00:26:51 · Tyler, John Coogan, Luca Ferrari, Jordi Hays
[00:26:51] S2: I mean, so in this article, he doesn't there's nothing really new here. They're just kind of talking about his the points that he's been making over the past couple years, which is just that, like, LLMs will not bring us to AGI or Yeah. Or anything. [00:27:03] S0: And Yeah. [00:27:03] S2: Like, even if you keep scaling, they don't they're not, like, actually intelligent. They can't reason or whatever. [00:27:09] S0: Yep. [00:27:10] S2: Which I don't know. It's like, these models are much better than me at math. Like, they can do they can get IMO gold medals [00:27:17] S0: Yes. [00:27:17] S2: Which, like, I cannot do. And that's like it it's like, does that take reasoning? Never doubt yourself. [00:27:21] S0: But, I mean, like yeah. The the I mean, the the there is an element of, like the computer has been able to do good math fast forever, Like, since, like, the eighties. Like, if you were, like, what is 7642 Yeah. I mean, I I think there's difference [00:27:37] S2: you can you can make between just, like, raw calculation and, like, how to, like, think about solving a math question generally. And, like, you can say that he's been right about, like you could say that he predicted, like, spiky intelligence maybe of models. [00:27:50] S5: Yeah. Yeah. [00:27:51] S2: Yeah. And that's, like, sure. Yeah. [00:27:53] S0: Yeah. [00:27:54] S2: But I I I yeah. It's like I I I I think it's not true to say that he's been right about AI for the past forty years. [00:28:00] S1: Oh. That seems man's never had a bad take. Zone. Forty years. Not a single bad take.
00:28:00 · Jordi Hays, John Coogan
[00:28:00] S1: Oh. That seems man's never had a bad take. Zone. Forty years. Not a single bad take. [00:28:05] S0: Single bad take. Now he thinks everyone's wrong. And, yeah, it's very it's very funny for me because oh, does he think George Hotts was wrong when George Hotts said that GPT six will not be AGI and the GPT paradigm will not scale on the Lex Friedman podcast in 2021? Does he disagree with that too? Does he disagree with Andre Carpathi saying it's sloth and that we need new ideas? Or, you know, there's, like, seven other people that have kind of echoed the same thing. He doesn't think everyone's wrong. He just thinks some people are wrong. [00:28:35] S1: But Sort of. I mean, Tyler's take I don't know if now is the right time I can go through this Karpathy post. [00:28:41] S0: Oh, yeah. Please.
00:28:43 · Jordi Hays
[00:28:43] S1: Karpathy posted yesterday. He said sharing an interesting recent conversation on AI's impact on the economy. AI has been compared to various historical precedents, electricity, industrial revolution, etcetera. I think the strongest analogy is that a is that of AI as a new computing paradigm, software two point o, because both are fundamentally about the automation of digital information processing. If you were to forecast the impact of computing on the job market in the nineteen eighties, the most predictive feature of a task job you'd look at is to what extent the algorithm of it is fixed. I e, are you just mechanically transforming information according to rote, easy to specify rules, I e, typing, bookkeeping, human calculators. Back then, this was the class of programs that the computing capability of that era allowed us to write. With AI now, we are able to write new programs that we could never hope to write by hand before. We do it by specifying objectives, I. E, classification accuracy or reward functions. And we search the program space via gradient descent to find neural networks that work well against that objective. This is my Software two point zero blog post from a while ago. In this new programming paradigm then, the most predictive feature to look at is verifiability. If a task job is verifiable, then it is optimizable directly or via reinforcement learning, and a neural net can be trained to work extremely well. It's about to what extent can an AI practice something. The environment has to be resettable. You can start a new attempt. Efficient, a lot of attempts can be made. And rewardable. There's some automated process to reward any specific attempt that was made. The more a task slash job is verifiable, the more amenable it is to automation in the new programming paradigm. If it is not verifiable, it has to fall out from neural net magic of generalization, fingers crossed, or via weaker means like imitation. This is what's driving the jagged frontier of progress in LLMs. Tasks that are verifiable progress rapidly, including possibly beyond the ability of top experts, I e math, code, amount of time spent watching videos, anything that looks like puzzles with correct answers. And while many others lag by comparison, creative strategic tasks that combine real world knowledge, state context, and common sense.
00:30:59 · Jordi Hays, John Coogan, Tyler
Software one point o easily automates what you can specify. Software two point o easily automates what you can verify. [00:31:05] S0: Okay. You gotta go to this other post for the perfect example of what's hard to verify from Nat Purser. This is my personal benchmark for AGI, and looks like we're a ways, a ways, boys. A ways away, boys. Can you come up with 10 jokes in the same format as the you're telling me a shrimp fried this rice joke. You're telling me a shrimp fried this rice. And the, and GPT five pro reason for one point, one minute and thirty one seconds, it says, you're telling me a hamster drove this car? You're telling me a pigeon delivered this mail? You're telling me a Roomba cleaned this mansion? You're telling me a goldfish coded this app? You're telling me a squirrel filed these taxes? [00:31:51] S1: I swear I swear swear AI just has a different sense of humor. Sense of humor is just bad jokes. [00:31:58] S0: Yeah. Oh, GPT five thinking did a little bit better. [00:32:00] S1: It's just you're telling me [00:32:01] S0: a chicken fried this steak? You're telling me a handmade this pasta? Hand pasta? Handmade pasta? That's like [00:32:09] S2: That actually is a good one. Look at number 10. I think that one's good.
00:32:11 · John Coogan
[00:32:11] S0: Number 10, you're telling me a ghost wrote this book. That's good. Ghosting book. That one does make sense. You're telling me a star crossed these lovers? That's actually pretty good. Okay. So we're getting somewhere. We're getting somewhere. At least it, like, understood the prompt on this one. You're telling me a beer battered this fish. I like that. Babe? You're telling me this Figma thought bigger and built faster? Figma helps design and development teams build great products together. Get started for free. We gotta whip through a bunch of these posts. I I I will not leave this show before we talk about the Apple the Apple iPhone, sock. What is up with the sock? Have you seen this? Apple launched a sock. So it's a it's a, fashion accessory, and everyone's debating it. Do you see this? It's this blue light blue sock that you put over your shoulder or over your hand, and people are very, very upset about it. Aditya Agarwal says, when a company releases something that is so obviously underwhelming, then the natural question is, did no one at the company see how bad this is, or did no one have the courage to speak up? I'm not sure which is worse. And someone else says, look. Apple has a lot of fumbles. This is not one of them. They knew exactly what they were doing and exactly who would buy it. Also, it's okay. Fashion accessories are not for everyone. And the news, of course, is that, the it's called the iPhone Pocket, a beautiful way to wear and carry iPhone. Not carry the iPhone. Remember, they you don't say the iPhone. You say iPhone. And so born out of a collaboration between Issey Miyake and Apple, iPhone Pocket features a singular three d knitted construction designed to fit any iPhone. Jaya, who I actually did a collab with on Instagram, very fun tech commentator. She says a lot of tech bros prematurely dunking on this release because they don't get why it's a big deal. So let me translate. You're not the only consumers Apple designs for. This is a huge designer and the mind behind Steve Jobs' iconic black turtlenecks. I didn't realize that. People outside The US wear phone straps and slings all the time and would pay for this. They're tapping into an existing trend. Apple has infiltrated music entertainment, but not high fashion even though the Tech X fashion is exploding. Tech built into Met Gala looks, etcetera.
00:34:40 · John Coogan, Jordi Hays
The three d knitted construction reduces material waste and shows a push toward more sustainable made to shape production. This will absolutely sell. Sell, what do you think? Are you bullish or bearish on the Apple iPhone pocket, Jordy? [00:34:56] S1: I feel like I know a lot of people. Most I feel like most of I feel like my mom would love this, to be honest. [00:35:04] S0: It seems like it's a great Christmas gift. So interestingly, how much do you think this costs? [00:35:09] S1: I don't know. Like $230? [00:35:11] S0: That's wait. Did you look it up? Yeah. Yeah. The long one's $2.30. The the short one's $1.50. But I I think most people would look at this and be like, okay. It's a sock. Like but it's from Apple, so it's probably, like, $30, maybe $50. Some people were surprised that it was a little bit more expensive. But, you know, it's from this famous designer, and it's this interesting status symbol. The question is, like, this could be like I, like, I don't think I'm gonna be using this thing no matter what. The question is, is this gonna be like libooboos and gonna be, like, super popular or, like, Stanleys? Like, will this become actually like a very, very popular form factor in America specifically? I don't know. It's hard to tell. I I'm not really the person to like, you know, handicap it. I I think Apple knows what they're doing. I think they'll I think they'll make money on this certainly.