Tech Consolidates Stock Market Growth, Lawyer Uses Hallucinating ChatGPT, XR Wars Heat Up

0:28The truth is AI security is identity security. identity security fabric can help you secure the next generation of identities, including your AI agents. Visit Okta.com. That's O-K-T-A dot com. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge, and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks. It doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade in value. Advanced, intuitive, and deployed. That's how they stack.

1:09That's technology at Capital One.

1:18Welcome to Big Technology Podcast Friday edition, where we break down the news in our traditional, cool-headed, and nuanced format. We are here with you on a very big Friday. We have Apple's big WWDC event coming up on Monday, so we're going to preview that. But first, we're going to talk about how the tech stocks have started to consolidate much of the gains in the S &P 500. We're also going to talk about AI copyright, what's going on in Japan. Actually, there's a story brewing that you might want to hear about, and AI hysteria, all these signatures about preventing the worst from AI. We're going to get to all that and more.

1:49Joining me today, is as always is ranjan roy here with us from margins on substack ranjan welcome to the show it's friday good to be here great to have you so let's start right away talking about our big story because you sent this to me and i said we have to talk about this and what's happening right now is the market is absolutely ripping the s p 500 is up almost 12 on the year which is extremely impressive especially given the terrible down year it had last year and who's making up the growth You would think that with interest rates rising, you would have a much more even distribution of the growth.

2:26But no, this is from Axios. The five that are responsible for the vast majority of the stock market's 2023 gains are Apple up 36 % this year, Microsoft 37%, Alphabet 39%, Amazon 44%, and of course, NVIDIA up 159 % this year. What's happening here, Ranjan? Because it doesn't seem to make sense given everything i heard about how higher interest rates are supposed to bring down tech valuations and maybe open room for other companies in the economy i thought rotation was the key word but we're back to technology i think we're back to that point where again the jobs number came out today it was strong employment is strong inflation is falling economic growth looks stable and then you have the ai hype cycle so all these factors coming in together are perfectly placing these big tech companies to go up at this scale and i think they're the two biggest questions investors have to ask themselves now are one do they continue the dominance we've talked a lot about you know is open source a threat are they really be positioned well to capitalize on the AI hype?

3:41Is there a business model, a feasible business model for the AI hype? But I think another thing, I'd seen one comment around, like when do investors start trimming? You see gains like this, you see levels like this, you see PE ratios like this. At a certain point, can it keep going? And obviously that's - Very dangerous question to ask. The most dangerous question to ask. But at a certain point, again, out of the S &P, 490 stocks have been flat or down. This is what Jeremy Siegel, the Professor Wharton said last week. As you said, if not for the seven companies, if you add Meta and Tesla to the five you mentioned, the S &P would be down on the year.

4:23So I think from an investor standpoint, is this too much is the main question. But then also, you know, from a general health of the economy standpoint, we're in an even worse position than we were a few years ago when we were talking about over concentration among a few companies so i guess like i'm trying to figure out why tech is having the surge now you can say ai and all that but wasn't i mean this is your whole thing right wasn't the fact that interest rates and i mentioned this in the open the fact that interest rates are going up we're not supposed to see these valuations i was checking the valuation of apple it's nearly three trillion dollars again so explain that one for me ron john i'm truly puzzled by it yeah no I think this is the amazing part.

5:07Again, the fact that we are currently in this Goldilocks scenario, falling inflation, strong growth, and it's stable and strong employment. I did not expect that. I think most people did not expect that we would be in that kind of situation. So the question is, can that continue? So I think on the general macroeconomic side, everything is still very geared towards these big tech companies. But then also, again, I think it is the AI hype cycle is at the basis of all this. It's bringing energy back to these companies. And yes, Apple is not directly benefiting from that, but Apple is just kind of caught in this wave as well.

5:47But it's given this narrative, I think, that once again, there's untold economic potential and untold growth right around the corner thanks to this new technology. And who better to benefit from it than the incumbent players? It does. I mean, we've talked about it a lot on this show about how that economic promise is not certain at all yet. And we're going to see what's going to happen here. But it does make me a little bit worried about the state of the economy. No, no. That's one of the dangers I think I see here is that, again, once all resources and energy and investment start flowing back into these seven companies, maybe nine companies, then the other stocks, the 490 other S &P, and imagine mid-cap and small-cap stocks, or just other companies are not going to be receiving that and can easily get killed through this.

6:41And this is basically what happened in the mid 2010s in some way. But that's where I think it's in terms of like the overall innovation potential of the economy. I think I've said it in the past. And I think, again, we're back to that place where it's a problem. Okay, let me make the other side of the case, then we'll move on. The other side is these companies all fell when interest rates went up and the inflation problem is being tamed. Interest rates are going to come down. The market is investing not for what's happening today, but for what's happening 12 to 18 months from now. and in the s p 500 you have the strongest companies in the economy strongest companies and if you're the s p 500 and you're being buoyed by an apple and an amazon and an alphabet and nvidia which is a new technology new innovation maybe that's not the worst thing in the world yeah i mean nvidia is trading at 172 times earnings where you know traditionally call it 20 times is solid.

7:40Like, yes. And again, I have been bullish on NVIDIA for a long time, and they are well positioned. And as we talked about with John Herman last week, like NVIDIA has actually shown every time there's this kind of potential around the corner, they're able to go and grab it. Again, starting with video game graphics cards and actually being able to evolve into, you know, just microchip processing and now moving into actually being the kind of like infrastructure for AI. So, so again, these companies, they show that they can, they've shown past experience they can get there. The opportunity it's a, it's a clear idea of how they get there, but still the execution side of it, I think it's rich.

8:23Hey, Kathy Wood said, NVIDIA's rich. When Kathy Wood says your stock, obviously she sold it right before. Yeah. She lost out on all the gains. Right. I don't know if we can really listen to Kathy's investment advice anymore. That was a bit tongue in cheek right there. No, no. I think that you're right. It is a good signal if she's calling the top. But then again, I don't know. That actually makes me want to almost buy into it. You got to be in, obviously. 137 times earnings is cheap. 172. 172. So let me talk to you about this quickly, and then we can move on to the AI stuff. there's uh in 2001 there was a paper uh from mi from uh wharton that talked about the asset market meltdown hypotheses and that was that you had a lot of boomers in the economy who had much of the investment and they were all going to retire and here's the philosophy 10 to 15 years from now millions of boomers will retire and instead of putting money aside for their old age they will begin cashing in and it stands to reason that this will reduce demand for stocks and as any beginning economic student knows lower demand means lower prices it's kind of a bit of field from what we typically talk about but i actually wanted to get your your thought on this which is that i'm curious how much of this market right now because the market is going up despite some of the negative conditions that we're seeing here and i do wonder whether there's part of the market that's at risk of people retiring and spending that stock or cashing in those stock investments and then lowering demand?

9:56It's tough when you think about these kind of very aggregate demand style theories around the stock market. I mean, one thing I was just reading about was, I think it was close to$2 trillion in outstanding student debt. The student debt pause has still been going on since COVID. So once that resumes, how much liquidity gets taken out of the market. That's something no one could have predicted. I mean, general, just overall influx of money during COVID via stimulus and everything else, no one could have predicted. So I think trying to predict this stuff based on the demographics and what typical behavior could be, I think it's tough.

10:39But it was interesting because even on that historical perspective, I was looking at in terms of concentration, is it a good bet to still go into an Alphabet or a Microsoft or an Amazon. And it was at like an AT &T in the early 1930s made up 13 % of the entire stock market. General Motors in 1928 was 8 % of the entire stock market in 1928. This was all from Ben Carlson, A Wealth of Common Sense, which is a great blog. Like this idea that at points in history, we have seen this kind of concentration, but those concentration usually happens before bad economic times. And I think that's where this question of like, is the economic healthy?

11:21Look at it. It's healthy for seven to nine companies and it's not healthy for the other 490 or however many else. So I think, I think it's one of those where trying to bet on demographic predictions around behavior is going to be tough. Cause I mean, who would have guessed millennials and meme stocks? Like that's definitely not something we could have predicted, but, But even more so, I think this question of like, where would you put your money right now? Would you buy, like, would you buy NVIDIA right now? Hell no. It's a tough one. It's an easy one. All right. All right. You would. I still, I buy into like the story.

12:03I've owned it in the past and made some money off it and then took profit. And again, I regret not continuing to hold it. And they're one of those companies that they've very quietly and cleanly executed and realized multiple visions, but all based on the multiple evolutions of the same vision. Again, computing power is the core thesis. It's not like Facebook, you're a social network and you want to make a headset, which we'll get into in just a moment. It's all a really clear path, which is what they've done and they've done it well. So like on that side, I would want to own NVIDIA, but I wish I owned it a month ago or two.

12:43Yep. So, okay. You mentioned that AT &T used to be a big chunk of the S &P 500. I need to ask you this before we move on to some of our AI stories and then headsets. AT &T stock is down 4 % today after Amazon said it's been in talks with wireless carriers about offering a low cost or possibly free nationwide mobile phone service to Prime subscribers. according to bloomberg what the heck is amazon doing here and is this like why we're seeing consolidation in big tech companies is because amazon is moving into mobile apple is moving into banking alphabet is moving into ai is this what's what we're seeing well i it is interesting because i hear a lot of sentiment of like whenever you see these moves what is lena khan doing where's the ftc where's antitrust i thought it was supposed to be here by now but there's been plenty of strong moves, you know, like Facebook, Giphy now, or Activision and Microsoft from the UK, like they're more so than we saw in the last 15 to 20 years.

13:47They've been plenty of aggressive antitrust activity. But yeah, I agree that Amazon being able to go into mobile to lose more money in e-commerce because they have AWS and can funnel profits over from there, I think is not the best economy or the most innovative economy we can live in. However, I don't know. I think Amazon is playing a riskier and riskier game. This was our entire conversation last week around Timu and Shein and how they're threatening Amazon in ways that they've never seen it probably in their lifetime, or at least since they've achieved some kind of dominance. So I think Amazon making more high risk bets around trying to just give you an impossible package as a prime subscriber in order to just try to keep growing rather than looking at actual unit economics and trying to make their e-commerce division genuinely profitable, I think is a very risky thing to do right now.

14:50What is this mobile move in reaction to? Is it the fact that they need to offer people sell service to keep growing prime is it possible that prime has sort of now saturated the middle and upper end of the market and they now need to basically throw in mobile service to you know maybe the lower end of the market to make it something that i mean i can picture this slide in a presentation it's the average lifetime value of a prime customer is you know like 5x the non-prime customers. So how do we get more prime customers? All right. People on average spend however much money on mobile phones every month on their mobile bill.

15:32So how much of a loss can we take that we'll make up on e-commerce? I'm sure, again, I can see either a management consultant pitching that one or in general, even a corporate strategist outlining that. It's a pretty simple case but again it's uh i think they're running out of ideas it does well also it doesn't sound desperate to you at all no it does a bit i agree i think i think they are realizing the limits of like well actually hold on it's desperate in the sense that i think they are realizing more and more non-subscribers again non-prime subscribers that entire market maybe they are of recognizing that Sheehan and Timu are a genuine threat to them.

16:19Oh, so do you think this could be like a direct outgrowth from like our discussions over the past couple of weeks? I think Andy Jassy was listening to big technology podcasts. Wait, Timu? We got to give the cell service. Yeah, I think Bezos was listening on the beach all buff and jacked somewhere and put the call in and said, hey, we got to do something about Timu. Up the prime subscribers. Yeah. I mean, one other thing that someone floated to me was that mobile networking is like a pretty high margin business once you have the infrastructure installed. But it looks like they're working with other cell carriers.

16:53But I'm curious, do you think that that is part of the calculation for Amazon as well? Oh, maybe I could see this is one thing where what's the Amazon like local area networking across houses? They had some effort where it was like network. the mesh network but you're like your network is it called arrow uh your euro i actually have yeah that that's for home is yeah that's for home they had something where your home network could help power access for external people to then power their own network um basically the idea basically what i'm thinking is here maybe the idea would be you sign up they use other people's mobile infrastructure.

17:40But over time, again, in this slide deck I'm envisioning, they can say, you know, year one, prime subscribers by 5x more. Year four, we're starting to make high margin business because we built our own infrastructure by that point. And I mean, if anyone, they certainly have enough real estate and infrastructure around the country that they could, if anyone could toss up an entire mobile network, I could see it being Amazon. Very interesting. Okay, let's just touch on a couple of these AI topics and then we'll take a break and go to headsets. So this is one of the things that you've been talking about for a long time is that eventually, you know, the thing that this runaway AI development is going to run into is copyright law.

18:24But it only takes one country to say there's no copyright law when it comes to training AI here to sort of flip that equation on its head. And that seems to be what's happening in Japan. Now, the reports are a little bit shaky. The one that everybody's talking about this week comes from a website called TechnoManners.ai. So caveat for what it's worth, but it doesn't seem like this has been shot down yet. So here's the story. Japan's government recently reaffirmed that it will not enforce copyright on data used in AI training. The policy allows AI to use any data, regardless of whether it is for nonprofit or commercial purposes, whether it is an act other than reproduction, or whether it is consent obtained from illegal sites or otherwise.

19:11okay uh so basically this is um the japan minister of education culture sports science and technology confirming this at a local meeting and saying that japan's laws won't protect copyright materials used in a ai data sets to which yan lakoon the chiefs ai scientist at meta says japan has become a machine learning paradise. Ranjan, what is your reaction? Well, one, I continue to take this with a grain of salt because, and I'm glad you kind of gave the disclaimer as well, that there's one blog that basically says that the English language coverage of this topic is sparse. And then that was retweeted by someone.

19:58And then Jan LeCun of Meta retweeted that or quoted, commented on that. So I think I would be also a bit cautious about what exactly is happening. But I think that question of global coordination on the topic is a good one, because could one country being this machine learning paradise, ruin it for everyone else and actually make it so then to stay competitive, you have to allow for unlimited access to all content. but okay here's where i think this almost feels like the crypto argument that if you regulate us we'll go elsewhere and then they will be the paradise and the place that makes all the money is i it goes we always we've gone over this many times it's like i don't believe that size matters in ai models and some sam outman's gotten into it and like the idea that you have to have access to ingest everything is the only way to build these models.

20:58So I actually think this could be a case where the same way a complete lack of regulation can almost hinder crypto's development because it leads to chaotic outcomes that make it impossible for mainstream adoption. I think this is a similar situation where, I mean, we see, and we can talk about like hallucinations and training on too big a data set can create models that just aren't applicable and usable in real business situations. I think we're going to move to, if we move and when we hopefully move to a world where it's more specialized models, I think you won't need to ingest everything all at once and have unlimited ability.

21:40And it's better for the overall ecosystem as well, if there's some kind of compensation and some kind of understanding of how economics are distributed throughout both the content creators, the ai models the ai trainers everybody interesting so you don't think that people will inevitably flood i mean like this is part of this is japan does want to be an ai leader you don't think that there are going to be countries that will say we want to be an ai leader and you know okay there might be some rules established elsewhere about compensation for training who cares about the size of your model you know you can just go take them if you want well hold on but there's also the technical aspect of this right that there's like the scraping and and i i do think i one thing i really have been thinking about is like both in terms of api access is the simplest way to kind of like uh you know cut off unfettered access to your content but then even preventing scraping in more at like more scalable ways or just trying to think about how people protect their own content, I think is going to increasingly be part of how the web is built and how people publish and think about content.

22:53Yeah, how would that play out practically? No, even if in Japan, I am not liable for training a model on all content on Reddit, if I'm unable to easily take in all the content from Reddit, then it's no different for me. So like, just because there's a lack of legal liability, there's still the technical question. I think on the technical side of it is going to be just important as the regulatory and legal side. That's interesting. Yeah, that'll definitely change things. So something to keep an eye on. We'll just sort of keep hitting on it as we get more and more concrete stories coming out of Japan or wherever it might be.

23:34Did you see the story about the AI lawyer guy, the lawyer that presented a number of fake cases to a judge that he just sourced from ChatGPT? It's sort of interesting because it comes at this moment where we're talking about the power of AI and you just have complete jackasses like this guy who's running to the judge and saying, ChatGPT did my homework. he deserves the the backlash he's getting like i think this was hold on what there was very oh the the court filing cited six cases that completely did not exist yeah so this is from ars technica a lawyer is in trouble after admitting he used chat gpt to help write court filings that cited six non-existent cases invented by the artificial intelligence tool chat gpt see lawyers Stephen Schwartz of the firm Levidow Levidow and Oberman by the way amazing law firm name quote greatly regrets having utilized general artificial and generative artificial intelligence to supplement the legal research performed herein and will never do so in the future without absolute verification of its authenticity I just love how they're like lawyering it away being being not being like you know you know clearly we messed up but like talking about how it's so so like weasley to supplement the thing and like we we've been adding more ai in our practice to get the cutting edge like come on guys just admit what happened here no but this to me is the most this is one of the most important stories that came out in the last week because this is what i've been like ranting about for months now is that getting actual usable ai high-generated content that is risk-free, that can be used in high-risk situations and MIM born situations is difficult.

25:28And that's why the gap between, again, making a funny limerick in ChatGPT and getting to somewhere where in a business setting, you're going to actually create predictable, repetitive content. There's still such a gap there. And I strongly believe that we're going to get there. But I think this idea that everyone was just going to plug in something or just ask chat GPT something and everything would work out perfectly, I think was ridiculous. for context in starting to prepare for today's podcast on our topic of the concentrated big tech companies and the returns of the S &P this year. I asked Google's BARD, their chat bot, you know, like I just wanted to quickly see what were the returns of the 10 biggest companies in the S &P.

26:27And it told me that Meta is down 58 % this year. It says as of June 2nd, 2023, Meta is down 58%. That's so embarrassing for Google. We got to get Jack Krafcik back in here and make him answer for what's going on here. And the thing is, I would even see like ChatGPT would probably at least have like a bunch of disclaimers like financial information should not be taken from this site or whatever like google's just like straight up for the record meta is up 119 this year sorry thank you because i i knew it was more than doubled i knew but but here's what here's what i don't get about like let's take that question doing a simple mathematical today's price minus the jan first price fix you know would figure out the percentage like that is like you know sixth grade math or whatever.

27:21I don't know. The idea that this all knowing chatbot that's going to lead to human extinction didn't do that simple calculation or got it somehow completely wrong. I don't even know where the data came from. That's just a reminder to me that there's so much distance between what we're promised and where we are today. And again, I'm still very bullish on this, but it's a reminder of where we are. It actually ended up being the perfect follow-up story to this discussion of Japan as the AI paradise. Maybe not as much of a paradise as Jan and some others might hope. I am starting to come to your standpoint that maybe we do need some rules and actual sophistication around how these models get trained as opposed to a free-for-all.

28:08But even still, even as we're having such issues with something like ChatGPT, we're like having this it's a very weird and interesting wave of like very impressive people starting to come out all together and talk about how like ai could potentially you know destroy us and this uh came out of the center for ai safety earlier this week it was the biggest tech story of the day actually on monday where you had hundreds of ai scientists sign on to this statement from the center for ai safety and it's just a sentence it says mitigating the risk from extinction mitigating the risk of extinction from ai should be a global priority alongside other societal scale risks such as pandemics and nuclear war and the people that signed it i mean jeff hinton yashua benjo from two-thirds of the ai are the people that basically invented deep learning the other third is jan lakoon demis hasabas of of deep mind sam altman you know the list goes on bill gates is on there i'm sure elon is on there somewhere what the heck is going on here rajan i I mean, why are these people feeling the need to come out and put their names on these statements?

29:16I mean, what do they know? Again, Google Bar just told me that Meadow is down 58 % this year. A lawyer tried to use ChatGPT and cited six non-existent cases. Like, this is all kind of like out-of-the-box technology. We're figuring out what – and we talked about this, I think, a couple weeks ago when Sam Maltman testified in DC right it's like what's what's the game here because I still cannot remove actually though I guess in this case when you have some people who are more pure academics signing on this it makes me take it a little more seriously because when it's purely someone who owns the technology to sell for their business like a Sam Altman there I can never move past the idea that obviously obviously overstating the fears of what is possible is good for business because then everyone will go to it's like you know only i can save you everyone will go to you but when uh when an academic who you know essentially invented this stuff has the same fears all i can think of is is there some is there some chat group somewhere that everyone's like sharing like Like the computer's telling that they're going to kill us all and just not letting us in on that information.

30:37Well, it's obviously like they're not talking about ChatGPT, which I think is important. Like their fear is not about it, but their fear is about where AI could go. But it is just so interesting that this is all being heralded by ChatGPT and other LLM technology that makes these fears seem so tangible to people. And I find that interesting. One of the other interesting things you asked about the chat room. First thing I thought about when I saw the statement is who is coordinating this? Because it's not coming out of nowhere. And the Center for AI Safety is not an apparition. It's an actual nonprofit that is being funded by someone.

31:14And I thought maybe we can get a little bit towards the motives if we figure out who's funding it. So I went on a little bit of a treasure hunt this week, trying to figure out where this money's actually come from. And I'll talk to you a little bit about how that went. So first thing first, so you go to the Center for AI Safety's website on Monday when the statement comes out and you want to figure out where's the money coming from. Here is the closest you're going to get. By the way, founded this center, which got all these people to sign, was founded in January 2023. They talked about their funding.

31:48They say the Center for AI Safety is a nonprofit organization entirely supported by private contributions. our policies and research directions are not determined by individual donors ensuring that our focus remains on serving the public interest like okay what does that say it says nothing and i emailed them and i say can you please reveal who's funding you i didn't get anything i know the la times picked this up and emailed them and they didn't get anything back okay i checked back later this week clearly they're like well there must have been a scramble because they updated the website on their funding and they're like we have to put something up there Still no reply to the email.

32:23Here's the update. The Center for AI Safety is a nonprofit organization. Currently, over 90 % of our funding comes from Open Philanthropy, an independent philanthropic organization. We would not accept funding from stakeholders, which would compromise our mission of reducing AI risk. We are currently bottlenecked by funding and seeking to diversify our funding sources. So please consider donating here. I mean, okay. So they updated. They said that the money is coming from this group, Open Philanthropy. You go to Open Philanthropy, you have to click the governance to figure out who's funding that.

33:00Support for the fund comes primarily from our main funders, Cary Tuna and Dustin Moskovitz. And Dustin, of course, is the CEO of Asana and the co-founder of Facebook. So it's in his interest. Now, I invited Dustin on the podcast to come speak about why he is interested in this, why so many people are coming around, and to help fill the information vacuum around it. And currently, I'm being passed to his handlers, you know, and he's saying he's not doing much press, but maybe he'll make an exception. I would love to have Dustin on the show to talk about why we're seeing this happen. Now, we know about open philanthropy.