How To Make Games In 2025 (Jabrils) — Full Time Game Dev Podcast Ep. 041

Look, I'll be the first to admit I'm obsessed with AI in the game industry, but not for the right reasons. See, I'm scared of AI. All of this makes me wonder, is the game industry going to be completely destroyed with all of this disruptive new tech? If there's any game developer out there who can help me, I don't know, see the forest for the trees, it's Jabril. Jabril is a gamedev and highly successful YouTuber. One that I actually had the pleasure to meet along with the Brachie team in San Francisco a handful of years ago. See, Jabril's has amassed almost 500,000 subscribers, and he has a slew of content that quite frankly tends to go way over my head. So, I couldn't help myself in asking Jabril to come on the podcast and talk about AI in the game industry. We discuss a ton of stuff. We well talk about why AI is so misunderstood right now. Most people don't even know what real AI is. We talk

all about that. Whether AI is going to take a bunch of gamedev and tech jobs. We talk about why AI and automation are actually totally different things. We also talk about why 2D and 3D art generation are potentially going to fall flat. We also talk about where AI fits into more personal indie titles and whether indie games are going to be disrupted by AI. You're going to learn so much about AI and honestly there's some disagreement between me and Jabril as well in this conversation. So stick around for that. So I start out the conversation with an important question. Are personal indie games going to go away in this new era of gamedev or do they actually become more valuable? Let's jump in. But before we move on, I want to thank this video sponsor, Bezi. They are the creators of this new tool that could be considered your development partner for your project. The tool uses AI to write code and scripts and has a real-time context of your entire project like the codebase,

the asset library, scene graph, and components, which means all recommendations are made specific to your project. It helps generate exactly what you need, building on what's already existed in your project. It can help with optimization and is great for handling repetitive and boilerplate tasks like debugging to let you focus more on the creative side of your game in the complex parts of the creation. It's important to mention that your project's integrity and data privacy. They're of course completely secure. Bezi is currently in early access, so make sure you go to join the wait list to be one of the first to get access simply by clicking the link in the description. I think it only increases it probably by an order of magnitude to be honest. Uh as people are looking for authenticity. Uh I don't know if you've had this experience but there have been increasingly corners of the internet that I've visited where it's just AI slop is the output and I'm like oh no

it's what's going on right? and that feeling you're like I want to be connected to human experiences you know. Yes. So I think like if that does continue to increase like it's only going to increase by order of magnitudes I believe. Yep. Becomes more and more valuable too because it's going to become more scarce. Yeah. Like this this idea of a handcrafted game or a handcrafted product that comes from a solo developer or something, you know. But at the same time, I'm not fully convinced that we're going to have like AI made games, at least to agree. I'm not convinced yet. I Well, what's the vision, right? Is the vision that you type out, I want Grand Theft Auto mix with Minecraft and and Tetris and it's going to spit out the game and then you can like sell that to a publisher. Like, is that the vision? That I do not see. I do see potentially like textures being used, but like at that point like like AI generated textures, but at that point is it not just a tool, you know, like Yeah. I

think about like on demand textures. So you're playing a game and the textures change based on what the user wants kind of thing. And this is a new feature that released in the new game Enzoy. I'm not sure if you're familiar. It's the competitor. It's called Enzoy. I think it's I N Z Oi, I think it is. Okay. It's a uh it's a Sims competitor that people are really excited about and they have a feature in there where you can use like a uh image AI image generator to create textures on the walls and stuff like that. That's cool. U but that that's about as far as I've seen it so far. And like I I stay pretty close to like a lot of the AI stuff happening and I think there's some headway happening with AI generated models, but even then like the texture wrapping on it, I'm sorry, the UV wrapping on it is really bad. It is. But that's not to say it can't improve, right? But even then, if we add AI generated models into games, I don't know. It's just I just don't see it at all. Well, surely surely if it's

got if if we've gotten AI generated 3D models, which they do exist, by the way. They're not great, but they do exist. Yes. Surely. Wouldn't it Don't you feel like it's been exponential, though? Like all of this stuff is going so fast. Surely it's just going to get faster and faster and faster and it's going to be more than just textures on walls, right? Um I I mean it's possible, but I don't think I think the curve is slowing down at this moment. We've had we had AI generated models back in 20 like 21. I remember playing with some something in 2021. It's now 2025 and they're still not production ready, I would say. So like when I just draw that delta of 2021 and 2025 like we need some crazy breakthrough which I I I'm not convinced we have yet to get to the point where I can do text to video game which which is the fear right. So you feel like the curve is slowing down not doing this. I think

it's tapering. I think it's tapering. Okay. So let's figure out this curve because I know you like uh graphs and curves and stuff. is the curve. What What do you call this one, Jabril? Is it It's like this. It starts slow, zooms up, and then and then tapers off. It's usually S-curve, sigmoid, you know. So, we've got an S-curve for AI technology. Do you think that that's the situation we're in right now? And we're at the top of that Scurve. Now, obviously, this isn't with every AI tech, but in particular, art generation and asset generation. I I do feel that way. Yeah. With with our generation specifically. I hope you're right. I really do because I don't want I don't want it as an artist. I want I want to do it myself, you know, and I want that to be what makes my games stand out, you know. I have a question for you. Do you do you use Photoshop? I do. You

do? H I'm curious where you draw the line. Once upon a time, there were uh there were a lot of people vocal about the um negative uh the detriment of of Photoshop getting in the market and how it would like replace artist and you know we need real artists drawing art instead of using these digital products. Yeah. I'm curious where do you draw the line? Boy, I I wish I Okay, I'll tell you where my heart draws the line and then I'll tell tell you where logically my heart draws the line. Okay. Okay. My heart draws the line at AI art in general. Hard stop. If you if this is, and again, this is my flesh talking. This is my anger and frustration and fear talking. Sure. Which is if you're using AI art outside of for concepting purposes. Okay, I don't like it. I don't want to see it. I don't want to see it in games. I don't want to see it even as AI. I

don't want to see textures being drawn with AI. I don't want 3D models. I don't want none of it. That's my heart talking. And I know it's illogical because welcome to technology, right? You know, that's that's just like it's like somebody who's who's really really angry and they're like the the roads can't support uh um cars. The roads support carriages, horsedrawn carriages, and can you imagine the infrastructure we need for for roads and cars are so dangerous and blah blah blah blah blah. And it's like, well, sure, patient, be patient, and then the infrastructure will be there, and suddenly everyone will be working at tire shops instead of, you know, putting on uh sho horseshoes, you know. So, the the economy changes, the industry changes, and I know that, but my heart it's like sad. I'm just It's more like I'm just sad, man. I'm like I don't I don't want to see games where the artist his hand was not directly involved. It

just makes me sad. Do you know what I mean? Yeah, I do know what you mean. I just So, I guess like my biggest issue when it comes to having this conversation is often times I feel the conversation is like we have to have all AI generated art or none at all, you know? Yeah. And I don't think it's that that dire you know like what is it called the um the bone what's it called 2D bone uh like the skeleton you know there's software that you can do for animating right we don't have all games are made using that you know people still draw handcrafted you know stuff to this day and it's beautiful you know like you're a great example of that I think there's something to be said about not all. Oh man, this is potentially controversial. Controversial. Yes. Come on. Give me the controversial episode. And And it's selfish because I'm sitting here in a selfish self-righteous

uh like the the highest form of art is two-dimensional paintings in Photoshop. And I actually, by the way, believe that in my flesh. My body feels that way. Sure. I feel it. I'm like handdrawn or like music. It's like that's the highest form. code that's way down the line when it comes to like the highest form of art. You and I, we got to fight there. I know, right? And someone, you listening or someone listening right now would go, so you're telling me, Thomas, that code is not an art form. And I would say not in the same way that and I'm I I I believe this. I know I'm wrong, but I believe it. So I can actually operate that way. I can believe something and know it's wrong. Sure. I can feel I can feel in my heart that a custom handwritten story for a game and it's written by Ken Lavine that that is more of an art form than the backend code of a game.

Yeah. And I want people I want to hear what people have to say in the comments. Do you do you agree that there this is an art form and this is a a this is a skill? They're different. What do you think? I agree they're different. Okay. I agree they're different. Um I think what is beautiful about what you just expressed is that you're an individual and you are entitled to have your opinion, right? Like irregardless of what I believe like that's your opinion and I respect it. You know, it's not correct but but I respect it. Hey, what is what is your view of the correct opinion? I want I want to hear what you have to say. I have a boring answer. I don't think they can be unpaired to be honest. Like you can't have one without the other, you know? Like are they both art forms in the same way? In the same way? No. Cuz if they're not in the same way, then they are unpaired. You've got one in a different category,

right? Okay. I I I will accept your distinction there. I will accept What's the question again? The question is, is are is every art form? Okay, it's this is a difficult conversation because No, I love this. I Let's go deeper. It's it's it's it's basically like why do people get mad about AI art but they don't get mad about Claude? Why? And and I think it's because this one is an art form and this one is not. It's art. Art is different than than code. They're different. One's more human, one's more one is saying something poetic, something from the heart, and another one is building something. They're different. Uh, the arguments I've seen is because a

lot of these, uh, image generators just steal a bunch of art and then spit it out without crediting. That that's a lot of the arguments that I've seen. That's one side of it. The other side of it is you humanity is lost. There's no more humanity in art. It's gone. There's no person behind the art, right? So, one is one is yes, stable diffusion is effectively stealing. The other one is okay, but let's even say it's not stolen and all of this material that it's it's being licensed by like Adobe or something. But you're still you still have a problem which is that the game itself is not human and it doesn't feel human and it's not it doesn't have a heart. But I arts in the eye of the beholder, right? Like if I go take a walk out in a forest and I see an interesting configuration between the trees and the leaves and the branches and the sun and the clouds and I take a picture that that becomes art, right? Because of my perspective. Um and I I don't know. I like I do not think that AI art is something that should be used in games in this current state.

That's not my position. I think that I agree with you that should be used in the concept phase um as inspiration. But at the same time, I don't think I'm the the person that would be anti- art in games. If it works well, it works well. And regardless, however you and I like it, the market ultimately decides, you know. Correct. song. Man, it's um isn't it funny how this has become a massive ethical dilemma? Like AI in general is just this massive ethical dilemma. And I don't know how you solve it. I don't I don't know how do you how do you solve any of this? I mean, I I don't even know if any nation or country or government has come up with any legislation around it, you know? I mean, I'm sitting here stuck on uh Steam Steam, they're basically I don't know if

you've seen Steam's um like licensing now. It's like you have to tell whether you've up whether you've used AI and it's like who's not who hasn't used Copilot, who hasn't, right? who hasn't used clot at least once, right? Um, how's it defined? Yeah, how's it defined? And and I'm just really curious how this all ends up, you know. Um um one of my AI predictions is that um humanmade art will get some kind of certification process. Sure. that there will be um technologies that come about maybe Jabril will make one where it can identify whether or not it's humanmade and it it will be a certificate or a badge given to products that are used maybe on the back end. So, for example, uploading a a game to

Steam, it will go through that certification process and it might take like an hour or two and then you'll get the badge and that will that will provide value to a product. So, that's my prediction that consumers will actually care if it's certified humanmade. I think that's an interesting prediction. I can see it happening for sure. Uh the there's a lot of question on how the verification process is going to work, but I can see it at least being attempted. Uh but at the same time, yeah, you know, I don't know, man. I don't know. As I think deeper, there there has to be an issue, right? like Steam is not just going to pay all these salaries to develop this without there being an issue to solve and without like an outcry of like hey we need to know the distinction between AI versus human games because I guess we

can't tell. Yeah. Um and again back to my point I I don't think we'll ever get to that point. Um but real quick quick aside if you don't mind. Yeah. My friend made this uh this piece. Can you check out in Discord? Yeah. And tell me what you think about the art piece. Mhm. That's amazing. I I do. I've seen this though. Dang it. Yeah. And I will say I'm I am not lying to you. I I'm going to tell you right now. I know this is AI. I can smell it. Okay. I'm telling you. You can tell. Am I wrong? It is AI. I was trying to keep you. Yeah. Um, I've seen this one and it's beautiful. I actually saw it at the at MidJourney. I know it's Midjourney. Um, I saw it on MidJourney's website. So maybe that's because that's the reason I know because I saw it at the top of Midjourney. Fair enough. Um, but the reason why I bring this up is because like this is kind of like the

task that we have. Like you say that you can tell it's AI. I don't know anything about art, so I wouldn't be someone to go to to ask these qualifications. Yeah. But how many people have that skill that you have and wouldn't just verify it as human, you know? Well, that that's the thing, man. It's like that on one side it's people on one side there's people who can smell it and they can just know they it's like uh I don't know for I always smell I always smell my sandwich meat before I make a sandwich. Same. I'm paranoid about it. And it's you can just you go it's that little funk and it's the same is true with AI. It's like you just it a little bit of funk there. Now, I don't know if they'll be able to get rid of it. Um whether MidJourney will, you know, adapt the model to get rid of that funk, whatever that funk is. I don't even know what it is. But that's one side of the equation. On the other side of the equation, it's even if you can't tell, let's pretend people can't tell that there's AI, that it was made

with AI. What if you told them as a marketing tool that little green badge that says everything in this is fine-tuned and perfectly made by human beings. There's no LLMs, there's no stable diffusion, there's no clawed, there's no nothing. My prediction is is that that will be a marketing tool and it will provide value and add value to a product. I agree. I I completely agree. Yeah, I completely agree. I just think that the the issue is going to come up eventually where someone is able to uh defraud the process and all verification badges are no longer valid. You know, man, it's going to become a game to some people like they're like, "Yeah, I'm I'm going to figure out how to pass the verification." Probably a YouTube video for you. Yeah. You and I know you'll figure it out. That's That's why I think it'll be a huge market because it'll be a Here's

what I think. I think there's going to be an eternal golden goose egg or gold mine for whoever can figure out how to build a software company that is constantly identifying AI created content. Maybe it's a legal, maybe it's for legal purposes, too. But what whoever can come up with that software that can identify what's AI created, they're going to have a a sustainable business model for the rest of eternity because they're going to they're going to, like you said, they're always going to have to be keeping up with the people who can hack the system, right? So, I feel like it's just going to be this eternal race of trying to identify or maybe you think we're eventually just not going to care. I I do believe that the path that's the path we're wrong. I I look I I view the

AI generated stuff as Photoshop. You know, once upon a time Photoshop hits your timeline, you're like, "Oh, that's clearly Photoshop." But nowadays, like you don't even question it. You know, I I do believe we're on that path. Um, I'm not totally convinced that we'll be able to get onetoone with like how like some top human artists can do. I'm not totally on that. There's there'll always be artifacts. It's just part of the style, the aesthetic. Okay, man. That's controversial. I like it. I mean, I think I think you're I think you might be right. And that was actually Well, go ahead with Do you have a Do you have a prediction? I do. I do. Now, listen. Okay. I'm very hot and cold when it comes to AI, okay? Like, I love it one day, I hate it the next. Okay? Today, I'm on a I hate it day. Okay? Listen, listen. Here's my prediction. Okay? We're finally going to get all of the investors out of the AI space and AI generated, what they call it, Gen AI, is going to completely just just dry up and we're going to finally

get back to the real AI stuff, which is AI classification. That's my prediction. Okay, that is my prediction. One of my biggest uh issues that I have with AI, the current AI culture, is that it's all focused on generative AI, like all of it. Like generating images, generating video, text, the text stuff is cool. I like tech stuff. Uh generating uh um what you have it, right? Music, sounds, but that stuff doesn't really have a lot of use case. Like if you if you really think about it, like you you said the images for concepting or the textures in a game, you know, but even then they're funky, dude. They they're funky. They're weird. I know LLMs are useful for code. I think LM are really good technology. I I don't ever bash LLMs. I love LM. Okay. But when it comes to asset generation, it's like it's just we it's just funky, you know? It's like you can't really use it in production unless like you really don't care. I won't never I won't get on that

side tangent. Do never mind. You can't. You can. I can cut it out if you want. Uh, okay. You do editorial juice. But yes, as I'm saying that, I'm thinking back to the Poppy Playtime and there is some kid in his bedroom right now generating a bunch of funky textures, a bunch of funky sprites, a bunch of funky everything and making one of the greatest games of all times right now. That's right. And it's going to completely change the the culture of like this this AI style. It's the next Minecraft but like AI, you know? It's it's just different. It's and we can't even predict it, you know. Yeah. And that's the part that's the the sad part about being old is that like you and I we got stuff to lose, you know? Like we can't be out here experimenting with this stuff. But there's some kid in his bedroom in his basement that has nothing to lose. Like everything to gain. Why not? Why wouldn't he use these free tools, you know? That's right. And

that's what somebody said that during a live stream today. I was streaming and they said, "Thomas, all you do is use old game design techniques for your games." Like I I like uh Twisted Tower. It's got Legend of Zelda: Ocarine of Time uh temple game mechanics. Sure. Um levers and switches and keys and maps and it's like old school. And I said that's what I love. And also it's to your point, I'm not going to risk I mean I'm I'm not going to make any risks right now. I'm going to make a solid game that I know is going to work. Yeah. But like you said, there's some I mean I guess Gen Alpha is the the next generation. You know, you got Gen Z and Gen Alpha. They're gonna come up with some crazy idea. Yeah. Using generative AI and and the image and the asset generation stuff. And who knows, maybe it will be valuable, you know. Yeah. Yeah. I mean, that's the only way I can really see the current Gen AI stuff like being adapted is like some cultural there's some cultural shift toward towards it. Uh, but interesting enough

about the the person who left that comment, like I'd be so interested to like drag them into a room and just like pick their brain like what do you mean by this? Like what what how do you define, you know, like I think I think what they and I I engaged with them a little bit on the stream and I I think essentially there was something like they wish that I would be more innovative. Um, but I'm not innovative. What does that even innovative? What does that even mean? Well, it's kind of like it's kind of like stable diffusion. It's like Thomas is really good at stable diffusion. He he takes all this this stuff that he's experienced from his childhood and all the games that he's inspired by and he puts them into a game. Um, which I do, you know, I don't come up with anything like Thomas Maller. He's the he's the creator of uh Celeste, I'm not Celeste, uh, Ori in the Blind Forest, and he was talking to me on the podcast. Um, and he's like all about advancing genres forward. I mean, he he he wants to take

like uh RPGs forward. He he wants to take platforming games forward to the next what's the next evolution, right? Sure. For me, I'm like I don't want to I don't want to evolve anything. I want I want to put my style and my aesthetic when it comes to art forms and put it into a genre that works. Yeah. You know what I mean? Yeah. Yeah. I I understand that. I don't think there's anything wrong with that, you know? Yeah. Yeah. Like the the truth of the matter is Damn. I'm sorry. Sometimes I'm about to say something that involves a stat and then I realize I read an article that like throws that stat off. I I'm sorry. My brain works in weird ways. What I was going to say is is that there are 8 trillion people. Trillion. 8 billion. Let's take a look. How about I Google it? Wait, is ition people or trillion? I think it has to be billion. It's got to be billion, right? There's no way there's a trillion people. Yes. Okay. 8 billion. 8 billion 8 billion people on the planet and like if you put in the work you find your people no matter what

you're making you know. Yep. That's what I That's right. That's right. So the prediction was what was that? that all of the investors are going to leave the AI space and all of the money for Gen AI is going to dry up and we're get back to the real AI application which is AI classification. Tell me more. Sure. So, uh where do I begin? So, I think one of the greatest components of artificial intelligence is being able to classify things, right? Um some of the greatest innovations in the last few years have been using classification AI. Um this is a controversial somehow but like the vaccines was was founded in a in a very fast period because they used AI to classify different um proteins and stuff like that. Uh there was what was it? Snake venom recently was you can

only if you got bit by a snake the only way that you can cure it is by using antivenenom which was extracted from snake venom. However, using artificial intelligence they were able to find like an enzyme or something like that where they can artificially produce antivenenom now. So it's like completely revolutionized. Is this the same thing as I heard about an app that could potentially teach you how to talk to your dog? And the way that it's done is by it has a huge bank of sounds that it can interpret for each species or each breed of dog and then it can go through and it can you you record the dog. This is all theoretical. I don't think it exists. But because of the ability to quickly pull and classify the different sounds from a dog, it can it can translate what the dog is trying to say based on this huge library of labels for the dog's um different sounds. Is that kind of what

you're talking about? I mean, ideally that'd be a use case for it. I'm a little skeptical if that will work, but ideally I don't know if a dog is actually saying anything. Right. Right. Right. In the sense that we can understand, but uh Yes. Yes. But yeah, I I think there's a lot of like missed opportunity of innovation with using AI to classify things that are just like beyond our capabilities, our human capabilities to classify. Like we're talking about taking something with like a thousand dimensions of data and then figuring out like what if if in this uh uh node where the the venom amount is 30%. But in this node the saliva amount is 50%. And a thousand different per uh dimensions. What what does this mean? Right. Yes. AI can classify that and help scientists like do really insane stuff. Side note here, dude. What I struggle with is is what did what do we mean by AI? Yeah. We act as if it's like an alien that came

down to Earth in 2017 or something. And and it's just it is AI. It is AI. And I'm like, well, how long have is LLM the same as stable diffusion? Do they use the same technology? They all AI. So, so LLM is a branch term. Uh, it means large language models in which it's just a like a bunch of text that's trained to predict like the next token is what they call it's not next word, it's not next character. Sometimes they're bunchkens, but stable diffusion is a model itself. I think that might be the name of the company as well. Yeah. Um, and I believe that's a totally different model. So, they're not they're not they're they're called AI just in general like by news outlets, but they're completely different things. It's it's a mess. Honestly, it's a mess. Uh I was at a conference that Amazon was putting on about robots and stuff and I was talking to some PhD AI students and they were trying to convince me that the AI in Pong in Pac-Man is not AI. Of

course it is. I was like, "Wait, wait. You guys are PhD AI PhD students and you're saying this is not AI? It's a mess. It's a mess." So, this is what I'm trying to figure out. This is a sincere question. Maybe you know the answer. Why did this all of a sudden explode? Yeah. Uh, Chad GBT was one of the one of the pioneers for this explosion. Um, well, it depends how far you want to go back, but specifically Chad GBT was in development for a couple years. It was it was actually open source. It was already on the market. There was GPT2 I think was the first one and then GPT3 was released and then they figured out a process where you can and the way GBT2 and 3 worked, it was nothing like chat GBT. It literally would predict the next word and it would sometimes give you uh um just random garble. I don't know if you ever watched some of the videos like back in 2000, I'm sorry, in uh 2020 2020 2019 of videos of like I used AI to

generate a script. So here's how the matrix would work if Superman was in it. Like that was using the GPT2 and 3. Yeah, but then they figure out this process where you can essentially like get humans in a loop and then you can uh get the GPT2 to generate like a corpus, I'm sorry, generate text and then you tell it if like the text was good or not responding to your question, right? And that just like magically worked and it boomed it bmed the industry from that point. I what I'm trying to figure out is why is it that okay, we've got three categories of AI that we've just discussed, right? We've got LLMs, we've got stable diffusion or whatever is being used. It's stable diffusion for for art, right? Then we've got what you just mentioned, which is categorization. What was it called? Uh classification. Is that linked to these two or is it its own new technology? Uh it's I I'd say it's its own thing. The the only real difference is all of them have inputs like all of them. All

of them have the black box, the middle network. It's the output that's different, right? So like when it comes to image generation, the output is often like the size of an image, whatever image you're outputting, right? But if it's a classification, you can have a thousand inputs, the black box, and literally just one output that's like yes or no. You know, do do I have answer for example, here's a thousand inputs, blackbox, yes or no, right? Okay. Um, unfortunately, we have not invested enough money, I believe, into using classification AIs, which can really like improve so many lives. Like here, here's an example, right? You have a car, right? I'm assuming. Please don't answer that. I do. I do. You have a car. Yeah. Yeah. And your your car is starting to have problems, right? You don't really know what's going on, right? Like you don't want to take to the mechanic because maybe you might get uh swindled or something like that. Yeah. Yeah. And that process for a lot of people is very grueling. If we were to use AI, classification AI correctly, someone can develop an algorithm where

all the different components of your car is fed into an algorithm and then outputs what's wrong with your car and it could classify at a high probability of what's going on. So you don't have to have a mechanic go in and say, "Okay, it looks like this is doing that and this is doing that." And then they plug it in and they write it down into the algorithm. You're saying instead it can just read the data through automatically automatic computer like have the the model built into their car and it just does probabilistic math. Correct. To figure it out quickly. Correct. What makes this different? What makes this an AI as opposed to just a really fancy, you know, library of data that we could have done in the early 2000s. Sure. Sure. So once again, it's a mess, right? But AI specifically is when you're utilizing neuronet networks, right? So the black box in the middle, it's a neuronet network, which I'm not going to go into too much detail, but essentially it's like a very complex uh uh like linear

equation that you essentially give data input, you feed it into there, and then there's output. Oftentimes, you need labels. So this will be up to Honda or Ford or whoever to train these models with the correct data. As the input goes out, the blackbox itself is like slowly corrected to be correct on the input. So when you buy a Ford or a Honda and you're out and your oil keeps leaking or something like that, your check engine light comes on, you can just say like, "Hey, based on this model they train with over a million different customers data, what's going on with my car?" and they can take all the inputs that they have and just say like, "Oh, well, it looks like that you might need to replace your your valve cover or something like that, right?" And then you can go straight to the mechanic, say like, "Hey, look, Ford gave you this model. I don't want to hear your This is what's happening. Do it." You know? Gotcha. Wow. It would improve so many people. And that's just one example. That's just one example. I'm thinking about the medical field like Yes. No more getting screwed, you know? No more getting crazy or crazy diagnosis where it's like

it's much easier to just you can just cross check and go like cuz like I've been I'm fine with doctors but it's like I've been to three different doctors and they say completely different things and so you're just like what are you supposed to do with this? Okay that's amazing. I love that. But do you see my frustration? Like there there's so much money that goes into this gen AI stuff. They're like does it really have any use case? Like why are we investing? Like I love chat GBT. I love OpenAI, but they have a $300 billion valuation and all they've really given us is generative like image and like and I will say, you know, I I will say like a lot of the cool stuff I was like really really impressed with their their chat bot or something. It was like you you can it's an app where you can just talk to it and it it talks back to you. It's really cool. Yeah. But I don't use it. Sure. because there's there's some there's some sort of uh issue with its functionality and it not being productive. Um so I agree with you that there's something missing there and that they're the investment might need to be

redirected. People are scared I people are scared that it's going to cause industries to collapse um and that we're going to lose jobs because of it. My theory is that it will instead of it being like a pie chart where it's like 50% of jobs in America are AI and 50% are human. My theory is that it doesn't fill up like a typical pie chart. Instead the pie expands. So instead we get just more jobs and more economical opportunity. So I don't think people need to be as scared and believe me I've been scared. I don't think people need to be scared that AI is going to take jobs. I think it's going to make everything way more productive. It's going to increase the value of currency. It's going to make uh energy cheaper and therefore we don't have to work as much because of AI, right? Therefore, I I just feel like if if every everything is more productive because of AI, that means that we

probably won't have to worry about working so much. you know, everybody spends their time and myself included, um, plugging in data and scrubbing through emails and it's like, well, I I think AI is probably going to take those jobs and a lot more fun jobs will be created, more creative jobs. Yeah. So, that's my prediction. What do you think about that? I think it's interesting. Uh, I think once again, I'm not really convinced. So there there's a there's we need to make the distinction of AI versus automation, right? If we're talking purely AI, I'm not convinced. If we're talking automation, I think I'm more a little more convinced um that it could replace jobs. Uh I have played with a lot of like the LLMs and and some of the latest greatest state-of-the-art models and they are not as competent as you would might think. I think on the surface level they're really good at doing surface level stuff. when you task them to do like

more deeper tasks like they they're not competent at all right and the question is like okay cool we're on this acceleration path AI is getting better right but at what point will it get to where it will be constant enough to do like more serious like human level tasks right and then no you project it out 10 years whatever the case right then you go okay cool what is that going to cost right because open AI is already fundraising what is did 30 billion I think the other day with a $300 billion evaluation and it's like now we're starting to get into the territory of like fundraising you're going to need nations to come in right and it's like then that that's when it starts to like break for me it's like the funding to get to that that point in the acceleration curve it starts to become like financially infeasible and then you're talking about nations need to collaborate to do one AI that that doesn't seem likely to it sounds like you it sounds like you feel

I know you mentioned you're kind of up and down about AI um but it sounds like and that you're feeling negative today. It it sounds like you're you're almost you feel like it's it was all a charade kind of like the emperor has no clothes. Yeah. And we're all realizing now that it's not as exciting as we thought. Yeah. Kind of. I I think Yeah. Yeah. I think I do believe I think the only caveat I'll say is that it's hard to see things that you don't know, right? Like it's very possible in five years from now quantum computing drops and it completely changes everything. Yep. They've been saying it's going to drop for the last decade. Uh you know, it just doesn't seem to ever happen. Um maybe I'm wrong. I have a swindle story with the computers. Yes, please. I'll I'll make this short. I don't want to talk forever, but uh I did a sponsorship. I'm not going to name who, but they had said that they had a quantum computer they wanted me to make

a video about. I was like, "Oh, yeah. This sounds exciting. Cool. Cool." And I was I had no idea what it was. So, I asked them like, "What's some of the stuff I can do with it?" Like, "Oh, you can make, you know, algorithms with it. You can use our actual quant computer, process something, and then bring it back to the computer and use it, and it'll give you all these speed ups, and it's great." And I was, "Okay, cool. Here's what I'm thinking about doing. Can I do this?" Like, "Yeah, absolutely." Yeah. Yeah. And so they give me this program where I can like train to learn how to use the their con's computer when the time comes. And so I'm sitting there trying to figure this thing out. It makes no sense to me. And then it it hits me like I'm like 3 days in, 16 hours a day trying to learn this thing so I can make a cool video. And then it hits me and I realized the capacity in which they gave me was literally like equivalent to like eight bytes or something like that. And it's like there's no possible way that I can do this project pitched to you to make this thing. And then the time came for me to use it. And it turns out they put me on a simulator for the I was like, dude, I feel like I'm being swindled.

What is going on? Did they ever come out with the product? Uh, no. No, not to this to this day. It's still not out. I can tell you who it is after. I don't want to blast. I will say this. I've haven't heard someone as I wouldn't call you cynical, but you sound cynical. I I don't know what the right word is, but I haven't heard someone as cynical as you about AI, and it's making me wonder if we're being tricked. I don't know. I will say this. I will say this. I started working with AI in 2017. This is before all the boom and stuff, right? The communities back then were very fun. They were very hopeful. Uh we didn't have anything like LLM, none of that stuff, right? There was some progress. It was all still very cringe, but it was fun, right? And there was this big big big debate back in the days about like open source versus uh privatization of of AI. And and the name open AI came with this promise like, hey, we're getting

all this money. We're getting funded by Elon Musk and blah blah blah and we're going to make these models, right? And they did. And they released GPT, I think one, two, and three. I think they did. Uh open source models you can download. You can play with them. But then came Chat GPT. and then they privatized at some point. They're now called Open AI. They shouldn't be called Open AI. Um, but the the point that I'm getting at here is that at some point and they were also doing other stuff as well. They weren't just doing text generation, image generation. They're doing other cool experiments as well. The point I'm getting at here is that my frustration comes from the gen generative AI stuff is such a small slice of what makes this technology cool. That's what I was telling you about the classification stuff. That stuff literally can change people's lives. Literally change lives. But we choose to invest all this money in the tech the the generative tech stuff. And I think that it's largely because it's what people know and it's just easy to fund raise from it, you

know. Would you uh Jabril, would you would you compare this to something like cryptocurrency? crypto. I don't I don't have a any any issues with crypto other than the fact that it's not currency, right? I I don't think I would only because like there there are use cases for and again back to our issue of like AI is a mess, right? Like right like there's so many different like subsections of AI and uh so I would definitely wouldn't say the entire umbrella is is like cryptocurrency. I think there are like certain segments that are like that. Uh for example, I do not understand video generation. I don't understand why people want it. I don't understand what people plan to do when it drops. I don't understand. It's great for content creation. You can get some terrifying videos out of that. Um but that's about all I've seen it for. But even that, like what percentage use case is that? Like we're investing hundreds of millions of dollars for three people in

their bedroom to make some memes. I think I here's my thought. I think that if society I think eventually society will stop caring whether it's AI or not. Um use cases are dependent on whether people actually care. So for example, if I think people still care if AI is going to be used in a Marvel movie. I think people will care and and people will view a Marvel movie as being not as good and a Marvel movie sucks and it trends on Twitter as sucking because there was one character who was AI generated. Go ahead. What's up? Can I cut in here? Yes. Fun fact, they already use AI in Marvel movies. I know. And it's causing people to get pissed. No, no, no, no, no, no, wait, wait, wait. I'm not talking about the Secret Invasion intro thing. I'm talking about to to recreate Thanos's face. They heavily used AI to make that happen. Okay. And is it the same kind? This is

the problem is like is it the same kind of AI as midjourney, right? Or video generation. Yeah. That's right. And so that's why I think once people get so confused and it's been around for so long, they're just going to stop caring. And then suddenly suddenly Marvel is going to use it all the time. and no one's going to care. So that's where I see the use case is when people stop caring, then the floodgates open and then they can use uh video generation to their heart's content without any any worry about what the backlash is going to be. That's my theory. I have to call you out on this though. Okay. It sounds a bit contradictory. I know it does. Does it come before or after the verification badge process? Right. And so here's what I think. You ready? Yeah. I think just like there there's people like me who hate Marvel movies. I hate Marvel movies. I don't care. I don't What about the Avengers? Do Do you like the Avengers? I don't care about any of that

stuff. I don't like I know. I know. I know. Look, post Avengers, I'm with you. Okay. So, I don't think I don't care. I don't like pop music. I don't like sugary anything. Okay. I don't like I don't even like Coca-Cola. I like things to have depth and and and have a heart. And so, and I'm one of those pretentious probably I I only watch A24 films, right? I'm kind of one of those one of those kind of guys. Good films. Good films. Right. And so what I'm saying is is that there's going to be just like in in film, you've got plenty of money on the Vanguard side and then you've got plenty of money on the, you know, action flick or pop music or it's just made. It's clearly supposed to be a blockbuster and it's made for money. You've got an audience for both. It relates to what I said in my second point, which is it's not that the pie gets filled up with AI or not. It's that the pie expands. You've got a huge

market for human generated content and people only love that stuff. And then you've got a huge market for AI stuff and nobody cares over here. So that's kind of where I think all of my predictions, they fit together. Uh I don't think it's a contradiction. I think that you're going to have you're going to have tons tons of entertainment content available at your fingertips that's AI generated and no one's going to care because that audience never cared to begin with. And then you've got all this other stuff because they're hanging out watching Marvel movies. And then you've got then you've got all this other kind of content that is for people who actually do care and that's where the verification comes in. That makes sense. That makes sense. So we'll see, man. I have no idea where my head is. Yeah. I'm so curious. Uh name like what what's a film that you like? Oh man. Uh I really really enjoyed um The Lighthouse. The

Lighthouse from Robert Edgars. And I actually I watched um Nasaratu. He made Naseratu and I didn't like that at all. So it's not like I'm just sitting here if it's Vanguard or if it's if it's got this artsy vibe, I'll like it. Um, I do like movies that that it you can almost sense that it's got the fingerprint of the the aur or the creator. I really value those a lot. Um, I love a ton of might Shyamalan films because even though they might suck, they have that thumbrint. Um, you can tell that it's it's made by a human. Um, it's got a human feeling. It's It's very um Well, it's it feels like it came from a singular mind. I really appreciate that. What about you? What? Go ahead real quick. This is hypothetical. Okay. I want you to to work with me here. Okay. I want you to imagine, okay, in the lighthouse, one of your favorite scenes heavily used AI and you just learned

about it right now. How does that change your relationship with that film? It changes it a lot. It does. And the reason why is because I I think I I did a video about this. I don't think human beings are utilitarian. I think we are social. Everything we do, everything we do is driven by this social desire. And the social desire is to be heard and seen and felt and understood. To be told that we're not crazy. to be told that that we mean something and what we think matters. Some of us are crazy. That's right. So, so when it comes to when it comes to consuming content, I don't consume a a Netflix show or like for example Severance, I don't watch Severance because I'm entertained. I think I'm entertained because I feel like I'm heard. Like when I when I see a

character and and I relate to that character and I want that character to character to succeed, it's because I kind of identify with that character and I I I see myself in that character. The reason I say all this is because the moment the moment that a viewer finds out that that character wasn't created by a human, it's something just is offputting for me. It immediately turns me off and I'm like, well, I can't be understood by nothing. I can only be understood by something that is a pure. Does that kind of make sense? I'm not sure if that's too abstract. Yeah, I mean um or annoying. I get the general gist of what you're what you're getting at. I get the general gist. Um that's fascinating. So, like obviously like we had different experiences when it comes to the the mess that is AI, right? Yes. And often times when I hear AI, I hear nothing but like the

optimization process of data, right? That's where my mind goes when I hear that's where your head goes. Gotcha. Right. And and I understand where you're coming from and I like have no no like counterargument to it whatsoever. I just it's a yes and yeah. I wonder how much of art is the optimization of data in itself, right? Yeah. Because if you would define like the evolution of art as the optimization of data, that would make AI that is a mess. I wouldn't necessarily say art, but a tool to help art. I see what you're saying. Oh man. We have this deep philosophical conversation on a Monday. Who are we? Welcome to the Thomas Brush

podcast. You know, I I I think that um it's a good point. I think it's a really really good point. Um there is nothing more exciting than having a real human being. This is why I think I this is why I believe in a higher power because there's something about human beings that are divine. And when I consume art, it's like a hug. It's like a cozy little hey, something with a divine consciousness is interacting with you and you merge. When I watch a great film, I merge. When I hear great music, I merge with some other consciousness. Yeah. And this is why I think that the AI art, you know, that stench I was talking about, the smell, that's what I think it's I think it's something like it's

almost like spiritual. You just see it and you're like, nope, I'm not getting that divine warmth. I'm not getting it anymore. Yeah. Um with this this AI generated piece of art. I don't know. I I just feel like there's something about it that doesn't feel human. And at the end of the day, everything we strive for as human beings is not comfort. It's not power. I think I think human beings strive to feel valued and connected with their peers. Um I'm not striving to to gain power over every dog on the planet. I'm looking at my dog right now. Like my deepest if I was if I had an evil desire, my evil desire is to have dominion over humans. Not dogs. Why humans? Why not dogs? Wait, hold on. What are we saying? I'm saying dominate dogs dominate humans. Wait. I'm saying I'm saying that human beings human beings crave to have either a

relationship or dominion over other humans. We care about our peers. We care about humans. Yeah. We don't crave dominion over like other things. It's typically I want dominion like evil. Evil is about dominion over human beings. And so that's why I think that art has to be human for it to be art because that's what we care about. We care about it being from other human beings. That's what makes it valuable. I could I could continue to press you on this all day. This is fascinating. I don't know. I don't know if there's a solution. We're going to have to just see and watch it watch it play out. You know what I mean? We don't have an option. Like it's a we can do man. Dude, that was that was a really good conversation. Yeah, I enjoyed it. Yeah, dude. Totally. Thanks for coming on, Jabril. That was fascinating. Really really fun. You want a link in the description for anything? Yeah. Yeah, please. Uh go download Ultra Bouters. Uh it's the one of the best

turnbased fighting games that you've never played yet. Go check it out. Awesome. We'll do that, man. Well, thanks for hanging out. Really, really fun. Thanks for having me, man. I I really enjoyed it. Also, hey, don't forget to check out Bezi to simplify and accelerate your development process. You can sign up for the wait list to be one of the first to get access by simply clicking the link in the description.