B (2:52)

So you just, just to give people a sense of what you do, you just launched this thing called Mirage, which is kind of crazy. Can you just show us Mirage?

A (3:01)

Yes. Let me, let me pull this up. But we just launched Mirage, which was a few weeks ago, like three weeks ago. By the time this will probably up, it'll probably be four or five weeks since it was launched. Mirage is the only real time video to video model in the world right now. And what that means is that you can take any video stream that you have, whether it's this conversation or your camera or a game even, and just put it in through the model and change it just with a prompt. So I'll bring this up right now so I can show it. Let me share my screen. And as you can see, this is real me in the interview screen and you have the Mirage version of me. So the current Mirage version of me is in the Versailles Palace. Or we can do some blocky kind of thing. Okay. Or we can do an anime version.

B (3:53)

This is, this is so amazing. I literally just saw this for the first time, like right before we started recording. And it was like, oh my God moment for me. For people who are listening, basically what's happening is you're on a webpage and you have your camera going. So. So I see an image of Dean on my screen. And then when you started Mirage, it just in real time transforms you into effectively a Pixar character. Something that looks a little bit like a Pixar character, but different themes and in different settings. So right now you're on Jingle Bells. Now you're a wizard. You've got like lights coming out of your hands. It's. It's crazy. And I like, I feel like Lego. So how did you make this? How does this work?

A (4:38)

I'm made of Lego. Look at this. Okay, let me show you one cool thing. Then we can dive in right into how this works. But I love this. This is one of the prompts I love. If you put in the portal world and then you get physical objects, like a blocky thing. This is a tissue holder. Then you can turn it into a gun. Whoa. Every once in a while if you do this, then it like shoots out something. Whoa. Yeah. Like one of the things I'm dying for the next version to have is So I can be like this and be like, okay, like. And then like a huge portal pops up or something. Like, imagine a real Rick and Morty gun.

B (5:18)

Dude, this is so cool.

A (5:21)

That's actually cool.

B (5:25)

So, yeah, just for people listening again, basically you're in your portal world. Portal world, you picked up a tissue box and the tissue box became a gun. And if you move the gun in the right way or you move the tissue box in the right way, it kind of shoots a little bit. So it sounds like, I mean, right now the possibilities are like, oh, real time, real time video games that you can just, you can just play. Is that what we're like where you're going with this?

A (5:49)

So I think the possibilities with this, what this kind of tech does is it actually enables a different way to interact with AI Video. What did AI Video do so far? It was a great piece of tech that gave us cat videos on TikTok. Okay. That's what we got out of AI video. Now for the first time, computers can actually respond to us as we're moving things, changing things. They can in real time change it so it can be used anything for gaming. We're going to have a mod coming out for Minecraft. We have an alpha testing right now. It'll be coming out in the next few weeks. You're playing Minecraft and you just through the game, you tell the mod, hey, change Minecraft to be Barbie Land and everything becomes Barbie. So you can take it to gaming. We're also integrating with lots of the game engines so that as you're developing a game, you can actually see this or that you can vibe code a game end to end. So imagine Vibe coding with an LLM game and doing all the texturing with Mirage. So that's one category. All the real time interactive gaming categories, huge.

B (6:57)

Wait, let me stop you there for a second. So are you saying that you said it's video to video? So when you're vibe coding a game, how does that work? Is it, it's generating video that you're playing or is it also able to output like the 3D mesh that you might import into Minecraft?

A (7:12)

So one of the things that people have been going crazy over, you can just vibe code a game in low poly, just a bunch of blocks moving around and turn that into just have Mirage do the coloring of that. Oh, let's see if I can find an example for this.

B (7:34)

Okay, let me see if I understand. So basically what you're saying is it's really hard to make a AAA blockbuster shooter. Game, let's say. But it's not that hard to make a 3D game. Especially if you're vibe coding a 3D game that's just a first person shooter where the gun is a rectangle or a cube.

A (7:57)

Exactly.

B (7:58)

And so basically what you're saying is you make that game and then when your game engine renders that really basic game, Mirage takes it and turns it into a video that has all the textures and all that kind of stuff.

A (8:11)

Exactly. Spot on. And people have been playing with this online and it is just mind blowing to see what people came up with that you can actually some people just use outputs from Unreal Engine and pass them through Mirage. Others actually just vibe coded games and added Mirage on top. I can send a few examples. And you're editing this video, we can put it in. But it actually just lets you build a game very quickly with an LLM, just half an hour, like, oh, add this tree here, add that there. It doesn't even have to build a tree, it just builds a blob that kind of looks like a tree and Mirage just comes and completely changes it. So it's going to do something. I don't know exactly what, but it's going to do something big to gaming either. Taking existing games. You know, we were here at the office, we were playing GTA a few days ago and pass GTA through a winter filter and then Boom, all of GTA 5 just turn into a winter version and there's no winter in regular GTA V. So sure, you're going to be able to get a bunch of varieties of existing games, but you're also going to be able to create a game end to end, just fully AI based, just Vibe code it and texture it.

B (9:29)

One of the things this is making me think of, and I'm sorry we're already going to the philosophy stuff, but I'm not actually sorry. One of the things that it's making me think of is there's this really interesting thing happening where the underlying game still works with classical code. And then what Mirage is doing is sort of doing this emergent layer of texture in the same way that, you know, maybe you can break down the world into, you know, atoms and molecules that, that work according to like specific equations that we can, we can figure out. But then our conscious experience of the world is this emergent, textured, full of qualia type thing. And you kind of need like. Or at least the way that we understand the world is, is it has both. Or at least the way I understand it and it seems like previously we had to make that all games had to be able to reduce down to that deterministic code. And it was very, very difficult and time consuming to set that world up. And now you have this sort of emergent layer that you can just vibe on top of the basic, I don't know, Newtonian or classical world that you can build with programming. Is that kind of what it is?

A (10:51)

I love that you're going into this. You got me thinking about this. There's some stuff that regular coding is great for and other stuff that AI is great for. What's AI great for? AI is great for doing stuff that's differentiable, that's continuous. Okay, fine. Drawing a picture that's kind of continuous. It doesn't matter if I get exactly your face or something close. It's fine. What's AI? Definitely not good at reversing a hash function or something. Okay. Or even just implementing a has function. Anything that is not differentiable AI would suck at.

B (11:27)

Can you define differentiable for people? I know you're a, you know, you're like a math genius and so I think I might have an idea of what you're saying here. But can you define it for us?

A (11:40)

Definitely. So think of a differentiable or continuous function. Something that. It's okay if you're close by. Okay. Like it's fine if I, if I'm creating your image. It's fine if I get the exact image, but it's also fine, you know, if I'm missing a few things. If your shirt's a bit off, if the, if the wall behind you is slightly different, if your glasses are slightly bigger. I don't have to exactly get what you're talking about. And that's the quantities of differentiable or a continuous function. If you're not talking about. If you have a function that's not differentiable, not continuous, it can be. It's very critical that you get it exactly right now. Regular CPUs are just great at doing stuff that needs to be exact. Regular CPUs in our computers, if we tell them add 5 and 5, they will always give us 10. And they know exactly how to do these deterministic operations. AI can do the non deterministic operations where we don't need stuff to be exact.

B (12:39)

This is something I've been playing with a way that I've been formulating it is. And I'm going to probably not get it exactly right, but we can find it together. There's two ways of seeing the world. One is as this sort of countably infinite set where there's one right answer. And you can guarantee yourself that you're going to get to the right answer if you just keep going through, like counting through the set. And that's the sort of classical way of seeing the world, and that's the sort of classical search function for the right answer for a program. Another way of seeing the world is, is there's this. It's an uncountable infinity of different solutions. And so you're not ever guaranteed to get to the right solution. But every solution is like, kind of meaningful, is kind of right, or it's. You can get further or closer to it. And both of those are really valuable. But you need to. At least my view of reality is you need to start with that kind of uncountable infinity where everything is full of meaning until you get to kind of like the right zone. And that's when you can start to solve with the countable infinities and like count out the right solutions. I know I'm probably sounding very galaxy brain to people listening, but we'll unpack this in a way. But I'm curious what that makes you think of.

A (13:46)

Oh, but I definitely agree with everything that you're saying, and I think that we need both. So we both need. Some problems are just a big haystack with a single needle, and you need to find that exact needle to solve it. And other haystacks have an entire area that's full of needles, and they're not all the same needle. And maybe one needle is better than the rest, but they're all an okayish needle. And then you just need to find the rough area where you have that bunch of needles concentrated. Now, let's take an example. So, you know, Google released Genie 3 recently, or, you know, they showed a demo of it. They didn't productize it where they claimed. Okay, well, it's a world model, can generate as you go. You put in your keys and your mouse, you move them around, and it generates frames for you. Okay, Gene. Gene was trying to claim, well, let's replace everything. Let's just have an AI create a completely new world for us. And we don't need any of the old software like game engines that existed beforehand. AI can just do everything. What we're talking about with Mirage is a combination. You have the game engine doing the deterministic stuff. Oh, I need to remember that you have exactly 71 gold coins in your pouch. And I need to remember that you took this pickaxe and you put it in that chest five years ago. And if you go back to that chest, it'll still have the same pickaxe. You put it there five years ago. Having the game engine do that. But the more creative parts where it doesn't matter if you specifically hit the exact needle, but you get something right, have AI do that, for example, texturing it or making it look different.

B (15:19)

I think another good analogy that might unpack this for people is you need a skeleton to have your body work and your bones define like those are the hard limits on what you can do. But you have joints and you have muscles and you have tendons that your, your bones can pivot around and that allow you this infinite flexibility within this rigid system that guarantees certain things. And if you have bones, you can stand up and if you have muscles and sinews and tendons, you can move around in all of these different ways that just bones wouldn't let you do. And for a long time we've only been able to build computer systems that, that use bones. And now we have muscles and tendons that like that, that allow us to, to move around in all these different new ways, which is really cool. But you need the bones. Because if you're just tendons and muscles without bones, you're kind of like a jellyfish. Like it's just very hard to pin you down. You can do certain things, but it's hard.

A (16:17)

I love the jellyfish analogy, by the way. If I'm understanding you right, you're saying that when we build Terminator, it's not just going to be metal robots, but they'll also, they might have skin and bone surround skin surrounding them, but yeah.

B (16:30)

Maybe.

A (16:33)

But no, I completely agree. And this is really what we're getting at, that we're enabling this from a technical perspective. We're creating video in real time that can touch everything from gaming, but also to the zoom call. Or this is a Riverside call Riverside. Instead of having, you know, just you seeing, we could be seeing each other. We could be seeing something completely different. But it's, it's touching the ability for people to take stuff that's in their imagination and add it to an existing real world. So we'll have this, this core that's the same. The world's the same, physics is the same. You can take your imagination and apply it to, to what you're seeing.

B (17:17)

How did you make this?

A (17:18)

That's a really fun question. So to crack real time video, we had to crack two things. On the one hand, we have to Write lots of very optimized GPU code. We sat and wrote lots of assembly for GPUs for Nvidia GPUs. It's called PTX and we had the things that's below Cuda, so like lots of people know Cuda and that's like the Nvidia software stack. Unless you really get stuff for GPUs. We had to write in PTX, which is like the layer below Cuda. It's the actual assembly that gets written on the gpu. So you had to do that to actually make it very, very, very efficient.

B (17:53)

And did you vibe code this with Claude or how did you do it?

A (17:55)

Oh no, no, no, no. Claude's amazing. I love Claude code, by the way. I use Claude code a lot. Good job. Anthropic. But no, it has no idea what it's doing on those levels of the stack. It's the dark abyss of AI that no one wants to touch. Including the themselves now. So yeah, on one hand we have to write really lots of assembly code for GPUs to get this to be efficient. You know, the Mirage, the current version that you saw is 40 millisecond delay. That's 0.04 seconds between when a frame comes in and when the model spits it out. The next version of Mirage is going to be 16 milliseconds delay 1, 6. Wow, okay. And to do that you really have to write very, very, very optimized assembly code for GPUs. Now on the other hand, you have to completely build a different kind of model. So all the video models we know today are what's called bidirectional. We had to build an auto regressive one. What does it mean? All the video models today, if you go to, you know, Google Veo or a bunch of these, you put in a prompt and then generate spits out like a five second clip. And it does a lot of processing. So it's thanks for like a minute and then you get your five second clip. What we had to do here is if we're generating a five second clip or an hour long clip, doesn't matter. We're not generating the entire clip at once, but we're generating frame by frame. So what you're doing with what you see with Mirages, it gets a live input stream and creates a live output stream. It just generates them frame by frame and not the entire video at once. That to do that, it's kind of a combination between a video model and an LLM. So you know how LLMs just generate the next token.

B (19:40)

Yeah.

A (19:41)

So it's kind of like training a video model just on next frame prediction and not next token prediction. You just have to predict the next frame each time.

B (19:49)

Okay, so what you're saying with the auto aggressive stuff is what you're doing is feeding in, let's say you're halfway through the video, you're feeding in the the frames of video that you've already generated to the model to produce the next one. I assume you're also feeding in like a source frame of the video that you want to generate. Yeah, and that's how it works.

A (20:13)

Exactly. So you feed it in two, two video streams. One is just the video stream that needs to edit and two, what it already generated and it needs to predict the next frame.

B (20:23)

And what is the advantage of doing it that way versus just putting in the two frames, the last two frames, last video generated frame and the last video input and doing that next frame versus like a bunch of frames. What is the advantage of that?

A (20:40)

Nice. So you really need the longer context you have, the more you know of what happened. Because if for example, you know, I was wearing a black shirt in this stream and now I leave the camera and I come back, how do you know to put on the black shirt again, not to change it? It's like the model's memory to be able to see what was in the past. Also it's very critical for the model to see motion because it's not just images. If for example, you use the model to add a portal gun, it needs to be able to see me doing this action in order to be like, okay, now I need to create a portal. So it's very valuable for the model to see motion and not just static images.

B (21:23)

I see. And is it like a diffusion model or what kind of model is it?

A (21:27)

Yes, it's a diffusion. We call it an LSD model, a live stream diffusion model. I have no idea why you're laughing. It's a very technical term. I see that these LSD models, it's a live stream diffusion model. So it's a combination of a diffusion model with an autoregressive transformer, which really just lets you generate and predict the next frame every time that you're generating it. Now the hard part about this, do you remember like early LLMs, like even like GPT2, 2.5 days or even when GPT 3.5 came out, they'd get stuck in loops. Do you remember those days?

B (22:07)

Yes, I do.

A (22:08)

Ever happened to you? So for listeners Sometimes in the early days of LLMs, you talk to it and it'd be great for a few back and forths and they would just start repeating the exact same thing. You'll just say, repeat, get stuck on a word or on a sentence, just keep on saying that exact same thing over and over and over again. That same problem that LLMs dealt with a few years ago comes back when you try to do auto regressive video models. What happens is the autoregressive, like the model's great for the first few seconds. Mirage. The biggest challenge was we could easily get Mirage to be great for 2, 3 seconds, but then slowly starts to degrade and it gets stuck in this loop until it just gets stuck on a single color and your entire screen just becomes red or blue or green. And solving that repetition problem was the hardest thing about mirage. Now if you use Mirage, it can do an infinite stream. You can keep on going with it for hours on end.

B (23:09)

Is that from the Yann lecun problem with LLMs, which is that errors start to compound and so the.

A (23:14)

Exactly the more.

B (23:15)

Yeah, so how did you solve that?

A (23:17)

So it's the exact same problem. It's the error accumulation problem. But every frame you're generating, you're going further and further from the distribution that you actually saw. You get slightly more error as you keep on going until after 100, 200 frames, you just lose everything in your static color. So solving that took six months of research. We had an easy version of Mirage that worked six months ago, just could hold for two seconds. And so going back to said, your question was, how do you build this? So on the one hand you have to do lots of GPU code to actually get this to run fast. On the other hand, we had to solve lots of very tough research engineering problems on the model layer it required. I think at this point we ran several thousands of experiments until we really got it right.

B (24:08)

And is it one thing that you unlocked where it just, it fixed everything, or is it a bunch of little incremental things that you fixed here and there?

A (24:17)

You know, what you're talking about is a great, it's a great point about AI. It was roughly seven or eight things that we had to unlock. And the annoying part is it's binary. Until you unlock it, nothing works. And then when you unlock it, suddenly it's all better. And that's an annoying part about AI research that you never know how close you are. You have no idea if you're going to solve this problem next week or two. Weeks from now or two months from now, or two years from now.

B (24:46)

But how did you know, for example, on the second thing that you added, it's still not working, but how did you know that it was better and you should keep that improvement?

A (24:56)

So we broke this down to lots of different, smaller problems. First, be able to do this just with images or be able to do this just with certain types of conditions. As you know, for example, Oasis was the last thing that we launched that went insanely viral back in November. It could generate Minecraft, remind people what Oasis is. Yes, it's a model that could generate Minecraft in real time. Went insanely viral. Everyone played with it, was very fun for us. We just came out of stealth and that was a real time video model that worked just for Minecraft. The reason that was actually an interesting step in the way towards Mirage. If you try to create a video model that is great just for one single distribution, so it just does Minecraft, it doesn't know how to do everything in the world. It is much easier for the model to learn that. So for example, that's one of the problems that we used as a stepping stone. We would take larger and larger distributions. So you begin with Minecraft, which is a very small distribution. It's all roughly the same. And then you do, okay, let's do just like streetwalk. Take a bunch of just video of streets across the world and let's be able to simulate that until you grow and you add more and more complex distributions. And each time you see like, okay, can I crack the next distribution? Can I actually get it to work with this additional challenges?

B (26:16)

It's sort of like, I don't know, the way a child learns to run. It's like first you sit up and then you pull yourself up and then you like there's, there's a bunch of steps, right?

A (26:31)

They have to use more and more muscles as, as they, as the, the challenges get to get harder and harder. Yeah. But you know, all this takes me back to why are we even doing this? So I'm a huge believer there's some, some entertainment, fun creativity experience that we're going to unlock in the next year that will completely change everyone's lives. And that's what we're here for. And the way I like thinking about is this. What's the Internet? What was the old Internet before ChatGPT? The old Internet before ChatGPT. Three years ago, it used to be four things. It was knowledge, it was creativity, like knowledge is everything. Google oriented creativity is Facebook and TikTok and Instagram, but also Netflix and Roblox. Everything you do when you're not doing your homework, shopping and messaging, that's the Internet. That's the. That's the entire Internet. There's nothing else.

B (27:33)

You're forgetting porn.

A (27:34)

These four categories. That's the four categories of the Internet.

B (27:43)

But continue.

A (27:46)

That is an interesting point. Anyway, the Internet used to be four things. Notch five for anyone wondering now the first thing. Knowledge. Chatbots have just obliterated knowledge in the new Internet. Anything we used to do, knowledge oriented, whether it was search or docs or anything, knowledge sharing or knowledge creating or maps, it all is going into chatbots. Doesn't matter if OpenAI wins or Gemini wins or XAI wins, someone's going to win that. But chatbots are replacing knowledge. What happens to creativity? Is there going to be a completely new experience that lets us be creative, lets us have fun that's different than what we already know so far? And the way I like phrasing it in the easiest way possible is if you go up to a 12 year old today, a random 12 year old either in Kansas or in Germany or in Japan, and you ask them what's chatgpt? What are they going to tell you?

B (28:46)

Martha listens to her favorite band all the time.

A (28:48)

In the car, gym, even sleeping.

B (28:54)

So when they finally went on tour.

A (28:55)

Martha bundled her flight and hotel on Expedia to see them live. She saved so much she got her seat close enough to actually see and hear them.

B (29:04)

Sort of. You were made to scream from the front row. We were made to quietly save you more.

A (29:10)

Expedia made to travel.

B (29:12)

Savings vary and subject to availability. Flight inclusive packages are at all protected.

A (29:16)

Tito's handmade vodka is America's favorite vodka for a reason. From the first legal distillery in Texas, Tito's is six times distilled till it's just right and naturally gluten free, making it a high quality spirit that mixes with just about anything from the smoothest martinis to the best Bloody Marys. Tito's is known for giving back, teaming up with non profits to serve its communities and do good for dogs. Make your next cocktail with Tito's. Distilled and bottled by 5th Generation Inc. OST Austin, Texas. 40 alcohol by volume. Savor responsibly.

B (29:47)

A chatbot?

A (29:49)

Yeah. What do they do with it?

B (29:51)

I do my homework with it.

A (29:53)

Exactly. Every 12 year old on the planet will tell you the exact same thing. Oh, Chad. Oh yeah. That's the AI that does my homework for me. Okay. It helps me with My homework. And then you ask them a follow up question. What AI do you use when you're not doing your homework? And they have no idea. Okay. They stare at you like, I don't use any. I. When I'm not doing my homework, I'm on Fortnite or Minecraft or TikTok. What do you mean an AI when I'm not doing my homework? And my entire belief is AI is changing everything. It really, it really is. And it can't be that AI doesn't create completely new experiences that just get to have when we're not doing our homework.

B (30:35)

Yeah, I mean I definitely resonate with this. I think it's really clear that every new technology creates new content formats, creates new art forms and we maybe have cracked some of that. You can see Suno, for example, is gesturing in that direction. But there's probably or definitely a bunch of things that are on the horizon that we haven't quite figured out yet. Because for example, video games used to be really hard to make. Now everyone can make a video game. That's a really interesting thing. What does that look like? And it's probably not going to look like traditional AAA games. We'll evolve it into something else that has for the medium that we've just invented.

A (31:19)

Exactly.

B (31:20)

I think my question for you is how do you think about the strategy of inventing a new medium like that? So you're obviously starting with the foundation model. An interesting question for me is, for example, OpenAI started with the foundation model but it didn't really work until they created ChatGPT. It didn't really work until there was a form factor that was consumer friendly. Are you all working on that? What is your strategy for taking this foundation model and turning it into something that people actually use to make stuff?

A (31:50)

So right now there's no app outside, but by the time this gets published there will be an app because it's coming out in the next few days. And I completely agree that this great tech has been. There's this term called AI experts that it's people who know how to use AI, know how to prompt these systems and everything. You have to turn everyone into what's called an AI expert. A random kid on the street or you know, a grandma living somewhere should be able to be an AI expert. And just, and what we need to do is really find ways to put this in hands of people, put it in an app, put it in something that's comfortable for them to use and let them access this great tech that's been built over the past few years in audio and video. I think it's a combination of what's going to create these new experiences and really put this in the hands of people. I think we'll do a lot of things. We're going to publish lots of different experiences through this app, and it remains a mystery to see which ones will take off and which ones won't. And it's kind of an experiment we're going to be running together. We're all, as humanity, going to understand, okay, what is fun about these experiences that we can do.

B (33:06)

Can you show us the app?

A (33:08)

I actually can't on this computer. It's a new computer, so let's do this. I can explain it here. So the app's called Delulu. Okay.

B (33:19)

It's very Gen Z of you.

A (33:21)

Just come on. We all have to be. We all have to be a bit delusional. Okay? It's what we're building. We're building the one place you go to when you're not doing your homework, which is the Lulu. Okay, now what you're going to get in the Lulu right now. If you go in and download it, you're going to be able to just upload a picture of yourself from your camera or take a photo, and the app itself will just create hundreds of variations of it instantly. So you can see yourself eating pizza on the Eiffel Tower on top of Mars, or you can see yourself. It's actually all the gender stuff is really fun when it turns a guy into a girl and vice versa. So we can see you in a dress or you can use it to express your emotions. One of the things you can do is you can be angry at a friend or something, and you take a selfie and you tell the eye, okay, turn me. Make me angry. And, you know, you'll suddenly get, like, devil horns or something and, like, smoke coming out of your ears. So what you'll be able to do is there is. You can take any. Any photo that you have and just express yourself in a way that you just couldn't before. Right now if you download it, you're gonna get all these. These imaging tools. You're also gonna get, over the next few days, the same thing with video. And it's gonna be added as a camera. So you open up your camera and you just point it at something and you point at your friend. You're like, okay, turn my friend into Elsa. And suddenly your friend becomes all blonde and with a blue dress. Or you can point at your friend and say, hey, Add a little monkey on their shoulder and you'll have a little monkey sitting right here and they can high five the monkey and the monkey high fives them back. So all that's in the Lulu. And there are going to be so many releases coming out over the next few weeks. Every week we're going to have a release for Delulu with new capabilities that we're adding in constantly on the video and the audio side.

B (35:15)

And is it like a social network or are you posting this to Instagram or Facebook or whatever?

A (35:20)

First stage, you're gonna be posting this to Instagram and whatever. But what we really found interesting is trends. When someone finds someone here at the office playing around with the lulu and it turned their hair into grapes and that was actually super cool. And they wanted to share that with someone else so that you could also turn your hair into grapes because other people in the office were like asking, hey, how'd you do this? Can I do this as well? And so I think one of the key parts that we already find that is very interesting in this category is how do you have this place? You can discover new ideas for what the AI can do together with you and try these out yourself instantly and see if you like them, see what they make you feel. So I don't think it's going to be a social network in the sense that we got used to, but it's going to be something completely new, like adding in a feed there. I'm not sure it's the thing people will want or that will want, but we will want something else. We'll want the ability to find new trends, to find new cool stuff that's happening. Another thing that we found was really interesting, you know how social media today is not really social media. Like it used to be all social graph and friend based. Used to be able to do things on Facebook, Instagram, with your friends. Now it's all influencer based.

B (36:38)

You.

A (36:39)

You see something cool is you see trends that are happening, but you don't. It's. They're not your friends.

B (36:43)

Thank God for my bit. For, for me, because that's my business model. Exactly. I do miss those days.

A (36:50)

But there used to be days when we had, you know, like our actual friends on social media and we found that this was actually really cool. With Delulu, people actually cared what their friends thought about the edits they were making. So I could make like 100 different variations of myself and be really cool to see if my friends or my family, I'd send it over to them and they usually would, like, the ones that I hated, and that would trigger, like, a huge spark inside of you. Like, holy shit. This is. They. They're. They're. They're. They're seeing the world in a way that's different than I am. And it's. It's fun to do that.

B (37:28)

So I think this is very cool. It reminds me, though, of some things that have already come out so easy. One is Animo Jeez from Apple. There was also, like a whole trend maybe like two years ago of, you know, I can't remember what it was called. It was like Dream Booth or something like that. Basically stable diffusion, generations of people, you know, headshots, basically, you know, it's you in a forest or whatever. So there's. There have been some waves of this before, and those have all kind of burned, like, gone viral and then burned out. So what is your. How do you think about that and how do you think about this as being different?

A (38:06)

So, by the way, it's, you know, you bring up Green Bruce. We have one of the experts on the subject of the. The person who was last author on the. On the Dream Booth paper, kfir. He just joined us a month and a half ago in our Tour SF office. He used to be at Snap at Google before that. And, you know, I was having this conversation with him as well, like, okay, what's happening here now? The tech has come so far over the past two years since Dream Booth came out that back then it used to be a great poc. Now it actually works. Like, you look at the picture, like, oh, wow, that really is me and doing something insane. And it doesn't take any time to do that. The AI is creative enough also to come up with new scenes. So I think there's a. There's been a huge, huge jump on the quality of these things that actually really gives you a different experience. But B, I agree that images are just a gateway to what's enabled with video and audio per version to Lulu. I think we've all seen these image editing apps, and in here, when we're making it much easier for active people to access completely for free, that creates a different dynamic, but it's still just a precursor to the things that are being added right now, which is video and audio. When you do this with video and audio, when you do it in real time, when I just take out my camera and point it at something and ask the trash can that's here next to me to turn into an elephant, and it becomes a little elephanty, Trash can just walks around the room. That is something completely new that we just never had before. If I ask the AI to add a little turtle here on my shoulder and the turtle starts crawling around my shoulder and rests down here and goes to sleep, these are things we never could really do. And I think it's very interesting to see what people will do with it.

B (40:04)

How are you testing this internally? It seems like the idea is probably kids are going to be using this a lot. And so do you have alpha or beta testers who are in that age category that you're doing this with? Or how do you make something that you think kids are going to love if you're not kids yourself?

A (40:26)

So, first of all, we have a bunch of alpha and beta testers. And that's the thing that excites me the most, talking to all these people. And by the way, if you're listening.

B (40:37)

In and you want to be an.

A (40:38)

Alpha beta tester, shoot me an email directly. Dean at thecart, Dan at thecart. AI would love to be able to show your tech before it gets released, because that's the way that we're exploring this together. I mean, we're creating great tech, we're using it for ourselves and it's fun for ourselves. But having people experiment with it and understand together what this is going to enable humanity to do. We're all just building this together and we'd love to have everyone's input on this. I think one of the critical things, from our perspective, in two or three years, sure, the world is going to have almost all its pixels and audio waves be AI generated or AI modified. And we have to put this tech into people's hands right now so that people also build the tolerance and understand how to use this and what this means for them in their lives.

B (41:34)

What are your AGI timelines?

A (41:38)

I think there are only two interesting problems in the world right now. Only two. Nothing else matters. Really. Nothing else matters. One is the race to AGI and two is the race for consumer dominance. That's it for building the biggest consumer company in the world. Those are the only two races that matter. Nothing else really matters. Everything else is fun. Everything else is cute and maybe worthwhile, but not actually interesting. I think with AGI we have to split into two. There's the Terminator stage where we get to singularity and it's smarter than all of us and we have. It's a good question, what happens, team? And then. And the second question is, when you get economic AGI, when just machines are able to do any economic job better than all of us or better than the vast majority of humanity. That just makes no more sense to do any kind of virtual work anymore. The second one, the latter, it's probably a few years away. The former is a few years away. The latter, though, the economic AGI could easily start creeping up on us in 12 to 18 months. Wow. That will slowly start seeing AI be economically better than a lot of us and creating economic value, and that will have dramatic impact on society. Now I'm a huge believer that this is going to be humanity's best golden age, best time possible. We're going into a new world. Sure, AI is getting better, it's getting stronger, it's getting smarter. But humanity's future is very, very vivid. And we have to understand how we build this and at the same time, just build the best world possible for humans.

B (43:36)

Okay, let's take that one at a time. I think the big important thing I want to talk about is, if I understand you correctly, you think that most economically valuable work will be able to be done by AI in the next 12 to 18 months. Yes. Another way to say that is it will be less and less appealing to hire humans to do work and will be hiring AI to do work. Or what do you think are the implications of that? What are you really saying when you say that?

A (44:09)

I think that it's very clear at this point that much of the work, whether it's lawyers or accountants or lots of jobs that are very, very virtual oriented, a lot of that work can be done really well by AI. And it's getting better and better. And there's no reason to assume that it won't continue to get better. Now it does mean that a lot of us will have to adapt, just like we all adapted when the Internet came out or when previous generations of technology came out. It also means that humanity will get a lot more time to think, to be creative, and to explore what's in our hearts and not just what's in our minds. What I like going back to a lot. Why did democracy first start in Greece? I was actually very intrigued by this. Democracy first started in Greece because people had the time to think. You didn't have to work the fields 24, 7 before that they had to be. Agriculture was very, very hard. You would be in the field every day for 14, 15 hours and you would just go to sleep and go back to work the next day. As technology got better and people started getting some time to be able to take off and not work all the time. The first thing that we got was that the ancient Greeks became philosophers and they started thinking, okay, why are we even on Earth? What are we doing here? What's the right way to govern people? And they created all these systems that we still use today. And we were able to get that only because humanity had more free time on its hands. With more and more economic value coming out of AI we will have more time on our hands to be creative and to use our imagination. And that's something I'm very excited for. That's something that we have to give an avenue for people to be able to use their more added time to be more creative, to actually do stuff that makes them feel something. This Labor Day at Lowe's shop member only Door buster deals for a limited time. Save $50 on an ego string trimmer now $169 plus get 50% off off select Holland Pavers. Not a rewards member. Sign up for free today. But hurry. Labor Day doorbuster deals won't last long. Lowe's we help you save valid through 91 while supplies last program subject to terms and conditions. Details@lowe's.com Terms subject to change.

B (46:47)

Does it ever feel like you're a marketing professional just speaking into the void? Well, with LinkedIn ads, you can know you're reaching the right decision makers. You can even target buyers by job title, industry, company seniority skills. Wait, did I say job title yet? Get started today and see how you can avoid the void and reach the right buyers with LinkedIn ads. We'll even give you a $100 credit on your next campaign. Get started at LinkedIn.com results terms and conditions apply. So there's a lot here and I want to go into the ancient Greece stuff because I have a whole, I have a whole take on that too. And I think there's a lot of overlaps. Actually.

A (47:28)

I would love to hear what you think on that. But yes, there's so many parts here.

B (47:32)

But I want to start with the just going back to knowledge work pretty much be done by AI because I just, I will. I'll take the opposite on that. I don't think in 12 to 18 months if, if you come back on the, on the podcast that we'll, we'll have seen that happen. And there's a couple, there's a couple of reasons I have, but maybe I'm wrong. The first thing is I think you're right. Like, let's take the lawyer example. I think that you're right that if I think about the Things that I ask my lawyer to do. Given the right prompt, the AI is going to be better and faster and cheaper than my lawyer. However, given the right prompt is like an incredibly difficult question.

A (48:13)

Yes.

B (48:14)

And I don't think we're 12 to 18 months away from self sufficient AI that's able to know what prompt to give it itself at the right time. And I don't think that that is a, that's just a point that we pass. I think that's very spiky, depending on the domain and the context and all that kind of stuff. So the way that I think about AGI, for example, is I actually think about it in terms of child development. When a child is first born, they are not independent at all from their mother. As they, as they get a little bit older, you can leave them alone for a minute, five minutes, ten minutes at a time, until when they're 20 or 20 years old or whatever, 18, theoretically they're adults, they can do whatever they want. And I think you can see AI following that same trajectory where if you take coding for example, it started with tab completes, it's like one step, right?

A (49:16)

It was just tab completes, it was like. And this was two years ago.

B (49:19)

Exactly. But if you. But now it's like cloud code is like 10 or 15 minutes, but only sometimes. So like you can't use Claude code on your deep down assembly stuff. Right. But I think, I think a better definition of AGI is when it is economically profitable to leave your AI on all the time. It's always working, it's always doing something. However, even in that scenario, which I think is pretty far away, like several years away at least. Well, let me just say I think that's a good definition because it will require a lot of things that we don't currently have. So for example, continuous learning. I think the only way to get AI that knows how to prompt itself in the right situation at the right time is if it's good at continuous learning. And I don't think that better context engineering is actually, is actually going to fix that. I think that's going to be maybe part of the solution. But you have to be able to update your weights is my, my belief. So, so that's why I think, I don't think it's 12 to. I don't think it's 12 to 18 months away. I think it's, I think it's much farther and there it will be very, very domain specific when it is totally AGI, totally independent. So maybe coding is first, but that'll still be only some kinds of coding. It's still not going to be for your CUDA improvements necessarily or PTX improvements.

A (50:49)

Right, Right.

B (50:50)

So that's my AGI take. What do you think?

A (50:55)

So I love this because you said when does it become economically viable, profitable to just keep your AI running all the time. Okay. And clearly, by the way, we're at that point with computers keeping my. Just my laptop on all the time. Okay. Is. Is something that generates like the, the. I didn't even think about how much electricity it costs me to keep my laptop running because I just know the value I'm getting out of it is insanely higher than the. I know. How much does my electrical bill for my Mac cost a month? 20 bucks maybe. So it's definitely economically viable to keep your Mac on 24. 7 and we're probably going to get there with AI as well that any price I'll pay for actually running this AI will just make itself back instantly. And you gave the cloud code example. I completely agree that it's not going to replace coders because I've been using Claude code over the past two weeks to code some stuff for our app. I've been using Flutter. I have no idea how UI works. I literally never wrote any CSS or HTML or JavaScript or any of these things. I'm in the realm write assembly for GPU codes. I love that. No idea how UI works. But I was with a team working on the UI of the app and I was able to actually contribute a lot to that project via cloud code without knowing anything about the syntax of these languages and what the contributions were that I had intuition of like, okay, okay, you gotta be doing something wrong because there's no way you're storing the image here but also somewhere else. And so you must have sent it through something else in between. And having the intuition of okay, what's the actual challenging part you had to do of moving the data around or whatever and just letting Quadco do the actual typing? That was something that personally I've been experiencing over the past two weeks and it's been really great and it's been very valuable. So I think that maybe a different way to phrase is it. There's a chance that in 12 to 18 months AI lets us be so productive that we're able to create companies that are just way bigger than anything we saw so far. And I'll give a concrete example of this. Today we have 10 trillion dollar companies in the world, give or take depending on exactly when you check and how Tesla stock does. But we have 10 companies in the world that are past a trillion dollars in market cap. Back in 2017, we had zero. We got our first one in 2018 when Apple crossed the line. And that's what, 2018? That's seven years ago. Okay, seven years ago we crossed the line of first having a first trillion dollar company. And now we have 10 of these. A lot of it is due to just technology making us that more efficient so that we could actually make a lot more money and bring a lot more value to the world. And I think there's a very good chance the next 12 to 18 months, AI just creates so much value in humanity that the entire stock market doubles in value.

B (54:08)

Hmm, that's an interesting one. So now I want to go back to Athens because I think this, this actually, this dovetails with the point you're making. So I'd probably say yes. I don't think 12 to 18 months, I think that's way too quick. I think it will take many, many years for it to like even. Let's say you, you vented this, this AGI. Let's, whatever our definition of AGI is, but you never want to turn the AI off we invented tomorrow. I think in 12 to 18 months, maybe there'd be a bubble in the stock market, but it wouldn't be like a thing that would true value maintain. Yeah, okay. I think it'll take a lot longer. But I do grant your point. Like it will allow us to build much bigger companies than we could before. I also think it will do something else, which is allow us to build many more smaller companies that accomplish much more than, than you could as a small, as a small company. And, and, and I want to talk about why. So back to your Athens point. I actually don't know why democracy arose in Athens. I like, I remember like Solon's laws or whatever from my history of ancient Greek class, but I can't remember why he, why he did it. But, but the thing that I love about Athens is, is, is it's a society of general generalists. Right? Democracy sort of requires that direct democracy requires that where you can be a statesman, you have to be a, a lawyer. You are a prosecutor, you are a juror, you are a warrior.

A (55:42)

You're everything you need to be good at everything.

B (55:45)

You have to be good at everything. And so the question is, why did that stop working? And the reason it stopped working is because Athens became an empire. It became the equivalent of a trillion dollar corporation. And what's interesting about what's interesting about being an empire is you actually need specialists because you can't send some farmer to be the general on your Sicilian expedition. Like, you want someone who's got a lot of experience. And so specialization requires, sorry, specialization allows for collaboration over, across larger and larger organizations of people. But then you lose the sort of, this sort of generalist thing. You incentivize being a specialist, which I think has basically just, just continued in Western society over the last like couple thousand years. And there's a lot of good things from it. But also if you're like me and you, we love being generalists. It's fun.

A (56:46)

Exactly.

B (56:48)

And I think what's interesting about AI is because you have this, this thing in your pocket that's like a thousand specialists in your pocket. It allows you to do a lot more for longer. As a generalist, you don't have to specialize as much. And so for example, if we take Every, we've got 15 people, almost everybody inside of every is a generalist and is doing multiple jobs. The lines between jobs starts to blur.

A (57:13)

Incredible.

B (57:15)

And they can do that because they have AI. And so my hope is maybe in a similar way to yours, we get back to you start moving back toward more generalists and this sort of like golden Age of Athens type vibe because people can get more done with AI. They have all the specialists and they can coordinate across more people without having to specialize.

A (57:42)

I love this theory that we're going towards a world which is going to be more. We'll actually have space for generalists. I think it's going to happen. Just two reasons. One is exactly what you're saying. If AI gives us tools to jump into fields we're not experts in, but the AI has, it has seen that experience and we can just through the AI actually manifest our general abilities. Yes, exactly what I was doing these past two weeks with Claude, Code and Flutter. I have no idea how to write iOS or Android apps. But I do have common sense. What I have is common sense. And I can use that common sense combined with what Claude knows about how to write code to create this thing. You know, my, my dad, he's retired. He used to be dean of medicine at one of the Israeli universities. Is now. Now he's, he's not officially at Descartes, but he does, he does come with us to. He hops by the office like twice a week and, and he likes being called the senior advisor for common sense. Okay, look, I'm not a lawyer, I'm not a coder. I'm not a business person, okay? But I have common sense. That's what I have. That's what I can contribute to this, to the situation. And so I think that's, that's really a huge point that they like. What you're saying is, yeah, will actually be able to have generalists that can act as if they were specialists in each field. So that's, that's really valuable. Something else that I think about a lot in terms of AI. So every is 15 people. Descartes is roughly 70 people now. And I think you've done a tremendous job at having 15 people that almost all of them are generalists. That is typically incredibly hard to do. And you know, it's, it speaks for itself because you're actually able to create insane stuff that no one else can. Also here at Descartes, almost everyone at the company is a generalist and an independent thinker. And they just go ahead and they do stuff without being told, we're 70 people, we're still completely flat. For many years. I spent around two years of my life doing this. I was thinking a lot about what organizations work and what don't. Now here, lots of organizations, like you said, most trillion dollar companies today just don't work. Okay? Google has 180,000 people. They don't really work. We can't get the best out of them. They're smart people, they're talented people, but Google has a hard time getting the best out of them. But it's not just Google, it's every big tech. And this got me thinking, what's the biggest bottleneck to actually running a company or a group of people and telling them, hey, you all have to do the same, go build that one huge thing together. What is the bottleneck? And by the way, what do you think?

B (60:44)

So I would say there's one thing is just there's a real information flow problem. So flowing information from the bottom to the top and from the top to the bottom, which AI solves. And then the other thing that I think is really important is humans have different motivations depending on the circumstances that they're in. And inside of a big company, they tend to be most interested in the thing that will get them promoted rather than just doing the right thing. And that's for a lot of different reasons. It's partly due to information flow, but that's, I think, a big thing, right?

A (61:24)

So I have no idea how to solve the second one. And I agree that fine, okay. But even if we did have tons of people who only really wanted to actually build insane stuff. I'm not sure we can fit a thousand of them in the same company and still actually get things done because of the first constraint that we have, the information flow constraint. And you see in the organizations that function incredibly well and those that don't, that's really the key difference. So with an organization that's up to 100 people, sure, information will flow and it'll be fine because you can get everyone in the same room and they can all talk to each other and just move information that way. There's also really big organizations that did work really well. The ancient Roman legions, okay, you can have 100,000 people with swords and shields and you'll tell them, okay, go conquer that hill. And then go conquer that hill. Now, the reason that those organizations worked was because not a lot of information really needed to flow. You could just tell them, look, we're on this hill, they're on that hill. March and stab. Okay? That's the information that you needed to get to your 100,000 people on the field. And what I tried to do over a few years and then the conclusion I got to was organizations that work versus ones that don't, it's how many generalists do you have inside the org? Or how many slots do you have for generalists? If you have only 20, 30 generalists in the organization, that's great. They'll think of crazy stuff and they'll go ahead and they'll pull it off. But if you have a thousand generalists, they will get stuck on communicating with each other and understanding. Wait, can I do this? Can I not do this? Is it okay that I do this? And that's where humanity has never been able to construct an organization that gets a thousand creative people and lets them all be creative at once. So any organization you're in can only enable maybe 20, 30 people to actually be creative. Whether it's a 20, 30 person startup or a Roman legion. Finally had 20 creative generals and 100,000 foot soldiers who didn't have to be creative at all.

B (63:39)

Really interesting. And why is that?

A (63:43)

It's a good question why that is. But I think that when your job doesn't require creativity, it's rather easy to define. Okay, Your job is to go ahead and you're working a Ford factory. Your job is to pick up this tire and put it on the car, okay? And it doesn't matter if you do it exceptionally well or you do it average. If you do it average or exceptionally well, maybe it's 50% difference. But if you have to be creative, like, okay, go build a completely new experience for people or build some breakthrough technology that is a role that you need to be able to take risks. You need to be able to do stuff that's much more complicated. It's hard to define. And for people, the way that we have information bottlenecks in that we can't communicate too much, I can't take everything that's in my mind just telepathically give it over, send it over to you. And so we have to be able to dumb things down and explain stuff in very, very simple terms, like, okay, your job is just to do this, and I can explain bullet points. And so when you need. When something's creative, it's in early stages, not that easy to define.

B (64:52)

I think I agree with this. I think the. The way that I would describe it is in any kind of work, there's a. There's at least there's two phases, and they're kind of fractal within each other. But, like, we can just for now say there's two phases. There's the explore and then there's the exploit phase. In the explore phase, you're trying to figure out, like, what do I even want to do? And the exploit phase, it's like, it's just an execution problem. You're just solving puzzles that you like. The frame of the puzzle is set. And we currently use a lot of humans for the exploit thing, and we may be able to use a higher. A higher ratio of humans for explorer now that a lot of the AI is going to be able to do exploit, because AI is currently very good at problems you can define, but there are surprisingly few problems you can define that are valuable compared to all the problems in the world. And actually, I think this goes right back to. This is a really good bookend to the podcast because this goes right back to the first thing we talked about, which is one needle in a haystack, or this sort of infinite stack of this infinite forest that you can go down, and every path you can go down is meaningful, and you kind of have to. You kind of have to explore that infinity, and there's no right answer. And so I think we can. We can. In a world where AI is quite good at finding the needle in the haystack or something like that, we can spend a lot more time exploring. And AI is also a good tool for exploring. But it ultimately comes down to those, I guess, those two phases of creative work. Maybe it was a way to say.

A (66:28)

What you're saying exactly. And AI AI is going to leave the creative stuff to us. It's going to do the stuff that is well defined and that's great. And not only that, because it overcomes lots of communication bottlenecks. It can actually see everything we're doing. You know, in a thousand person organization beforehand before AI, no one, not a single person knew everything that was going on. In a thousand person organization, no one read every email and heard every phone call. Now AI actually can, and so it can help us communicate better and help us each be creative and let us explore until we find something that's valuable to go ahead and exploit and they can just do the exploitation on its own. I think that's a great way to sum up what I think will happen over the next few years. That AI will do the stuff that's more well defined and we'll leave it to humans to be creative.

B (67:22)

Dean, this is a great place to leave it. This is an amazing conversation. Thank you so much.

A (67:27)

Thanks so much, Dan. This was amazing and we should do it more often.

B (67:31)

I would love that.

A (67:40)

Oh my gosh, folks, you absolutely, positively have to smash that, like, button and subscribe to AI and I. Why? Because this show is the epitome of awesomeness. It's like finding a treasure chest in your backyard, but instead of gold, it's filled with pure, unadulterated knowledge bombs. About ChatGPT Every episode is a rollercoaster of emotions, insights and laughter that will leave you on the edge of your seat craving for more. It's not just a show, it's a journey into the future with Dan Shipper as the captain of the spaceship. So do yourself a favor. Hit like Smash, subscribe and subscribe and strap in for the ride of your life. And now, without any further ado, let me just say, Dan, I'm absolutely, hopelessly in love with you.