The Shintaro Higashi Show - Judo and Innovation

Episode Date: April 29, 2024

As some of you may know, Peter has been a PhD student since he quit his job as a software engineer back in 2020. During this time, he’s gained a lot of insight in scientific research and innovation, and how they are related to Judo. In this episode, Peter and Shintaro discuss how Peter has utilized his learnings from AI research into Judo, and vice versa. Join our Discord server and start chatting with us and other grapplers by supporting us on Patreon: ⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.patreon.com/shintaro_higashi_show⁠⁠⁠⁠⁠⁠⁠⁠⁠. Any amount helps!

Hello everyone, welcome back to the Shintaro Higashi Show with Peter Yu. Today is going to be the Peter Yu Show with Shintaro Higashi. We're going to talk a lot about systems and ideas and innovation, where Peter's an expert. We have to thank our sponsors first, Jason, Levon, and David. Thank you guys, without you guys we can't do it. Judo TV, discount code SHINTARO. I'm going to Georgia for the Grand Slam to commentate. So please check it out. Use my code. Also, ShintaroHigashiBrand.com.

Sponsoring my own thing. He'll be the new face of Judo TV, guys. Shintaro will be the new face. That's the plan, right? Yeah. That's so exciting. That's how it goes. So today, I think we wanted to talk about how judo like some of the things

you know people talk about how people you learn so much from judo and then how those things translate into their real lives and yeah i'll talk about how i took the learnings i got from judo into my own work which is currently research ai in ai i'm trying to get a phd just been a crazy journey so yeah that's uh that's the topic today i will it's a kind of out there thing we're not really talking about judo exclusively hopefully are we talking about like parallels and thinking or are you thinking about like judo in your work environment where someone doesn't fund you and you slam them away? Oh, yeah.

We can talk about both. We can talk about both. When my undergrad assistant doesn't do the work, I'll give a stern talking to. Do you have an undergraduate assistant? I used to have a couple, but I stopped taking them on because i think the biggest problem is that they are so busy with their classes and that should be their priority unless they want to

go to on a research thing so i didn't get much a lot out of it out of them so i kind of stopped taking them on that's like oh that's too bad i know yeah but anyway so i think so i've been talking thinking a lot about how to conduct how to be a good researcher like and then what does a researcher do a scientist it's basically solve identifying it's not just about solving problems you have to identify the right problem to solve and that's very very hard that's i think even harder than actually solving the problem and and i guess that's something relatable to judo martial arts too yeah so i can't finish uchimata and then it's like hey can you teach me uchimata right right what's the problem yeah exactly and i think it's not about just like oh

for example we talk about this a lot right like when you watch a youtube video and then you're like oh my god the thing that shintaro does with the georgian grip is so cool i'm gonna learn it that's not the right problem to solve necessarily you know that might not fit your system you're maybe creating a problem that doesn't really exist yeah you know what I mean? Yeah. So how do you find the right problem to solve? So I think the best way is you have to know the basic techniques, the fundamentals.

So you can't just go out and like, say you can't even do a Seoi Nage and you can't just go out and try to learn like the variations on the Seoi Nage, Korean Seoi Nage, you know, all go out and try to learn like the variations on the seo inage korean seo inage you know all this yeah reverse seo inage or like the way you enter different things drop seo inage without actually knowing the intricacy of the actual the just the textbooks throughout and i i realized that i feel that a lot when i do research now you can't just go out and for example just read research papers that's just like just watching youtube videos wait but isn't that the fundamental to learn the actual text how do you where's the repetition in my field it's not just about reading research papers and also and just like learning the material you also have to know how to make them like how to implement the

research ideas like you have these theories right like oh this height uh you know the model architecture should be like this and then you have to use this type of data and then this type of loss function whatever i'm kind of throwing all the jargon i love it, but ultimately you have to, it's more of a, computer science is thought to be the show me field of, in science because you have to actually. It's like the judo or the martial arts world. Yeah, I kind of like that. It's like a show me field because, yeah, you can establish all these theories or whatever,

but ultimately you have to make it happen. You have to write the program and and then show run you know experiments to show that it actually works you know okay quick question so computer science 10 years ago five years ago one year ago now right like the trends change right yeah and there's some meta that happens in grappling right with georgian grip and this and that and what's popular now drop knees yeah right so like is there something similar in of course be in tech right so where was it 10 years ago five years ago one year ago now yeah so i've been in the field 10 years for about a little over than 10 years now because i graduated from college in 2013 and back then the big thing was the meta back then was big data and distributed systems so

that's when the internet was popping that's right girl that used to have a t-shirt say data is a new bacon yeah yeah exactly yeah so at that point people are just just gathering data because internet was really popping off i mean of course it was growing before but it was more like social media were getting social media companies were getting really popular so they were and the people are lay people were interacting with the internet more it wasn't just about these nerds going on this message board and talking about nerds nerd stuff like your grandmas were coming on facebook you know sure that's true it's like with judo and then now jujitsu then you got like yeah it wasn't just

all the olympians training in judo it's more like you know so 10 years that's why i actually my first job was a advertisement technology company which mean like it's a that was another meta like how do you monetize these free services you run ads and then there's a lot of things that happens in order to run ads like show your ads yeah so that was before app nexus that was app nexus that was that was okay app nexus was the company that, yeah. So what happened trend-wise after that? So that around 2012, so I guess that was like the peak of 2012, 2013 was the peak of big data distributed systems.

But then something crazy happened in AI. Basically, until then, from the 80s till then, AI was kind of a second citizen field. Because, again, it goes back to the idea of being a show-me field. AI researchers put out all these new systems, but just didn't work. They couldn't show people they worked. So funding dried up people stopped paying attention it's called the ai winter um but then in 2012 this guy actually the chief scientist at open ai ilia suskiver yeah and his uh lab mate published paper, but developed a new system that could recognize classify images like identify what's in the image really well. It was like

it was almost as if Their model was almost like I don't know. It's like I'm trying to think of a good analogy in judo like it's it kind of like think of a good analogy in judo like it's it kind of like oh no show hey kind of like they were just running through people that model beat the next best model by like 10 or something i can't i forget that yeah that's when people started paying attention back to ai so that's the question you know how like are you a human logging in pick all the squares with bicycles in it? Yeah, yeah. Does that have anything to do with training AI or what is that?

I think it kind of does. My understanding is that you're actually helping label the images a little bit. It's not, so there's a lot of different ways. You actually don't even, you know sometimes you don't even, it wouldn't even show you an image. You just have to click on it. So that works because it's differentiating you from a bot by just tracking the mouse movement.

Like the way you click the button. Because the human, we click the button yeah okay because the human the human we click the button differently so but then if the confidence level is low it starts showing you the image and my understanding is that sometimes they'll show you like images that are already labeled but sometimes they'll it'll show you images that are not labeled so you're almost like kind of helping yeah label the image like where then how will they know if you picked it right or not like pick the traffic lights i don't know the details but i think it's more about like you can show the same image

to multiple people right and then see what what the agreement is between them and then see what the agreement is between them. Oh. Something like that. So it is AI-related? Yeah, so then that data, that labeled image, can be fed back into the AI system. You know what would be funny one day, man? You could pick the one that's Uchimata versus Hanegoshi.

Oh, dude. You know what? I'll break the internet. That's actually a good idea. Yeah, like a judo throw classifier you show a video we're definitely not solving a problem that is actually a good problem

is it a good problem? let's stop wasting our time on Hanegoshi Uchimata no one gives a shit guys just train please but yeah so those are the meta changes so then AI once that system showed a big improvement,

people started saying, oh, wow, all the data we've been collecting for the past 10 years can now be processed through this new paradigm. It's an old paradigm that's been rediscovered and then shown to work again. So that's the meta change. All becomes new, new becomes old, right? Like even in judo, BJJ meta, right? Yeah, of course. Recycled.

So that's kind of the meta change for the past 10 years that I've been in the field. So what problem are you trying to solve now so i am solving my thesis will be around videos and large language models so i can go into more details but the basic idea is uh you know you know chat gpt right yeah yeah so I'm trying to make it so that ChatGPT, something like ChatGPT can also understand videos. Since when are you working on this? Because I thought you were doing AI chatbot stuff for a little while.

Yeah, so... Natural language. Yeah, so I used to... When I entered the lab, that's another meta change. When I started my PhD in 2020, the pure language models, just the AI system that only process language were thought to be unsolved. Still, there was a lot of challenges involved.

But then open AI basically took everyone by surprise in a way. So ChatGPT, their ideas weren't all new, but they took all the learnings from other researchers. A lot of it came from Google, actually. And then they just took it to the max, and it turns out that's actually the key. You just have to do it in a larger scale and then it just works really well to the point where a lot of the pure language tests I didn't basically made me think you know also

this disclaimer is that there are researchers out there who are all still working on the pure language test and I'm not trying to like, you know, diminish their research. But for my personal interest, I thought there was no more room for me in that pure language world. So I thought, you know what, maybe I need to incorporate other what they call modalities like uh language versus plus image language plus sound pivoted sometimes right when you're doing standard stand-up two-handed judo and you're going against the japanese or somebody right in their whole life you kind of have to find your way to force new wazza or something yeah i mean that's that's exactly what i did i i i thought i had to pivot because

i mean look at jackie netzka he's like uh has judo specific new wazza that works really well for him and he wins a majority of his matches that way you know yeah yep i mean yeah right it's it's he's not gonna lock up and stand up and do high uchimata and things like this right you know i mean i'm sure he can but that you gotta pivot and find your own niche and then you gotta stay with it to kind of win and get ahead of the pack yeah that's my niche like this uh new this idea this videos and language that's my special move i guess this is that why you started the d script situation is that how you got no i actually i got into that first actually uh doing doing the podcast, it kind of helped. It's kind of funny how life works, right?

Like, because I started dealing with video editing, I'm not doing anything crazy, but during the process, I had to learn about how videos are actually saved on your computer and how they are streamed and i kind of had to learn teach myself about these details and that's when i realized you know what there's a lot of challenge there's a lot of challenge that uh you know that not a lot of researchers are looking at you know because that's like black belt thinking right there because for me it's like compress make it big take it drag it put it in youtube i mean in the process like you know if you keep

doing i mean that's this is one other thing that i that's why i i think you need the fundamentals i'm not saying uh i wouldn't have yeah uh realized that this is a problem by just reading research papers. Like same as like this know-how, like this hands-on knowledge is, I think, very valuable. A lot of times people overlook that. It's not just about reading books. You got to actually go out and do it. It's not just about watching YouTube videos. That's what they say, right?

F around and find out. Yeah, it's actually kind of like that. Like you can't just mess around just reading things that other people have done. You actually have to do it. So that's why even in pure language tests that I didn't think, which I didn't think eventually concluded that it wasn't for me. If you actually sit down and do it, there are still a lot of challenges, you know, this know-how, but it just didn't interest me that much anymore. So when did you pivot to the video?

It's been about a year. Wow. I've been... I'm sure there's a lot of people working on this too, right? Yeah. Yeah. So, but then there's... And then but also my niche even with videos that's like the big umbrella but even within videos I'm like going more into like a specific niche that tell me tell me yes are you

allowed to say it yeah yeah of course I mean I I have a paper out so it's um where can we find this paper? How come you're not sending me this paper? You don't think I can read? Well, it's... Of course, I don't think... It's pretty niche, so I don't know if you will find it interesting. I'll send it to you.

I can read the abstract, right? I mean, it says it in English. Yeah, I'll send it to you. Yeah. So, my thinking is that... So, I was involved in a project that dealt with the HoloLens. You know the HoloLens? No.

You know Apple Vision Pro, right? Yeah, of course. It's awesome. Microsoft had a similar thing way before. That's the problem with Microsoft. Apple is amazing at it. They wait until other people do it and then make it better but anyway that's a side but I had I had a project that I had to work with video oh this headset wearable that will transmit videos back to the

computer and I had to do something with that video so now that's a different challenge now because if you go back to how you look at how you use chat gpt it's more uh turn based right it's like you type in a message and it responds go back and forth try to take turns yeah and some people are very very good at this and it's a field called prompt engineering yeah that's right this guy he's uh he's he's getting his hands dirty on this and then he's gonna figure something out all because of you though you've been inspiring me to take a look at these things and uh i'm not gonna lie it's helped me write a couple of captions for instagram i'm not gonna lie to you yeah i i use chat should be all the time to write my you know

descriptions for the podcast and all that yeah if you guys haven't seen like me fighting the junior world champ on my instagram there's like a little snippet that's written and i kind of just told how i was feeling to the chat gpt and i like make it more legit and caption friendly for instagram and it does it right instantly did it and i was like okay my work is done yeah how and now you can do other better things how great is that that's the you know anyway so in this so that's term based but in wearable everything gets transmitted in real time it's more interactive like videos there's a stream of videos coming there's stream of sound coming you know and my niche is that like instead of turn-based interactions how can we make these models work better in this more real-time streaming actions and that's my

eventual goal and there's some other really niche technical challenges that i need to solve and that's the paper i just uh wrote um it's under review you know so like hopefully it'll get in but yeah so can we vote on it? yes yes yes I know you can vote on it there's a

whole programmer social media kind of thing where you can host your code so you can give me a star if you were like hey anyway so that's that's my niche right now so but I could find I I was able to find this niche because of course I had to read a lot of papers just get the basic knowledge but up but the same time I had to really get my hands dirty and it

worked like I had to really do a lot of Randori basically yeah try to figure out what the problem is what is the actual problem that needs to be solved and then I could go back and read more like papers that are more appropriate like more related to the problem I'm trying to solve so it's like me going I I did I was like going with this for example like with Shintaro then man like he's so good with this position I just couldn't figure out how to break his high collar grip or something and then I like would go back to your channel and then try to look up specific videos that address that you should

try to get inspired like oh yeah what kind of things and then when you apply to to your specific situation you have to tweak it a little right and that's i think it's research yeah and you know you watch it you learn it you go do it and then you have other problems and then you go back to it maybe they answer that question maybe they did not maybe they answered it but you know it just didn't log properly but now you have a different perspective different idea you internalize that information now you have more questions and then yeah you watch something else maybe the ideas connect right yeah exactly it's got to be the same and it's exactly the same and i i it's kind of uh doing research uh helped me enjoy judo more because I've been seeing all these parallels and vice versa so I'm honestly I honestly think I've gotten better at judo

just by doing research yeah because I I see that oh this is how it works like I can do the same thing for judo and then back and forth so let's hear about your processes for the PHE going in reverse. What are some ideas about Judo theory that you have right now? So right now I've been briefly mentioned that I've been working on my Uchimata a lot. So I used to be a, my niche was Serenage and Tai Otoshi. You know, I'm not a tall guy. I'm not a strong guy. So my niche was that I need to rely on my technique and speed to get under people in the throw.

But I realized that as I got older, my speed has decreased. My knees are not as good. So I've been having knee pains a little bit. You rotate on the elbow and shoulder too sometimes, right? So I needed a better way to I didn't want to excise like Seoi Nage or Taiyo Toshi from my game that's my bread and butter like I but then I needed a way to really save it until the end like it's like my finisher, right? Yes. So I needed a way to stay on my feet in the beginning stages and then finish off with, you know, Tayo Toshi or Seoi Nage.

And then I realized that Uchimata actually is a very good throw for that. So I've been playing around with it a lot. So, of course so then i have a problem right now i identified a good problem i needed a i need to develop my uchimata yeah incorporating to my judo game in the earlier stages so i watched a lot of uh a lot of your videos obviously i went back and watched all your oshimata videos i went to watch a lot of videos of maruyama oh no and i wanted to watch the middleweights a lot i didn't really watch the heavyweight uh uchimata i'm gonna send you a link remind me the one of the paid video stuff the

uchimata instructional the ot uchimata. I put it on YouTube as unlisted. Yeah. I'm going to send you the link. I think I might have watched that. As you know, I have an editor. Oh, yeah. That's right. I watched it.

You pirating mother fucker. Yeah. You need to pay for this, but you know. Good. Purely special privileges. Yeah. So I basically went out

and read the prior works, you know, on... Prior works. Prior works work that's what they call and then and then i started trying to implement it of course like so i will try uh oh no set up and maruyama set up in randori you know i wouldn't i wouldn't even go for any like i had practices where i didn't try a single sewinage, for example. Nice, nice, nice. Because I would just force myself to play. And then, of course, I would... This is another thing that you talk about.

I wouldn't go against black belts a lot. I would go try to do it on white belts and green belts. The intermediate people. You'd load on every white belt in there? No, people so that they loaded every white button i was very gentle of course that was very gentle yeah then i'll go back to the drawing board then you get to like refine the problem better like what's the more niche thing like oh i wasn't pulling it out should i try the different entry different way to unbalance people it off should i try the different entry drug different way to unbalance people because a lot of that stuff is not explicitly taught right now a lot of it is feel for you too because only you know you could say like all right right first left they're putting this weight on this lead arm yeah

how do i lift it you could force it but sometimes like finding a way to like let them not think about that arm and do something else all right so. Then you can lift it when they're not thinking about it. That stuff is all. A lot of that stuff is feel. That actually happens in coding. And engineering and research too. You need this intuition. I don't really believe that some people.

Are born with it. I think. Maybe some people are more talented at identifying. Maybe they have more interest in it. So they just watch or read more. you know I think yes maybe some people are more talented at identifying maybe they have more interest in it so they just watch or read more but ultimately what you need to do is you just need to do it like you have to code you have to actually code or try to build something or in judo you have to do randori or more like if you can't do randori more uh you know the you're the drills you talk about more like with the resist people that give you more

resistance the role that might be analogous to like the dojo or the teacher yeah you're at michigan right now yeah yeah you mish you mish yeah in ann arbor yeah in Ann Arbor yeah and then I was a you mesh and then having the guy who's the professor that is in charge of your research what is that called advisor my advice is she's a she's a yeah she's a great advisor yeah and so I get to our bounce off ideas with her they so would she be the sensei equivalent so it depends there are a lot of different types of advisors i would say mine's more of a hand off advisor so she's she's tenured already she has already had some uh a lot of research experience but then the like i said the meta has changed so she'll be more like the

traditional japanese judo sensei kind of thing that's how i view it so she has a lot of knowledge herself but it's the meta has changed so she may she's not as familiar with my meta right the videos right the videos yeah but she can she helps me how to frame the paper how to frame the story how to set up the experiments so that you know to make my story more convincing to other people because research also is not about just like oh you know let's put out the discovery and idea you need to convince other scientists that's the peer review process you need to convince you need to be able to convince other scientists that your work is worth it yeah yeah same in judo right you gotta go out like a competition yeah bomb and white belts doesn't do anything yeah exactly you gotta prove your thing you prove your theories ideas

so you're but you seek this advisor out but it's not the same advisor right i thought uh so basically how i got in was i had two advisors i was co-advised by two but the the older gentleman and my current advisor yeah uh because i thought i could do something together with all three but ultimately all the older gentlemen uh retired and i just decided to just work with this uh i remember before you even went in you were emailing the guy hey i read your paper hey blah blah blah yeah i was like there's someone emailing me hey bro i watched your videos amazing i want to come train with you i'm like yeah yeah okay yeah i'm gonna show up yeah that's exactly it so for me because i didn't do a lot of research in undergraduate years i needed to kind of take a

more personalized approach in my graduate school admission process yeah so i reached out i read i picked out i went through the list of professors at university of michigan because i knew that i would only want to go to the university of Michigan because I was going to be in Michigan. Yeah. Yeah. And I picked out these two professors and I emailed them. I read the papers and I was like, hey, I think I want to do something along the line. Like kind of like, yeah, you know, like an Olympic hopeful would come and email and say, hey, can I be, can i live in your apartment and then train with you kind of

thing so when you finish this thing and then you become a phd and you have your niche eventually you to become a professor right you're some professors do research and some professors teach right so both industries no money so so right uh there So there's teaching positions, there are research positions, but even in research positions, you're kind of required to teach also. Okay, yeah. But some professors, they teach one class. Yeah, usually they don't teach that many. I want to talk about full professors, not like adjuncts or anything like that, right?

Professors on the track to get. Like a tenure track. Yeah. That's a whole different thing, right? I don't. So I'm not too familiar with it because my plan is not to stay in academia. I want to go back to the industry. So I'm not going to be a professor.

What would be the equivalent of that in judo? I would say opening up. I guess being a professor is more like, I don't know if the analogy would be correct, but maybe going to the Olympics would be like, or like trying to stay as a competitor maybe. Staying in academia. But opening up a dojo yourself might be like going into the industry maybe

it's more lucrative in the industry yeah i guess all right open up your own dojo will be the i don't know if it's like exactly the right uh analogy but something like that there are many paths i don't think staying in academia is my path. Why? Because there's no money in it? No, I mean, there's that too. But I just don't think my talent will be right for academia. I think I'm...

All right, so once you decide on this niche that you've already decided on, and do you see like a path? What is the path after that that like in an ideal world like you finish your phe in the next two years is that right next year hopefully in a year sure and then what do you do right after that i'll look for a job so my plan i mean i mean you can come to new york and teach time champs at my dojo yeah yeah that's my backup so So my plan is to, like, again, I want to build up my street creds, basically. I want to do more, like, I want to do more hands-on work. So I want to join a research lab in a bigger company in the industry so that I can, I already kind of, I can already tell people that I'm a good, I'm a decent engineer.

Cause I've, I've, uh, you know, seven year experience at pretty and with good work experience. So now I want to kind of say I did my PhD. I did some more further research in industry so that I can tell people that I can, I know how to do it. And then maybe join a smaller company as a higher up person. All right, so let's just say, right, someone listening who works in that field,

if you were to give your like two-minute resume, right? Oh, I worked at AppNexus, went to Princeton, got a look at Etsy, right? You were a programmer at Etsy. Yeah, I worked at another startup. Blah, blah, blah, PhD. If you were to give your like one-minute elevator pitch, maybe someone's listening. Probably not.

It's just a bunch of grapplers. You know what I mean? Isn't that like a thing? Like a lot of tech people are doing BJJ now or something? I don't know, man. Maybe. My elevator pitch would be that I am... Maybe Mark Zuckerberg is listening because he likes Jiu-Jitsu now.

Hey, he's got decent BJJ. All right. Sell yourself to Mark Zuckerberg is listening because he likes oh yeah he does hey he's got decent BJJ I sell yourself to Mark Zuckerberg alright uh Zuck

you know I tell him that you came even before that movie came out you were like drop the the

like you already said that been saying that so you know one thing I can really say about to distinguish myself is

that i i can uh i can wear the engineer hat as well as a researcher hat so i'm i am not only trained to find good problems but also i can solve them in a and in a more practical sense like because a lot of times in academia, when you identify a problem and then you solve it in a very haphazard way, you don't have to set up the right infrastructure or anything because the goal is to prove that your problem is good and can be solvable.

But I'm more of a generalist in the sense I can identify the right problem to solve. I can propose a good solution and I can even implement it. Basically, I can really show my work. That's awesome. I think that's what I'm trying to build myself up as. Yeah, I think everyone who does grappling needs to be able to kind of have that too right yeah they have to be some people just go in and then just practice the techniques that have been taught they never really make the

connections you know some people learn you know this guy's teaching this and that and they're just kind of copying it and then trying to do those things and making their body but some people are truly innovators right they just like kind of in the judo sense, yeah. Like my goal would be to, like not only can I do the techniques properly, like in the whole like pretty way, right? But also I can identify new problems and then actually implement it in randori.

Like I want to be a good, I want to stay as a judoka that can do randoi with anyone. Heavy, light, tall, short. I try not to shy away from... I don't ask people for their weights, for example. Oops, I almost fell. So that's kind of in the judo,

and then there's some parallel to my professional career i guess yeah i love that love that that's so interesting man yeah so i think that's what that's what i got out of judo yeah especially around innovation and yeah identify how to identify the right problem yeah how to then do the research to find a solution and actually implement it nice that's great stuff yeah this is this is fascinating dude yeah interesting yeah wait maybe you should uh did it did i inspire to become a more of a researcher i mean i've been doing a lot of prompt engineering stuff yeah i know so now you this is your you know now at the phase where you're like learning you're doing all the uchikomis, you know?

Yeah, yeah. And then maybe you're dabbling in their andorra a little, like trying to solve problems that you actually have. And now you can come up with new ones, like the new problem that no one solved before. Sure. I was asking it to do like woodblock prints and like Monet, Manet, Impressionist style things. Like experimented with that because now they could generate images yeah yeah you know yeah but like not hyper realistic images they can't do that at all and human body is very difficult but like you said landscapes things there's archives and archives of art that have already been created and you can say hey make this color different or make

this more right right whatever it is and make it futuristic throw this in there throw that it's like so cool what you could do. You could be kind of like an artist without actually having the skill of it, you know, just like prompting it. So I think that'll be

a new form of art too. A lot of people are already doing that kind of work. Yeah. Yeah. And that was kind of my idea with Higashi Brand.

I put like a Japanese woodblock print, Ukiyo-e style, like the big wave and like impressionist style with very vibrant colors and slap it on the back and you know

call it my own design even though AI came up with that right I mean it's like you're collaborating you're using a tool I don't think

it's like you know people drawing on the iPad instead of like a canvas that's true you know in a way

but I'm experimenting with it it's pretty cool yeah there are other things too other models out there too so you know you should definitely experiment that's like doing uchikomis right now right yeah i mean even like copywriting that i get solicitations for that all the time hey can i copyright for you hey can i do this hey this is an email thing that i've already copywritten for you like would you be interested in my services if i could give you three emails

for free can you give me a call if you didn't didn't know that. I get that all the time. Every day I get like two or three solicitations like that. It's usually some kid on the other side using chat GPT. You know, I, I'm almost certain that nowadays. Yeah. Yeah. They're not writing themselves. Like this kid, you know, didn't even go to college and he could write better than me.

Get out of here. Maybe, but like, so I've been kind of experimenting with that and uh you know sending out emails to you know if you guys haven't already go to shintarigashi.com and sign up for an email list you know right i'll send out emails very rarely but every now and then i'll be inspired and by me sending out emails i mean le Lewis. Who is an actual copywriter. Wonder if he uses ChatGPT. I don't know. I don't know. Maybe he does.

But he's a little bit older and he's been doing this for a long time, way before this. Right. ChatGPT stuff. So he probably actually writes it himself. He probably has more like the hands-on knowledge that ChatGPT can replicate. Yeah. And you know what?

I think once you start implementing something like that, you could scale and you have much bigger capacity to do more stuff in terms of like workflow becomes better, productivity becomes better. You could prompt with the proper context and the specificity to produce what you actually wanted to produce in a much more shorter amount of time.

Yeah. So I think that's kind of... Yeah, I mean, i don't know how it's become but like i there's a i'm prompt engineering is i think it's becoming its own field more like a skill yeah yeah uh i mean there's some people like hey dude prompt engineering is just proper grammar and proper speaking and proper requesting things and communicate right it's like just communicating yeah it is like we have we've invented this thing and now how do we get the right information out of it so it's like yeah there's no manual yet so you guys like we're all

figuring things out you're figuring it out man you know what maybe i'm like a yellow belt in that yeah i i am not good at prompt engineering myself. So, yeah, I'm right there with you. Some of these videos are ridiculous, though. They're like, here, let me teach you how to be a prompt engineer. And it's like, what you need? Context. Hey, I'm a judo YouTuber.

What do you need? A script for a video that may go viral. Specificity, like in a one to two minute thing. I didn't even know that. No shit. No shit. That's like i didn't even know that no shit you know no shit that's like who doesn't already do this to begin with you know but that kind of video is so if you go back to the uh to find the analogy in judo it's like boutique judo dojos versus like the scaled up dojos right yeah yeah that would be like the judo curriculum and then you just of course

it's not the best like it's kind of rigid and then maybe you know it's not as personalized as your instructions and your boutique gym but yeah you could that's a good idea i guess i'm gonna ask chat gbt to create a 10-week judo curriculum, a 5-week judo curriculum, a 20-week judo curriculum. Just sell it. Sell it. I mean, yeah. That's unethical. Well, you'll probably have to give a lot of information about it.

I don't know. I mean, it's going to be a conversation, right? For sure. That's so interesting. If you don't want to find Peter, where do they find you? You can find me on Instagram. I don't really post a lot on Instagram, but I do interact with you guys, you know, if you message me. Yeah, send him a message.

He will always respond. I'll put it down in the description below. Peter likes talking to strangers on the internet. It's by Gunwoo Peter. Gunwoo is my Korean name, and Peter is my middle name yeah k-e-u-n-w-o is it peter guno or kuno peter i forget i forget let's see it look it up yeah for the people man it's peter guno i think p-e-t-r peter and k-e-u-n-w-o-o yeah nice and you can reach you can find me there.

And of course, you know, I'm the producer of this podcast. So I like, you know, you can find me on Discord if you're a Patreon supporter. Yeah. And if you leave comments

on YouTube for podcast videos or like, I think Spotify, you can leave comments. I read them. Yeah. Nice.

Yeah. If you want to hear more about Peter please leave a comment let us know and then we'll do more of these this is so interesting I hope you guys liked it I loved it

so I'm sure everyone else would like it too I mean why wouldn't they you know yeah I hope so yeah it was judo related

it was judo related I really think there's a lot of parallel tons yeah so any pursuit of excellence yeah all related exactly hey I really think there's a lot of parallel. Tons. Yeah. So, that's right. Any pursuit of excellence.

Yeah. All related. You got to put the time in. Hey, that's a quote. That's a quote. Any pursuit of excellence. Did you just come up with that?

Yeah, it just came to me. Oh my God. It's genius stuff. I know. That was pretty, that was good. Any pursuit of excellence is the same,

you know? Very true. Very true. All right. Well, thank you everyone for listening please sponsor us and we'll see you next time yeah bye