Endgame with Gita Wirjawan - Kevin Chandra: Indonesians Cracking the Code of Silicon Valley

Episode Date: June 29, 2023

Come along on an exciting journey with Kevin Chandra, a daring young entrepreneur from Indonesia, as he recounts his startup odyssey in one of the harshest business environments—the Bay Area, if not the most. In the conversation, Kevin graciously imparts his insights gained from working closely with influential individuals and renowned organizations in Silicon Valley. About the guest: Kevin Chandra is the co-founder and CEO of Typedream, an Indonesian-founded no-code website builder. About the host: Gita Wirjawan—who is also from Indonesia—is an educator, entrepreneur, and currently a visiting scholar at Stanford University’s Walter H. Shorenstein Asia-Pacific Research Center (APARC). #Endgame #GitaWirjawan #KevinChandra ----------------- Episode Notes: ⁠https://endgame.id/eps144notes⁠ ----------------- SGPP Indonesia Master of Public Policy: [e-mail] | ⁠admissions.sgpp.ac.id⁠ | ⁠wa.me/628111522504⁠ Other "Endgame" episode playlists: ⁠International Guests⁠ | ⁠Wandering Scientists⁠ | ⁠The Take⁠ Visit and subscribe: ⁠Youtube SGPP Indonesia⁠ | ⁠Youtube Visinema Pictures

When when still still more you must fail fast, fail cheap, fail forward. This is endgame. Hello, Treman, today we're in Kevin Chandra. Belial is founder from Type Dream.

This is a startup that is based in Bay Area California. Hey, Kevin. Hi, Pat. Thank you. You're up, sir, sir. You're up.

And I'm a bit better from many of people Indonesia that I know and thought you. You're coming from Indonesia, busar in Surabaya, then,

come to be English, not know what, but you've done well. How that? Thank you, thank you, thank you.

So, so, yeah, so, I was 18-al-a-old-a-old-old-a-canty. So, the school has never got to take, so that's the school that's the same asem a Christian Gloria one.

But, but the school from the tech up to great, always at Gloria. So I'm, so I'm, yeah, more embower to, like, that's school national,

more like middle, maybe low end of of, But but I'm always always always a little unique in school. So, mom my mom's like a guy who's too, so I'm a lot of people, so I'm going to school in USC, so.

Okay. So, from the kiddickle, I'm still instill if, oh, the education is important, you're going to be able to, you know, to be more. More than mama, if I was. Okay.

Papa, too, solace, because I'm sorry, put, so much more than the U. So, mama that's more push. And it's important. Science, mathematics. Then, then, busaing English is important. But, yeah, that's, not there's not a lot of people

that's about. So, yeah, bleopat-and, until, actually, after, after, after, up, up, My name's from here too, so they're going to be able to surabaya, like, this is, like,

this is, this is, when I'm able to ask you know, so, I'm going to be able to, like, the out one-up, you know. No, no, who's a lot,

and then, there's a story, like, I, like, I was, I, actually, I'm, usually, I'm, really, not sure,

not like, that's like, yeah, can't be, language, English, this, you, like,

kind of, connotation, like, , You're in the Uralbay? Yeah. In Surabaya? In Surabaya.

Wow, okay. So, they're even like, live in here, bestar in here. If you're in, when I'm like like, like,

Kberatang, so, yeah. So, so, yeah. So, so, growing up, plan, so, you know,

I've been, kind of, kind of, kind of, like, playing with them, like, oh,

they've got, playing, game, computer, that I'm that I'm not put, but,

but, but, but, I'm still, plon-plagre to be communicates with the English, that I'm not-garet, that's as penting that. I, too, I mean,

so, I'm just, like, so. So, so, from the little, I've been, with, nonton-n-n-div-video people-Bas-A-M-Gris, belajared in-de-pac-cacacacage,

that's, you know, then, from S&S-M-A, I, I, every day, like, religiously, to, know, video-key-NOTS-N-N-Jew-Jobbs. Like I'm like, like, word by word, he's going to be like,

just so, I'm so, I can't be like storyteller that's, like, if I'm going to be able to, can be able to, that's, that's true,

that's like. When you're not, the U.S.C., how, so, so, so, so,

so, I, actually, if, when, if, So, so I'm going to beable Valley College, in B.R.A.A. in Pleasant Hill.

So I'm, so I have to go to two-tawn, then, then, then, then, I'm going to transfer to USC. Yeah, all that, that. Okay. And then, that's, you know, when you're going to be able to

or, you've got to know when you've got to do you're doing? To, What's the jury? Okay. So, what's it for me. Why?

Because, because it's because it's good. So, for me, for me, I think it's much deeper than that. So, I was a huge, Star Wars fan. I was a huge Star Wars fan. Star Wars, not even Star Wars,

not Star Trek. Not, not Star Trek, so yeah. I was from the kids, but I was the more than I'm going to Star Trek, more realistic, more interesting. Star Wars is more gontok, gontok, can?

Yeah. Adventure. Because Star Trek, more exploratory. Right. Yeah. I'm actually, I was going to beaise English, many from Star Wars, that. And then, and I was, this, space, this is kind of,

yeah, that the technologies they use, the gadgetries, the spaceships. Like, I like, I like, so much, that. And then from, from, like, oh, I'd really be able to make a personawatt, but, yeah, it's a pretty tall order, it's a pretty tall order, like, so it's really to rocket,

like, wow, in Indonesia, no, there, got, like, rocket. Yeah, all, it, did get a little, like, and, actually, I'm gonna', like, robotics competitions, that, this, this, this, this, this, and from robot, I'm like, like, like, oh,

if you're like, yeah, yeah, you know, what do you, this robot? This is what? It's going to do you're going to doxecute task what, how not there, I'm interested, oh, okay. Programming is pretty cool, yeah? And, maybe this is one of one thing

the most of the most of the most the model of the small that's the world, dialed down to programming until, I'm actually, I'm one of one generation that inspired from film, like social network. I saw Mark in his dorm room friends, like, wow, we can't be able to move

the world, just just using laptop with friends with people with the whole world, like like working group projects, but who's the whole world, and then from, and from the whole world, And then, from there, I realize, okay, you know what,

with limited model this, you can actually do something really great, then I, from before injacking kaki in U.S., I've been like, I have to find like-minded friends

who want to make something together, that's. Nare. Nour. Then. And then. And, then. And, how.

From Diablo to U.S.C.? Because mom that's not, that's because I'm able to get 4.0, Gpa. So, so. So, so. So, so. So, so the system education, to, if you know, if you're going to make sure, you know, either for people in America, it's either for people not mampu, or you're, you know, in SMA. Like, academician, that's still is the other than the other than the other than the other than student. So, so, so that's, um, so much more than you, um, so long as well,

it's, if you're, you're limited option. If you're, if you're going to transfer, yeah, transfer to university at America, you know, you're not able to, move, to start, like, it's near impossible, almost. Almost no, there's almost. Ampire. Now, So the either you either, like the university of California UC Berkeley, UCLA, UCSD, or you know, or you must school private line that's not

derogalong Ivy League or Stanford MIT level. So, so, when I was in Diablo Valley College, it is mandatory for you to get a 4.0 in order to come computer science at Berkeley. That was the best option, that we can't. This is written or unwritten?

This is unwritten. But most people who transfer if you're open, applicationing, like even my friend, I'm transfer, co-founder my two who's two, must be able to Berkeley,

computer science, to, has 4.0. Perfect. No blemish. So, yeah. So, I'm one of the ones that,

yeah, academicing, not, not too so jagued, to be able to get in 4.0. Yeah. I was like 3.8-8-9.

But I missed the mark. So, so you can't get to go to the best option that was USC, you know. After the school is good. The Trojans. The Trojans, exactly. And then C.S. Nambil computer science.

Nambil CS. Okay. How, do? It's after lullos or when lulles, perjalan of the life? Perjalan. When the U.S.C. it's really, it's bad, ma'am.

So, yeah, ma. So only, if, to private university, that's a credit that's transferable. So I was like, I'm taking class computer science from freshman. So, whatever I'm able

in the Abelelil all, unless you're transferer to school in a country, who's made for that. Because there are quotal from community college to the University of California system, it's transferable.

But if, if not, yeah, along all. So, so, I'm going to, semester to like, like, four-same-five class core, heavy, heavy-science, so, yeah, it's a bit more enjoy, like, because, because of the schooling,

I'm gonna, yeah, the, I'm gonna take class freshmen, and, uh, upper division at the same time, so, it's very, it's not a seamless experience, that's, like, oh, cumulness with the kids, freshmen, but at the same time, with seniors, like, yeah, a bit weird,

But but that's like that's like that's I'm always keep in touch with the people from my team from DBC from the Hiavo Valley. Ever since I've been injacking kaki in here, can, you know, try don't,

people who, people who, who, make make something, and we're making something. Yeah. Yeah, even when we're transfer,

the time we're doing two to Berkeley, one to UCLA, one to UCSD, even when we're transfer, we always in impetting time, like, okay, yeah,

guys, every time we have a little, we're up to comepull, we're just to try something, and what we're getting ready to, like, when still still still

muda, you must fail fast, feel cheap, fail forward, and this, you know, you're just, just leave just

just leave away just, just, it doesn't require a lot of capital, move on, so, so,

so, wow, make assam, garam it, we're, we're, like, like, when we're going to get this whole tech industry,

dap funding, how do you know, how do you, who's going to be to get to know, so, and then, and then, we're,

to be outreach to, like, to venture capital in Indo. This is while school, yeah. Right.

Tinkat, this? This, this, the sophomore, like, the year. Okay.

Yeah, still in this. We're still in this. Yeah. We've got, reach out, then, then, yeah, belajara, the hard way, from there,

if it'sa-cadish-cadishol, so matter-bent. So, many people, too, even, even, to look at least, the Males, like, this, like,

what are you, what are you? Can, usually, you know, if you're not, or, Stanford,

MIT, like, oh, go-minded, brand-minded. Yeah, so, we're,

we're getting, from all these investors, oh, try, maybe you know, transfer through school, after school, try, maybe you,

try, maybe, you know, in Silicon Valley, in companies, company that big, and then. Try, you know,

maybe, you know, to get to accelerator, like, Y Combinator. Abis, the goalposts keep shifting,

to the point, where we're like, oh, yeah, there's already so much. Mekin shiute just. Yeah, yeah, we're like,

not able to do, not can't continue to, at a certain point, and then. And then, start-up first did begin.

Start-up first that's, likena, we're 2015, semester, yeah. 2016, sorry, semester two.

That's the first we're first time we're trying. Okay. We're, in total, I've done done to do you,

maybe, eight different projects together. Wow. But, yeah, so,

yeah, I think that's, like, gagal it's like, gagal it's like, like, yeah, yeah. Gagal the first, it's just, it's just, gagged the two,

like, oh, we're gonna get in user, because if you just just make do'an, yeah, you know, you must be in

marketing, like back to the old traditional values, you know, beauched, don't, you, now,

you're, Beleasure, oh, how, how much more? Yeah, right? Oh, ternata, what you know, not that matter, like, especially there's seat stage. Storytelling, it, more matter, like, credentials, it's more matter,

as long as you can execute in there, maybe you're got, and then, then, and then, then, after we're coming into to WIP combinator, And then, the way that's like, and the way of the way. So, after the time.

So, the type of dream, the 7? Maybe to 8, yeah, maybe. Yeah, maybe. And the 5 this, still?

Massey. Yeah, but... Morebiasa. The original cast. It's not gampang, to maintain the lequatatant. Apalagi,

maybe what I'm going to I'm not because of the gagagalant that's the same thing, that's the same thing. Benar, we're just the successan that makeslequatat. Yeah. Not, right,

the better, the bapacal backer, it's like, so, what, yeah, it's, it just doesn't matter, that,

what we're doing, move forward just, try, yeah, It's exciting, and it's en-escendenged again, but it's been like to try to find out, but maybe, we're still,

the time that's just, like idealist, idealists, the other from Indonesia, not get to America, not get what, and then, so,

so, the story of the life's really, the bigger goal is, it's, so, nothing else, that much.

And we're, And we'll learn the way to tolerate a lot of the things, because because a big in a business especially five people, this is like, we're like,

this is more than the average marriage in the US, so. So, so, like, partner, got married to each other.

We've got to workiness of each other, like, the the lastisnesses now, like, like, I'm, already, like, and, yeah, be able to make-a-I-lac-lac-all-al-a-lac-lac-old, yeah,

yeah, we're really, really, yeah, yeah, yeah, yeah, yeah, the, yeah, the co-founder, to, um, can't push each other? Mm-hmm, that's true, not, so,

comeitraan this, or, or, maybe, this, this, it's not, this will be the samean-in-in-ing-ing-ing-a-oh-lottizanity also maybe the other than the other than the other. For the challenges, I'm a life time. Because to try people other than people who've been lewatin

like this, it's just very often from Indonesia. But maybe if the Sisi Opportunities Agatte, yeah, if I say we'll be able to slavage with, like, yeah, there's a lot more than, like, oh, I'm going to pursue a few things that's different,

yeah, later. This is better than Beatles, like, Yeah, if you're gonna becathes, yeah, maybe peccas but can't beckxed, but it's not too,

yeah, but I'm right. But, but I'm notherst. So, tell you, about time dream. So,

type dream this, like I said, I'm going to maybe, the 7, to 8. Yeah. Now,

basically, this is this is a product that we've been begin in English better. So when we made Type Dream, it was actually because of the learnings of our previous product, it's called Kotter.

It's a passwordless authentication platform that helps companies send OTPs, email magic links and all that. And we were trying to move down market. Okay, let's try to go somewhere that's easier. Maybe like something like prosumers, right? Still above consumers because consumers don't really need that. So prosumers are people who don't know how to code, but they have a business. And they usually use website builders and no code tools to run their business.

And they need these authentication platforms to sign in their users, authenticate the transactions and all that. So we did that. And so we built a lot of different plugins for website builders out there, like Wix, Squarespace, Shopify, you name it, WebFlow. And we had like a lot of requests coming in for people wanting authentication on Notion, which is actually a note taking app. So that piqued our interest. What are you doing with a note taking app that you need authentication for? How did you break into the YCOM mold? The YCOM mold is like it's not easy, right?

It's not easy. Like they always publish this acceptance rates of like below 1%. below 2% and they always call themselves people call them on the internet like the Harvard of startup accelerators right the people that get in there are like the founders of Airbnb

Stripe Reddit Instacart DoorDash coinbase all that cool companies and for us it was just honestly we just apply it guts yeah we just

it's not normal for Indonesians to do that especially like Indonesian students here in the US to just try to apply to to to YC, like, they already have this misconception of like, we can't compete against these people. And so when we applied, we didn't even think about- You're talking about these people being from maybe India, China, or the rest. Mainly the U.S. first.

Okay. The U.S. first. And then the Chinese and the Indians, yeah. Wow. Which predominantly dominates the standards, right? Yeah, they have. They really have. And so for us, it was more like, oh, you know what?

if we want to do this, since we can't raise from any investors at all, nobody believes in us. Might as well take a shot, you know? It's just an application, guys. Like, it takes 30 minutes to fill in. And if they notify you that you get an interview, it's just 10 minutes away from getting funded. And we had no other option. Nobody's willing to fund us. And mind you, we were on a student visa.

So if you're not working for a company, if you just quit your jobs, you have. have, I think, I think, about 60 days before you get deported at the country. Clock sticking. Yeah. So we were like, guys, we really want to do this. So we need somebody to fund us so that we can really work on this, like seriously. Right. So we just applied. It took us about, I think, 30 minutes to apply, recorded a video, telling them like,

oh, why we want to do what we're doing, explaining to them the dynamics between the founders, It was like how we were with friends. And we waited, I think, maybe almost a month after submitting the application to getting an email saying that, hey, you guys got an interview. And when we got that email, we were like, whoa. At the time, there wasn't even 10 Indonesian founded companies at YC. We were number nine or something. Okay. We got the interview and we practice for the interview like religiously. Every day from the moment we wake up, the moment we sleep.

It's actually not good advice. We were supposed to keep running the company, like whatever we're doing. But we just thought like we really can't fail this, guys. Because if we fail this, then it's probably day jobs for us, right? Yeah. We have to go do. It's pretty existential. We got to do something else.

Like life moves on. So we practice it religiously. And then we even know the details of like how the YC interview works. It's usually only 10 minutes. You're faced with three group partners. And then they'll ask you rapid fire questions. There's a template for this. Like if you Google it, there's like a somebody made a quizlet, something like that for

YC interviews, just rapid fire questions. They're like, it seems to be like really hard questions that they ask. But they're not really. Like after going through it, the common theme is just asking you about, so what's your business? What have you done? What are some of the learnings that you've learned that you can teach me? Hopefully, from this conversation, you walk out the door, I learn something new for you.

And then the other theme is just about, to tell me why you, this team, right? Why are you the right people to do this? And whether we want to work with you. That sounds like a very common. Very common question. Anybody else would have asked. Exactly. But at the time, we just thought it was like,

it was very hard because the way that they phrased these questions are very specific. Because they understand a lot of different industries and they usually pair you with a partner that understands your domain really well. So for us at the time, we were trying to bring U.S. stocks back home from the U.S. to Indo and other alternate investments, hopefully. We were paired with this one partner called Gustav. He's somewhat the head of like for Southeast Asia. He interviews a lot of the Southeast Asian founders and companies. And he just knew directly where to strike. So our conversation wasn't even 10 minutes.

It was a lot less than that because he read it. He was like, yeah, I know the landscape. I've prepared all the questions. So I'll just ask you straight on. All right. You're trying to do fintech. regulations is very hard. All the licenses for you to get is hard. And yeah, I just want to know why your team is the one to get all those things.

Like to move in a regulated industry is not as easy as like just pure tech. You can just push code. Sure. Like every code, every new feature has to be evaluated by the regulators, the financial authority, whether you can do this product or that product, whether it requires a new license or not. So it just struck head on. And we were like, yeah, we just answered to our best. Right.

We were in the process of going through at the time was like OJCA Sandbox, the local regulators. We just showed proof that, you know, we were, it's ongoing. The conversation's ongoing. Yeah. And they liked that, knowing that we were still here. But, you know, we just kept on trying our best. And I think the thing that they value most that YCS has taught us once we're accepted is that It's really not about the idea that you're working on.

Because ideas change. So their philosophy and investing is they invest in the people and the team instead of the idea. Like the idea can be shaky. Yeah. But as long as the founding team's foundation is really strong, if they see that you've been working together for a very long time, grind it through, you know, all the hardships together, that's the company that will succeed. You've got what it takes to execute the idea. Exactly. So the market can always change, but a team has to be really solid.

So yeah, I think all in all, we just applied. They really liked the team. And then we got in. And yeah. And how long has the relationship been? With YCOM. We got in, I think it was in late 2019, and we began the program. I think it was in early 2020.

So it's been about three years, roughly. Did you ever meet with Sam Alton? Yes. How is he like? He's an unstoppable force of nature. I think you don't need to take my word for it. You can watch all the different videos on how other YC partners think about him. Even the founder of YC himself, Paul Graham said that if any founder from YC can really make it, he's got to be this guy. He's like relentless. This guy, he's like, he's something else. He's like a being from a different planet. I want to use as a way to segue or pivot to AI. I mean, you strike me as somebody who understands AI. Sure. Tell us what your view is. Yeah, for sure.

So I think first broadly, then we can peel the onion a little bit more. For sure, for sure. So I think talking about AI for me, I'm not an expert in it, but I play around with it. You sound like you know a lot about it after talking to you sometimes. Me and my co-founders, we sort of hack things together to try to find, like, what are the end-user applications? So what's possible? Right. And the way that I view it right now, a lot of people view AI as something that's like, what product can I make out of this?

It's the old internet days. I'm trying to do a sell dogfood.com type of thing. Like, it's a product. But the way that we or like I personally view AI, it's not really. just about a product. If you want to actually develop something that is related to AI, you got to think about it as a power up. Like in a game, right? If you were to be a superhero, what superpowers would you give yourself? So for example, chat GPT, right? Something that's really popular amongst a lot of people worldwide. When they launched this new thing called

chat GPD plugins, the ability for chat GPD to actually use an actually use an actually access your applications. That was mind-boggling to a lot of people. But what really stuck with me is that chat GPT is able to use more than one application at a time in parallel. So that is a superpower that humans can never have. You use your phone, your computer, you can't use more than one app at a time. You've got to look at one thing and just manipulate the bits in that particular software. This thing can use more than, you know, it can even use 10, 100. It doesn't matter. It's just a compute power thing, right? So right now, what I'm seeing in the mark is that people are developing a lot of power-ups that enhances human productivity. So imagine if you can,

it can use 10 different apps together. Imagine if you ask it to do things like, can you read 10 books at the same time then? Sure. Can you watch 10 different videos at the same time, listen to 10 different podcasts at the same time, and be able to extract all the learnings from it? So it's like a technology that enhances your productivity to levels that has never been seen before. And I think it is also one of the first time. that humanity has invented something that can also have judgment. That you trust its judgment. It's not there yet, but it's getting there.

Like for certain tasks, you'll trust it more than a human. Right. Right. So usually we've always been creating tools to, you know, the human judgment aspect has never been touched. Like, you'll always need humans. But this is the first time ever in which you're like, hey, you know what, actually,

if my car, if I drive a Tesla and it says, you know, the tires are going to blow up in a little bit, I'll stop the car, right? If I'm flying a plane and they say that the engine is going to have some problems, I'll land the plane. Like, you trust the machine more than the human. And it's becoming more and more prevalent in, like, you know, the healthcare industry and a lot of other things as well. So, yeah, I think the way I view it AI all in all is like it's a super powerful. that you're giving humans to do the grunt work of a lot of jobs. What could go wrong?

What could go wrong is that the development of AI is so closed. Like even open AI, the name itself is not really open. You see? It's closed source. It's close source. It was. It was. Exactly.

But they change. And now it's for profit. Yeah. It's for profit as well. It should have been nonprofit. Exactly. So it's a black box. You don't really know what's going on inside.

You don't know what data is trained on. And it's close into really big companies. So right now, if you want to develop fundamental AI models, you want to improve upon it, you want to develop like new cooler models, it's actually a capital and resource problem. It's no longer about the technology. Like, they've solved that aspect for this particular large language models. And whoever has the biggest compute. power and the most money to train these AI models because to train them, it costs a lot of money.

It depends on the amount of parameters and the amount of data that you train it on. It could be something as simple as, you know, as cheap as like $100,000 one run, or it could be something as high as, I think, Open AI is like something north of more than $5 million just to train it. And for you to be able to train with that kind of capital costs, it's not for everyone. Because it is so close source and it is concentrated in these big companies with a lot of resource, a lot of people whose industries are affected, whose jobs are going to be affected, doesn't really have any say in it. Unfortunately.

Right. And I hope that they're trying their best to be more inclusive. Yeah. So they do this thing called RLHF, which is reinforced learning from human feedback. They'll include these people, right? That are beyond. I hope so. I want to believe that. I mean, a couple of observations, right? The fact that he changed it from open source to close source, nonprofit to for profit. Is that a manifestation of his personality? or his inability to manage the bigger stakeholders. Yeah, I think I wouldn't want to speculate on this, but I'll repeat what Sam said. He explained it, like why Open AI became a for-profit organization. And the reason is because they required capital.

And they couldn't raise enough to eventually reach their goal of building AGI, which is artificial general intelligence. requires a lot of money. and they can only get those kind of money from really big corporations like Microsoft, right, in which doesn't really invest for charity. They want to get something out of it. And exclusive rights to the model, the weights and the training data and the consumer products and all that before any other company can.

So he had to switch to raise the capital in order. But he could have stuck on, right, with somebody like Elon Musk. who could have, he could have supplied him with the necessary capital. He could have, but I think. To maintain it, you know, close source and nonprofit. Yeah. Sam and Elon, they had a fallout even before the deal with Microsoft. Yeah.

I think it was before the for-profit thing. And because of that fallout, I think it's pretty hard to find another guy that's like Elon that's willing to risk a lot of capital to train something that. I mean, he put in quite a lot of capital, 50 to 100 bucks, right? Supposedly. 50 to 100 million. Yeah. Yeah. But, but, but tell me this. I mean, if it's, I get this effort. Mm-hmm. To try to attain AGI, which could happen in the next few years. Some people say 10 years. Some

Some people say sooner. Yeah. It really balls down to scalability, right? And you kind of alluded to this earlier. It's not so much a challenge in programming. It's a lot more. engineering because when you need to scale up it becomes an engineering challenge yeah talk about that yeah so i think right now the challenge and and the AI community right like you said earlier people are debating about when aGI is going to happen some say it's a few years some say it's in a decade now the thing is humans haven't really discovered how the human brain works right they don't even fully understand how these, you know, transformer architecture works. Like the way that they train it, they sort of set weights.

They don't actually, you know, understand that if I do this particular thing, I'll get this outcome. It's more of experimentation on their end. They know how to train it, but they don't know what outcome will I get. Even when they train it, if you check out the latest research, paper from OpenAI and Microsoft, they were surprised at some of the capabilities that it was able to do, like few shot problem solving. Never seen the problem before, but given a bit of context, they can solve it. And now they're working on maybe zero shot problem solving, which I don't need any background at all. You give me a problem, I'll solve it. So the problem is that nobody actually

knows what we're playing around with. They're just by luck, right, they found this architecture that was actually used for language translation by giving AI a language interface. Like it got really big because you're going to actually talk to it now. Back then it was more like, oh, it is really good at this one task, but now it's kind of like generic enough so that you can talk to it and stuff. Right. And yeah, it's a lot. It's basically nobody actually knows what they're doing here. And they're just discovering new possibilities out of it. And this architecture is not broken to this point

in which they just kept increasing the parameters and the amount of data that they put into it. And it's still not breaking. It can still do more stuff. but people who say that AGI might might take longer are saying that okay these LLMs you're seeing like crazy capabilities like in the short term but maybe in the next few years it'll plateau that's all that it can do it is nowhere near AGI it doesn't have vision it doesn't have like a lot of things that say a human has as input it just has text right so probably not be scalable like maybe AGEI maybe AGE is like a multimodal interface for AI in which it can can see, it can hear, it can feel, it can read. So yeah, a lot of mixed and different things. My concern is, and I've had a separate discussion on this, is that, you know, OpenAI or LLM

has gone through this face of hallucinations, right? Yeah. And it's, it's upon us to make sure that, you know, AI will hallucinate in a good way. Right. So the feedback or the feedback loop, you know, ought to be done in a very robustly judicious manner. Yeah. So that it leads us or leads humanity to a good place as opposed to not so good place. Yeah. Right? Because, you know, this is hallucinous. This is going to continue hallucinating. We just got to make sure that the hallucinations going forward are spurt by some sort of hypnosis that's done by us. That will lead us to a good place. There is a risk, though, that the hypnosis is not being done in a good way. To the extent it's not done in a good way, then I think we might end up in a soup. Right? Yeah. And the other observation that I've made is that, you know, much of the discourse is propagated, led and managed by the technologist with very little, if not, no involvement of the other disciplines. And at the rate that this is not as multidisciplinary as it should be, I think that exposes us to a greater risk. I believe the same thing. That's why, from my point of view, I really do wish

when they develop these technologies moving forward, that they'll be more inclusive, that they'll have people from multiple different disciplines to participate in the development, participate in the training, and reinforced learning by human feedback, which is not just focused on, you know, a couple of Silicon Valley technologists in an office. Yeah. Or, or, you know, or computer scientists in general. Like, you got to include. Especially, you know, it just kind of seems intuitively that it's going to get further elitized. Because who's going to be able to engineer this thing?

Yeah. More extensively is the guy that has more money. Yeah. And the guy that has more money, I think we'll be vested in making a lot more money for himself, for themselves or themselves. I agree. Right? Yeah. I just kind of intuitively think that this will take us to a place where,

sort of like what the internet has done. The internet was supposed to flatten the world, but I think kind of deflatten the world because it hasn't really democratized ideas. It's democratized information. I mean, we've seen how inequalities have gotten bigger, right, in most parts of the world. Yeah. Developed and developing. I think AI is the next level of the unequalizer, potentially.

So my question is, how do we mitigate this risk or potential risk? I think from my end, it's just to, I guess, support more of the open source communities. There you go. Open source developments like they are right now. And I hope companies like open AI, like all these close source developments, will be made open so that anyone can participate. Yeah. Anyone can, you know, modify the models and not work with a black box.

It's not an API hit. I don't know what it does. It just spits out things that, you know, works for me, right? They got to really see the insights of it. And I think at the moment, I think companies like stability AI, is doing really good open source work. And if you check out a lot of the different models from Hugging Face,

which is like the GitHub repository for AI models, you can run on the cloud as well, the open source community are getting better and better at it. And what's impressed me the most is that some of the models that they've built is with less training data, less parameters, but the performance is, you know, getting there

to equaling what OpenEI is doing. But yeah, I hope for the future, it's more of these open source models, gaining traction, rather than those of the close source ones. But I do understand why it's necessary for these close source models to be developed because with a lot of compute power, resources,

and you have all the best talent in the world, maybe for specific tasks, right? you can do it better if it's close source. But yeah, I just hope that they're good people as well. They don't develop it in a certain way that... Fingers crossed. But explain this. Again, I guess you've explained this to me earlier,

how the future is going to be more with using our two thumbs as opposed to audio for, you know, the usage of LLM forward. So I don't think the future is going to be two thumbs. Okay. I do think that, so the history is like, now we're computing at the speed of our fingers. Yeah. It doesn't matter if you're on a computer, it's typing on a keyboard.

If you're on a phone, it's your thumbs. Right. Right. I think the future would be computing. Well, I type with two thumbs. No, I'm just kidding. Right. Yeah.

So I think the future would be whatever is convenient, right? Whatever is the best way for you to communicate that idea, it could be through voice. If it has to be through text, it will be through text. If it's through touch, then it might be through touch. Which is more likely to be the case. I think the most... So I have this point of view. One of my friends who are...

Who's not a technology. He's actually a musician. So he told me that whatever technology is trying to do in the music industry is that it's trying to replicate whatever is in the real world. It's just trying to be more and more real and less synthetic. So I think in the future with the advent of LLMs, I do believe that typing with your fingers and your thumbs, it's kind of limiting to certain tasks. It could be through voice. Maybe that's the most natural way that humans converse with each other, deliver ideas. It could be through that.

And today, that technology already exists. So Open AI has this thing called Whisper. There's a bunch of other open source models who can already capture audio, turn it into text, flawlessly, even with the ums and butts. And a lot of different languages, I think almost all, it can be translated into any language. So, yeah, I do believe that in certain tasks, it should be voice, in certain tasks, should be text, whatever is convenient, the fastest way to deliver the idea. Well, there are people who claim that they can think better when their fingers are moving. Yeah. There are people who claim that they can think better when they're talking.

Yeah. Right? Do you envision a near where you'll see a massive dislocation of real jobs? Like, I think, I think one of the debates that, you know, I've had with some people is, will AI be able to write novels as good, if not better than what J.K. rolling would have done or Stephen King okay would have done got it so those are two different questions first dislocate dislocation of jobs right i think we will see a lot of jobs getting a place in the next few years i do believe so because there are a lot of like wide-collar workers now at the office whose jobs are just to punch in numbers which is which isn't really different from like decades ago. It's just through a different interface. It's spreadsheets, documents,

presentations, which are the things that most white-collar people do. That'll be easily replaced by AI. But what I do believe as well is that it'll create new kinds of jobs that didn't exist before. Humanity has always adapted to the invention of technology. For electricity, after electricity, industrial revolution, before and after. We'll always find new things to work on. And what I find really pleasing about this AI development is that now as humans,

we can actually work on the really important things, like curing cancer, space exploration, solving world hunger, something. That's not just trying to manipulate bits and pieces, creating weird content on the internet just to make a quick fuck. You can actually focus on the really important problems now. And then the second, you asked about can AI write better novels than J.K. Rowling or Stephen King? At the moment, it can't.

It can't create something that humans haven't already created. At the moment, it's a really powerful assistant. It helps you clear out the grunt work. It makes, I think, a lot of people who are in the workforce, they can sort of get promoted to a managerial level. You don't do the groundwork. Maybe entry-level jobs aren't about getting told what to do anymore. You've got to be able to think for yourself.

It could be in the next decade or so. Every software engineer has to be a CTO level. We're thinking about architecture, right, about how to scale these systems instead of moving buttons. Right. So I think at the moment, no, but there will come a time in, I don't know, the next few years, next few decades, where they will be able to do that. I think what people are discussing right now is that within the decade, they could be co-authors and research papers to that level. I think sooner. than a decade.

I know some writers were already talking about stuff like that. It's scary. What about singers? What about actors? See, this thing about... Singers, I think, would be easier than actors, right?

Yeah. So, lately, there's been a lot of development in the media industry tools, like the AI tools for the media industry. There was this song released, I think, a couple of weeks back that was trending. It was like Drake and the weekend. Yeah, I heard that.

That was awesome. I was mind blown. If I didn't know it was created by AI, it would have been easily a Spotify top 10. I would believe that that's something that Drake and the weekend made. Maybe we got to do an Indonesian version of Drake. Maybe, yeah. I've seen a lot of content around President Jokoi, singing Dangud's song, which is like pretty cool, pretty funny as well.

But yeah, I think to me, it just raises the bar for people in the media industry. Like if you're a movie director, you can't just keep milking franchises like what the maybe, this is my personal opinion, but maybe like the Fast and Furious series. Just keep milking that content or any other franchise that's transformers. Just keep milking the content. Same template, but different kind of storyline, but still the same thing. thing overall. It raises the bar to which, if you're a movie director, at Christopher Nolan's level. You can't make those kinds of things anymore. But the good part about it is, as well, what I see in these developments is that it's democratizing the creation

of music and movies. It's not just Hollywood anymore. Anyone in the world, if you have a good enough camera like a phone. You could edit it to be a late 2000s or early 2010s type of movie with CGI and all that from the web browser on your computer, which is really awesome. For me, the future of entertainment would be like very, very personalized. I'm feeling this way. I want to hear this kind of song sang by this person. It's just kind of, it's scary in a sense. that it's just going to make humanity more lonely. Right? You can artificialize connectivity with anything of superlative qualities.

You know, that sort of like pulls you away from the real interaction with real people. I hope it's the other way around, but I get... Oh no, I mean, I think there's a lot of positivity that... that AI is going to bring about such as education. I mean, we can talk about our part of the world, Indonesia and Southeast Asia, right? I mean, I've been thinking, you know, there's less than 5 to 10% of the population of Southeast Asia that could speak English the way you do, right? Imagine if 50% of the population of Southeast Asia can speak English, can speak Mandarin, can speak Japanese, can speak, speak, whatever international languages that allow themselves to be able to internationalize,

you know, that will just open out possibilities for the others to be like you, be able to fly over to L.A., the Bay Area, and then coexist, you know. Yeah. There's a lot of people in Southeast Asia that have not gotten a good taste of the democratization of opportunities because of that simple, lingual, you know, obstacle. And I think at the rate that, you know, we don't have that many teachers, that many great or good teachers, AI can be very helpful. I think it can accelerate process. I do believe so. It is happening today. If you go to, I think websites like, I think it's character.com. A.I. It's developed by, I believe, it's one of the team members from Google's deep mind. Exited Google. There, you can choose, like, learn a language. I think there's Chinese, Japanese, and all that. And if you want to learn a new language, right? For me, it's been about if you converse with actual people. It's the easiest way for you pick up a language. And these tools, they do exactly

that. You can talk to it about your day. They'll ask you questions like a normal human would, but in the language that you want to learn. Will it correct you also? Yes, it will. Wow. It would. Yeah. And there are some, I think it's like called speak.com that teaches English. It can even do it through audio. So it's like you're even talking to, you know, a non-existent. friend that speaks English that helps you correct a Siri like a Siri yeah but a lot better wow it can fix your intonations word choice vocabulary it is amazing what these tools can do specifically for education because I think growing up right for me I've always had a problem when things get hard as a student yeah I would kind of like want to run away from that thing

So I'm not going to go through with it. And I've recently discovered that that's because when I was a kid, my mom pushed me a lot, especially in all these Asian kids go to Kuman. And when I was in kindergarten, I was doing multiplications and divisions. And it was kind of hard for me to comprehend. And then when they start to introduce me to algebra with all the letters,

I got freaked out. And I'm always used to see. seeing that at Kuman, you get this 100s with the red marker and the circle around it. And that was the first day that I didn't get it. And I didn't know that it would be such a profound moment for me growing up that I would keep on running away from really difficult subjects. And the reason why I tell you this story is because a lot of humans, they actually aren't dumb, but they give up when they learn in certain conditions and they don't have a teacher to guide them through that. So when I use these AI tools, like when they teach language or even mathematical concepts,

scientific concepts, other subjects, they can identify the moment where you're about to give up and give you that extra boost that you've always needed. When you get confused in class, you know, when you were a kid, you're like, I don't know how to to do this and I don't know what to do, right? It identifies that exact moment and it gives you what you need. And it speeds up the learning process by quite a lot. And that's what I'm really impressed by these technologies when it's used the right way, especially for education. I agree. Totally. I just think that the low-hanging fruit that the Indonesians need to get over is that lingual proficiency.

And I've seen with my own eyes, I've said this many times over, where at such time when a person overcomes that obstacle, lingual obstacle, he or she just feels very differently. Yeah. It just is an injection of confidence

that translates into this person's becoming, you know, someone to the power off. I agree. Right? Yeah. And that, I think, is what we need. Exactly.

Just imagine of a hundred million Indonesians speak an international language. That's economic unlawful. That's game-changing, man. And I think AI can accelerate this. Yeah. You know, one would think that it's probably going to take us a decade to get there. I hope not. I hope not, too.

And speak your Javanese, speak your Bahasa, speak your sundani, speak your whatever. But speak an international language because we need to become a more relevant part of the international community. Because we're big. For sure. We're not a small potato. Yep. You know. I've experienced it firsthand, right?

From my end, when I actually raised from investors here, one of my investors asked me, So you're from Indonesia, right? Yeah. How come I rarely meet your people? You're the first Indonesian that I've invested in. And I've invested in a lot of companies like Pinterest, Coinbase, you know, big international, a lot of immigrant founders as well. But why, where is Indonesia? Where are your people? And I couldn't answer that. And he kept asking me questions wanting to understand more about the background. He's like, okay, let's not talk about tech. Maybe it's too high-end. It's too elitist. How come in America? I don't, I can't fly to your country, like one direct flight away. But I can fly in a lot of countries in Southeast Asia,

like Singapore, Malaysia, Vietnam, or even in South Asia, India. There's a direct flight to Hanoi from San Francisco. Exactly. And he can, and he can, kept asking me even further questions like in any corner in California are like in the big cities right New York LASF I could find Vietnamese food Korean Japanese yeah even here Singaporean food now right oh yeah you got Kopitiam yeah Killin Lee Killingee how do I find your food it's normal in my diet to eat pot Thai and potse you know they even can call it by the Thai name of these things and asking me about things like culture, where are they? And I just couldn't answer any of them. And I felt like an anomaly. And he told me that, so this guy is Indian, right?

The way that he made it is through the other Indian immigrants before him, who came to the U.S., who paved the way for him. And it's always better for the next generation. And he told me this one really profound statement that, whatever that I do, even if you failed, but you guide the next generation, you're not really a failure. You become their stepping stone for a better future for the next generations. And he told me that it should be part of your mission to bring more of these Indonesians to learn here and explore and make it here and become stories like how the Indians and Chinese back home can, can derive from their role models.

Yeah. All the CEOs of these tech companies are Indians these days. Man, you know, there's this, you know, public secret that 30% of the new ideas in Silicon Valley are coming out of Indians. Yeah. And there's a staggering statistic about the H-1B visas. 75% of H-1B visas are given out to Indians every year. The Chinese only make up 10%. The Indians only 2.

The Indonesians, 0. whatever percent. And it's somewhat empirically correlated with the representation of each country in all the campuses. But if you take a look at the mainland Chinese, there are 400 to 450,000 all across campuses. The Indians are about 200,000. Disproportionately smaller than the mainland. in Chinese, but they disproportionately got much more of the percentage of the H-1B visas.

I think something to think about. Definitely. You know. Hey, you've basically proven to yourself, to your family, into the world that you grew up in a place where you spoke Javanese. Yeah. You spoke Bahasa. Yes.

You may speak Mandarin. you speak English, you survive in one of the most competitive tech environments in the world. Yeah. Right. And I think you could be an inspiration to a lot of people. I want to ask this final question. What do you think you would have done differently? All along.

I'm that type of person that doesn't regret anything. And I believe that I learned things at the right time that I could take. it but I think if I could do it earlier I would have learned languages earlier honestly that is dude you're you're doing all right I mean I would being fluent is one thing being articulate as another dude you're doing all right I hope so I hope so so I think if I could learn from you know maybe if I was like in kindergarten to be able to speak all the international languages. That is not just limited to, I'm only limited to, I guess, English that I can speak this way, as well as Bahasa, maybe. But I wouldn't limit myself to just those two languages. Why not speak Chinese? Why not speak Japanese? And learn from all these, yeah, from all these developed countries. And, you know, I'll be able to grow a lot faster and learn a lot of mistakes faster because one of the greatest unlocks, right?

learning English at a young ages that water or not, the fact of the matter is that a lot of research, a lot of books are written in that language. And even content on the internet are spoken. The way people learn is mostly in English. Yeah. All the books. Exactly. The way people undertake economic activities, it's mostly done in English. Yes. It's not a statement of race or nationality or geopolitical, you know, interest, but it's just the fact. Yes. So if you want to undertake greater degree of economic activities, if you want to learn more, I think you've got to learn English.

Yes. Now increasingly Mandarin. and some of the other European languages, too. I think, number one, it's definitely language. It unlocks, you know, you learning from other people's mistakes. Yeah. I got to YC today. I forgot Japanese.

That's also another important thing. I got to YC today because I watched their videos growing up as well, even though I didn't follow it as religiously as Steve Jobs. But if I weren't able to speak English, I wouldn't be able to understand all these new technological developments and whatnot. And I think the second would be, I would try to be more confident. Because I think the culture is that growing up from, I think this is not idiosyncratic to Indonesia. I think most Asians are, we're humble by default. We don't want to speak our minds. We tend to shy away from conversation.

put in our heads, talk behind their backs. But I think what I learned from people here, when they do that, the Indians take advantage of the moment. I tell that to my Indian friends. I see. Yeah, so I think for me, confidence and, you know, just dare to make mistakes. It's not to be ashamed of. Because growing up, I think at school, I was always taught that, you know, if you get a bad grade, it's kind of shameful. But it's not. It's sometimes. it's part of learning. Isn't it an irony that you actually would have attained more confidence out of your failures as opposed to successes? It is. Yeah. Right? I didn't realize that. Yeah. When you're so comfortable, if you fail a thousand ways, you just don't know how else to fail

because you've done it all and subconsciously you're being guided to the right direction. And yeah, you become more confident. Like, I've done that. I've made that mistake and I'm not going to repeat that again. Yeah. Yeah. Just language and dare to do more things. What I've learned now is that being me at this very moment, being surrounded by these kind of people, I feel like nothing's impossible. Yeah.

Right. If I ever want to do space exploration one day, you know, be like Elon. why not? Yeah. It's not impossible to have an Indonesian flag flying a rocket that goes up and down. Yeah. Deliveres goods that way. Or cure cancer.

Yeah. Like you've interviewed tons of amazing Indonesians who's done that. It has never been highlighted. Yeah. Yeah. Keep that fire alive and burning. For sure. I want to be there to see it.

For sure. Thank you. Thank you, Kevin. Thank you, ma'amacis. That was a good conversation. Tremantam, that's Kevin Chandra from Type Dream. Thank you.

This is an end game. I just say is a lot about the honor. And this guy could have been burned zero shot problem or something which I don't need any background at all.