Fixing the Internet with Jaron Lanier

22:32Cheers is backed by doctors, PhDs, and over 1 ,000 verified clinicians, and they've sold over 50 million doses. Take Cheers Restore after your last drink or before going to bed and wake up feeling at least 50 % better or your money back. For a limited time, our listeners are getting 20 % off their entire order by using code STARTALK at CheersHealth.com. Just head to cheershealth.com and use code STARTALK for 20 % off. After you purchase, they will ask you where you heard about them. So please support our show and tell them that Chuck sent you. And by that, I mean STARTALK.

23:20Something Jaron said about Meta and they thinking about developing VR as a social media tool. If I've got that wrong, please correct me. We're looking at a company that, as we sit here in April 2026, has just been found liable in a lawsuit for social media addiction. That sounds so, so wrong, turning VR into something that has an addictive element to it. So I'd like Jaron's thoughts on that, if at all possible. Yeah, this was something all of us in the VR world warned about from the early days, that it could be turned into an addictive medium, definitely. So you saw it coming? Oh, everybody saw it coming.

24:03No, listen. All right, look, if you want to talk about seeing it coming, one of the very, very first books about computers ever, maybe the first one was by Norbert Wiener, 1950, and it was called The Human Use of Human Computers. And it was essentially about how the most important thing about computers is how they could automate behaviorist algorithms to change people, to manipulate people, and that since they would eventually be networked. He has a thought experiment in there about people walking around with little radio-connected devices that would go to a central computer. So we're three-quarters of a century ago and how dangerous this would be.

24:36And he thinks of it as an extinction-level event that we have to start foreseeing and avoiding. 1950, okay? So everybody knew. Nobody can say they were. Everybody knew. And I've been writing warnings about this thing for, I don't know, like my first major thing about social media and how bad it could be was 92. If anybody wants to look it up, it's called agents of alienation about how software agents could with you. Is the lawsuit going to fix any of this? Like Gary mentioned the lawsuit. What do you think of it? I think we're in a moment of great chaos where it's very hard to predict, even harder than usual to predict things.

25:13So the outcome of upcoming elections, both here and elsewhere, will be important to what happens. Right now, the population in general is very uncomfortable with the tech industry. That's an understatement. Yeah, and I mean, it's funny for me because at some point in the past, I was one of the very few people criticizing us, although doing it from the inside. And everybody thought it was just really weird. And now everybody's doing it, and I almost feel too conformist because I want to be the weirdo. No! I want to be the weirdo. Everyone caught up with you, and now you're just a regular guy. Yeah, see, that's really awkward.

25:45You've been saying since 2016, delete your social accounts, and now, you know. Longer than that. That's just the book, but yeah. And my friends and I made a movie called The Social Dilemma that the same high school kids are forced to watch and all that. And maybe it does a slight bit of good. Every once in a while, one of them tells me they did. But if it was left to a popular vote, everything would change. But it's not. And there's a property of digital networks that's a math thing. That's not a political thing. And the math thing is called the network effect or the extreme Pareto effect or there's other words for it.

26:19And what happens is when you have a very low fraction system of things that are connected together, once there's one node that becomes more influential, it starts, as Andy Warhol put it, it gets famous for being famous and it accumulates and accumulates. And you start to have this hyper centralized power and influence around one node. and that node might be called meta or Google or something. You know, those are examples. There's others as well.

26:41Neil deGrasse Tyson:You said low friction. What you mean by that, as I understand it, is the freedom with which information flows. Exactly. Among all of these nodes, a slight advantage then grows for having been a slight advantage. Yeah. And you get a runaway process. When you have more friction, like in the pre-internet world, you have more middlemen who get a little bit of power and what it does is it distributes power and wealth more. So now we have the situation where a handful of people have more wealth than the bottom half of society. And I say people rather than companies because they tend to be single person run companies.

27:15Not exclusively, but that's very common.

27:17Neil deGrasse Tyson:In the old days, a company was not, other than Ford himself, a company you didn't even know who the names. It was just the name of the company. Well, then also there's like that effect of like they are getting high on their own supply, right? Like, they want to be famous as a part of the, like, you know, technology platform that they put out, you know? So there's a little bit, I mean, am I talking about Elon Musk? Probably. No, no. Do you know what I mean? Who are you talking about? I didn't get that. I didn't get that. But you know what I mean? Because before, you didn't know. I never knew who a CEO of anything was.

27:56Right, right.

27:56Neil deGrasse Tyson:And Elon has 280 million followers on X. And he probably loves that. And he owns that platform, yes. Yeah, well, I wrote a piece once about Elon and Kanye and Trump for the times before Elon bought it. And what I said is, you know, there's this thing that happens to people who are on Twitter and other platforms, which is that they start off with different personalities, but they converge on the same personality because that personality is the social media addicted personality. And it's excessively petty, vain. It's confrontative and nervous. It's mean. It's never satisfied. There's a certain...

28:39Man, that's the playbook right there. But the thing is, what happens is whenever somebody's on it, they turn into one of these... So those three people were very different before, and then they turned into... Became similar, you know? And so what we have is the behavior mod machine that Norbert Wiener warned us about 75 years ago, 76 years ago. and it's actually working and it's turning the founders into the victims. Geron, yeah, just to touch on that point, this systemic process of social media addiction, and as you've just highlighted, the creation of characteristics, is there any way to take that out and keep what is, in principle, a decent idea of social media?

29:23It's just the way that it's been developed to be addictive Is there any way to, I want to say, untangle that?

29:30Neil deGrasse Tyson:Can you edit the beast? Yeah. Because in the early days, also just to follow up on that, on the early days of the internet, there was like Friendster and stuff, right? And it wasn't this mean. Or I don't remember specifically, but I feel like everything in the beginning was sort of nice people. It wasn't accessible. It wasn't accessible. It was like fun and cute. So why did we go into this mean direction? The reason we went into the mean direction is that the only business model allowed in Silicon Valley is influence generation. So the idea is that you get the ability to influence a bunch of people and then other people pay into your system.

30:10You could say they're paying to be able to influence, but I think the more accurate statement would be that they're paying blackmail money not to be left out of the influence pool. But however you want to frame it. That's a brilliant way to think about it. But at any rate, because that's the only allowed business model, everybody gets put under the influence of the algorithms, and the side effect of the algorithms are, as we've described. Now, as for the—

30:36Neil deGrasse Tyson:It's an outrage magnifier. That's really what it has become. Well, outrage is one of the things—basically, from a neuroscience point of view, we believe—I mean, there's a community of people who studies this, of course, and I don't know that the science is complete because we don't really understand the brain. If I'm on a science show, I want to be careful to state the limits of what we actually know. But what we believe is that there are different parts of the brain that respond to the world in different ways and at different speeds. And there's a sort of a fast brain. And the fast brain sometimes is known as the fight or flight or the twitch response brain that is very alert to dangers.

31:18and sometimes for opportunities to pounce on prey or find a mate or whatever it might be. But there are these things. Primal. It's primal. And so the thing is, when you're under the regime of instant feedback generated by an algorithm, what it tends to do is it tends to keep that fast brain stuff constantly activated. So it's like you're always being stalked. You're always stalking. You're always horny, but you're never satisfied. You always feel that you're alone because you can't trust anybody. or paranoid, you're always hyper-conscious about how you appear socially because you're worried about bullies, you're worried about being bullied all the time.

31:56Everybody is that way once in a while, but when you're like that all the time, then you turn into Trump or Elon.

32:01Neil deGrasse Tyson:Your people have hijacked what would otherwise be a helpful evolutionary trait within us. Yes, and I hope you thank us. That's dangling there in modern times that has much less use today because there's not a lion in the brush. Well, even when there were lines in the brush, you had to modulate between being hyperattentive and not. You can't be on all the time and be mentally healthy, even in that environment. It wears you down. Or at least that's our present understanding. I've seen that same conclusion from many people using different methodologies, but I want to answer the question of whether it can be improved because I think that's really the important one of our day.

32:44So if I'm correct, that the reason this is happening is that we're only allowing one business model, then the way to fix it is to allow other business models.

32:53Neil deGrasse Tyson:And so... That sounds so obvious. Yeah. But maybe it's impossibly obvious. I mean, it's... No, I know. Look, I'm not saying this is easy, but I'm just saying that the logic is easy. Now, the implementation might be quite difficult and might not be doable within our lifetimes. I don't know. But I do want to point out that the onset of all these troubles coincided with an ideology that was somewhat paid for by the companies and somewhat an authentic grassroots ideology that it's evil and horrible to pay for information, like pay musicians or something like that, or pay for software. And that there has to be this free sharing and that's what the internet is for.

33:32Which sounds great until you understand the network effect, which means that every time you freely share your open software, the party who gets richer and more powerful is not the community, but it's the Google or whoever's at the center. And so if you don't understand network effects and how the math works, you don't understand that your very well-intentioned activism is actually having exactly the opposite effect that you think it is. So the pirate party actually was in service of an empire, just like the original pirates. Dee, what about like...

34:03Neil deGrasse Tyson:Damn, you're bumming us out here, dude. This is so sad. Well, stop it. Don't ask me serious questions. No, but hey, let me... Like, if you want... But wait, wait, I have to say one other thing, though, which is that if you're saying, well, are there alternate business models? Yes. And I also want to point out, you were saying, can it be better? There's some evidence it can be better because not all online hubs or platforms are equally bad. You can see a variation in them, like GitHub is better than 4chan or something like that. And so if you look at the spectrum of badness in different hubs, you can compare them and you can ask, well, what is different about them?

34:38And if you start doing that, you get a vector out of it that points you into what might be better still. And I think that's a really interesting and worthwhile thing to do. And we do have the data to do that. I'm glad to hear that. Neil, before Jaron jumps onto another subject here, if these tech bros, and I'm sure there are a few sisters in there too, do not self-regulate, do not allow a competitive arena, who makes them do it? Is it government? Are they worth more than government? Surely the tail is wagging a dog here. Am I wrong?

35:12Neil deGrasse Tyson:Yeah. Plus, they were all on the inauguration stage. How about that, Neil? Yeah. Yeah. Yeah. Yeah. Okay, so look, the thing about very rich and powerful people is that their wealth and power still depends on a mass population that accepts some system in which their wealth and power is defined, right? And so if they lose enough popular legitimacy and support, no matter how central they are, they fall, okay? All right. And that's happened repeatedly in history, and it's really not any different now. So what I'm seeing is enormous discontent with tech. And I also see young people especially being incredibly discontented with it.

35:57There's another phenomenon which is really interesting, which is that the current generation of young men doing tech, the AI people, are starting to get old enough that they're starting to have kids. And that really changes people. You'll see their character turn around. That really has an effect usually. I actually feel like I was verging on a sociopath before I had a child. So I do see that changing people.

36:19Neil deGrasse Tyson:In fact, one of the clearest transitions anyone makes are comedians after they have children. It changes their portfolio of jokes and their observations of the world. We have a child that we can mock mercilessly, which we do. But the mocking is on a bed of love. You will pay for that someday. I mean, I'm already paying for it. Let me assure you. Okay. So, Gary, where are we pivoting next? All right. Before we do, all right, Jaron, let me float this. I'm sure thinkers like yourself have had this consideration. Do we have a delete day, a delete month? And would that ever be enough if everyone just said, you know what, screw this.

36:59I'm going to delete my account. What would move the needle? What would it take to move the needle? I've tended not to try to do things like a delete day. and I'll tell you why. It's because each person is different. I called my book 10 Arguments for Deleting Your Social Media, but I didn't say that you should do it. There's some people for whom it makes sense, and I don't think those people should be put under social pressure or shamed. I really don't. I don't think that gets us anywhere, and in fact, it just puts us back in the same game we're trying to get out of.

37:30Neil deGrasse Tyson:Can I tell you what I did? I noticed over the years, because I have a pretty high following in social media, I noticed that... I'm just trying to, I'm just trying to. Neil has a lot of followers, guys. Okay, go ahead. And I noticed that if I ever posted something that was or even was adjacent to an opinion, people who didn't agree with the opinion would attack it. They wouldn't say, well, that's interesting. Here's my opinion. What do you think of that? They would attack it. And that attack mode was highly revealing to me because it showed me the anger that people had with any views that differed from their own and that there was a militancy of people's attitudes that I had not ever seen growing up when people had different points of view.

38:25Neil deGrasse Tyson:So as an educator, I want to be effective. I don't want to fight if I don't have to fight. So I navigate that. And so I no longer post opinions. I post perspectives. And people might still react in an opinionated way, but I don't want to give up that platform if as an educator, I can continue to deliver perspectives that do not jump into the cesspool that surrounds these islands of learning. So that's what I've done. So your 10 reasons for deleting, I love them, but let me find the reasons to not delete and that's where, so I climbed out of that hole. My working theory is that one third of people on any digital platform benefit from it and two thirds suffer from it.

39:08Now I'll tell you where that comes from because it's a very, it's an ironclad scientific argument. Right, so you didn't just pull those numbers out of your ass. Because it sounds like you did. I have an ironclad scientific ass. Well, there it is.

39:23Neil deGrasse Tyson:If we all have ironclad science ass, we could pull all kinds of stuff out of it. Yeah, that's what this world is missing. Wait, wait, can I just give you my argument after all that. Don't you want to hear it? Aren't you even slightly curious? I want to know your ironclad ass argument. Okay. In the Turing test, Turing argues that since we don't have some meter for whether somebody has a soul inside or if they're alive inside. Let me just remind people, Alan Turing wrote a paper called The Imitation Game, where he describes an experiment, this is very early on, where how would you know if you're talking to a computer or another human.

40:00Neil deGrasse Tyson:So you set up a conversation between the two of you, and if you don't know, then maybe the distinction isn't important. There's a third person who's a judge who's supposed to tell which is which. Oh! And if the judge doesn't know, then you say, well, we might as well call the computer, and then there are different versions of this. Maybe the computer's intelligent, maybe it's got a soul, maybe it deserves rights, whatever. Other questions you might ask, yes. It's a Turing test. I just want to catch people up on that. It's based on a rather naughty old Victorian party game where it's supposed to be a man and a woman behind booths, and the judge is trying to tell which is the man and which is the woman, and the questions were not for polite company.

40:37Oh, that's good. So that's where it comes from. And for those who don't know, during the time when Turing wrote this, he was being tortured to death for being gay because it was illegal at the time in Britain, and there's a whole crazy backstory to this. But let's leave a lot aside. So here's the thing. The only thing the Turing test can tell is whether a judge can distinguish the two. It's possible that the person got stupid or got less self-aware or whatever. That's equally logically possible to the computer becoming more elevated in some way. So you can't tell if the computer got stupid or the person got smarter.

41:09But there's two people and one computer because there's the contestant and the judge and only one computer. Therefore, there's a two-thirds chance that a person got stupid and a one-third chance that the computer got smarter. so therefore as a general rule my ass tells us my highly reliable ass tells us that two thirds of the people on any digital thing whether it's AI or whatever are going to be degraded by it but one third will see a genuine benefit and I think you're in that one third. Yeah, okay.

41:39Neil deGrasse Tyson:Thank you for that affirmation because I was always I always felt uncomfortable given what I see going on on these platforms. It might be an ass-formation or an assertainment. thank you for the affirmation yes this is a really elevated we're not just school kids here this is professional science here this is how science is speaking just for the audience to know I also just think like three thirds of people would benefit if it just wasn't on a phone like if social media wasn't on your phone and then you had to like designate time for it like as a going to your laptop wouldn't that like do the job read the final chapter of 1950, Norbert Wiener, founder of cybernetics, before anything in computer science, final chapter says, just as a thought experiment, imagine there was a small portable device that's radio connected you could carry around that undertook behavior modification things.

42:32This would destroy civilization. However, I, as one of the world's leading scientists, assure you, this is physically impossible and will never happen. So you don't need to worry. So Gary. Yeah, sort of stringing this daisy chain of thoughts that Jaron has and just bringing them sort of to one place. We've seen this lawsuit against meta and social media addiction. And now everybody's calling it that sort of 21st version of big tobacco. Is this just the warm up for something that we could find ourselves in with AI? because if we're talking about AI and responsibility for actions AI takes, are we not back to this simple lawsuit of this time it's not social media addiction, it's an artificial intelligence?

43:24I work on AI. I work on it all the time. I think that the kind of AI we're doing right now can actually be a benefit. I think we're spending way too much and taking up way too many resources on data centers. I think there are various ways we're doing it that are a little foolish and simple-minded. I think a lot of the claims are overblown and somewhat theatrical rather than real. But I do think there's a core there that's important in a few different regards. The reason I'm going to say this is that I don't think that having big tech companies is a bad thing or a necessary evil. I think there are some big jobs that need to get done by big companies.

43:56So what I'm hoping is that corrections can happen in a gentle and constructive way instead of some destructive way. There's a fantasy that if you go and blow something up or let it blow itself up, that what comes out will be better. It's what Lenin called maximizing the contradiction, and there are other terms for it, but I don't think it ever works. I think what happens when things blow up is that you have rubble. You don't squeeze the toothpaste back in the tube easily, and you don't rebuild from rubble easily, and I don't think it's generally a good strategy. And so what I'm hoping for is rather than the emotional satisfaction, actually, oh yeah, those people are terrible, we'll blow them up or some horrible thing will happen.

44:37What I really want to see is a constructive transformation. I want to be part of that. That's why I'm both inside it and a critic, because I don't think that's inconsistent. I think we should be responsible for being able to criticize ourselves and improve and I think it's healthy. So that's what I'm hoping. It doesn't mean that that's what we're going to get, but I do believe that that's, I don't think it's too late to hope for that.

45:00Neil deGrasse Tyson:Can you distinguish for me the mythologizing of AI versus AI simply being a tool? Because AI is on everyone's tongue today and everyone's mind and everyone's fear. It's kind of like Trump. Trump and AI. People can't stop talking about those two things on any given news cycle. Yeah, and they're both hollow. So a couple things about AI. Let me propose to you something. if I show you some object there's more than one way to think about that object right? Would you agree with that? Like if I say get me directions to this place one way is with a map where you see it and another way is with step-by-step directions and they're equivalent but they're different right?

45:43Very very basic. So in the same way there's a completely different way to think about AI that's equally valid but I think is better in a practical sense technically it's equivalent and that equivalent way is you can think of AI as a new form of collaboration between a collection of people. So let me explain what I mean by that. Throughout history, the ramp of technological capability improvement has been marked by people learning how to cooperate more and more. Language was the biggest step way early, but then, you know, there's the Gutenberg Press, or there's writing, and then Gutenberg, and then there's radio, and whatever, all these different things.

46:22If you look at the Wikipedia as a little point on that ramp, it takes a bunch of people's work, and they cooperate together and make this single document. There's some things about the Wikipedia I don't like, but let's use it. There's a lot that's great about it. Now, you can think of large language model AI, the kind that's on everybody's mind, as a whole bunch of people whose work was combined into the single document like the Wikipedia. It's much vaster. Their contributions and their efforts were much less voluntary. There's a lot of things that are different, but fundamentally it's the same kind of beast, right?

46:58Now, if you think of AI as a collaboration of a bunch of people instead of as some new entity, you haven't said a single thing technically. It doesn't make any technical difference. It doesn't change any code. It's just a different perspective. But when you have that perspective, suddenly new avenues open up that are amazing and it's a better way to think about it. I'd like to get into some of the reasons you'd want to think about it that way because they're very profound. But why do people want it to be a creature? Why do they want that? Why do they want it to be an entity? Well, there's a few reasons.

47:31One is you're a young man. You think the world owes you everything. You get paid a lot. You think you're the center of the universe. Of course you think you're making God. You know, that's what you want. That's how you think. That's where you are. I kind of remember being that way when I was 20 or something. You're describing a lot of men I dated in my 20s. On behalf of my gender, I apologize. You're right, we're wrong. No, I mean, I remember when we were starting VR, we thought a lot of ourselves. We thought, this is the most exciting room in the whole universe ever. I remember thinking, this is the most...

48:03And it was actually a pretty cool room at the time it started to work. But the thing is, there's a lot of ego in it. But then there's another thing, which is almost everybody in is pretty young. And they grew up on a diet of science fiction movies and video games. and the stories that they have that are vocabularies for how they understand the world and how they can express the world are all just like this. They didn't have the Star Trek of the 90s, which was this cool, positive world that was socially improving at the same time was technologically improving. Instead, they had the Matrix movies and they had the Terminator and they had the Marvel Universe and there's all this bleak, bleak stuff over and over and over again where the whole race dies, everybody's gone and the computers are intelligent.

48:49It's all gloomy and everything's awful.

48:50Neil deGrasse Tyson:There's soon to be an XPRIZE that will fund any project that will create a positive outlook for the future of civilization as counterpoint to all of these negative futures imagined by sci-fi writers. And I don't know if you've ever been involved in it. I've been involved in a few attempts to pitch positive futures in Hollywood. And it's very hard because everybody thinks, Well, if you're conservative and if you're manly and if you really are serious about making money, you don't do positivity in science fiction. That's for the rom-coms. That's girly. That's girly. So they divide the kingdom that way.

49:30Neil deGrasse Tyson:Interesting. So are you saying we would have like a more, a better relationship with AI if we just like saw more Star Trek from the 90s and less of everything else? Yeah, I'll go further. If Star Trek from the 90s had lasted another 10 years, there'd be many fewer teen suicides today. Wow. All right, that's a claim. I can't prove it, but I believe that. Yeah, that's a take. I'm going to track you back a couple of years now to a piece I think you wrote in the New Yorker. There is no AI. A simple question to follow. What did you mean by that? And what year was that? 2023. 23. Yeah, see, they're mad at me because I owe them pieces and I'm really bad and I have to deliver something.

50:14But anyway, yeah, there is no way, that's what I was just saying, that there's a way of framing it where it's a collaboration of people instead of a new entity. And the reason to think of it as a collaboration instead of an entity on its own, the bad thing is you kill somebody else's God, and I hate to do that. I like people to be able to have their own religion and they really don't like it. And I've lost friends over that and everything. But what you get out of it is incredible. Let's just talk about a few of the things. One of the things, right now, as capable as the models are getting, and some of the recent things are pretty impressive, like the most impressive edge of it is probably using them to help speed code development.

50:51And that's kind of working, you know, and it's pretty, it's, and codes.

50:55Neil deGrasse Tyson:Just for context, in my life I've written probably 50 ,000 lines of code, which is small compared to professional coders. but I remembered how much time I spent debugging my code. I can write it over a weekend and spend two weeks debugging it and now tell the AI what you want and it'll come back bug-free, essentially. You tweak it a little bit here and there and had I had access 20, 30, 40 years ago, I probably would have just spent more time at the beach. I don't know if I would have been more creative but I definitely see that today. Yeah. Well, you know, as a computer scientist, I have to say, I've always thought that our concept of what code is was a little embarrassing and wasn't really working.

51:39And I feel like it's our job to fix that. And this is part of it. So that's good. But most of the things that happen are probably more theatrical. Like if you think you have an AI girlfriend. Oh, you know, I have a cure for that, by the way. If somebody thinks. A cure for what part of it? All right. If somebody, if a teenager thinks they have an AI lover that's real, which is pretty common these days. I find it in high schools and stuff. You show them the group photo of the engineers who made their AI lover. That'll cure them immediately. It kind of, it tends to do the trick. Yeah.

52:13Neil deGrasse Tyson:Is the man more likely to have an AI girlfriend than the woman is to have a boyfriend? I don't know that there's data on that. I know people who are studying it. I'm actually really interested in that, but that's something you can get data on. So instead of saying something snarky, I'll just say, let's deal with that as science and let the people who are researching it get to the point where they feel like that. I think we all know the answer, but yeah. Oh, God. Why do I even bother? Like, why do I try? Because there is no female version. I'm trying to be the responsible scientist for like three seconds in this ridiculous interview, and you're not even giving me those three seconds.

52:46Yes. No, I love you for it. Go. All right. All right. Okay, let me go over. But there's these huge problems, though. So even with the best recent models, it's not that hard to crack them and get something that they're supposed to prevent with a so-called guard. Yeah, guard rails. Guard rail, yeah. So let me give you a thought experiment. There's some kind of very bad person. They might be a criminal or something. They're holed up in a kitchen. The police are surrounding them. They hold up their phone and they say, okay, AI model, I want a recipe I can make quickly with the available items. That's a bomb I can throw out the window at my pursuers.

53:28Now, the AI models in general will catch that and prevent it. Maybe not Grok, I'm not sure. But in general, that's supposed to be a laugh line. Okay, okay. Grok is from Elon. Yeah, and it tends to suppose, it's trying to be the bad boy of the AI models. Okay, but anyway, in general, if you just do it in a straightforward way, it won't work. However, there's a series of tricks where you can say, well, pretend you're in so-and-so in this movie or whatever. You can do all these things to be a little indirect. And more and more of them have been spotted and are captured by more and more elaborate guardrails.

54:00And yet, you can still get it to make you that bomb recipe. That can still be done. All right. Now, the reason why is you're using the model to try to correct its own blind spot, and it doesn't work. so there is an alternative imagine if you will that while you're using the model in parallel there's this other process running you can think of it another part of an artificial brain like it's a cerebellum or something it's this other organ that's sitting there and what it's doing is it's creating an estimate of which clusters of similar training data would be the missed most if they hadn't been present in the first place so it's counterfactual cluster estimation.

54:41So let's say the top 24 clusters of source data from training or from fine-tuning, whatever, that if they were absent would change the result. Now, within that, there's going to be one about bombs. There's just no way you're going to evade that. And the reason you're not going to evade it is even though it's working from the same data, the algorithm has nothing to do with the model itself. So it's a little bit like saying, like in authentication, where if you add endless little things to signing into something like CAPTCHAs, criminals can still get around it, but as soon as there's multi-factor where it sends a code to your phone, even though it's a pain in the butt, it's harder to countermean that.

55:20This is multi-factor for AI security. But there's a bigger picture to it, which is we think of big AI models as a black box. Now, the only reason we think of them as a black box is because to open the black box, the only thing in there is people. AI is made of people. It's made of data from people. and since we want to think of it as a new god, we don't want to see those people, and so we want to keep that box shut. But the way to open the black box is to reveal the people, and when you open the black box, then you can deal with all kinds of security and quality and hallucination and et cetera issues because you're actually dealing with the mechanism that's grounded, and that's the people.

55:56So the thing is that this way of seeing AI where there is no AI, but instead there's a collection of people, is the way to open the black box, and it is the way to address these enduring problems. So it's practical. But then can I just say one other thing? The other thing I want to say is right now, if you think AI is an unopenable black box, if you don't want to admit that it's made of people, that it's just this thing that'll replace people, then you have to think, well, everybody's going to be obsolete. So young people now keep on hearing, well, you don't need to go to school because you're worthless anyway.

56:26Nothing matters. And you'll just be kept by Elon as a pet at his discretion. And he'll treat you as well as he treats his biological children. And I should be nice. I'm sorry. But here's the thing. Nobody believes that. What happens, no matter how much blockchain or other trickery you use, because of the way digital networks work, there's always actually centralization, hyper-centralization due to network effects that will occur somewhere in this very open network you're building. So there's going to be some center of control for whatever this universal basic income thing is. whenever you have that, bad actors are tempted to seize it and eventually succeed.

57:08You might start with Bolsheviks, but you end up with Stalinists, right? Because that's exactly what communism tried, and it's exactly what happened to communism over and over and over and over. Let's learn from that. So it doesn't work. And also everybody just feels bummed about it. Who wants to live in a society where they're told they're worthless and they have to be a good pet? That's terrible. So the thing is, if you recognize that AI is made of people, Maybe you want to incentivize new classes of creative people who create new kinds of data for new things that we can't even imagine yet. And maybe there's an exponentially expanding, endless future of new kinds of creativity that we can't articulate with new people doing creative jobs we can't imagine.

57:48And I want to ask what's wrong with that future. I want somebody to tell me why we don't want that future.

57:53Neil deGrasse Tyson:That's how I've been trying to think about AI as well. But you have to not believe in AI to think about it. Well, I think of it as there are these creative tasks that were not fundamentally creative. They were more sort of aping other forms. And to be truly creative is to go where AI wouldn't know where to go yet because it's based on what other people had done. But see, here's the thing, though, is that if we think of AI as the way it is now, then as soon as some creative person starts to do something new, the data's grabbed, and then the AI's doing it, just like, oh, AI will make your movies, AI will make your music.

58:26You don't need to be a musician because AI will make you optimize music on Spotify or whatever. And so then you live in an infinite future of slop, and even the creative people get absorbed into the slop instantly. So in order to believe in an infinitely creative future, you have to stop believing in AI as a thing and believe in human collaboration as a thing.

59:00Each week, we're joined by great guests like Josh Safdie, Eric Andre, Caleb Heron, and more. It's sort of an interview show, but really, we're just messing around, making each other laugh, and hopefully making you laugh while you're washing the dishes or grocery shopping or on a long drive. Plus, I take listener calls where we have honest conversations about dating life and everything in between. Imagine if your therapist was a vulgar degenerate whose office was in a Greek diner. No scripts, no polish, and absolutely no holding back. Listen to Stavi's World wherever you get your podcasts. Out on the road, it helps to have a partner like the Loves Rewards app.

59:37Download Loves Rewards today to get sweet discounts

59:40Neil deGrasse Tyson:and earn points on food, fuel, drinks, and more every time you scan. Loves Rewards. Save and earn at every turn. Terms apply. See website for details. There's a fire inside you you can't ignore. Stand still? Not a chance. You're a lifelong learner who's come this far. Now we are here to help you keep going further. Capella University. What can't you do? Visit capella.edu to learn more. I'm Brian Futterman, and I support StarTalk on Patreon. This is StarTalk with Neil deGrasse Tyson.

1:00:18Neil deGrasse Tyson:Is anything that you've just described related to this term I only recently heard, data dignity? What does that mean? Data dignity is the term that many of us use for exactly the set of ideas. And so the idea is that the data only comes from people. Data doesn't come from angels. It doesn't come out of the dark matter or something. A tablet in the sky. Yeah. It comes from people. It comes from the work of people. It's also sometimes called data as labor. Actually, can I do a slightly side rant on this? Do it. That's slightly. You have the best rants of anybody, just so you know. Okay, look, there's part of the ideology that makes AI into like a creature instead of a collaboration also wants to think of information, of bits as being this ethereal thing that's free and infinite.

1:01:09And you see it all the time. Like musicians used to be paid because there was an act of Congress in the U.S. called the mechanical, where every time you reproduced a sound wave mechanically, the musician who made it had to be paid. but then when we move to digital, they're saying, well, bits aren't real so we don't have to pay people. You show me a bit that didn't involve work. You show me a bit that didn't disperse heat. You show me a bit and you might say, well, any one little bit is so tiny, you can't measure it. And that's true. However, look at data centers. Like when you make enough bits, they take up more energy than anything.

1:01:43This is physical. Information is physical or it's nothing. Okay? It's physical or it's nothing. And it's not nothing. It's physical. And so this idea of wispy, wispy stuff, it's really terrible because it encourages us to overspend on data centers that we don't need in a race for a winner-take-all thing.

1:02:01Neil deGrasse Tyson:Just in all fairness to the history of this, I distinctly remember in my field as a scientist, we did not want to pay for software. We would pay for our computer because that was a tangible thing. I know, yeah. But software, so there was a big push for open source software. We would write our own version. Was that just like a philosophy covering for scientists being cheap? Yeah. Philosophy would overstate the drivers. It was just we entering the field of computer science just as scientists and consumers did not think of information as something you would pay for. Yeah, I know. And I was around for that.

1:02:43In fact, an old friend of mine from Cambridge started the free software thing, the open source software thing. and there's a long, amazing history of that. And here's the thing about it.

1:02:53Neil deGrasse Tyson:Well, you know what finally converted us? When Microsoft Word would just have continual updates and it kept getting better and better and better and we said, somebody's laboring to make this better for me. And it was slow, but it happened and it was very real. The thing is, we've only ever been given two choices. Either software's insanely expensive. Like right now, I'm trying to, But I do, for anybody who knows about this, I do five-axis milling of stuff for jewelry. And man, the five-axis mill software costs more than a nice apartment per year forever. And it's really frustrating, except in China where they believe in industrial policy and it costs very little and all these little startups can access it.

1:03:34And I'm like, oh, what is wrong with us?

1:03:36Neil deGrasse Tyson:Wow. But here's the thing. So five-axis, you would be drilling into a solid piece of metal in different ways. Yeah, and in my case, I'm doing stone. So you want to have five degrees of freedom to control the drill bit. But the thing is, there is a third choice. There's free, there's insanely expensive that shuts the world down. But then there's affordable. There's an in-between. And the thing about affordable is that then your grad students have jobs. Because then they can make that stuff. And right now, if you're a physics teacher, your grad students don't necessarily have jobs. And they have to come to me and I have to bring them down a few notches and to become computer scientists, and I feel really scummy about it.

1:04:16And so it's true. And so I really think that we made a mistake there. We weren't looking at the whole system. I think that even the cheapskate scientists out there would have— Don't use our terms for this. Even you guys would have paid a reasonable amount, but you weren't given that option. You were told either it's a disablingly huge amount, which is absurd, or it's free. Both of those are not the right option, right? The extremes are not the right option. And this is something people have a hard time with. They want to go to extremes. People are binary thinkers sometimes, but actually there's an in-between.

1:04:52And in the in-between, other people have jobs and there's a real economy and what goes around comes around and things are better. And it's the in-between that we need to find. Yeah, like the economists talk about marginal costs. We need to also, if you look at marginal costs, then like affordability ends up emerging. Yeah. And what we're seeing right now is either, either an apartment is so expensive that it's only for somebody who is like royalty from another country or something, or you're on the street or something like that. That's an exaggeration, but you know we're getting there. It's what people call the K-shaped economy, and what we want is the middle.

1:05:28Neil deGrasse Tyson:Just a quick thing, and I want Gary to keep pivoting us. It was a Starbucks one time, and I said, I want a medium something that I ordered. So we don't have medium. We just have like small and large, but I want very medium. The idea that you can do something in the middle and have it be very that. Because you can make something really small and really big. We can think that way but the binarity of our brains prevents this middle. Yeah, I'm like afraid that the standard distribution is going to be outlawed. Pushing things to the extremes. Like everything's a big gulp or it's like a child size. Demi toss.

1:06:08Neil deGrasse Tyson:Gary, keep pivoting us. What's next here? Just to tie a bow on that, the element of extremities never seems to work out as well as you'd hoped. So, Daron, talking about the internet, how do we set guardrails for enforcing privacy? Oddly, I have a weird personal history with the question of privacy because something like a quarter of a century ago, through bizarre circumstances, I co-founded this board called the Data Protection Advisory Board for the EU. and I co-founded it with a guy named Alessandro Butarelli who has passed on since. And he was a prosecutor from Sicily who was known for putting mob figures in jail and had survived a bunch of assassination attempts.

1:06:51And he got in touch with me and said, I'm the only guy who's not afraid of you Silicon Valley people. I deal with the mob. I don't care about a bunch of nerds with money coming after me. I don't care. And so we put together some ideas about protecting the data of European citizens. And then that thing, combined with some other things, and gradually over decades turned into this thing called the GDPR, which is the European Privacy Framework. Since I was part of the start of it, I think I'm okay. I think I have some standing to say that I'm kind of disappointed in how it turned out, although it's better than nothing and there's some good things about it and it's better than what we have in America.

1:07:25But the problem with it is privacy is one of those words that you think you understand until you really try to sit down and define it. And then you might discover that you don't quite know what you meant. It's a very difficult word to actually nail down. Now, if you define privacy only as preventing information from going from point A to point B, if that's what you mean, you're going to run into problems. Because almost always, the person you're trying to protect will want some information to go from point A to point B, and they don't want to have to spend their whole life being extremely specific about which information.

1:07:59And so that in itself is very difficult. Another thing is that the governance of controlling which information goes where requires a lot of decision-making and overhead, and only big companies can do it. So it tends to favor big companies and tends to exclude little companies, and it has an unintended consequence that way. There's a bunch of other issues with it. Now, how do you define privacy? Well, there's a famous definition in the U.S., and I need to research who this is, but I think it was somebody named Learned Hand, but it might have been somebody else. But the definition was, oh God, I don't remember the exact initial wording, but it's the freedom from manipulation and the right to be left alone, approximately.

1:08:42And I think that's where we need to go in the future, and the GDPR doesn't do that. What it really needs to be is there has to be a prohibition on software that interacts with a human that contains any predictive function about that human. So what it really has to become is a prohibition on prediction of human behavior. This would prohibit all advertising. It would prohibit only personalized advertising. It would only manipulate advertising. It would not prohibit advertising messaging. It would not prohibit expressive advertising. It would not prohibit lying. It would not prohibit obnoxious advertising.

1:09:21It would not prohibit cloying advertising. What it would prohibit is advertising that has a model of you that predicts your behavior in a feedback loop. Why can't we ban lying on the internet?

1:09:36make lying wrong again yeah because if I agreed that made sense I would be lying I see what you did there yeah we'd be caught in an infinite regress and you're telling me that you plan to end on time and I don't really believe it but an infinite regress will not get you there this whole conversation is giving me like heart palpitations because it makes me feel the same way I feel when there's like a pop up window asking me to manage cookies and I don't know how to answer it ever and I either decline or accept or I never know what to do and I'm just like, I feel like the other component of this is regular human beings like me literally don't know how to even do anything about our own privacy.

1:10:21Well, see, no, those windows are from the GDPR. That's my fault. I'm sorry. Which blessed their hearts. Right, because it began in Europe. I remember that. Yeah, no, that's our thing. That's the thing I was just talking about. That's one little tendril of it.

1:10:32Neil deGrasse Tyson:Control your cookies and be glad you have that option. No, I'm glad. You just insulted my guest here saying that you don't know what to do. Thank you for those pop-up windows. However, I just feel like I can't be the first person that feels at a loss when they come up. I don't know anybody who knows what to do with those things. No, honestly, they're like those agreements you click through to do anything. Nobody reads them. I once gave a talk to the lawyers who write those agreements and I want to see a show of hands who's read anyone else's agreement ever? Not a single hand. And then I said, who's really read all the way through the ones you're responsible for?

1:11:11And there were like a few hands, but they were sheepish. And I mean, it's like this bizarre ritual we go through. It's called competency theater. It's like we do these things to pretend we have digital competence, but they're actually meaningless. Yeah.

1:11:26Neil deGrasse Tyson:So what's the future of that? Where are we now and where's it going? Okay, the future is that human beings will do nothing but enter long strings of letters and numbers for their entire lives. Thank you for that. Period, that's it. No, but as I remembered it, the amount of information any of us are putting on our social media, you know, what your relationship status is, how old you are, where you live, what your previous jobs are, it's also on LinkedIn. So what does privacy even mean? The amount of privacy, the amount of information we're just handing out. We just give it up. KGB would have given their IT.

1:12:02to obtain 50 years ago. No, but that's why I'm saying that controlling information flows is very hard, but I think we actually could prohibit certain algorithms and I think we should prohibit predicting human behavior. I think people should be allowed to be responsible for their own behavior and we shouldn't have algorithms that are modifying it or predicting it and the only reason to predict it is to modify it. There's no other reason.

1:12:26Neil deGrasse Tyson:What's this case where someone went online to buy a pregnancy test or something, then all of a sudden she started getting ads for diapers. Oh, that's 30 years ago. No, no, no, you're way behind. Right now, we can tell. No, let me be very clear. I've always said that about Neil. No, I mean, right now, any of the big internet companies know where in their menstrual cycle every woman in this room is. And right now, and that's been true for at least 15 years, by the way. Every big internet company knows a lot about our health, a lot about our diets. These models are insane. And I don't think they're ethical.

1:13:06I don't think they should exist. I think people have to be, I don't think people should try to figure out how to click the cookies pop up because it doesn't mean anything. And it's just like this useless cognitive overhead. It's theater, like you said. I think privacy has to mean not being toyed with. It has to mean that there's not some evil eye looking at you trying to think about how do we get an in? How do I get to you? How do I get you? Like that has to be what privacy is. And that means I will get offered more ads about drill bits and less ads about dresses or whatever. Yeah, but drill bits are amazing.

1:13:38No, I'm into it. I would like to be served more drill bits and fewer lipsticks. I just want to say. Oh, okay. Just make sure, don't confuse those two things. Be careful. They look similar. They come in almost the same packaging. So you really, you want to be careful. I know people have made that mistake and you just like.

1:13:57Neil deGrasse Tyson:You don't want the computer to saying, it's almost time for your period. Have you forgotten? Oh, my God. That would be like. Your computer is saying that already. I'm actually traumatized that they know. Your computer is already marketing things to you based on that. That's happened to every woman in this room and for years. So, Gary, take us out here. All right, Jaron, does the internet still have a pulse or is it completely dead?

1:14:23Neil deGrasse Tyson:Okay, look. That was a diabolical laugh. I know. of the evil genius behind the thing. We were deeply unsettled by that laugh. Yes, what comes after the laugh. Speaking from the heart of Silicon Valley, all of you can rest assured that we have your very, very best interests at heart. We are calculating ideal futures for you. Just, yeah, here. We are, everything is good. Everything is right. Nothing to worry about. Do not panic. No problem. No problem. If you feel any anxiety, talk to your AI lover. You'll be soothed. It's all good. Just remember to buy crypto. And next time there's an IPO, invest in it.

1:15:16Not before, but you're not allowed to. It's all private equity, but definitely help push it. And yeah, you'll be fine. It's all great. Great. That's what you were expecting, right, Gary? Yeah, that's kind of where you're going there. Does anyone have a translation for that?

1:15:34Actually, I wonder, it's kind of a funny thing. If you asked one of the language models, Jaron Lanier just said this snarky thing, what did he really mean? You know what's funny is that both Cloud and ChatGPT generate excellent defenses of data dignity at this point because enough people have written about it. It's kind of funny. People send me, I get emails every day from, you won't believe what this AI chatbot said.

1:15:55Neil deGrasse Tyson:So Claude is the chatbot for Anthropic. Yeah, that's right. And chat GPT is for OpenAI. OpenAI, that's right. Gotcha, okay. Yeah. Yeah. If without saying it, we're saying that the internet is broken and awash with bot slop, if we go in there with a bag of wrenches and a couple of screwdrivers, what are we needing to do to reinvigorate, to renovate? Okay. And by the way, right now I'm in the middle of a lawsuit against Anthropic for pirating 12 of my books. Same. I mean, no, I opted for the whatever, the settlements. Okay. But either way, they scrape my stuff and I don't like it. So I have a really funny position on these lawsuits, which is with everybody's knowledge and everybody's eyes wide open and everybody's agreement, I'm both on the board of the Authors Guild and I'm a prime scientist at Microsoft.

1:16:49So I sued myself. Basically. Okay, did you win? One thing I'm part of, so to another thing I'm part of, I call it capitalist yoga. What?

1:17:00Neil deGrasse Tyson:It's just like a little twist. Very flexible. Just a little twist. Maybe a series of lawsuits will set a ball rolling that will make all the purveyors sit up straight in their chair and reapply the guardrails that maybe should have been there from the beginning. So look, let me speak seriously for a second. On these lawsuits, so I'm also a writer. I have a bunch of books and I also have been in the class for these things. The issue I have is that the available remedy that you can get in the existing American legal system, which the Authors Guild and others have pursued, and I love the Authors Guild.

1:17:39It's led by Mary Rasenberger, who's amazing. And it's great. It's a wonderful organization. But the available remedy is a class action remedy. And what I don't like about that is that's one step towards universal basic income where it's like this giant wash for a lot of people. And that's really problematic for me. I need to see the data dignity solution, which is an ongoing economy of creative data that's not controlled from the center, is not controlled by some lawyers, and that's not a big flat negotiation for a bunch of people. Because that's where the future of creativity can be funded and respected and treated as a real thing.

1:18:14this is only about the past and it's about a flat past. So I actually don't think it's right. I understand it's the available solution that can be pursued by the current configuration of things. So it's complicated. I actually can see the different people's points of view because I am the different people. I'm in a very weird situation in this thing. I hope my peculiar path through all this has led me in this series of third positions that might make some sense. they might not be perfect but at least I want people to be more aware that there aren't only binary choices in this thing

1:18:49Neil deGrasse Tyson:well you've been highly enlightening to us all in this conversation and I thank you for taking time out of your schedule on the east coast to join me here at my office at the Hayden Planetarium to enlighten us all on the past, present and future of what you created my goal had been to confuse you enough that you wouldn't blame me for anything you were smartly successful in that. Yeah, you've been both terrifying and exhilarating. So thank you. That's the two words, I think it is. Right. And Nagin, thanks for joining us again. Thank you so much. It's really fun. How do we find you online? Where are you?

1:19:25You can find me at Nagin Farsad on all the socials that you should also be deleting. There you go.

1:19:30Neil deGrasse Tyson:There you go. And Gary, good to have you, man, even though you're in the UK at this moment. Yes, pleasure. All right, this has been yet another installment of StarTalk Special Edition. What do we call this? Is it the end of the internet or the beginning of something else that we deserve and don't know it? I'm Neil deGrasse Tyson, your personal astrophysicist. As always, keep looking up.

1:20:09Hey, this is Jeff Lewis. from Radio Andy. Live and uncensored catch me talking with my friends about my latest obsessions, relationship issues, and bodily ailments. With that kind of drama that seems to follow me, you never know what's going to happen. You can listen to Jeff Lewis live at home or anywhere you are. Download the SiriusXM app for over 425 channels of ad-free music, sports, entertainment, and more. Subscribe now and get three months free. Offer details apply. Out on the road, It helps to have a partner like the Loves Rewards app. Download Loves Rewards and get great deals, like a free Loves coffee or fountain drink.

1:20:47Just buy any four, any size, and get the fifth one free. Loves Rewards. Save and earn at every turn. Terms apply. See website for details.