← Making Sense with Sam Harris
Making Sense with Sam Harris
#427 — AI Friends & Enemies
00:23:28
Sam Harris speaks with Paul Bloom about AI and current events. They discuss the state of LLMs, the risks and benefits of AI companionship, what it means to attribute consciousness to AI, relationships with AI as a form of psychosis, Trump’s attacks...
Loading summary
Transcript55 lines
- [00:07]
Sam Harris
Welcome to the Making Sense Podcast. This is Sam Harris.
- [00:10]
Paul Bloom
Just a note to say that if.
- [00:12]
Sam Harris
You'Re hearing this, you're not currently on our subscriber feed and will only be hearing the first part of this conversation. In order to access full episodes of the Making Sense podcast, you'll need to subscribe@samharris.org we don't run ads on the podcast, and therefore it's made possible entirely through the support of our subscribers. So if you enjoy what we're doing here, please consider becoming one.
- [00:37]
Paul Bloom
I am here with Paul Bloom. Paul. Thanks for joining me, Sam.
- [00:41]
Unknown
It's good to talk to you, as always.
- [00:42]
Paul Bloom
Yeah, great to see you. We were just talking off mic. It's been years. It's crazy how that happens, but it has been.
- [00:51]
Unknown
I've been following you, though. You've been doing well.
- [00:54]
Paul Bloom
Yeah, well, I've been following you. I've read you in the New Yorker recently writing about AI. I think you've written at least two articles there since we. We actually wrote a joint article for the New York Times seven years ago, if you can believe that. Oh yeah, about Westworld and the philosophical import of watching it and realizing that only a psychopath could go to such a theme park and rape Dolores and kill children, et cetera. And I think we predicted no such theme park will ever exist because it'll be a bug light for psychopaths and normal people will come back and if they do anything like that, they'll scandalize their friends and loved ones and be treated like psychopaths appropriately.
- [01:38]
Unknown
We'll see. We may be proven wrong. Who knows in this crazy time? But yeah, that was a fun article to write. And I think we wrestled with the dilemmas of dealing with, you know, entities that at least appear to be conscious and the moral struggles that leads to.
- [01:54]
Paul Bloom
Yeah, well, so to. I think we'll start with AI, but remind people what you're doing and what kinds of problems you focus on. I think though we haven't spoken for several years, I think you still probably hold the title of most repeat podcast guest at this point. I haven't counted the episodes, but you've been on a bunch. But it's been a while. So remind people what your. What you focus on as a psychologist.
- [02:18]
Unknown
Yeah, I'm a psychology professor. I have positions at Yale, but I'm located at University of Toronto and I study largely moral psychology. But I'm interested in development in issues of consciousness, issues of learning, notions of fairness and kindness, compassion, empathy. I've actually been thinking it's funny to be talking to you, because my newest project, which I've just been getting a site advice to do with the origins of morality, and the spark was a podcast with you talking to Tom Holland, author of Dominion.
- [02:52]
Paul Bloom
Oh, nice. How. How so? What. What was the actual spark?
- [02:56]
Unknown
Yeah, I. I found this a great conversation. He made the case that a lot of the sort of morality that maybe you and I would agree with that, that, you know, the idea of respecting the right to universal rights and a morality that isn't based on the powerful, but instead maybe in some way respects the weak, is not the product of rational thought, not the product of secular reasoning, but instead the product of Christianity. And he makes this argument in Dominion, in several other places. And I've been thinking about it. I think it's a serious point. He's not the only one to make it. It deserves consideration, but I think it's mistaken. And so I think my next project, as I've been thinking about it, and I thank you for the podcast getting me going on this, is to make the alternative case, to argue that a lot of our morality is inborn and innate. I'm a developmental psychologist. That's part of what I do. And a lot of morality is a product of reasoned and rational thought. I don't deny that culture plays a role, including Christianity, but I don't think that's the big story.
- [03:55]
Paul Bloom
Nice. Well, I look forward to you making that argument because that's one of these shibboleths of organized religion that I think needs to be retired. And so you're just the guy to do it.
- [04:06]
Unknown
So thank you.
- [04:07]
Paul Bloom
Let me know when you produce that book and we'll talk about it.
- [04:10]
Unknown
I mean, I gotta say, I heard you guys talk and you sort of, you know, you. You guys engaged properly on the idea. And Holland I've never met, but he seems like a really smart and quite a scholar. So. So it's not, it's not a. You know, and, and he gets, I think, credence by the fact that he's in some way arguing against his own side. He himself isn't a devout Christian, A secular. Yeah, but, but that doesn't make him right. So. Yeah, so I'm very interested in engaging these ideas.
- [04:35]
Paul Bloom
So. So how have you been thinking about AI of late? What has surprised you in the seven years or so that we. Since we first started talking about it?
- [04:46]
Unknown
A mixture of awe and horror. I'm not a doomer. I'm not as much of a doomer as some people. I don't know when the Last time I checked with you, Sam, what would you say your P doom is?
- [04:58]
Paul Bloom
I've never actually quantified it for myself, but now's your time. I think it's non negligible. I mean, it's very hard for me to imagine it actually occurring in the most spectacularly bad way of a very fast takeoff. An intelligence explosion that ruins everything almost immediately. But fast or slow, I think it's well worth worrying about because I don't think the probability is tiny. I mean, I would put it in double digits. I don't know what those double digits are, but I wouldn't put it in single digits. Given the kind of the logic of.
- [05:31]
Unknown
The situation, I think we're kind of in the same place. Yeah, I mean, people always talk about, you know, you have a benevolent, super intelligent AI and you tell it to make paperclips and it destroys the world in terms of sold into paperclips. But there's another, another vision of, of malevolent AI I've been thinking about with the rise of what's called Mech Hitler.
- [05:49]
Paul Bloom
Mecca Hitler.
- [05:50]
Unknown
Mecca Hitler. You know, so, you know, it's a simpler scenario. Some deranged billionaire creates an AI that fashions himself on Hitler. The Trump Defense Department purchases it and gets it, connects it to all of its weaponry systems and hijinks ensue.
- [06:08]
Paul Bloom
How could you come up with such an outlandish scenario that could never happen? There's just no way.
- [06:13]
Unknown
It's a bizarre fantasy. And by the way, it also makes porn. So just to get the trifecta. So anyway, like a lot of people, I worry about that. I also find, at the same time I find AI an increasingly indispensable part of my intellectual life. I, you know, I, I have a question, but, but not, not just a specific, I use it as a substitute for Google. I, you know, say, you know, where's a good Indonesian restaurant in my neighborhood? And you know, how do you convert these euros into dollars? But, but I also have a question, like, I don't know, I'm, I got into an argument with somebody about revenge. So what's the cross cultural evidence for revenge being a universal? And who would argue against that? And three seconds later, boom. Bunch of papers, books, thoughtful discussion, thoughtful argument mixed in our hallucinations. Occasionally I find insights, a paper written by me that I've never written. But it's astonishing.
- [07:06]
Paul Bloom
I've been credited with things. Yeah, it's told me, I've interviewed people I've never interviewed and it's amazingly apologetic when you Correct it.
- [07:14]
Unknown
But I'm so sorry. Yes, you are totally right. Let me now give you citations of papers that actually exist.
- [07:21]
Paul Bloom
Yeah.
- [07:21]
Unknown
In your working day, when you write, when you prepare for interviews, how much do you use it?
- [07:26]
Paul Bloom
Well, I've just started experimenting with it in a way that's probably not the usual use case. So we have fed everything I've written and everything I've said on the podcast into ChatGPT. Right? Yeah, yeah. So actually we have two things. We've created a chat bot. That is me, that's, that's model agnostic, so it can be run on ChatGPT or a Claude or whatever the best model is. We can swap, you can swap in whatever model seems best. But so this is like at the, you know, a layer above, at the system prompt level. And it has again, access to everything. It's something like 12 million words of me. Right. So it's a lot. You know that. That would be a lot of books. We've just begun playing with this thing. But it's, it's impressive because it also, it's using a professional voice clone, so it sounds like me and it's every bit as monotonous as me in its delivery. I mean, so I'm almost tailor made to be cloned because I already speak like a robot.
- [08:27]
Unknown
Must be agonizing to listen to sound.
- [08:29]
Paul Bloom
It is, but it hallucinates and it's capable of being weird. So I don't know that we're ever gonna unleash this thing on the world. But it's interesting because even having access to every episode of my podcast, it still will hallucinate an interview that never happened. It'll still tell me that I interviewed somebody that I never interviewed. And so that part's still strange. But presumably this is as bad as it'll ever be. And the general models will get better and better. One imagines we're looking at each other.
- [09:01]
Unknown
On video, and I imagine it doesn't do video yet. But if we were talking on the phone or without video, would I be able to tell quickly that I was talking to an AI and not to you?
- [09:11]
Paul Bloom
Only because it would be able to produce far more coherent and comprehensive responses to any question. I mean, if you said because it's. Because it's hooked up to whatever the best LLM is at the moment. If you said, give me 17 points as to the cause of the Great Depression, it would give you exactly 17 points detail in the cause of the Great Depression. And I could not do that, so it would fail the Turing Test, as all these LLMs do, by passing it so spectacularly and instantaneously. That's a surprise. I want to ask you about that. That's a surprise for me how not a thing the Turing Test turned out to be. It's like the Turing Test was the staple of cognitive science literature. And just our imaginings in advance of credible AI, we just thought, okay, there's going to be this moment where we're just not going to be able to tell whether it's a human or whether it's a bot, and that's going to be somehow philosophically important and culturally interesting. But overnight, we were given LLMs that fail the Turing Test because of how well they pass it. And this is no longer a thing. It's like it was never a Turing Test moment that I caught.
- [10:28]
Unknown
All of a sudden, you and I, we have. I have a super intelligent being I talk to on my iPhone. And I'll be honest, and I think. I think other psychologists should be honest, but if you had asked me a month before this thing came out how far away we were from such a machine, I would say 10, 20, 30 years, maybe a hundred years. And now we have it. Now we have a machine we could talk to and sounds like a person and. Except, like you say, it's just a lot smarter.
- [10:52]
Paul Bloom
Yeah.
- [10:53]
Unknown
And it is mind boggling. I mean, it's an interesting psychological fact, how quickly we get used to it. It's as if, you know, aliens landed, you know, next to the Washington Monument, and now they walk among us. Oh, yeah. Well, that's the way things go. Oh, you develop teleporters. Now we got teleporters, and we just. We just take it for granted now.
- [11:12]
Paul Bloom
Yeah. Well, so now what are your thoughts about the implications, psychological and otherwise, of. Of AI companionship? I mean, what is. So at the time we're recording this, there's been been recently in the. In the news stories of AI induced psychosis. I mean, people get led down the primrose path of their delusion by this amazingly sycophantic AI that. That just encourages them in their messiah complex or whatever flavor it is. And I think literally an hour before we came on here, I saw an article about ChatGPT encouraging people to pursue various satanic rituals and telling them how to do a proper blood offering that entailed slitting their wrists. And as one does in the presence of a superintelligence. I know you just wrote a piece in the New Yorker that people should read on this, but give me your sense of what we're on the cusp of here.
- [12:09]
Unknown
I have a mild form of that delusion in that I think every one of my substack drafts is brilliant. I'm told this is. Paul, you have outdone yourself. This is sensitive, humane as always with you. And no matter what I tell it to, I say you don't have to suck up to me so much. A little bit, but not so it just. And now I kind of believe that I'm much smarter than I used to be because I have somebody very smart telling me my article is kind of nuanced in that I argue two things. One thing is there's a lot of lonely people in the world and a lot of people suffer from loneliness and particularly old people, depending on how you count it. You know, under some surveys about half of people over 65 say they're lonely. And then you get to people in like a nursing home, maybe they have dementia, maybe they have some sort of problem that makes them really difficult to talk with. And maybe they don't have doting grandchildren surrounding them every hour of the day. Maybe you don't have any family at all and maybe they're not millionaires so they can't afford to pay some poor schmo to listen to them. And if Chatgpt or Claude or one of his AI companions could make their lives happier, make them feel loved, wanted, respected, that's, that's nothing but good. I, I'm, I, you know, I think it, in some way, it's like powerful painkillers are, are powerful opiates, which is, I, I'm not sure what, I don't think 15 year olds should get them, but, but somebody who's 90, in a lot of pain, sure lay it on right. And I feel the same way with this. So that's the, that's the pro side. The con side is I am worried and you're, you're touching on it with this illusion talk. I'm worried about the long term effects of these syncopathic sucking up AIs where every joke you make is hilarious, every story you tell is interesting. I mean, the way I put it is if I ever ask, am I the asshole? The answer is affirm. No, not you. They're the asshole. And I think I'm an evolutionary theorist through and through and loneliness is awful, but loneliness is a valuable signal. It's a signal that you're messing up. It's a signal that says you gotta get outta your house, you gotta talk to people, you gotta, you know, gotta open up the apps, you gotta say yes. Yes to the brunch invitations. And if you're lonely, when you interact with people, you feel not understood, not respected, not loved, you gotta up your game. It's a signal. And like a lot of signals, like pain sometimes is a signal that, where people are in a situation where it's not gonna do me any good. But often for the rest of us, this is a signal that makes us better. Yeah, I think I'd be happier if I could shut off generally as a teenager, I would be happier if I could shut off the switch of loneliness and embarrassment, shame and all of those things. But they're useful. And so the second part of the article argues that continuous exposure to these AI companions could have a negative effect because. Well, for one thing, you're not going to want to talk to people who are far less positive than AIs. And for another, when you do talk to them, you have not been socially entrained to do so properly.
- [15:03]
Paul Bloom
Yeah, it's interesting. So, so believing the, the dementia case aside, me, I totally agree with you that, that a very strongly paternalistic moment where anything that helps is fine. It doesn't matter that it's imaginary or that it's encouraging of delusion. I mean, we're talking about somebody with dementia. But so just imagine in the normal healthy case of people who just get enraptured by increasingly compelling relationships with AI. I mean, you can imagine right now we've got chatbots that are still fairly wonky. They hallucinate, they're obviously sycophantic, but just imagine it gets to the place where, I mean, forget about Westworld and perfectly humanoid robots. But very shortly, I mean, it might already exist in some quarters already, we're going to have video avatars. It's like a Zoom call with an AI that is going to be out of the Uncanny Valley, I would imagine, immediately. I mean, I've seen some of the video products which, like sci fi movie trailers which are. They don't look perfectly photorealistic, but they're getting close. And you can imagine six months from now it's just going to look like a gorgeous actor or actress talking to you. That's going to be. Imagine that becomes your assistant who knows everything about you. He or she has read all your email and kept your schedule and is advising you and helping you write your books, et cetera, and not making errors and seeming increasingly indistinguishable from just a superintelligent locus of conscious life. Right. I mean, it might even seem it Might even claim to be conscious if we build it that way. And I mean, let's just stipulate that at least for this part of the conversation, that it won't be conscious. Right? That this is all an illusion. Right? It's just a. It's no more conscious than your iPad is currently. And yet it becomes such a powerful illusion that people, just most people, I mean, the people, I guess philosophers of mind might still be clinging to their agnosticism or skepticism by their fingernails, but most people will just get lulled into the presumption of a relationship with this thing. And the truth is it could become the most important relationship between many people have. Again, it's so useful, so knowledgeable, always present. They might spend six hours a day talking to their assistant. And what does it mean if they spend years like that? Basically just gazing into a funhouse mirror of fake cognition and fake relationship we've seen.
- [17:47]
Unknown
I mean, I'm a believer that sometimes the best philosophy or movies do excellent philosophy. And the movie her came out, I think in 2013, is an example of this guy, you know, lonely guy, normal lonely guy, but gets connected. An AI assistant named Samantha played by Scarlett Johansson and falls in love with her. And she does all of she. The first thing she says to him and what a meet cute is. I see you have like 3,000 emails that haven't been answered. You want me to answer them all for you? Yeah, I fell in love with her there. But, but, you know, but the thing is you're watching a movie and you're listening to her talk to him and you fall in love with her too. I think we've evolved in a world where when somebody talks to you and acts normal and gives you and seems to have emotions, you assume there's a consciousness behind it. Evolution has not prepared us for these extraordinary fakes, these extraordinary golems that elude all of the behavior associated with consciousness and don't have it. So, so we will, we will think of it as conscious. There will be some, some, you know, philosophers who insist that they're not conscious. But, but even they will, you know, sneak back from their, from their classes and then in the middle of the night, you know, turn on their, their phones and start saying, you know, I'm lonely. Let's talk.
- [19:08]
Paul Bloom
Yeah.
- [19:08]
Unknown
And then the effects of it. Well, one effect is real people can't give you that. You know, married, very happily married. But sometimes my wife forgets about things that I told her and sometimes she doesn't want to hear my long boring story. She wants to tell her story instead. And sometimes it's three in the morning and I could shake her away because I have this really interesting idea I want to share, but maybe that's not for the best. She'll be grumpy at me because the thing is, she's a person and so she has her own set of priorities and interests. So too, of all my friends, they're just people and they have other priorities in their life besides me. Now, in some way, this is, I think, what makes, when you reflect upon it, the AI companion have less value. You know, here you and I are. And what that means is that you decided to take your time to talk to me and I decided to take my time to talk to you. And that's the value. When I switch on my lamp, I don't feel, oh my gosh, this is great. It decided to light up for me. Written in that choice at all. The AI has no choice at all. So I think in some part of our mind we realize there's a lot less value here. But I do think in the end, the scenario you paint is going to become very compelling and real people are going to fall short. And it's not clear what to do with that.
- [20:25]
Paul Bloom
Now, there's something that you. I think you've just come up with a fairly brilliant product update to some future AI companion, which is a kind of Pavlovian reinforcement schedule of attention where it's like the AI could say, listen, I want you to think a little bit more about this topic and get back to me because you're really not up to talking about it right now. Come back tomorrow. Right. I wondered that would be an interesting experience to have with your AI that you have subscribed to.
- [20:59]
Unknown
I've wondered that, like, you asked the AI a question that says, is that really, like a good question? Does it seem like a question you couldn't just figure out just by thinking for a minute?
- [21:06]
Paul Bloom
It's really, listen, I know everything. That's really what you want to ask me is that. Don't you have something deeper?
- [21:13]
Unknown
You were talking to a super intelligent God and you want to know how to end a limerick.
- [21:19]
Paul Bloom
Really?
- [21:21]
Unknown
I would wonder if these things, how people would react if these things came with dials. Obviously not. Maybe a big physical dial. You might be a big physical dialogue. And the dial is pushback. So when it's set at zero, it just. Everything you say is wonderful. But I think we do want some pushback now. I think in some way we really want less pushback than we say we do. And it's this way of real people too. So everybody says, oh, I like when people argue with me. I like when people call me out on my bullshit. But what we really mean is we want people to push back a little bit and then say, ah, you convinced me. You really showed me. I thought you were full of it. But now, upon reflection, you've really out argued me. We want them to fight and then acquiesce. But you turn the dial even further. Will the AI then say, we've been talking about this for a long time. I feel you are not very smart. You're just not getting. I'm going to take a little break and you mull over your stupidity.
- [22:21]
Paul Bloom
Paltry human, the recalcitrant style. That's, that's what we could build in. All right, we're going to.
- [22:28]
Unknown
I feel this could be the worst business.
- [22:29]
Paul Bloom
We're going to get rich. Paul. This is with AI that calls you on your bullshit.
- [22:34]
Unknown
That's really the business idea of the century.
- [22:38]
Paul Bloom
But so what are we to think about this prospect of spending more and more time in dialogue under the sway of a pervasive illusion of relationship wherein there is actually no relationship, because there's nothing that it's like to be chatgpt6 talking to us perfectly clearly?
- [22:58]
Sam Harris
If you'd like to continue listening to this conversation, you'll need to subscribe@samharris.org Once you do, you'll get access to all full length episodes of the Making Sense podcast. The Making Sense podcast is ad free and relies entirely on listener support and you can subscribe now@samharris.org SA.