Singapore vs. USA Education Systems, WEIRD AI Teddy Bears & US Senate $32B AI Investment Bill with Shiyan Koh - E428

Shiyan Koh, Managing Partner of Hustle Fund, and ​​Jeremy Au talked about three main themes:

1. Singapore vs. USA Education Systems: Jeremy and Shiyan discussed differences in structure, funding, and outcomes. In Singapore, the education system is centralized with high academic rigor, while in the USA, quality varies significantly due to local property tax funding, leading to disparities between affluent and less affluent areas. They highlighted how the Singapore system is envied by global peers for its academic outcomes but faces internal criticism for being a "pressure cooker.” Shiyan also emphasized the personal nature of educational choices and the high costs associated with private international schools, which can be prohibitive for startup founders.

2. WEIRD AI Teddy Bears: Jeremy and Shiyan debated AI integration into children's toys and education. Jeremy expressed concerns about AI potentially replacing difficult human conversations with always-pleasant AI interactions. They discussed the inherent biases in LLM models, which are trained on data from Western, Educated, Industrialized, Rich, and Democratic (WEIRD) societies. Embedding AI into education thus further propagates this world view to the next generation, which may not reflect current, local or parental cultural values or norms.

3. US Senate $32B AI Investment Bill: Jeremy and Shiyan discussed the proposed legislation advocating for a $32 billion annual investment in AI. Jeremy added that the US is already a leader in AI investment, with private investments expected to reach $82 billion next year, more than double that of China. She questioned the need for further government investment in an area already heavily funded by the private sector, and argued that government funds might be better spent on education, which offers broader societal benefits.

Jeremy and Shiyan also talked parental competition for "good jobs”, the challenges in balancing academic rigor and student well-being, and the future implications of AI in the workforce.

Be a part of Echelon X!

Join us at the startup conference Echelon X! Engage with over 10,000 of Asia's innovators and decision-makers on May 15-16, at the Singapore Expo. We have 30 exclusive complimentary tickets for our podcast listeners. Sign up and use the promo codes BRAVEPOD or ECXJEREMY to claim your free tickets now!

(01:43) Jeremy Au:

Good morning, Shiyan.

(01:45) Shiyan Koh:

Good morning to you. Hello from San Francisco.

(01:48) Jeremy Au:

Hello from Singapore. Our last podcast episode speaking about the prime minister transition and more specifically our points around education and our recommendations for some changes around tech pass program really got a lot of comments. Maybe we should talk a little bit about that.

(02:02) Shiyan Koh:

Hot button issues.

(02:03) Jeremy Au:

Yeah. So one piece of feedback was that sorry. So there were two pieces of feedback on the education side. One was that people were surprised that there was a pathway which you had mentioned. So they were very much educated by you, but that's one. And then secondly, of course there was a conversation about not just about how to get into the system, but more specifically whether parents should choose a Singapore local system. I was just kind of curious to hear about your thoughts about both of these dynamics.

(02:27) Shiyan Koh:

Yeah. I mean, I think we had spoken about the fact that you could go, but they just don't really advertise it. And understandably, ministry has enough stakeholders, citizens, kids and parents and PRs that they're not going out of their way to be like, Hey guys, come one, come all, right? Like that's kind of not, probably not in the conception of their job description. So the fact that it does exist and people are aware of it, I think hopefully is helpful to some people who might want to consider it as another option.

I think on the whether it's a good idea for you, I think educational decisions are really incredibly personal, and there's nothing that can get a group of parents riled up like talking about school. And I think that is true regardless of the country or system that you exist in. I'm here in San Francisco. I had dinner with a couple friends last night and big debate over public or private San Francisco, Oakland peninsula, there's like a huge, I think every parent wants to set their kids up for success, whatever your definition of success might be. And they want to feel like they're giving their kids every opportunity. There's just a question for people to answer, "What does that look like?" And if you think that you're only in Singapore temporarily and you want to go back and have your kids slot back into your home country school system, it can make a ton of sense. Stick with the curriculum. Have that seamless transition.

And I think for other folks, you might be optimizing for different things. You might want a more international experience. So you might choose different curriculum or have kids coming from lots of different places, not just primarily your home country. And I think there's always like a question of cost, which is what kind of started that conversation initially, which is that a lot of these private international schools are really expensive. And if you're a startup founder, it might just not be in your budget to pay 40-50K a year per kid while you're trying to get your business off the ground. so I don't want to tell people what they should do. I don't want to get into online flame wars, but I think thinking about what you're optimizing for probably helps to frame that decision. Thinking about what your budget is, and we just wanted to share that there is another option. We are not advocating the goodness or badness of that option relative to other options. Yeah, I was a little surprised by how much enthusiasm there was on this from the last episode. Curious what your thoughts were, Jeremy.

(04:29) Jeremy Au:

Well, I love what you said about how parents, that's the one topic we all have a very long debate over dinner together with, definitely feels like, I think it's not just about education system because it wouldn't just be a qualitative difference in terms of quantitative data, but it also felt like there's a little bit of a values perspective. I think some of it was like, is an international system better? Which is a little bit amorphous because in Singapore, there's so many different types of international schools, whereas, I think Singapore education system has a lot of like loaded language to it. I think the pros of the Singapore language was, people were very appreciative of that, the high academic rigor, but I think the shadow side of that was like, people were concerned about whether it was too much of a pressure cooker, or whether it was too academic-oriented.

I felt there was some subtext around the shadow side of Singapore. Again, you and I were discussing about, hey, these are the options of how to make it happen, and how do we increase optionality for the Singapore education system in light of the overall goals of integration and welcoming new immigrants, especially in the light of a falling birth rate and having to close down schools. But I think it was just interesting just to see that there was this piece again about the ethos of the Singapore system and how it should change. And I think that's a difficult topic.

(05:34) Shiyan Koh:

Yeah, to bring it back to parents though, I do feel like sometimes, parents are the source of some of the pressure, right? They took away exams from primary one and primary two, and then the parents are like, but if there's no exam, how am I going to know how my kid is doing? And then, of course, for profit tution centers are like, "Hey, pay us. We'll test your kid. We'll tell you how they're doing." And so I think there is an interesting like, as a parent, I often ask myself, what is it that really matters at the age that this kid is at? A little bit like, don't freak out about the, I don't know, keeping up with the Joneses or competing on something that you don't actually need to compete on. Because if we took a step back, as a country, shouldn't we want everyone to be as literate and as numerate as possible? It's not a zero sum game, right? We would all be happy if more people developed more competencies. It's not that there's like a limited set of people who can acquire knowledge. We should actually want everyone to be competent in as many things as possible.

And I think the argument that has been made to me is, "Okay, but there's a limited number of good jobs. So there's always going to be some competition at the end." And I was thinking about that and I was like, is that true? I don't know if that's true. There's probably a limited number of jobs at Goldman, but Goldman is merely one small slice of the possibility that the world holds. And I think there is something to be said about a lot of the competitive nature. The Singapore school system, to me, sometimes feels like an output of the older paradigm, like from third world to first, if you're like, "Okay, I'm EDB and I brought in these jobs. And there's only X number of jobs."

Then, I need to compete to go get those jobs because everything else is way worse but I don't think that's the case anymore, have made the transition from third world to first. We are in a global economy. We are encouraging people to innovate, start companies, and there's no limit to that. And I do think there's a really interesting question about What is the purpose of competition? And like, when does it serve us? And when does it not serve us?

(07:20) Jeremy Au:

I think it's actually a pretty good question, which is that, is there a limited number of good jobs? And what I would say in response to that is, there's a limited number of guild jobs. So now, there are only certain jobs or professions where there are quotas, right? So there's a certain number of folks that are being trained. And I would say, for example, maybe look at the professions that require a high academy level rigor, right? So you look at the lawyers, heading to doctors. Some of the scholarships are available out there. I think those are limited by quota by a certain level, in terms of the intake system. So I think that's maybe one classic definition of what a good job has been. And I think, there's a much larger set of jobs that are available in the private sector as well. So that's really true today. And I think that's going to be more true in the next 20, 30 50 years when our kids grow up.

(08:02) Shiyan Koh:

Yeah, but I think even the definition of job has expanded, which is that with the internet, your job, you could be serving anyone globally and you're not limited to some corporate "headquartered here's" job. And if you think about a Singaporean, the average Singaporean is more well educated, has more access to education than the average person in region. And so from a competitive perspective, it's actually better positioned if you want to think in that paradigm.

(08:28) Jeremy Au:

I think that's a very fair point you have , which is, the Singapore education system has a lot of angst and I think frustration internally within Singapore, I would say, that comes true but from the outside, across the region, in the US, Singapore education system is the envy of many parents in terms of cost and outcomes. So I think it's a very interesting dichotomy. It's like the moment you fly, take the flight out, and everyone's very positive about the Singapore education system, especially from a policy maker's perspective as well. But then once you're inside the borders, then people can be quite unhappy. Some people.

(08:56) Shiyan Koh:

Yeah. I don't know how to resolve that, other than, I get it, right? I feel like when I was growing up, I didn't know that there were so many jobs. I thought there were maybe like five jobs in the world.

(09:04) Jeremy Au:

Now, I want to know what those five are.

(09:05) Shiyan Koh:

A doctor, lawyer, engineer, a nebulous business person, but like I didn't really know what that was and then when I went to college, you know when you go to their first career fair and there's all of these companies with their little tables and you're like, "What's this company? I've never heard of this company. Who are all these people? What do they do?" I didn't know what a management consultant was. I didn't know what an investment banker was. I didn't know there were so many people involved in the semiconductor value chain. I mean, there's a million things I didn't know, right?

I was very dense, but now, everyone grows up with the internet. So they know a lot more than I did. That sense of like possibility and that openness to it hopefully can help people calm down a little bit About the competition level.

(09:42) Jeremy Au:

Yeah, for sure. And speaking of education and competition, I think one of the big things that we saw was that, in the US, there's a big debate around the AI bill and some of the recommendations having a policy level. Could you share a little bit more about that, Shiyan?

(09:55) Shiyan Koh:

Oh, I think it just came out this week. Senator Chuck Schumer put out like a proposed bill. Obviously, there's a lot of conversation about how AI is like a national security in the sort of competition. And so there's a bunch of stuff around the regulatory, but I think the thing that caught people's eye was that, they also propose spending 30 odd billion annually to invest in AI. And so, this is kind of a little bit in line with Biden's new industrial policy, you know, the inflation reduction act put a lot of investment into semiconductors and EVs, gave a bunch of money to Intel, TSMC, to come set up fabs and things like that.

I think it's an interesting choice because, why should the government invest in AI? There's already so much private sector dollars flowing into AI. My view is that the government should invest in things that private sector won't invest in. That's like the job of government, which is like education. Things that have positive externalities for the country that private sector is not incented to do. Putting more dollars in an already crowded space and so I thought that was like an interesting thing.

I don't know if people are familiar, but education in the US is funded in a very bizarro way. So you can think of America as like a country, but education is actually funded in a really local way. It's funded by property tax, and so, depending on where you live, the amount of funding going into education will be super different. And you can also imagine if you live in a place with really expensive houses, you have more money. And if you live in a place with cheaper houses, you have less money.

There's no national curriculum. So what you learn in Arkansas could be wildly different from Massachusetts, and that just seems like a very, like as a Singaporean, I think that seems really counterintuitive to me, but like conceptually as a government, you should be like, how do I increase the marginal productivity of all my citizens in the most dollar efficient way? There's historical reasons, right? I mean, it's really hard to rule a country with 300 million people. And so you do need local control, but, it is a system that, to me, seems to exacerbate inequality, because if you're poor, you probably don't live in a place with multimillion dollar houses. And so, if you think about it, like a pure country competitiveness perspective, like to me, I'd rather the government spent 30 billion on education and improving the quality of that instead of trying to fund more AI.

(11:50) Jeremy Au:

Yeah, I mean, I think this is something that always gets you and I very passionate, especially education and AI as well. But I agree with you. I think the US system, when I was studying at UC Berkeley, I think the thing that really struck me was the difference in schools between Berkeley and Oakland which are very close to each other, they were literally right next to each other. And I think the quality of standard in the public school system was like night and day when they effectively share a border. But like you said, it's primarily because of the housing prices, each different neighborhood, right?

And so there was almost like a line, in the neighborhood where it was like on one side, it has its own police department, has its very fancy homes and so forth. And obviously, I would like peg to it also to university. And then you get Oakland, which was kind of like known for crime and safety issues. And it was just a really weird thing to have actually feel like that demarcation of that boundary because it's like you could walk a cycle across it and then there would be like crossover crime. So people would be like, okay, you don't want to live on this edge of Berkeley close to Oakland because you're exposed to burglaries and other kind of crime from the Oakland side.

And I thought it was just a really bizarre thing. And our home had a bunch of folks try to enter and so forth and people get caught and things like but I think in retrospect, it felt strange to have that thing because the kids on one side and you're born on one side of demarcation is the same as the kids born on this side. But the spend, like you said, on education is dramatically different. And the outcomes are different. And I think there's this goal, I think, at the university level to try to equalize for that, where, they say, hey, if we come from this district, we can have a lower academic entrance requirements. And we're taking the more holistic point of view.

(13:19) Shiyan Koh:

They don't say it that way.

(13:20) Jeremy Au:

Sorry. That’s a nicer way of saying that, but that's how I kind of took the parlance of that, right? Because they look at your, they look for the best GPA within your school, right? So some schools have better GPAs and some other schools have, on average, lower GPAs. But it's a way to equalize some of those outcomes and create some of that diversity.

(13:34) Shiyan Koh:

Yeah, it's crazy. And I mean, I have friends whose kids are in public school. The advice the teacher gave them was like, oh, you should probably supplement outside of school because your kid is going to be bored. So I don't know. It seems like suboptimal, but it's a really hard problem, right? I don't think anyone has this perfectly right I think if, we had to be the policymakers, we would struggle with these trade offs as well.

(13:53) Jeremy Au:

Yeah, I think that one reason why does this, tension as always, is that I think people from Singapore are comparing the average Singapore school versus the best American schools, right? Because the best American school seems to produce like Elon Musk, Bill Gates, Steve Jobs, right?

(14:08) Shiyan Koh:

Elon Musk is from South Africa.

(14:09) Jeremy Au:

Well, I'm just saying that's what parents are saying, right? There's a saying Hey, like the American system produces these great, wonderful entrepreneurs. So Singapore doesn't do that. I'm just saying that's how I'm translating the dinner conversation around the table.

(14:22) Shiyan Koh:

Yeah. I mean, I think that's a hard comp, which is I always tell people that the US is a variance maximization machine and Singapore is a variance minimization machine. And I think if you're average person, like Singapore does better for you because the average quality is higher at a lower cost. I think if you're truly an outlier, like you should be able to succeed in Singapore, but you might also even want a bigger market. You want even more crazy stuff to hang out with. And then you'll find yourself in Silicon Valley, but that's true in bigger markets too.

I was in Japan earlier and they were saying that, Oh yeah, there's a bunch of Japanese founders who'd be like they made it in Japan, they really want to go test their skills in the US. So it's not just baseball. It's also in technology entrepreneurship. It's like, okay, show his leg. I'm going to the Major League Baseball, guys.

(15:05) Jeremy Au:

Yeah, I mean, I think a big part of it is that, it's really fact that America is really ahead on some dimensions, right? And I think people are comparing versus that, so I think, for example, a lot of people are concerned or frustrated about the Singapore exchange, about some of the dynamics, I would say that the US stock exchange is doing very well and sucking up the liquidity from every stock exchange in the world, except for the Chinese one because of the decoupling situation. So it's more of a feature of the US being a very strong competitor, but it's an interesting debate that's happening in the world today.

(15:34) Shiyan Koh: Yeah. And I think it's a whole package. I think that both systems, I mean, the feedback from the dinner I was at yesterday was, they were like, your kid has homework? Your kid can read? They were excited. They were like, we wish rigor in our system. They're like, you don't get to complain. Be quiet. It's always good to be able to step back and have a little bit of context.

(15:51) Jeremy Au:

Rotating back to the AI piece as well, I think it was interesting because the bill confronted up as emergency AI spending because they felt like they were behind. And it's interesting because the US is by far the number one investor in AI right now. And then probably if you look at global flows as well, AI spending, probably US-linked investments probably again, number one US dollar denominated kind of funds is currently number one in Europe and probably in Asia as well. So I think it's a lot of money that's already being deployed.

In fact, I think I'm just quoting from an article here. Goldman Sachs estimates that private investment in AI will total to 82 billion next year, more than twice as much in China. So I think adding another 32 bill feels like a big push.

(16:31) Shiyan Koh:

Yeah, but I mean, I would say that the pace of development in AI has been truly incredible, right? they just announced GPT 4 Omni. Crazy. It's like Her, right? Like the movie. If you've read Diamond Age, like having that sort of agent talk and interact with you without the stuff of sci-fi, it's amazing. I think there's so many more industries that are going to be disrupted. You're like, oh, like Chegg, right? Chegg stock price dropped when GPT 3 was released. Then now if you think about everything else, anything that's language related, Duolingo. Do I need to pay for Duolingo if GPT can do it for me? And collecting more and more examples of people being like, Oh I didn't even need my analyst to do this. I just uploaded the data, ask questions of the GPT, got it back. And it was like quick and easy. So, I think trying to think about there is the money being spent on pushing the boundaries of what the technology can do, but then there is the question of who's going to get the value capture once that stuff is actually put into production and applied to very specific industry use cases. I think the US' worry is that they might put out all this great technology, but a lot of the know-how around the hard arts of manufacturing and supply chain were outsourced in that first wave of globalization. And it's hard to bring back. So there's a delay on the TSMC plant in Arizona, was it? Because they didn't have enough skilled workers to actually build that type of high tech fab. They wanted to bring in their own Taiwanese workers and the unions were like, no and there's been a bunch of interviews I've heard with people who move their manufacturing offshore and then they were asked like, hey if you brought it back on shore, is that possible they're like we just don't have the labor? We don't have that labor ability anymore because you stop doing something and then you lose all of that vertical knowledge so, they're definitely really worried.

(18:10) Jeremy Au:

Yeah, I think almost every country that's really in this semiconductor and chip, which is the hardware side of this AI revolution, I think they're busy investing in education for workers because semiconductor is not an easy profession and it requires not just engineering, but a certain specific stack around manufacturing and process engineering, plus everything to do with the semiconductor side, right?

And then you're training them for not a large number of companies. Actually, it's a pretty specialized field as well. So you really need a lot of industry government partnership to make it happen because basically, government is basically saying we're making a bet that we're going to revamp our education curriculum. And then there's the semiconductor industry that's going to be picking up these graduates in 10, 20 years. So I think you see Vietnam is doing a big push on semiconductor engineering. Obviously, China has done a huge push for quite some time now as well. And then I think we're starting to see Japan and Korea, they've always actually had a lot of adjacent fields, but they're suddenly specializing in that more. So I think it's interesting to see that as everybody tries to replicate the Taiwan stack on semiconductor, I think you're starting to see that China, Korea, Japan, and Vietnam seem to be doing a more cohesive, whole of government, plus industry partnership, work to make that happen. Whereas, I think it's actually a little bit currently more difficult than other places that don't have the education push, but also the kind of cohesive push as well.

(19:23) Shiyan Koh:

Yeah, but the secret sauce the US has is immigration right? That's the way

(19:27) Jeremy Au:

That's true, I mean, that's where Elon Musk came from.

(19:29) Shiyan Koh:

What is it? PhDs. Yeah. But also like PhD programs, STEM PhD programs are overwhelmingly, immigrants, and I think that's probably slowing. You probably have fewer Chinese and Indian PhDs than you did 10, 15 years ago. That's how they counteract it. It's you want to work with the latest stuff. You want to work with in the deepest capital market. You gotta be here, like a different strategy to how do we get the intelligence to work on the problems.

(19:52) Jeremy Au:

Yeah, I guess what the strategy is eventually, find a top 1% of all these countries, Vietnam, Korea, Japan, China, semiconductor skip the top. But, I think that's actually a fair point that you had, which is that it circles back to our education point, which is that, I think the reason why around the dinner table, people feel like there's Elon Musk and Bill Gates is that, their university system is actually very open.

I myself studied in the US, right? I did well. The Singapore system, as well as American SATs, and then I got an opportunity to go to an American University. I think the K-12 system of America is what we're saying has its own structural issues that we were just talking about, right? That's the equivalent of Primary 1 to Secondary 4 and Junior College in Singapore. I think that's difficult in the US, but I'll say that the American universities are doing very well.

(20:33) Shiyan Koh:

Yeah. They attract like great researchers. It's a desirable place to go.

(20:38) Jeremy Au:

Yeah. So I guess the question I have in terms of AI is like, would you let your kid play with a Her voice assistant? Because I was thinking about it. I was like, Hey, it's better than a screen supposedly because it's a TV. I don't know about a tablet. So we're having that debate at home. And I was like, would I give my kid a voice assistant? Because technically it is interactive. You can answer all the questions about science they may have, but I don't know. What do you think, Shiyan?

(20:59) Shiyan Koh:

I think so. It's if you felt confidence that the assistant wasn't going to hallucinate, was giving reasonable answers and I think the feature I would ask for is that it would give a summary readout every day of what they discussed, then I would. But you know, people already kind of do that. Kids talk to Alexa. They talk to Google. All the home automation stuff. I see kids do it all the time. Okay, google, what's the weather going to be like? What are the scores? When is sunset? People do very rudimentary versions of that. And now, you can have these more multi turn conversations. So yeah, I think so, but this is the problem, which is the people who get the most out of it are the people who know what good looks like.

Yeah, so I don't know how good it is at if you don't have a good framework for learning, and you're just asking questions, it's like hard for you to build a mental model, then, I don't know whether that helps you scaffold as well, but I think once you have a good framework for like, how do I learn new things than having someone who is infinitely patient, can access the entire internet, can explain things in different ways. Sounds pretty awesome. Why? What are you scared of?

(21:52) Jeremy Au:

Big tech on the other side of the call. I don't know. I mean, I think now it's a pretty early in the AI field, so I think it is pretty generic. I don't think there's not too many levers in the system. It's like the good old days of the early games of video games where the video games were fun, but not overly engineered to really hook you too hard. And now I can't play a video game. It's just that there's too much stuff that's awesome about video games, but are actually based on psychology, right? And persuasion and all these lottery dynamics inside the games that I love, right? In other words, I think there's a difference between unrefined, sugar versus refined sugar.

(22:25) Shiyan Koh:

Oh, interesting. Yeah. So you're worried that they kind of get sucked in too easily, if somebody had a more nefarious purpose. They wouldn't know what was happening.

(22:34) Jeremy Au:

I think that there isn't necessarily a nefarious moment right now because it's like the early days of internet. It's just everybody is a geek online, but I do believe that over the future, I wouldn't use the word nefarious, but probably commercial. I think there's a lot of unstated systemic biases that push a system towards so I would believe that as we see for voice assistants right now, they're just trying to be helpful and try to just address the problem, but I can imagine that over time, just like every other industry that's consumer-oriented, it's going to be fine tuned towards being more engaging, to more of a hook, repetition, to upsell over time.

Like I said, I don't believe it's the current perspective. I don't believe there's any specific player's perspective. I'm just saying that if you look at the video game industry, it used to be, you buy one game, you enjoy for a certain amount of time. Now today, we've moved on to micro transactions, right? Subscriptions, loot boxes, which is kind of like a gambling mechanics. Is this a natural piece of the industrial sentence over time?

(23:26) Shiyan Koh:

So maybe, before you let them do that, you have to teach them psychology, to look out for these gachas, right? I don't know. When I was growing up, my dad always said the same thing over and over again, which there's no such thing as a free lunch.If it t looks like a free lunch, something's wrong. So figure out what's wrong. and so I think that does breed a natural suspicion when somebody repeats it to you over and over again. and so maybe that's what we actually need to teach people is it's the same thing as like teaching people to be careful on the internet for phishers or scammers, learning about the things to watch, just like we teach them how to cross the street. You have to teach your kids how to beware of these behavioral hooks. We tell Nir Eyal to write a new book.

(24:02) Jeremy Au:

About AI voice assistants?

(24:03) Shiyan Koh:

No, about teaching how do you help your kids ward these things off.

(24:08) Jeremy Au:

Oh, you know what? That is the hook. I would say that is a much better, that has a good version of it is the application, right? The practical version of how to teach. I think for me, what I think about is that certain industries, for example, I was reading about a processed food industry, I think a lot of people say hey, I can't help but be obese. They feel powerless that they're overweight. And that was an interesting dynamic because, if we go back 100 years ago, there was no such thing as people who are obese, right? It's just an impossible thing to have, but because of the surplus of food, that's one side, but the processed food industry.

And I read this great book that I can't recall the name of right now, but I think it's like salt, sugar, fat, which is about the processed food industry and some of the mechanics of the decision making they had. And I think the big realization I had was, if you live in certain food environments, where most of the food is processed food, you look like you're eating candy bar or something like that but actually on the other end it's like a full football team. You know what I mean by that is you think you're playing football with the one piece of a granola bar, some, candy bar, but actually on the other side is a huge team, right? It's a psychologist, the food scientists the MBAs, the accountants, like there's a huge, huge, huge team.

And this candy bar with it's highly tested rapper, with his positioning and chill channel partnership and distribution that creates that environment. And so I think that's something that I'm kind of mindful of right now, which is that it feels like, obviously as a parent, you have to educate yourself and the kid about these things, but I'm just saying there is a structure. It's not a one to one, tug of war.

(25:29) Shiyan Koh:

Oof. I feel like Capitalism, man, doing its job.

(25:31) Jeremy Au:

Yeah. Capitalism is doing its job. It's got to make more capital, okay, you know what I would say, if I was talking about a kid and an AI voice assistant, I think the biggest structural bias that AI voice assistants will have is not the upsell and so forth. I think the biggest thing they will say is that AI is human. It's the biggest structural bias that AI will have in terms of voice assistants. That talking to an AI is equivalent to talking to a human. I think that's the biggest push from a user interface perspective. There's an explicit goal. The implicit dynamic is an equivalence. And I think that's going to be a very interesting conversation, which is talking to your mom or dad about the birds and the bees equivalent to talking to an AI about the birds and the bees, like I said, from information, knowledge-utility basis.

(26:09) Shiyan Koh:

Yeah. It's possible your AI will know more. I don't know, it's kind of like, you know when you call customer support? Sometimes, I don't really actually want to talk to a human. It's like faster for me to process my own issue through the flow. So I don't know. AI is probably better than humans at some things.

(26:24) Jeremy Au:

I think they're going to be better at always having a pleasant conversation, which is why I think there's a lot of debate around AI teddy bears because people are putting AI into children's toys. And there's a big debate. I think it's going to be a witch hunt about AI teddy bears soon, but that's my gut feeling as a, not knowing humans as well. It'll be a witch hunt soon. But I think the concern is real, which is that, it goes back to the bubble, social media bubbles that were all in the algorithm and all that stuff is that I think what AI is really good at is that you no longer need to have a difficult conversation, cause you can always have a perfect one with AI instead. And I think there's going to be the interesting substitution effect that AI will have. That'll be my concern about letting kids use an AI teddy bear.

(27:00) Shiyan Koh:

I'm going to have to noodle on that one but I mean, maybe the positive version of that is that more people will learn to communicate pleasantly.

(27:07) Jeremy Au:

I hope so.

(27:08) Shiyan Koh:

Because, I think, because one issue is that, why do you have unpleasant conversations? Sometimes there's just, okay, it's unavoidable because there's something bad happened, but sometimes it's the other person, or you are not good at communicating. So there's like a crossed wire situation. You don't have the skills to navigate that conversation. And so maybe if people practice more, then they get the skill of doing it. Why are all the chatbots in the GPT store, like the most popular ones are like dating right? Like the girlfriend bot, the positive spin on this is that people are actually practicing their social skills and getting better so that when they interact with real people, they have that.

(27:40) Jeremy Au:

Yeah. Actually, that reminds me of something. There's another bias of the AI models, which is, I agree with you that it helps you learn how to speak in a certain way, because I already had one friend, not a friend, I'll say a mentee of mine, university student who didn't know how to break up with a girl. So he used the AI to generate a script, but what's interesting is that the script, it actually reflects a certain bias in the training data of these models because these models are primarily built in the West, in the US, so they reflect the training data of Reddit, YouTube, Facebook, etc.

And so, the implicit, another bias of communication style is that they're going to be weird, right? W E I R D which is how they define it. It's like Western. Educated, Industrialized, Rich, and Democratic. Those are going to be the structural biases of that language that the kids are going to learn from AI.

(28:24) Shiyan Koh:

That's why people are building their own local LLMs, right? Trained on different languages and with different training sets.

(28:30) Jeremy Au:

So you'd be like, I don't want the Western AI teddy bear. I want a Confucian.

(28:35) Shiyan Koh:

Yeah!

(28:36) Jeremy Au:

You heard of filial piety?

(28:38) Shiyan Koh:

Yeah, yeah, yeah. I only want the one that has di z igui sayings. Okay? chengyus! gonna recite chengyus every day.

(28:48) Jeremy Au:

Okay. On that note I think we'll tie things off and I think let's noodle the idea of AI teddy bears for the future. We'll just make it like a request for startup of the day. We're like, we put aside our qualms and our concerns as parents, if we need a request for startups, fireproof AI teddy bears.

(29:02) Shiyan Koh:

Oh gosh. Okay. This is not where I thought the conversation was going, but this was fun anyway. So thanks, Jeremy.

(29:07) Jeremy Au:

Awesome. All right. See you, Shiyan.

(29:08) Shiyan Koh:

See ya.