Mar 8, 2025 ยท 2025 #9. Read the transcript grouped by speaker, inspect word-level timecodes, and optionally turn subtitles on for direct video playback
Edit labels for this show, save them in this browser, or download a JSON override for the production folder.
Transcript Playback
I Built an App for Teens
Human Transcript
Timed transcript
Blocks are grouped by speaker for readability. Expand a block to inspect word-level timing.
Speaker 2
Hello, everybody. It's Saturday, March the 8th, 2025. We're moving fast through the year. And of course, AI is moving fast. Remarkable news every week. Alex Kantrowitz wrote... A substack this week acknowledging that AI can do his job after all. TechCrunch reports that a quarter of startups in YC's current cohort have a code base that are almost entirely AI generated. And Keith Teer, as always, as someone on the front end of progress when it comes to AI, has developed a That Was the Week editorial to talk about the speed of AI, how it's changing everything. He acknowledges this week, I built an app for teens and I did not write any code. The rise of what he calls autonomy touched on this last week, Keith, uh, but this week you write about it. Um, what do you mean by autonomy? I looked it up and, uh, of course the AI overview is the ability to make independent decisions and act according to one's own values and beliefs. What does AI have to do with autonomy, particularly in, uh, the way in which you built an app for teens?
Yeah. Well, I chose the word autonomy from the point of view of me. In other words, I was able to do something without having the skill to do it. And in that sense, I added a new skill without having to learn the skill. And what I did is I, you know, I've got three boys, as you know, they've all got the same challenges any teens have. In my case, I have one teen and two 28. 22 and 23 year old, they've all got their challenges. And I wanted to see if AI could build an iPhone app. So I decided to build an app for teens to reflect. There was a suicide here in Palo Alto this week. It's a common occurrence because the high school kids are under a lot of pressure and one high school kid from a local school threw himself under a train. So it's a big topic in Palo Alto, teen challenges and teen suicides as well, all driven by life, not by using Instagram or anything. And so I decided to write an app that a teen could use to reflect on themselves and get help. And, you know, It worked. It took me about a week, and I now have an iPhone app that, if I wanted to, I could submit to the App Store. It's called Reflect.
Yeah, and I'm quoting you from your editorial. It's a real app that has built-in support from open AI, anthropic perplexity, and deep seek. The app uses the AIs as a buddy, a coach, a therapist, and a counselor to the teen user. While I was reading this, Keith, I wondered, I mean, what does this mean for... coding does it really mean the end of coding that coding now is just like language it becomes accessible to everybody and we don't even use this word anymore yeah i i
think a lot of things are are at end of life um mg siegler this week wrote a article titled it's the end of the web as we know it. We've heard that one before, Keith. We've read that one.
But, you know, the end of Fill in the blank. And coding is one of them. Now, of course, it isn't the end of coding. It's probably the beginning of coding. There's going to be 100 to a million times more code in the future than there was in the past because it's becoming so easy to do it. It's a little bit like when Adobe PageMaker came out. It wasn't the end of design. It was the beginning of design. So it's the end and the beginning. It's both. It's both at the same time.
But what it does is it enables, and you're still a bit more sophisticated than most people, but it enables all of us to become coders or at least to produce stuff out of code that we would like to do but we weren't able to do in the past because we didn't have a knowledge of coding.
Yeah, code simply becomes a means to an end. It really isn't about the coding anymore. It's about the software. So now, if you can imagine something you think the world would like, you can probably bring it to life without coding. And that's what I did. I mean, it blows me away that I was able to do it, but I was. And anyone can do it. If I can do it, anyone can do it.
Well, that's not entirely true. But if you can do it, anyone can learn to do it. I was particularly struck by this TechCrunch headline. A quarter of startups at YC's current cohort of code bases that are almost entirely AI-generated. So again, the question is, what are these startups doing? How does YC decide which startups to invest in and which to ignore?
It's the same old story. It comes down to the strength of the team and the idea. And in the case of YC, they focus on traction during their cohort. So it's that as well. And it used to be that you had to have a technical founder. That is gonna become less and less true. Of course, you have to have a founder that's familiar with the technical ecosystem. Otherwise they couldn't tell the AI what to do they couldn't recognize whether I was doing a good job or a bad job. But actually writing the code is now less of a blockage, something that would normally, my son Dylan told me if he was given my app to build, and he is a coder, it would take him several months to get to the point.
And you and I have talked about this before. You've got a rather sad looking teen on the front of your editorial this week. I don't think this is AI generated.
I wonder whether it might have been a version of Dylan as a trained coder. He studied undergraduate. Isn't he the kind of guy who's going to get eaten up by this revolution?
Well, he could get empowered by it if he's open to learning how to include AI as a tool within his armory. I would probably say there's going to be a lot more engineers in future than in the past, not a lot less, but they're going to be doing a lot more individually because AI will increase their productivity massively.
So does this empower people like Dylan to become entrepreneurs, startup people? I mean, should you not be studying? I mean, your youngest son is about to go to college. Would you discourage him learning code? I know he's thinking of majoring in computer science.
No, I think knowing how to code is still important because it lets you understand what's possible and what to ask for and what not to ask for. So I still think it's important. It will eventually not be, but we're not there yet. Funnily enough, the strongest of my three sons this week helping me was Liam, who left high school in 10th grade. He gave product feedback. He found bugs. He described why screens were no good. He acted as a product manager, and he has no coding skills at all, but he wasn't intimidated to not have an opinion, therefore. So it's actually quite empowering. if you embrace it.
Well, another of your featured stories this week is actually a podcast show, the Ezra Klein Show, one of the best podcasts, certainly most popular and influential podcasts online, the New York Times podcast, is about the government knowing that AGI is around the corner. It features Ben Buchanan, who was Biden's AI guy. I was profoundly unimpressed with Buchanan. I don't know what you made. Did you listen to the interview? Did you read it?
And a bureaucrat. I mean, it's astonishing that if Buchanan indeed was the face... of Biden's AI, it's not surprising that he lost the election. It's not surprising that so many people are critical of him. I mean, this guy was so profoundly uninspiring. And I'm certainly no fan of Trump, but the AI people around Trump understand it profoundly more than Buchanan are willing to take risks.
Yeah, because, I mean, Trump knows what he doesn't know. And weirdly for Trump, he doesn't really care to have fine-grained control. So he's prepared to put technical people in positions of authority and give them the reins and benefit, bask in the glory of their successes. He cares more about publicity and media than he does about control.
Yeah, I mean, I don't want to turn this into I shouldn't have brought up Trump, but one of the things that struck me about Buchanan was that Klein told him, well, AJI is coming, and Buchanan said, well, it probably is, but we hadn't given it that much thought because we knew it wasn't coming in the next couple of years and we'd be out of power by then. I mean, it's just astonishingly short-sighted and unambitious and uncreative. I just don't know where they find these kind of people.
I mean, it's worth uninteresting. It's a disgrace to the Biden administration. It just reflects how... unambitious they were and how clueless they were. I don't know if it reflects on Biden or the people around Biden. Anyway, moving on, enough bashing of Biden. Another interesting piece when it comes to AI and about how it's changing the world is something from Dworkish Patel, what fully automated firms will look like. What does he say about that? What exactly is a fully automated firm?
i think i think it's a big deal the idea of a fully automated firm um i was discussing with a colleague this week who runs and he runs a company called cadence which does um software for hybrid workplaces and they started life during covid as a how to manage people and places during covid and uh The penny dropped at some point that the move away from the office was the first step in a move towards an automated office, an automated workforce. That software is aware of people and places, not necessarily places you own. It could be a coffee bar or a hotel meeting room as well. And that the future company will not look like the company of the past. So the whole move, which Google is in the lead of and Musk is in the lead of to go back to the office, probably can't work. History's already looking forward to a place where a workplace is an anachronism. And this article by Dwarkesh Patel envisages what that might, what the end game might look like. It's an interesting article. It isn't like totally science fiction. It brings some big things. overarching categories, if you will, to the evolution of the workplace, how to scale a workplace, how to merge pre-autonomous and autonomous into a timeline. So it's very thoughtful. It's not very long and it therefore isn't exhaustive, but it's a good introduction to the idea.
Yeah, Google doesn't need me to defend it, but this week, and you don't have this in the newsletter, but they announced a 3D product for communicating online, a 3D meeting product. Who did they develop it in alliance with?
Actually, I didn't catch that, so I don't know about it.
Words and timings
Actually,Ididn'tcatchthat,soIdon'tknowaboutit.
Speaker 2
But I don't think Google's... under any illusions that the future of the office is also in some ways virtual, and we will see on that. Meanwhile, there's another interesting piece that you linked to by Mika Sifri, very smart guy on the left, very critical of Trump, but at the same time, I think no great fan of Biden. On his influential Substack show called The Connector, He has a piece called Breaking Government in Order to Fix It. He's critical of, as I said, he's critical of Trump and Musk. But at the same time, he recognizes that AI and government need to be, that government needs to be profoundly changed by technology.
Yeah, this fits a little bit with some themes of prior weeks, which is why isn't somebody describing good government it's almost like there's only two points of view. Stick with the way government is bloated, expensive, not very effective. or fire everyone. There isn't a voice that's saying modernize it. And I thought... But there are voices, they're just not... Mainstream, yeah.
Right. Well, they're mainstream. They're just not getting in the news, given the hysteria. I mean, Ezra Klein and Derek Thompson are coming out with a book in a couple of weeks called Abundance, which focuses on this too. So this is not a story that people are completely ignoring.
and I thought it was striking that he was going against type in writing it, and therefore I found it very worthy, which is he's just independently trying to think through what needs to be broken and what fixing it means. And he's very critical of Musk for not having that dimension to his thinking, which is weird for a guy who's built all things Musk has built. Why wouldn't he want to build the SpaceX of government or the Tesla of government as opposed to the Twitter of government? He seems to have only taken one of his lessons into government instead of all of them.
Well, it's the ideology of Trump, and that's probably how he sold the whole thing to Trump in the first place, that he could destroy it. I don't think Trump has any great interest in rebuilding it. One of the things I, one of the people that Sifri touches on, I think, and certainly Ezra Klein and Derek Thompson talk about in their new book, Abundance, is Jennifer Barber. Palka, who was Obama's digital person who speaks about automating government. So I have a feeling, or I hope, that the Democrats will be a little more assertive than people like Ben Buchanan and actually recognize that they need to become the party of progress rather than reaction.
You mean like bashing billionaires? The whole framing of society as billionaires versus the rest is headline news, but not practical politics. In practical politics, you've got to build things. And to build things, you need technology and progress is what comes from that.
I think that that debate will emerge. I think Gavin Newsom this week made it clear that he's no great fan of woke. He's clearly throwing himself into the next round of and probably running for president. So I think this is going to get fought out in public. Meanwhile, another podcast that you point to is with Dario Amadai, the CEO of Anthropic, one of the more respected AICI CEOs who was on the New York Times' Hard Fork. What does Amadai say? I haven't actually listened to this. I need to listen to him.
Well, he... He's always been one of those people that gets accused of regulatory capture because he acknowledges the potential dangers of AI. Sam Altman is another one.
Yeah, Modi comes across much, much more solid, but he is in that camp. And this is, by the way, a video interview, which I put in the newsletter as a video, so you can listen, you can watch it or listen to it. But he's basically focused on preparing everyone, government and human beings, for what AI's success will look like. so that they're not surprised by it. That's more or less the theme. And he's not a doomster. He's more of a be aware of where this is going and be aware of where you want it to go to make sure that is where it goes.
Yeah, and of course, it's at a very abstract level. He came out with a long piece a few months ago outlining this as well. And it won't be the last we hear. of Dario Amadai. I mean, I use Anthropics Claude. I think it's excellent. Meanwhile, Vinod Khosla, one of Silicon Valley's most respected, prescient, successful VCs, told the information that most AI investments will lose money as the market enters a greed cycle. I'm not sure whether that... whether I ever lost that greed cycle. What is Vinod saying that's surprising or newsworthy? A lot of people have been warning about this for months, if not years, Keith.
Well, I think this is one of those where you have to read between the lines. The headline isn't necessarily the purpose. Vinod, and by the way, he was joined by Rulof Buter from Sequoia, kind of angry. that other people are getting in on AI investments. In the case of Rulof, he focused on special purpose vehicles where large amounts of money pile into investments that he is in. In the case of Vinod, he's having similar set of experiences. And what they're really saying is stay away from this pie because it's my pie.
That is my interpretation. Is that rather unkind, Keith? I don't think so. In the case of Ruloff, he was explicit. In the case of Vinod, he's more subtle. But I think they're both saying... By the way, Vinod's wrong. This is not the greed phase.
Yeah, but we're very early in the greed cycle. There's a lot going to happen over the next few years. We're not looking at a crash any time soon. And therefore, he's really trying to carve out the turf, saying he's trying to scare people off at the point where it's not really...
Are people really that? Is Vinod really that... conspiratorial that he's making these public announcements to scare people off investing in an area that he wants to invest?
I actually don't. I think the idea that you should trust anybody is wrong anyway. You definitely shouldn't trust anything I say, because just like everyone, I have self-interest. You listen, decide what you agree with, what you don't agree with, and form your own opinions. That's the right way to do it anyway. The whole idea of trust is outsourcing thinking to someone else. Well, no one should ever do that.
Changing the subject slightly, my Keen on America interview of the week was with Minashki Ahmed, who has a new book called Indian Genius, The Meteoric Rise of Indians in America. And one of the people she focuses on or she focused on in our conversation was your friend Vinod Khosla. She talks about something called an $8 club. Guys like Vinod came to America with $8 in their pocket. And they built themselves up that way. Of course, not everyone's quite as successful as Vinod Kostler. But Microsoft CEO, Google CEO, many of the top CEOs and investors in the Valley are Indian. What do you make of it, Keith? Is there something called Indian genius when it comes to technological innovation and success? And what is it about the Indians that distinguish them from other groups?
No, there's lots. Russian, you know, Sergey. Almost any nationality, because the Valley is, you know, above everything else, a melting pot of different cultures, who are recent immigrants as well, by the way, the Indians, many Indians still speak with Indian accents, many Russians still speak with Russian accents. It's not three generations, it's the first generation. And the valley is full of them, I meet them all the time. So I don't think there's anything exceptional, but I do think every one of those cultures has exceptional people, including Indians. And these are a fantastic group of people that she's focused on. Is it a she?
I think that might be an older idea. I think modern entrants tend to come out of Indian technical schools, which are superb, and have gone through college and probably have more than $8. But their parents probably were the $8 club. And I think if you come from Bulgaria, Romania, Poland, you might be in the $8 club still. It depends on the specifics. But the Valley rewards trying. It is a unique culture in that way. It rewards trying and it ignores failure. Not forever, but for a long time. And therefore, you can come here and do big things. You have to have the temerity put yourself out there, which is all about confidence, and then the ability to deliver. But if you've got those things, by the way, you can also just use your skills and get a job, and then you're not going to be a CEO, you're going to go through the ranks as Satya Nadella did, by the way.
Well, speaking of Satya Nadella, who went through the ranks, but very successfully, he's often exhibit A when it comes to Indian American success. I'm not sure if he was CEO of Microsoft. I don't think he was when they acquired Skype. It may have been Steve Ballmer. But one news story this week is about Skype being dead. I remember when it was acquired, you and I were at a startup, which was the audio equivalent of Skype. Sorry, the video equivalent of Skype, Santa Cruz Networks.
Firstly, what does this tell us about Microsoft? And secondly, it was obvious at the time that they were going to screw up Skype. So what does this tell us today that Skype's gone offline?
I think it tells us that Microsoft has done something good, actually, weirdly, the opposite of what you'd think. Microsoft Teams, which most of us hate, is actually a really good product, and it has all of the Skype stuff in it. And so Skype, the brand, is no longer required, and Skype, the app, is just a pale limitation of the features. Why do we hate Teams if it's a good product?
For the most part. But Microsoft dominates it. those big companies, and that's where the money is. So it really tells you Microsoft's done a good job, ultimately.
And what does it say about Dwarkesh's arguments that fully automated firms will eventually close their offices? It's going to take generations, and Microsoft will probably be the last company to shut its office door, won't it?
Well, AI has dominated the week, Keith, and appropriately enough, we have an AI quote unquote startup that is your startup of the week, CoreWeave, files for an IPO. They call themselves the AI hyperscaler. What are they up to?
So they're a cloud infrastructure for AI. I put them in because of the IPO. IPOs have been scarce. And those that do get through don't tend to perform very well. So this is... This is either the start of a new IPO positive era or not, we'll see. But the fact that they've filed, and by the way, they're losing a lot of money. On 1.4 billion of revenue, they lost 800 million last year. And they argued that they didn't lose it, they invested it. They would.
I mean, we've heard that one before too, Keith, right?
Words and timings
Imean,we'veheardthatonebeforetoo,Keith,right?
Speaker 3
And it's probably true. I mean, because of that investment, next year's revenue might be three billion. And so more than they paid to get it. So let's see. But it brings back the question, is it valid to lose money where the word lose is really invest? in order to grow. Is that still okay? And the IPO market is where you test that.
Well, let's end where we began on autonomy. Your final, your take on your editorial this week is you said, we're entering a phase of human existence where autonomous agents will increasingly help us achieve goals. Each of us has differing goals. Many of them will become achievable wherever they previously were not. It's real and exciting. And your post of the week comes back to that conclusion to the editorial by Box CEO, Aaron Levy, one of the smartest people, I think, in tech. And he writes quite a long post about AI agents. What does Levy tell us?
He's basically talking about the transformation of enterprise software from all of the modules and segments that we have been familiar with over the last 20 years and have all of that infrastructure replaced by new software layer called AI agents. That is to say software that carries out tasks free of humans in order to achieve the same goals that previously took clunkier software and humans. And he's right at the leading edge with, he's the CEO of Box.net, which is...
Right, and he says at the same time, there's going to be a new category of AI agent startups in spaces where the existing software is too slow moving or not a natural fit for agents. does this come back to your theme uh in uh in the editorial this week um not just for corporate use but for all of our use that we will we have agents that even will begin to know the kinds of apps we want to make and we'll make them autonomously or
independently of earth i think you could even today you could say to an agent i've got this problem I want to design an app for that problem, and it could start doing it, but it wouldn't be able to decide what the problem was. You'd have to tell it that. And yes, it does tie it all together. The only reason I actually wrote about the app that I wrote is because it's an example of what these conversations are. My app is just an app, right? But it is an example of the capabilities that we're gonna start to see everywhere. And I think Aaron Levy, his post is, For a skeptic, you should read his post because he has no upside in exaggeration. He runs a very big...
Yeah, I mean, he concludes incredible times, AR software going forward, and we're just getting started. He shares your excitement. Yeah. Well, it is exciting times, Keith. Even I agree with you. I'm a natural skeptic, but I think you're right. Things are changing very fast, and no doubt we will continue to talk... next week about autonomy, AI agents, and the rest of the tech world. Thank you so much. Have a great week, and we'll talk again next week. Bye, everyone.