Embrace AI Experimentation

Speaker

I've got sunshine on a cloudy day When it's cold outside, I've got the month of

Speaker

April the 1st, a new month, old subject, and an old show, and I've got sunshine on a cloudy

Speaker

day, and I've got sunshine on a cloudy day, and I've got sunshine on a cloudy day, and I've got sunshine on a cloudy day. The letter, I guess, your newsletter, was driven or inspired or triggered by an open letter signed by a number of people, prominent people, on pausing giant AI experiments. You beg to differ, is that fair? Yeah, not just me. I mean, I think that open letter got a lot of publicity. Gary Marcus, who was one of the signatures, I saw him on TV at least three or four times on Thursday and Friday. He's the gentleman I debated on IQ squared. Yeah, and he's been on Keynote several times. He's one of the world's leading and perhaps most controversial authorities on AI. Yeah. So, I mean, what I did is wrote a counter open letter, acknowledging many of the concerns, by the way, about the quality of large language models and their ability to be accurate, but instead of calling for them to be paused or stopped, called for them to be improved through further experiments. Yeah, and you wrote, we encourage all AI labs to continue exploring and developing AI systems beyond GPT-4 while prioritizing safety and ethical considerations. Did you get lots of television invitations, Keith? Were you on NBC and ABC with Gary Marcus? Well, for anybody listening, this is a draft open letter that has not yet been published and has not yet requested signatories. So, I put it out there as a kind of a boilerplate, if you will, of an alternative approach. And now the question is, should I put it in a place where people can sign it and gather signatures? Well, I mean, in that obvious prioritizing safety and ethical considerations, I mean, that goes without saying. What are you saying that's in any way original? I'm not really saying... Sorry, I'm recording live, Jenny. I can't actually do anything. That was my wife wanting me to answer the phone. That was convenient. So, you avoided my question on what you're saying original, Keith, on this open letter. Exactly. So, basically, yeah, it is common sense. And Sam Altman has been pretty clear, by the way, that they are very focused on building something of importance that benefits the human race, and they're not really interested in building something that is in any way a problem for the human race. But again, that goes without saying. And I mean, it also goes without saying that Altman's doing this to make huge amounts of money for him and his investors. It's a private company, OpenAI. It's not for the public good. Well, I think there's a correlation between the private value of OpenAI and the public good. I don't think it gets to be valuable unless it's a public good. Well, I mean, you can argue the same. Google came out with all this nonsense about the public good when they started, and it's certainly debatable whether Google's been for the broader of the Internet. But that's another issue. So, what did this letter say, the open letter asking for a pause? What was the point? Well, through the eyes of the authors, the point was to mitigate the risks of large language models influencing human behavior or thinking by pausing further development on the models. They weren't asking for chat GPT-4 to be closed down. They were asking for chat GPT-5 to not be developed. I mean, we know that Google, for example, are working on their own version of generative AI. Is this a pause letter for everyone or just for OpenAI? It's for everyone. I mean, it's a little bit of an incoherent letter. It's for everyone. Who was the lead on it? Who was inspiring it? The lead is this organization, what's it called, Institute for the Future or something like that. It's called the Future of Life Institute, which, by the way, has a somewhat shady reputation. And there was a lot of blowback on the letter because of who published it. And initially, at least, there were lots of fake signatories to the letter, high profile people that actually didn't sign it. So, it wasn't very well. You're wearing a Google hat versus a Microsoft or OpenAI hat. The call to pause this certainly would be for the benefit of OpenAI because they already have a lead. It's like pausing a race after a few laps when one person is way ahead of everybody else. Well, and it's also counterintuitive from the point of view of science. Most science goes through stages until it reaches a point at which it's as close to perfect as you can get it. To ask any science to pause midway through due to the fact that it has not yet reached the end is just an erratum. There is no end to this. I mean, I guess you could argue that this would be the kind of letter that might have been necessary with the development of atomic weapons. But it's hard to pause the science. You have some links in your newsletter, Keith, to a vice piece that describes this open letter as a huge mess. What I don't understand is how it even got any press. It's obviously a farce.

Speaker

Isn't the reverse true, that AI is not human beings? Is the open letter suggesting that the humans developing AI are flawed or that the systems themselves are flawed? No, that the users are flawed. For example, open AI could, in quotes, send misinformation out to Twitter using bots in large numbers, saying something that isn't true, and we humans will believe it to be true. That horse has already bolted. I mean, that's what Twitter is, full of lies and misinformation and propaganda of one kind or another. So why does open AI change anything? And guess what? We humans are not so stupid, and we tend to spot things as being lies, unless the lies are lies that we're part of, because it reinforces some ideological view that we hold. So this is really more about humans than it is about AI. AI is a way of increasing the capability of humans to do all kinds of different things. So what does the letter tell us about humans? That they're nervous, that they're pathetic, that they're divided, that they're disorganized, or that they're led by their own interests and they disguise those interests in ethical language? I think you have to distinguish between the authors of the letter as humans and the humans that the authors are concerned about. The authors of the letter are a little bit like Plato's philosopher kings. They believe they're superior, and therefore they have the right to call a stop, due to the fear that less intelligent humans will be badly influenced by AI. So they're basically taking a kind of a superior paternalistic view to the entire human race, saying you can't trust these people with this AI. And that at the core of it is an elitist point of view. I have to admit, I'm guessing, Keith, that this may be grist for your own mill, in the sense that you're against that, so you naturally associate this open letter. Aren't they also suggesting that we can't trust the companies? There was an interesting piece, you didn't link to it, but an interesting piece in the Wall Street Journal this week about the contradictions of Sam Altman, of course, who runs OpenAI, probably the most influential and interesting figure now in all of tech. Yeah, that's a great review, by the way. I mean, it echoes a lot of what we said when we talked about how smart Sam Altman is, which is it shines a light on... I mean, he's clearly smart, whether he's trustworthy. So the issue is also, and I think it would be fair to say of this letter, Keith, that we can't trust the Sam Altman's of the world with our future because he's full of contradictions himself. Well, that's the elitist view. It's basically saying... Because he isn't doing the things that we would do if we were running this. So it's all about... Well, that's counter elitism. I mean, if there is an elite in the world today, it's the Sam Altman's and the Keith Teers of the world. No, I don't agree with that. I think democratization of information means treating every human being as equal as a recipient of information. And the minute you start to diagnose some human beings as less capable of receiving information, you are turning them into basically animals that have no brain. So I think at the core of this critique of AI is an inherently anti-human point of view. Is this just the sensitive northerner in you, Keith, the Yorkshireman who's always been patronized by the all-knowing southerners like myself? Well, Andrew, you've hit a nerve there. I have hit a nerve. You know what we say to people like you? What do we say? Shine on. Although you don't need to sign on, Keith, because you're a wealthy northerner. I prefer the Pink Floyd version. Shine on. Well, this debate is not going away. Keith is very clear. He doesn't think we should pause. I agree, actually. But, I mean, everyone talks about prioritizing. No one's going to write anything suggesting that GPT-4 or generative AI shouldn't prioritize ethical considerations. But everyone dresses ethical considerations up in their own self-interest. This is certainly the theme of the week, Keith. Lots of interesting pieces you link to. Vinod Khosla, who, I don't know if he's a friend of democratization. He's certainly part of the new financial elite. Not always a man who does well in terms of marketing and PR to the rest of the world. You have a link with him on how he believes AI will free humanity from the need to work. What is Vinod saying here? Well, I definitely picked up on this one due to my own philosophical view of work. Where I think of work as a necessary evil until technology gets to the point where it relieves us of the need to work. So, I'm very aligned with Vinod there. You're right on the democratization stuff. I mean, he definitely doesn't want his beach democratized outside his house. So, he's a complex individual. But on this, he's 100% right that if it's possible for us to adopt large language models and chat interfaces and connect them to functional parts of the world like robots, it's very likely that the meaning of work will change. That we will all have a friend, if you like, that can help us achieve tasks that will lower the amount of time we have to spend on those tasks. And that that will revolutionize work. It revolutionizes where it creates new work. And of course, the age-old thing, we talked about this a million times on and off the show, is it's transferring wealth and power to the owners of this new platform system technology, whether it's Altman or Koestler, whoever. I mean, how does that... They're still going to be work. I mean, people operating these systems are going to be working. Yeah. Why is that different? I think if you follow the money, think of it this way. I'll just use one example. The California pension scheme invests in Koestler's fund. Koestler takes that money and invests it in AI, which then makes a big pile of money for Koestler Ventures, of which 100% of the original investment goes back to the California pension fund, plus 80% of the profit. And that goes into the pockets of pensioners. So actually, the idea that the rich get richer through this is partly true, because 20% of the profit goes to Koestler and his fund. But it's also true that society gets richer through these mechanisms. So now, in the future, and this is the Sam Altman profile in the Wall Street Journal, in the future, would it be good if 100% of the value went back to the people? Sam Altman says yes, and he has a project called WorldCoin. At least part of the inspiration of WorldCoin is to try to figure out how to make that happen. So it is true that as automation replaces human labor, more and more of the value created from automation should go back to society. Yeah. We heard this with the original internet revolution. I don't see much evidence of it. One man who's been following this clearly and a friend of mine and yours, Keith, is Albert Wenger. His last book was The World After Capital, which is describing this world. You link to a piece from him about thinking about AI. What does Albert suggest on this? What's his argument? So Albert wrote this blog post actually as a kind of a note to himself. And he published it so other people can see his thoughts. And it is incomplete thoughts. He didn't finish kind of an end-to-end narrative there. But he put some markers down. The most significant thing he does is that he focuses in on the fact that ultimately computers will do almost everything humans do better than humans. That's his core belief. And I agree with that belief. And I think it's a good thing, not a bad thing. And it's socially good because it frees up time for leisure and choice. But that's his core. And then he points to his book as a place where he elaborates on that theme at greater length. But I think it's a holding piece, not an end thought. I will say I have a private dialogue with Arthur about the application of AI to venture capital. And on that, he completely disagrees with me. Because you're threatening him. Your little startup, Keith, which is not so little anymore, is threatening Union Square Ventures. Maybe one day Albert will quote-unquote work for you. I don't think either you or Albert really need to work. You also link to a couple of interesting pieces on Stratechery, which is one of your favorite blogs or publications. This one, chat GPT gets a computer. What does that mean? Well, last week we talked a bit about this, that OpenAI released some plug-ins, all kinds of plug-ins. Expedia has one. OpenTable has one. Where applications can leverage the intelligence of chat GPT or its learning power for a specific purpose. And what that does is it basically links chat GPT to the Internet. And up until now, it wasn't linked to the Internet. It was kind of a closed environment. Now it's linked to the Internet. It can trigger events in the real world, if we count the Internet as part of the real world. So chat GPT getting a computer is really chat GPT being able to be linked to all kinds of other environments that have computational power. And that represents a shift at a high level. Now, this is a long article. There's a lot more in it than I just said. Well, it's more complicated. Can this actually be banned? You don't link to this because it came out too late. But the Italians yesterday banned GPT, which is an astonishing conceit, I guess, or hubris on the part of the Italian government to simply ban chat GPT. Are they banning, basically, if they're doing that, are they banning the Internet? You know, what they're trying to do is leverage Europe's privacy laws and use them to ban chat GPT on the basis of privacy invasion. I haven't drilled down into the details. It only was announced yesterday. And, I mean, obviously, it's not possible for them to actually ban chat GPT. But the fact that they want to is kind of interesting in and of itself. Wasn't King Canute a Northerner, Keith? King Canute was certainly an Italian, I would say. No, he wasn't. I'm joking. I'm talking metaphorically. I know you're joking. Wasn't he from Yorkshire, King Canute, or was he Norwegian? I actually don't know, Andrew. Is the King Canute metaphor appropriate that anyone now trying to ban chat GPT is like Canute, standing in front of the ocean and suggesting it should stop? Exactly, it is. It is exactly that. It's not going to stop. And as for the origins of King Canute, I am now in chat GPT and I will ask it. Where was King Canute born?