Bubbles and AI

Speaker

Hi, this is Andrew and this is Keynote, the daily NOW.tv chat show with some of the world's leading thinkers and writers.

Speaker

Hello everybody, it is Friday, February the 24th, 2023, it's that time again, that was the week, my weekly show with my old friend Keith Teer, as you can tell from my background I'm in a hotel room in New York City, I'm on the East Coast and it's 2.30pm, or a little after 2.30pm, I kept on changing the time and I think Keith was a little mystified but I wanted to make it happen, this show, at 2.30pm because of course it's been that kind of week, a 2.30 week, two major cases in the Supreme Court, I actually was in D.C. this week, not unconnected with what was happening, about the future of the internet, the future of section 230, Keith and I have talked about that, seems as if, at least according to the Times, the court seems wary of limiting protections for social media platforms, maybe that's because Google made such a professional performance, which I was certainly impressed with, their professionalism, and then the second day featured Twitter, who were perhaps less professional. Keith, now you get the 2.30, right? Exactly, now I understand, very imaginative, Andrew, I was also on the East Coast. This week leads with bubbles in AI, and I'm not convinced it was a bubbles in AI week, I think it's a section 230 week, a very important week for regulation and tech, so what happened in D.C. this week, Keith? Well what really happened is that those who wished to see section 230 either modified or removed, the protections that it brings, did a very poor job of presenting their case, and the judges, through their questions, and their comments, made quite clear that they were both confused, and wary of if they were to agree with the plaintiffs, what the consequences would be for the entire internet, which is the right thing for them to be concerned about. So it was a pretty good example of the American legal system doing what it's meant to do, which is A, figuring out what is legal, but B, interpreting the law, and when asked to interpret it in a way that would effectively change it, showing a lot of reluctance in doing that, which is what they should do, because the judicial system is not meant to make law. You mentioned the state performed badly, but perhaps the reverse is true, maybe Google performed well. Should we commend their lawyers, Keith? Well we know some of them, don't we Andrew? So we're conflicted. We can only say yes to that question, because A, it's true, and B, they would be very angry with us if we said otherwise. Well they certainly performed well, at least on the legal front. I don't want to give away too much, I wish I could, because we could have a really interesting conversation, but I'm going to keep that as pillow talk. One of the things that's interesting, I know that there was a lot of concern that the court would break down on political grounds, that the left would go after Google because it's a successful capitalist company, the right would go after Google and big tech because they see them as being sympathetic to woke culture, but neither of those things seem to happen. It seems as if both the traditional conservatives and progressives were both relatively sympathetic. Even Clarence Thomas, who is a notorious critic of big tech, seemed open-minded at least on this though. Yeah well look, let's give some context here. Section 230, the purpose of it when it was first brought in, and the only purpose I think it's good for, is that it distinguishes between a publisher and a platform, and allows the platform to be not liable for things said by the publisher. Now the word the publisher conjures up the New York Times, or you know organizations, but actually in this context the publisher means me and you, the individual. And so what it means is that Google isn't liable for what I publish on its platform or Facebook isn't liable for what I publish on its platform, and Twitter isn't liable. And that actually is common sense because you can't have a platform open to the entire population and make the supplier of the platform liable for what everyone says without effectively forcing the platform to close down, because those liabilities would be way too onerous for anyone to take on. And so that's the purpose, that's why Facebook and Google don't get sued all the time, and they literally would be sued hundreds of times a day if not thousands of times. Thousands, tens of thousands of times, I mean Twitter would be sued tens of thousands of times a minute. Exactly, so it's a totally common sense rule. I think it covers for example comments in the New York Times. The New York Times isn't responsible for what is commented online, however in print where it makes a choice which comments to publish, then it would be liable, because it then becomes the publisher. So this definition of who is the publisher is actually the key to the whole thing. And I think it also shows, I mean there's always the old cliche that tech moves a lot and that's obviously true, but I wonder whether this reiterates that in legal terms, that if tech runs so far ahead of government that these platforms are created and then 20 or 30 years later people want to change the law, it's impossible. You just can't do it, because you've got a multi-trillion dollar industry running on top of these platforms. So if the law, I mean you and I probably don't agree there. I mean I'm critical, I've always been critical of Sue Ferry, I can see the arguments, but now it's just not conceivable that the Supreme Court or Congress would blow up a multi-trillion dollar industry. Yeah, and they shouldn't, because they're not doing anything wrong is the key. Now interestingly enough, where they do do things wrong is when they choose to become editors, which is what publishers typically have to do. And so if you were to want to be critical of Facebook, more Facebook and Twitter than Google to be honest, because Google doesn't really have a platform that's like that. But remember Google is being, it's not Google that's being sued here, it's YouTube. I mean the case originally against quote-unquote Google or Alphabet was brought by the family of a woman killed in the Paris bombs, and their argument was that the terrorist who planted the bomb was radicalized by videos they saw on YouTube. Yeah, yeah, well you know, and that's the equivalent of that would be if this bomber went to a particular coffee shop in Paris every day, and you sued the coffee shop because he met people there. You know it's ridiculous, on the face of it it's just ridiculous. And also they couldn't prove, I mean that's the other absurd thing about this case, is that they couldn't prove it. I mean I think one of the things that the Google people and probably most of the tech companies feared was that the case seemed so

Speaker

flimsy and absurd on the surface, that I think they were concerned that there was such a strong political will to undermine big tech, that the law would be overridden by politics. But again it wasn't, it was a bad case, and as you say it was badly argued by the team on the other side, partly because they just didn't have a case, and their brief was bad, so everything about it was bad from their point of view. And I guess it doesn't augur well for the future of people bringing big-time legal suits against big tech, although that may change. I know you sent me a text earlier, an outrage, a key to outrage about your friends at the DOJ prepping an antitrust suit to $20 billion Figma deal. Why are you so outraged by that Keith? Well again on the face of it, it just is ridiculous. Figma has many competitors, Canva is a competitor, Sketch is a competitor, Adobe frankly is a pretty poor competitor, and by buying Figma it's possibly putting itself in a position where it can be a solution for creatives who want that kind of a tool, but even then it's competing with a lot of others. So what is the negative here? I don't see the negative, it seems to be there's almost a fundamentalist belief that large companies buying other companies is on the face of it a bad thing. Are you accusing your friend Lina Khan of legal fundamentalism Keith, or antitrust fundamentalism? Well she's anti-capitalist fundamentalism, she's basically a fundamentalist against capitalism and not in a good way. If she was fighting for the rights of poor people I would take her side. This is just a malicious attempt to stop normal business out of a desire to keep large companies from being able to grow. I don't get it, I literally don't get it, there's no harm here. Well we know you're not the biggest fan of Lina Khan, you've had a good week Keith, certainly Manchester United, West Ham had a less good week, they're famously associated of course with blowing bubbles, and that's what you lead on this week, bubbles and AI. As I said I'm not convinced that's the main story but you chose it, that's your editorial judgment. Isn't there something inevitable about that Keith, I mean we knew this was going to happen. Yeah you know I tend to choose my editorials based on whether I've got something to say and I felt like I had something to say about that and I felt like I'd already said everything that there is to say on 230 previously but you could be right. So what's going on here is that there is a widespread belief having just come out of what everyone describes as a big tech bubble and having gone through a correction, there is a widespread belief that AI represents the next bubble and so what I decided to write about because there were two stories, one by Hunter Walk and I'm blanking on who wrote the Hunter Walk one. The Hunter Walk one is good, he quotes Carlotta Perez, one of the smartest historians of tech, particularly of tech bubbles and their meaning, so that was a good piece. Yeah and the other one I can't remember who it was by but anyway they're both super good and what they point out is that bubbles, if you want to call them that, are rational in the early to have transformative capabilities. Why? So it's not like tulips where tulips really don't transform anything. This is truly transformational and lots of money is going to flow through AI in the coming decades. I doubt anybody would question that and therefore to dive in early and place bets is an entirely rational thing and the most intelligent people will do it first, the ones with the most insight and access and money and others will follow them and the inevitable outcome of that is an inflation in valuations and what everyone thinks of as a bubble. What Hunter Walk points out is that a lot more money will be made than is being invested but not evenly distributed. So some people are going to lose money in the frenzy to get in, other people are going to make a lot of money but if you grossed it all up a lot more money is going to be made than is invested. Yeah and the two, probably the two biggest tech companies, well certainly two of the four large tech companies are in on it too. You link with an excellent piece by Chris Stoker Walker, a young writer by the way, I think he was very good. I interviewed him on Keen On about his TikTok book, he's a smart young man. He talks about Google and Microsoft are now in an AI arms race which was pretty obvious given that Microsoft are bankrolling OpenAI and Google of course are launching their own platform. How big a deal do you think this AI arms race is? Well I think Microsoft are destined to lose it. I mean they've already made it. As they lose all wars, Microsoft I think. They made some really poor decisions this week, in reaction to the outcry over Sydney that we talked about last week, they restricted Sydney to only being asked five questions before it resets and you have to start over, which means that a lot of the value in it is being is destroyed. So they're caving in to the crowd in a way. I think that's a fundamental error and that's what Microsoft does. They're afraid of being controversial and to be fair Google's a bit afraid of that as well. That's held them back and obviously controversy is an inevitable part, especially in this space, of anything that's new. Does Google have the most technology here? I mean certainly

Speaker

Microsoft is outsourcing it in their investment. I'm assuming that this arms race will also involve Apple and Amazon in terms of acquisitions, Keith. I mean the M&A AI space is going to go berserk. I mean they're going to have insane bubbles, aren't you? Well that's going to be true in every vertical as well. I mean the venture capital vertical is the one I'm in and there's only two or three players who are leveraging AI in decision making and they're all, one of them this week, SignalFire, announced that it just raised 900 million dollars. Money is going to flow to good AI doing a good job of things that are valuable and in that sense it is an arms race but what Microsoft and Google are doing is providing infrastructure for one part of that arms race. I think it massively underestimates how big the value is to focus on those two because I think to the point here in the second part of the headline, changing how we use the internet is going to be in every field, in literally every field, business to business, consumer applications, medical, data analytics, I mean everything. And not just that but the whole idea of an interface and how we engage with our phones or our computers, everything changes. I mean our notion of ourselves and our relations with machines change as well so it is a profound revolution. All revolutions of course are profound. You linked this really good piece by one of the smartest people I think in tech, Stephen Wolfram, probably a little bit too smart for his own good. He's Mr AI before there was an AI bubble but this piece, what is Chet GPT doing and why does it work, really does a good job leading us through the mechanics of Chet GPT. I'm not a hardcore tech person, I understood it and I thought it was really good. What did you think Keith? Well it's really good in as far as you can understand it because there's a lot of math in it, equation math not simple math but I think even without understanding the math he makes a really good essay on how large language models make decisions and why they make the decisions they make. And somebody's making a good joke by the way, Dan Wang is watching us and he says whatever AI is forcing my camera to keep zooming in and out needs to be scrapped and I agree with him. I've been noticing it doing it every time I move my head. So not all AI is good but Stephen Wolfram does a great job of describing, if you really want to understand how Chet GPT works, I sent it to my son who's a computer science senior at Syracuse and he loved reading it. Yeah I mean what I found really interesting was and the subtitle of the piece is it's just adding one word at a time but that isn't a marketing spin that's literally how we describe it. I never really thought about it but basically it's making up language as it goes and just as you and I put our words together, I mean we don't think about each word that they somehow come out right, this is what Chet GPT is doing. Yeah exactly right and it's drawing on a massive body of prior knowledge in order to facilitate that and it's doing statistics, it's probabilistic scoring. And what struck me about that is that the amount, given that it's analyzing every single word, the amount of computing power it must require is phenomenal, astonishing and I don't quite understand how that's going to change the industry either. I mean maybe it's not uncoincidental that there are breakthroughs now on the quantum front too. Yeah well the quantum, you don't have a quantum piece this week but every week there's a usually a quantum piece suggesting some sort of breakthrough or other. Yeah the problem with quantum is that no one's yet figured out how to run programs on it that don't break for the most part. There was a story this week that Google had just had a breakthrough splitting data between I think they're called quarks but nodes in a way in a quantum computer because if all the data was in one it was unreliable that it would continue to exist which is to do with the quantum state of a quark. So it's super complex and it's very promising but you couldn't run Microsoft Word on a quantum computer. It's not capable of that yet. Well I'm sure Microsoft will eventually make the breakthrough in quantum or be involved and somehow screw it up. They always manage to shoot themselves in the foot. A couple of interesting political stories you linked to Keith this week. The first is Bessemer Capital calling on its companies to move more cash out of Israel because of all the political turbulence there and the undermining of the Israeli Supreme Court. Is that troubling? Is it good? What's your read of this? Well it's always difficult to talk about Israel because it's so political and politicized. But you're a big fan aren't you Keith? I'm teasing you. I'm teasing you. Go on. Yeah way too simplistic a question for me to possibly answer but what's happened is Netanyahu as we all know has struggled to get a majority government for the last three elections. He was out of government in the second one. He's now back in a coalition that involves some what I described as far-right coalition partners and Netanyahu himself is subject to various criminal cases and what as soon as they got into power they started to suggest that they should remove the right of the Supreme Court to have the ability to determine law when it comes to the judiciary which is basically getting rid of democracy effectively. Now that hasn't actually passed yet. It's a proposal. Okay so we take that but why should venture capitalists call on companies to move more cash out of Israel? What's that got to do with anything? Well so this is about the relationship between politics and economics. Basically these political moves have led to a run on the shekel which is the Israeli currency which most startups hold their funds in. Oh I see. So basically it's a pretty narrow self-interest based on the value of money that is causing them to suggest a that they shouldn't be in the shekel and b they should perhaps consider changing their domicile from Israel to somewhere else. Yeah it's an interesting story because it suggests that Israel's I want to say tech dominance but tech promise could theoretically be undermined by its very odd politics. I bumped into Yossi Vardy or I spent some time with Yossi Vardy at DLD, one of the great Israelis and he's deeply pessimistic about the political future of Israel. Another piece you linked to which was in some ways even more interesting was how China has become Saudi Arabia's largest trading partner. You remember George Bush's axis of evil. The Saudis, the Chinese, the Russians, the headlines today about Russia and China now joining the so-called peace talks. Is Saudi part of this axis of evil or anti-democratic axis? They've got terrible human rights record. I had someone on my show this week who a researcher on human rights who reminded us that Saudi executes more people than Iran. Well look there's a medieval aspect to many parts of the world which haven't had the benefit of independent path to capitalism that we had in the UK and the US and some parts of Western Europe and Asia. So the colonial hangover is that the monarchy was never replaced by democracy in most of the world and so you basically have feudal systems which have feudal legal systems and feudal punishment systems. I don't think you can particularly blame the Saudis for that to be honest. They are what they are probably mainly because of us. Now what's more interesting about this story is the decline of the US as their primary trading partner and its replacement by China which speaks to the wider global shift that I think is inevitable. I'm sitting looking at my bookshelves and The End of Empires. There's a whole kind of shelf of books that talk about the end of empires and you know there is yet to be an empire that survived. There is no such thing as a surviving empire. They all die and America is at the tipping point and China is at the birth point. It would be interesting I think if the Saudis do end up buying Manchester United will they execute your manager if they don't win any trophies? They would if he was in Saudi Arabia and he said something blasphemous. One person we haven't mentioned recently and it's not Elon Musk is our old friend Sam Bankman-Fried. You link to this wonderful piece on FTX co-founder Bankman-Fried facing four new criminal charges. What's astonishing from the photo, not everyone's watching this, some will be listening, is how criminal now Bankman-Fried looks. He looks like Bernie. He's increasingly looking like Bernie Madoff. He's pale, he's in an ill-fitting suit. What's happened to old Sam Bankman-Fried, Keith? Four new charges were added related to fraud and particularly preconceived fraud, wire fraud and money fraud basically.