00:42:34 · Haya Odeh, Ruchi Sanghvi
[00:42:34] S1: Yeah, I just wanna add something from my experience in building Replit, to watch out for power hungry people that can get like in, make you feel skeptical of what you're doing or questioning what you're doing. So it's finding like other people who are doing similar work like you and seeking advice like, hey, I'm noticing this, this, that. Am I crazy or am I not crazy? So advice there, this is something that we experienced, that we seek advice, and it was super valuable. [00:43:08] S3: So one of the things that I'm paranoid about is that I cannot keep up with the pace of information. So I'm curious, what is your information regimen? How do you keep up with the pace of everything that's changing, especially with AI, and also run a company at the same time?
00:43:08 · Ruchi Sanghvi, Amjad Masad
[00:43:08] S3: So one of the things that I'm paranoid about is that I cannot keep up with the pace of information. So I'm curious, what is your information regimen? How do you keep up with the pace of everything that's changing, especially with AI, and also run a company at the same time? [00:43:23] S0: It's hard. Look, I think that you need to have certain predictions or a working model about the world such that things should not be massively surprising. If they are surprising, you need to update your priors. So you need to have a model of a world such that if you're off of Twitter for a day or two or a week or two, it's not a big deal because you sort of understand where things are headed. And I constantly am trying to figure out where things are headed. I'll make predictions. I'll often make predictions publicly. And I think it's just like a skill to hone. So if you understand and try to understand the models and how they work and try to read some research papers when they come out. Again, it's a kind of buzzword, but if you have more first principles understanding of things, you don't have to be on top of That's actually the
00:43:23 · Amjad Masad, Aditya Agarwal, Ruchi Sanghvi
[00:43:23] S0: It's hard. Look, I think that you need to have certain predictions or a working model about the world such that things should not be massively surprising. If they are surprising, you need to update your priors. So you need to have a model of a world such that if you're off of Twitter for a day or two or a week or two, it's not a big deal because you sort of understand where things are headed. And I constantly am trying to figure out where things are headed. I'll make predictions. I'll often make predictions publicly. And I think it's just like a skill to hone. So if you understand and try to understand the models and how they work and try to read some research papers when they come out. Again, it's a kind of buzzword, but if you have more first principles understanding of things, you don't have to be on top of That's actually the [00:44:15] S2: best answer I've heard about this. I've asked a lot of people on stage that same question, and I think that I wouldn't have been so articulate about it, but a lot of what I'm trying to do is you are making micro predictions, and then if you see something that is wildly off, either you need to update the model, your own model, or you Or it's a scam. Or it's a scam, exactly. Most of apps say, oh, that person's stupid, right? Go ahead, I was gonna open it up maybe I after one more [00:44:45] S3: mean, do have a lot of questions, but we can open it up. [00:44:47] S2: Why don't you give one last one? [00:44:48] S3: Why don't you ask the kid question?
00:44:48 · Ruchi Sanghvi, Aditya Agarwal, Amjad Masad
[00:44:48] S3: Why don't you ask the kid question? [00:44:50] S2: Yeah, okay. So Shyam says six now, So our kids are almost like three, six, and nine ourselves. And you think a lot about obviously raising your children, giving them a great environment, trying to prepare them for the world that they will grow up in. But it is very hard to kind of have a high conviction, high confidence mental model of the world. What's the world going to look like in nine years when Avian has to go to college, right? That's kind of insane to think about. So how are you preparing your children for the coming, I would say, change in the world? [00:45:27] S0: Before we had our kids, I've been always like a ad tech nurse. I always liked education technology, worked in education technology, invested a lot in it, really bad sector investment. And then when we had our kids, I like tried all these tools on them, especially Shams, it was like he learns very differently. And it's sort of like, what is Mike Tyson's kind of quote where you have plans and [00:45:54] S2: then like- You get punched in the face. [00:45:55] S0: You get punched in the face. [00:45:56] S2: Everybody's got a strategy until they get punched in the face.
00:45:56 · Aditya Agarwal, Amjad Masad
[00:45:56] S2: Everybody's got a strategy until they get punched in the face. [00:45:58] S0: Yeah, I feel like having kids is like getting punched in the face by reality constantly. And so we're just learning as we're going. The one surprising thing is actually the Bay Area should be one of the places that's quite innovative on education. But it's like a Montessori school is like the most innovative thing you can find here, which is like 70 years old now. It's deeply frustrating how little change happen in education. I think what a lot of people with means end up doing is they end up creating their own schools like Elon. I think you guys might have experience with that. And then other people do just homeschooling because it really is not that great. Like you go look for schools, most of them are not that good. Ultimately, the main lesson from education technology is Bloom's Two Sigma problem. So this researcher found that no matter what intervention you can make in school, there's one intervention that creates two standard deviation improvement in educational attainment. And it is one on one tutoring. And there's nothing really better than that. So finding like one on one tutors for your kids, or finding AI apps that are really good at that, one of my investments, Synthesis, could actually be really helpful. But I think we're just trying to navigate.
00:47:28 · Haya Odeh, Ruchi Sanghvi, Aditya Agarwal
[00:47:28] S1: I can like talk about my thinking. I think about basics and basics will remain the same and values. And that means like introducing them to the basics of gymnastics or swimming, or like seeing through my children. And it's like a trial and error. Like, what are the areas that they are really good at? What are the areas that I think they should get better at? And it's like, you know, when you're building a product, it's like the same, and you notice and you observe. So it's like similar to that. And to what Amjad was saying, including going to school, we are dabbling with one on one mentoring and seeing where that take us. So it's just learning how they learn. It's very difficult and it keeps changing, and that changes how we introduce information and stuff to them. [00:48:22] S3: Sounds like AI models. [00:48:24] S2: Yeah, it's interesting. My big takeaway for my kids is actually just to remind them that the reason why we have all these computers and these AIs and these models is to actually be able to build things, and these models are there to serve them, and kind of like they have ultimately agency. I had this great experience with Aviain just like last week where he got a TI 84 for the first time. [00:48:47] S3: I charge him 4% interest, month Yeah, to [00:48:50] S2: because he doesn't have enough money to buy it yet. Ruchi was totally loan sharking him. [00:48:55] S3: And then he gets