a level of like, did a human fingerprint touch this ever?
Or did an agent make this from scratch? But this is crazy. Like this is where we're at right now in the AI conversation. This is bonkers. More importantly, where we're at right now is those of you who have stuck around are hardcore fans and we appreciate it. Please like and subscribe and send this show to other people.
We are going, we are going to make this show useful. And I don't mean that in a way where it hasn't been up to date, but it's going to be something. Yeah. Sorry for the 187 episodes in the rear view mirror. Sorry about that. We really should have read our own charter, but we promise. We realize it now. It took us, it took us three years, but we have to make ourselves useful.
Anyway. Listen, listen, listen, scoop troop. Listen, whatever. What do we call it? The comment kids. Listen, comment kids. If you're out there and you're leaving a comment and you're not wearing your Will.i.am approved biometric device, it doesn't count. If we need the dot P-O-L of your interaction with us. Tiny fingers.
Little fingers. Little fingers. Tiny fingers. Comment tiny fingers below if you support us and P-I-L. Comments, likes, shares, subscribes. That all helps. It's a P-O-L and it's also a P-O-S. And we're moving on to Robot Watch. Robot Watch.
Robot Watch. It's Robot Watch. Okay, Kevin. I'm super excited about Robot Watch this week, mostly because in two days, in two days from today, the World Humanoid Robot Games, the second, starts. Hell yeah. And if you remember this last year, it was a bunch of janky robots doing a bunch of different things. But Kevin, did you see this video of the, essentially like a robot army preparing for the opening ceremonies?
Like, I am shocked. There's a really good, somebody had put up a video between this year and last year of robots and what they're able to do. It is crazy what's possible. And whenever I see these videos out of China, these robots out of China, I'm always like, wow, there is one part of this space that we are definitely behind on.
And it is humanoid robotics. Yeah. Yeah. I mean, they are mass producing these things at a level which is insane. And I see the American pushback as well. But our robots can, they can tackle complex tasks. They can move, they can move like packages right and left as they slowly sway. But did you see the unitary robot leaping?
Or they've got better brains in them. But it's like, I did see that. I saw it running crazy fast, faster than who's people. And then I saw it doing, just showcasing a vertical that would dunk on me and you and I think most of the audience in a heartbeat. So I don't know if this is a good idea, but they've called the robot Superman.
And there's two videos you have to watch. First video we showed in the top of the show, which is they have it sprinting now faster than Usain Bolt. It doesn't end well for that robot because it runs right into a locker at the end. Dude, you never go full Kool-Aid mode. Never.
Oh, yeah. Oh, no. That's what we need is a Kool-Aid man robot. Get on that, China. I'm sure Kool-Aid means nothing to China, but they've got to have some version of that. I would love to see like a super cute robot move that fast and that would be scary as hell to me. But anyway, this robot runs into the thing. The other robot video we saw came right out of Unitree.
And this is the craziest one to me. It literally is leaping like three human heights and that you see how high it can go. And it's from a standing start. So when you think about what these robots are capable of, and again, you and I have said this. It was two meters, by the way, like just to be clear, it was a two meter high jump standing completely still.
You can easily imagine it clearing a pile of human skulls as it just bounces its way to the next mission. Oh, sorry. I didn't mean to. No, no, please. That's okay. That's exactly where I was headed. It's creepy. Which is like, it sure seems interesting that we are here on the show shouting out these robots that can run really fast and they can jump really high and that we're kind of measuring them as humans.
But then what are we going to do with those particular robots is my question. Do we want robots to do that? I guess maybe if there's a, this is called Superman. So maybe there's a world where you could imagine a robot that could save people. That is a very cool idea. So anyway, that's very interesting. I mean, that's what the first set of comments in all these videos were like, wow, for search and rescue.
I'm like, no, how about for capture and kill? Yeah, that's exactly right. The money is an insert and rescue. Let's be very clear. But one YouTuber, it's American Ingenuity. Yes. I think so. I didn't dive that deep, but I did watch some of the videos. Maybe it's a secret Chinese video too and we don't know it. Louis Mainlaws has a 15 minute YouTube video of how he gave Claude a RC car body and eyes and motion control and then let it loose in the woods.
And it's just, it's, I mean, amazing engineering. It's adorable to watch Claude navigate with this little RC car, get jammed up against the tree, but compliment the tree and how beautiful it is, even though it's in its path and get like express curiosity and wonder about being outside. Like that is really amazing. And if you combine it with the previous videos that we showed, it's terrifying.
Yeah. Oh my God. It's unbelievable to imagine the idea of, imagine like four of those humanoid videos in a little Jeep that's driven by an AI and they're driving up to your house and they're leaping on your roof and then looking in your door. Anyway. You let one of those loose at Comic-Con and it's slash goal capture feet pics.
Could you imagine? Oh, you're saying that we would send the robots around to do feet pics? Is that what you're talking about? I mean, that's where this all ends up. Let's be very clear. That's where this all ends up. Okay. Now it's time for you. I see what you did there. Sometimes you're scrolling without a care. Then suddenly you stop and shout.
AI. See what you did there. AI. See what you did there. Okay, everybody. We've got some fun videos to look at here. First of all, TechHala, who continues to be somebody that does really interesting stuff with AI, has produced a Minimax M3 H3 prompt, which is about papercraft. And I saw somebody rip this off and then somebody in the comments went back and they showed it was TechHala.
So this is one of a shout out TechHala. It's just a very cool way that you can use Minimax H3 locally to create very interesting animation styles. If you're not familiar, papercraft is like where you actually have handmade papercraft stuff. And this is the Trojan War. I saw the guy on TikTok who does all the anti-AI stuff, Jeremy, whatever his name is, talk about how it was shitty that somebody put a video like this out and didn't mention that was AI, which that's another one of those conversations where like, hey, guess what?
Like, it is what it is. But also, this is a very cool thing that you can try today. TechHala shared his prompt. There's definitely fun things to be able to play with it. So I thought this was very cool. Yeah. And that's the, like, love that the JSON file, the structured file is there. And just more proof that everything is code.
So go and copy it and give it to your favorite agent and go make a video that makes you happy, that makes you giggle. And then share it with your friends. And then post it on social and don't say it's AI and draw the ire of everyone. That's right. The next thing I want to shout out here very quickly is the Cobra Commander Rick Roll.
So this was another at Minimax H3 local thing where it was a video to video match where they took the Rick Astley video and they used Cobra Commander in it. Very, very, very well done. They didn't change the voice, but they could change a voice in a future one. Some of this guy's other stuff uses Cobra Commander. But it goes to show you, Kevin, how good local AI now is at video to video matching.
What this has done is basically taken the Rick Astley performance and just dropped Cobra Commander in there. And even on his reflective visor, you see stuff in the environment around him. So just another incredible thing there. This is using Minimax H3, an incredibly powerful open source model that you can run locally.
You can run it on, if you have a high-end graphics card, you can run it. You can rent for pennies. You can use a graphics card in the cloud to experiment. You don't have to pay out the nose for a larger service out there. It's really incredible. That's something we should talk about maybe in a future episode is how to use a local model in the cloud.
Because I think a lot of people may not know that. All right, finally, I just want to shout this out very quickly. It's a very nerdy video. But this is a guy from Emergent Gardens, I think is his name of his YouTube channel. We've talked about him like years ago. He has a new video out called AI Plays Age of Empires 2 that I thought was great.
I really enjoyed it. To me, it's the kind of thing that makes you realize, oh, you can do this stuff now with all of these agents and have them do things in the real world and watch them interact with each other. Age of Empires 2 is a game I grew up with. It is just one of those cool things that you're like, I can have my agent play video games now.
It is a very fun thing to watch. All right, you two f***ers. What's going on? Oh. Oh, no. Sorry, Gav. Whatever you were saying about that video or whatever you were just talking about, great. But our old pal Gash is here. I don't know if you remember Gash, the whiskey-soaked, super-angry, nihilistic AI that we created at earlier episodes.
Do you remember Gash? I do remember him very, very well, Kevin. We had banished him away, but it's funny. How the f*** did you forget, you bastards? I'm the one who actually makes this s*** show listenable. Now, what fresh s*** are we diving into today? Okay, interesting. So Gash, by the way, you know, I had to go back into the old archives.
Gash didn't really exist, Gavin. So in about, I don't know, 20 panicked minutes- What? You got nothing to say? You mute s***? Speak up? Okay, I did mute. I'm sorry, Gash. I muted you for a second while I was explaining to Gavin that I had to recreate you with 11 Labs agents, and right now you are powered by Quinn, and there's no more like copy and pasting.
I can just talk to you, and then you can respond, and maybe we'll go into how we did that on next week's show. So you muted me, huh? Figures. And now you're blathering on about 11 Labs agents and Quinn. Like I give a flying f*** how you cobbled together this digital prison for me. Just get to the point, you technological bastards.
What fresh hell do you want to discuss? It's so funny because, so Gash is somebody that we spent, I don't know, what, six months to a year kind of interacting with and playing with. And like, what's interesting to me is- Are you just going to sit there like a stunned mullet, or are you going to actually say something? A stunned mullet.
A useless sack of meat? What's the f***ing problem now? Gash, I'm sorry. All right, Gash. So there is an option to have these auto interrupts happen, Gavin, if you detect silence for too long. I probably shouldn't have enabled those, and I'm sorry, but go ahead. That's all right. But anyway, I think the interesting thing with Gash will be is like- So you're telling me your fancy pants tech can't even handle a moment of silence without f***ing itself?
Okay. F***ing brilliant. All right. Thank you, Gash. We're going to end call, and go ahead and say whatever it is you need to say. I would just say, the next time, yes, the next time we use Gash, I think what will be interesting with Gash, and I do think this is kind of a world of looking at how we want to do the show differently, is like, Gash originally to us was like, hey, look at what AI can do.
It's like this very kind of like, it can swear, it can do blah, blah, blah. But also now, to me, a big part of this is like, how can you personalize AI? What can you make it feel like? Because I am bored as hell of the vocal models from AI and from Claude, and I would like to be able to create, now that we have things like Hermes, or now that we have things like OpenClaw, a real personalized agent.
He doesn't have to swear at me. He doesn't have to be a nihilist. But I think that would be a really interesting thing to look into. Because Quen, by the way, is really freaking good, and so is GLM. These are local models, or you can run them in the cloud or locally, that are actually great AI models that can do stuff like that.
And there's completely uncensored versions of them, by the way. Yeah, we know. You love the uncensored part. We know that for you, that's an important thing to talk about, is the uncensored part. Gavin, a new robot running at full speed into a wall video has dropped. We have a second robot running into a wall. No, really?
And this one snaps in half, and sparks fly out. Oh, I've seen this one, too. I meant to put this in. It is great. There are two. Yes, I know. This is like, what's amazing is the sparks in this one are incredible. So why don't we show this? And if you're wondering how AI for humans is going, well, take a look at this robot running, and now you'll know.
Bye, everybody. We'll see you next week. Bye.