← The Political Scene | The New Yorker
The Political Scene | The New Yorker
How ChatGPT Will Strain a Political System in Peril
00:32:53
How ChatGPT Will Strain a Political System in Peril
Loading summary
Transcript53 lines
- [00:00]
Mint Mobile Advertiser
As summer draws to a close and the kids go back to school, I know I'm going to want to keep in touch with my kids at a price I can afford. Back to school. Shopping can be a hassle, but your phone plan shouldn't be. That's why I made the switch to Mint Mobile. For a limited time, Mint mobile is offering three months of unlimited premium wireless service for 15 bucks a month. So while other parents are sweating overage charges, I have a little bit more room in my budget for cool back to school threads. Say bye bye to your overpriced wireless plan's jaw dropping monthly bills and unexpected overages, Mint Mobile is here to rescue you. All plans come with high speed data and unlimited talk and text delivered on the nation's largest 5G network. Use your own phone with any Mint Mobile plan and bring your phone number along with all your existing contacts. Dish overpriced wireless and get three months of premium wireless service from Mint Mobile for 15 bucks a month. This year, skip breaking a sweat and breaking the bank. Get this new customer offer and your three month unlimited wireless plan for just 15 bucks a month@mintmobile.com New Yorker. That's. That's mintmobile.com New Yorker upfront payment of $45 required, equivalent to $15 a month limited time new customer offer for first three months only. Speeds may slow above 35 gigabytes on unlimited plan. Taxes and fees extra. See Mint Mobile for details.
- [01:17]
Tyler Foggit
Okay, so I've made an OpenAI account and I am getting ready to write a prompt for ChatGPT. So any advice, Josh, that you have on how to form this question so that it actually gives us what we want is welcome. Right now I have write an introduction to a podcast interview about how AI will affect political discourse online and beyond.
- [01:40]
Joshua Rothman
I think it will work better if you suggest doing it in a certain style, like in the style of NPR or even in the style of the New Yorker. ChatGPT does better when. When you give it some person that it can riff on or some voice that it can.
- [01:56]
Tyler Foggit
I'll do in the style of the New Yorker.
- [01:57]
Joshua Rothman
Oh, and you have to say the New Yorker magazine. Otherwise it will think it's just, oh.
- [02:00]
Tyler Foggit
Gosh, like Brooklyn accent.
- [02:03]
Joshua Rothman
Yeah, right. Like the Martin Scorsese version of the intro.
- [02:07]
Tyler Foggit
Yeah. Okay. It's typing in front of me. Oh, gosh, it's interesting. Okay. In recent years, the rise of artificial intelligence has had a profound impact on every aspect of our lives, from the way we communicate and work to the way we consume and create Media. One area where its influence is particularly controversial is political discourse, both online and in the real world. In this episode of our podcast series, we speak to Dr. Lisa Singh, a professor of computer science at Georgetown University and an expert in the intersections of AI, politics and society. Okay, so where. Where did Dr. Singh come from?
- [02:46]
Joshua Rothman
Is there a Dr. Singh? Is that a real person?
- [02:49]
Tyler Foggit
I don't know who Dr. Singh is. I didn't. I didn't mention her.
- [02:53]
Joshua Rothman
Dr. Lisa Singh is a professor in the Department of Computer Science at Georgetown and the director of the Massive Data Institute. So go chatgpt. That wasn't so bad.
- [03:03]
Tyler Foggit
Okay, so I'm going to try one more iteration. Try again. But focus more on the potential controversies and specifically how ChatGPT might be used by Republicans and Democrats. And replace Dr. Lisa Singh with Joshua Rothman. Does ChatGPT have the ability to find out information about you just based on the fact that you have a long history of writing for the New Yorker?
- [03:23]
Joshua Rothman
Yeah, it knows who I am, but it gets me confused with other Joshua Rothmans, so.
- [03:28]
Tyler Foggit
With the New Yorker's Joshua Rothman?
- [03:31]
Joshua Rothman
Yeah. Let's try it and see what happens.
- [03:33]
Tyler Foggit
Okay, it's working on it. Okay. What happens when political actors start to use AI for their own ends? And what are the potential consequences for democracy? In this episode of our podcast series, we speak with Joshua Rothman, a staff writer for the New Yorker who has written extensively on the intersection of technology and politics. Wow. That's actually exactly what I would probably say.
- [03:58]
Joshua Rothman
That's pretty good.
- [03:59]
Tyler Foggit
I feel like that was a win. I guess we should just start by, you know, ChatGPT is an LLM, which stands for Large Language Model. How do these technologies work? And, you know, what can they do and what can they not do? And what will they never be able to do? I'm sort of wondering what the. What the limits of these things. Things are.
- [04:18]
Joshua Rothman
Yeah, sure. I mean, they're totally. I don't want to say they're totally new, but they've recently crossed sort of a threshold of power. I think people are still grappling with kind of what the limits are and what's possible. So, basically, yes, ChatGPT, it's a large language model, which is a type of artificial intelligence. It's a type of neural network. And essentially it's a predictive piece of software that puts words on a page one after another in a way that reflects its ideas about what we normally say and normally write. It's processed 300 billion words or something like that. It's processed billions of words and it's sort of read and analyzed huge amounts of text written by people and it can now generate believable text in response to prompts. So like on a fundamental level, it's a lot like autocorrect on your phone, but it's that souped up in an incredible way where it can write text that seems creative, that can surprise you, it can respond to really subtle requests, it can impersonate people, it can answer questions, I think in a very powerful way. You know, if you ask it, you know, what would a libertarian think about this particular tax policy? It will look at the massive amount of information that it's ingested and, and do its very best to give you a realistic answer. So what can it do? What can it not do? What it can do is produce text that feels intelligent, that feels creative, feels insightful, but what it can't do is it's not actually thinking, it's just putting one word after another. So there are situations where what it says is just false or nonsensical or wrong or crazy and it doesn't have any real way of knowing that it doesn't know anything about the world. Yeah, it's not trying to be right, it's just trying to be plausible. To me, it's helpful to think of you're talking to someone who, when you talk to ChatGPT, you're talking to someone who doesn't know what they're talking about, but is really good at sounding like it. And the thing that's striking is there's a lot of overlap there. I think most people who use it feel that it's revolutionary. It's a whole new type of technology that simply never existed, where you really feel like you can have a real conversation with a computer and you're not encountering the rote pre written replies that have been hard coded into software, but you're interacting with something that feels very much like a mind, that has the same ability to move across different domains of knowledge and think through problems sort of the same way that you do.
- [07:05]
Tyler Foggit
Yes, I mean, it's revolutionary, but it also seems kind of frightening. So, you know, you've edited a lot of pieces about ChatGPT and in general there's just been a lot of discussion about how it's going to transform education, both good and bad. And there's been talk about whether it'll replace every journalist. But I feel like something that isn't talked about as often is just sort of its potential as a political tool. And so I'm wondering If you could, you could just talk a little bit about what you see there in terms of just ways in which it might become part of, you know, the campaigning system or just part of social media and ways in which, like, political information is disseminated.
- [07:41]
Joshua Rothman
Sure. So, you know, chatgpt. I mean, you know, to me, the thing that's most crazy about it is that, you know, so much of our political life basically unfolds online through text. So whether it's arguing with people in the comments of a YouTube video, which is like a national pastime now, or it's, you know, you know, obviously Twitter is text based. You know, Facebook is fundamentally text based. You know, we use these platforms and, you know, we put a huge amount of trust in the fact that text can't be autogenerated. Like, we never thought about it before, but we've always assumed that on some level, when we see text next to someone's name, it's been written by a person. Now, we know that there are bots and there are trolls who are paid to leave comments and post messages and things like that, but we also know that that's hard to do, that it's expensive, and that the messages look fake. You know, we have spidey sense that says, you know, that incendiary comment from, you know, Soccer Mom 2023 on this YouTube video probably isn't real. Doesn't seem right. And what large language models make possible is the wholesale generation of text that seems totally real and that is interactive. You know, it's not just one comment that exists. You can argue with it. It can talk to you. So, you know, first, one big consequence is just that social media will change. It's going to be a complicated process to try to and possibly an impossible goal to try to keep social media human only and a future world in which you see those comment wars unfolding, I think it's going to be reasonable to assume that a large number of those will be happening between AIs, like ChatGPT. You know, I'll have one and you'll have one, and if we're both running for Congress, they'll fight it out.
- [09:41]
Tyler Foggit
So you think you're imagining, like, politicians or politicians interns who are good with technology farming out social media conversations to ChatGPT. I guess I've been wondering whether it'll be more like average people or whether it'll be people in positions of power who are using it to get around maybe the more tedious aspects of their jobs, or whether it'll be super PACs or just sort of larger, kind of like political organizations that are using it in ways that we might not even really expect.
- [10:13]
Joshua Rothman
Yeah, in my mind, everyone who can use this technology will use it.
- [10:17]
Tyler Foggit
It'll become like just imessage or something. Just kind of ubiquitous.
- [10:21]
Joshua Rothman
Yeah, it'll be a window open on your computer. I mean, I think a good analog would be Wikipedia. A friend of mine pointed out once that it was possible to notice when other people online had Wikipedia knowledge. It's like you were having a debate with someone about something, and all of a sudden they seemed incredibly well informed about, you know, the inner workings of the French healthcare system. Well, they didn't know that that was because they had Wikipedia. All of a sudden this knowledge is available to us, and that it takes care of the chore of writing is incredibly useful. So ordinary people will certainly use it. Right. And we can expect to encounter it way more often in online contexts and interpersonal contexts of all types, like emails that we send back and forth. Professionals will use it. And in the case of political professionals, that means people writing campaign communications. You know, a good question would be if you work in a congressional office and you get letters from constituents.
- [11:24]
Tyler Foggit
Yeah.
- [11:24]
Joshua Rothman
You know, right now we assume that it's hard to write a letter and that it's work, but it's not work anymore. And certain types of communications that we now take for granted as intrinsically meaningful will become less meaningful.
- [11:37]
Tyler Foggit
I mean, I feel like there might be a period where you just have a ton of letters coming in to politicians that are all written by ChatGPT, but then eventually, you know, I feel like you just have the end of the letter to your senator if they aren't taken seriously anymore. Because it's assumed that so many of them have been AI generated.
- [11:54]
Joshua Rothman
Yeah, it poses, you know, anywhere where, as it were, the volume of text is taken to be a signal. A lot of online commenters are saying such and such. A lot of letter writers are demanding this and that. Any context where just the sheer volume of written communication used to signify something, I think it's reasonable to say that in the next few years, it will. It will become less and less and less significant because we'll understand how easy it is to generate large quantities of totally believable text. You know, this. This gets, I guess, to another political thing, which is that a huge amount of politics is about what people said. It's about, like, what they would think or what they have thought, positions they've had, attitudes they hold. How do you know what someone really said versus what is just like a quote on brainyquote.com it's already hard to distinguish that. And in this new world of unlimited text, maybe an even better term would be sort of unlimited plausible text. I think it gets even harder. It's not just that it's harder to tell whether something is true, you know, is something that somebody said or not, but a new space opens up for the things they might have said but didn't say. You know, ChatGPT and LLMs, they, they're really good at exploring sort of the gray area of the nearly plausible. You know, the thing that's like highly probable. And not just highly probable in the sense of like they might have said it, but like highly probable in the sense that analysis of billions of words has suggested that this is the kind of position that someone would hold or the kind of thing that someone would have said. So, you know, I don't have a, I have no crystal ball about this, but I'm really curious about, in the next, you know, several years, how our sense of, of the reality of text will, will change. I, I suspect like we'll get used to the fact that text now is a little soft focus and is everywhere. Like the other day I, I like tweaked my knee playing tag with my kit. So I went online and I was googling like what I'd done to myself and there's like a hundred million web pages. It's, it's a content mill. Yeah, kind of, kind of situation. You're reading all these web pages, they all have these names like, you know, Med Health Online or whatever. Is it a real website or not? Did doctors participate in the creation of this website? Who built it? And who are these people? Well, you don't know. It sort of seems like they're all recycling the same ideas and illustrations even over and over. It's like these very, very tiny differences that suggest to you that, you know, one is more trustworthy than the other. And it was still useful. But I would, you know, I've developed now, I think as many as many of us have a fairly sophisticated sort of nuanced way that we don't really articulate to ourselves of handling the fact that so much of that information is semi trustworthy, semi true. And this is sort of another step in that process where that feeling that we get navigating through the sort of content mill landscape now will be way more common. When you get an email from, from somebody or you read something online like an op ed or a Facebook post or a tweet Yeah, I mean, like.
- [15:21]
Tyler Foggit
Do you think that our society is almost like uniquely equipped to handle something like ChatGPT just because, you know, I feel like we all have had that experience of going on the Internet and seeing something and knowing that it might not be true, or that it's probably written by someone who's not a doctor, but like it gives you what you need in the moment and that's kind of what matters. I mean, you know, I guess it's just making even more judgment calls. Or do you think that just because so much of our, you know, the way that we do things nowadays involves text on the Internet, that it's just going to be too much and it's just going to be harder for us to navigate these things even if we, we've been able to do it in the past with decent accuracy.
- [15:59]
Joshua Rothman
I don't know. You know, it's such a tough question. I mean, I would say the political culture of the last decade which has been so shaped by, you know, inflammatory online life. Yeah, like we sort of live in like an always online country where arguments and ideas that are worked out in a web forum find their way into the, the mainstream and where, you know, discourse like words, ideas that start in a, a sub community online become dominant cultural forces. I mean, it can be totally fringy or you know, putatively fringy things like QAnon, but it can be sort of broader ideas from the sort of red pill world. You know, basically the conduit between words online and real life is a, is, is fully operational. What does it mean when words online are no longer being necessarily written by a person, they're being written by software. You know, software that's being directed by people on some level, whether it's been set up to automatically generate or whether it's been given prompts like the prompts we gave it earlier today. But what does it mean when software is in the mix and when ideas that we attribute to individuals or that we're used to coming from individuals are now coming from just like this font of words. It freaks me out basically is the bottom line. I find it hard to envision a world where ChatGPT and its ilk make online political discourse better. To me, the sort of overall arc of what the Internet has done to information has been that it's made information way more available. If you're a learner, if you're a person who's curious, it's incredible. It's like the best thing that's ever happened. But at the same Time, in a funny way, it's decreased the value of information because it's increased the volume and it's made a situation in which essentially you can find whatever you want to find. So we're going to collective kind of situation in which we have all this information available to us, we have all these views available to us. But our politics can feel at times disconnected from reality, almost as though fundamentally we have sides and we simply choose whatever information will help us feel jazzed about our side.
- [18:37]
Tyler Foggit
You'll hear more from Joshua Rothman in just a moment.
- [18:44]
Joshua Rothman
America is changing and so is the world. But what's happening in America isn't just a cause of global upheaval. It's also a symptom of disruption that's happening everywhere. I'm Asma Khalid in Washington D.C. i'm Tristan Redman in London and this is the Global Story. Every weekday we'll bring you a story from this intersection where the world and America meet. Listen on BBC.com or wherever you get your podcasts.
- [19:16]
Tyler Foggit
Is it inherently neutral? I've seen a lot online you noticed about ChatGPT being woke or. Which is kind of a funny way to think, to think about, but just this idea that if you. Yeah, that it'll give you information on one topic but not on another without, you know, an incredible amount of prying in certain situations.
- [19:37]
Joshua Rothman
Well, so there's two, there's a couple kind of levels to that question. You know, there's the question of whether it has rules in place that make it political in one way or another. So for example, Interesting. Yeah, you know, like ChatGPT or these LLMs, they're ideally, they're designed not to be able to, you know, use racial and ethnic slurs. There are some types of attitudes that it not supposed to be able to take to their users which could be construed as kind of a set of left leaning rules that it's following not to offend us. That's been. One charge that's been made is, you know, ChatGPT is woke because it doesn't like ethnic stereotypes, for example. On the other hand, if you ask ChatGPT for political views about, let's say like the war in Ukraine, it doesn't want to get into that because it's beyond its purview, which is another guardrail that's been put in place that can also be construed politically as a sort of great wall of silence, like a sort of pro Putin stance that basically says we're not going to talk about the war. That's going on. So there's one level to that question which is are there rules being put in place by the designers of the system that have political effects on its output? I mean, the answer's got to be yes. But are those political effects intended? Do they lean more one way than the other? Only time will tell. It's like, hard to say. Then there's the question of from the bottom up, as it were. Having ingested all the billions of words, does the model have a politics? Does it have a political outlook? Does it incline more in one direction than the other? And that's something that already data scientists are trying to sort out. And there have been various projects to create a more right wing GPT and a more left wing GPT and a socialist GPT and so on. The way a computer might understand someone's political views is just what words tend to follow what other words. And you know, it's basically kind of makes sense that if basically the word, you know, oil companies is usually preceded by a word like rapacious, then you're probably talking to a Democrat. But. But if a word like sort of like family is is followed by the words like in crisis, you're probably talking to a Republican. It's like there's this, you know, this aspect of politics that sort of is ultimately reducible to patterns in our speech that, you know, these systems are really good at grabbing onto.
- [21:57]
Tyler Foggit
You mentioned right leaning chatbots, that there's been an effort to create those. I mean, is that, I mean, when you say that, do you mean that there's been an effort to create a version of ChatGPT that doesn't have any sort of rules in place to, you know, to not say certain slurs and whatnot or like, is it right leaning in the sense that it's like all the speech with no barriers and that kind of like Elon Musk taking over Twitter and trying to install free speech culture their way? Or is it like Donald Trump but.
- [22:30]
Joshua Rothman
A bot like the. I think so. I mean, I'm sure that there will be GPTs that are like the bros on a podcast GPT. It's like they can just say whatever they want and it's cool. But no, I think the real idea here is to do something like in terms of the resources the network is drawing upon, maybe it draws a little bit less on the New York Times and a little bit more on PolitiFact or something, or on the Wall Street Journal. Maybe you can change what it thinks of as normal or you can adjust what they call the weights inside the network.
- [23:02]
Tyler Foggit
That's so interesting. Which I guess is already kind of an element of search right now with Google, it's just like I Google something that's happening in the world half the time. The New York Times is the first thing that comes up.
- [23:13]
Joshua Rothman
Yeah, I mean, what stands out to me as different from the current set of technologies we have is right now we have this idea there are algorithms and the algorithms serve us up material and we're sort of the passive users of these algorithms. We're sort of scrolling and it gives us more of this and less of that and we're not really in control. We can thumb up or like, or something, something and get more of it. But that's kind of it, you know, and we can say like, don't show me more of this. But this software is far more interactive than that. Every single session with it, every time you're using it, it's responding to what you're asking it. Right. Then it's not an algorithmic feed like we're used to. That's been like the dominant technology of the Internet for the last, you know, couple decades, really. This is much more a being basically that you're interacting with and you're talking to it and you're asking questions and you're prompting it and you're pushing it to tell you things. And when it's not making sense, you're, you're looking at it, you know, you're, you're phrasing the question a little differently. So I think the whole way of interacting with this technology is different. And when we think about a right wing or left wing GPT, I think those are going to be interesting experiments that happen. But the truth is, if you want to know what Margaret Thatcher thought and you want to have ChatGPT generate an argument from her perspective for why you shouldn't vote for Joe Biden in the next presidential election. It'll do a good job of doing that, whether it's left wing or right wing or whatever, because it's interactive and we're driving it. So it's kind of a different type of tool where at least to my mind, some of the algorithmic questions are not as they're not at the center. Yeah. I mean, in the fake text scenario, where really what we're interested in is the production of unlimited plausible text, what's cool about what ChatGPT can do is it can respond to us in such a nuanced way. We can tweak it, we can ask it to do something more and more specific we can say, oh, make sure you mention Hunter Biden's laptop and make sure you mention the role the New York Times played in bearing the story. And we can control the output very discreetly. We're not passive users of the software, we're active users of it and we're using it to generate content for us.
- [25:27]
Tyler Foggit
And do you anticipate, I mean a lot of people campaigns using it to just like spew out a bunch of disinformation or do you think that, I guess that part of the danger, you know, with sort of the disinformation element here, that part of it just stems from there being so much plausible text that it's hard to know what's real and you know, what's been written by a bot. And also the fact that, you know, ChatGPT doesn't necessarily give you something that's accurate. I mean, I guess I'm wondering like if you think this is something that's going to be co opted by a bunch of bad actors or whether we're going to have disinformation no matter what, because that's just kind of the nature of how this thing works.
- [26:05]
Joshua Rothman
So I think we're going to get it on every side because I think Great. On the level, yeah, I think on the level of kind of text on the Internet in general, there'll be a lot more of it and a lot more of it will be auto generated and of semi factual nature. But then you know those really bad emails you get from the congressional campaigns that are just like, hi Tyler, I need your help, it's urgent. So like this is such a blunt instrument now, but in the future it's going to be so much better because first that email will be written just for you, full of demographic and information gleaned from the data that exists online about you and what you care about, what you're interested in, where you live, what's happening in your life. That email from a politician could be so much better if it were auto generated by ChatGPT and if you wanted, you could write back and it might write back to you. Like an entire world of interactive text now becomes possible. Where, you know, we all understand what we're dealing with. You know, some of us won't, but many of us will grasp that this surprisingly specific missive from, from this congressional candidate wasn't handwritten by a staffer, but was instead created by an AI. But that doesn't necessarily change its power or interest.
- [27:22]
Tyler Foggit
Do you think that the AI might just wipe out the Political staffers or interns who would generally be sending those, those texts.
- [27:29]
Joshua Rothman
I think you can probably achieve something that no staffer can achieve because you could never, you know, no campaign could afford the number of skilled writers it would take to send a thoughtful note tailored specifically to the interests of all possible constituents. But in a, you know, unlimited text future, it will be easy. Like the mass communication that's personalized for the recipient is something that's like almost an oxymoron today. We see it happening algorithmically in the cheesy emails we receive from marketing companies. It's instant to spot, but also it's very low value, like it's not actually speaking to us. But I don't see any reason why that won't change now that generating text is so much easier and the system is so much better. So yeah, I think it's like whatever staffers can do a good job controlling the software will be.
- [28:22]
Tyler Foggit
Their job will be changed to ChatGPT, prompt writer or.
- [28:26]
Joshua Rothman
Yeah, I think a lot of jobs will be, will incorporate this now.
- [28:32]
Tyler Foggit
And you don't think that people are going to be just sort of naturally resistant to, you know, this like hyper personalized text, even if it is pretty compelling, simply because they, they know that it was written by, by a bot. I mean, I'm thinking about, did you read about the whole thing at Michigan State, there was a school shooting. The administrator sent out an email to the student body and at the bottom of the email it said like, this has been written by, you know, chatgpt. Sort of like, you know, sent from my iPhone basically. And people were really upset. And so I guess it makes me wonder whether people will either resent politicians for sending them an email asking them to donate, even if it's, you know, a perfectly crafted email, if they, you know, have a feeling that it's been sent by ChatGPT, or if, you know it's going to disincentiv politicians and other people to constantly be going out of their way to prove that they're doing something that is real and wasn't generated by AI. So just more videos, more in person visits, just things that can't be done by ChatGPT.
- [29:38]
Joshua Rothman
So I'm deeply fascinated by that Michigan State story. If they hadn't left that at the bottom, the emails would have gone out and no one would know and none of us would have said, man, those emails sure were generic because they're already incredibly generic. I mean, they're already essentially written probabilistically by a human being instead of by ChatGPT. Yeah, I mean, I think the question you're asking is such a good question. And it's a question that, you know, we genuinely live in times of suspense. How will we feel about text in the future when we know that it's entirely likely that it was written by computer instead of by person?
- [30:12]
Tyler Foggit
I feel like at the New Yorker we're gonna have to start using disclaimers like in the same way that a movie is like there were no animals harmed in the making of this film. It's like there was no chatgpt used in the, you know, writing of this article.
- [30:24]
Joshua Rothman
Well, you know, I've been wondering, I mean this isn't, this isn't letting anyone in on any behind the scenes secrets, but you know, a really good New Yorker article is a group effort. There's a writer, there's an editor, there's a fact checker or two, there's copy editors or two or three. Many people read it and offer suggestions. There's layers of editors and it runs under one person's name because there was one person around whom that effort has coalesced. But we don't feel it's dishonest that all those other names don't appear under the article. We understand, and I think readers understand that text is a product, let's put it like that. In the future will we feel that if ChatGPT plays a role in a lot of what we've written, but many other things also play a role. Like if we sense that, yeah, the university president's speech, ChatGPT might have written part of the historical section that explains that in the 19th century such and such happened on this campus. But we know that they read it over and they fact checked it. Will we feel cheated? Will the general value we place upon that text decrease? If I had to guess, I'd say we won't feel cheated, but the value we place in the text will have decreased. I think we'll probably develop some spidey sense around when writing seems too specific for us. Like when we can't imagine the material circumstances that could have produced such writing. So when it just seems implausible that our senator wrote us a note specifically, well, we'll know that probably an AI is involved and we might not care. We might find that to be totally fine. We might enjoy interacting with our senator's AI, asking it questions about her policies and her views. We might find it useful and illuminating and educational to be able to ask questions of these systems, all the while understanding that what we're learning could be wrong. We might develop some intuitions about when we're asking questions that will produce answers that are, that are not trustworthy, like when we've pushed the system too far. I think what does seem like something that you can say for sure is that the way in which we value what we read and the way in which we come to trust it or whatever, that's all going to change in the next few years, like fairly soon. And, you know, just the period of time in which we related to things we saw online as not fake is going to end because it's so easy to fake these things now. And we know what the political implications are in terms of what we think about what we read online. The truth is, it's anybody's guess. It's like the situation that exists today where online life drives politics to such a great extent. My own hope would be that somehow all of this will force us to be more educated about what it is that we see online and inculcate a certain degree of interactivity and skepticism about what we see, make us understand that we're not really dealing with other people necessarily. Hopefully it, like, lessens the power of the online world. I mean, I would say, I don't know if you've spent a lot of time in chatgpt. I've spent a lot of time with these LLMs over the last couple months. And it makes you really value in person conversation with real people. And it really, you know, it's a funny technology to have sort of first really, you know, arrived in like the sort of shadow of COVID because we've all become way more online and now this thing appears to make the online world sort of even more weird and uncanny than it was and makes you really think about, you know, what signals am I going to value?
- [34:17]
Tyler Foggit
Joshua Rothman is a writer and editor at this Has Been the Political Scene. I'm Tyler Foggit. The show is produced by Michelle Moses with help from Sidney Cobb. Our executive producer is Stephen Valentino. Our theme music is by Alison Layton Brown. Thank you so much for listening and we'll see you next week.
- [34:39]
Joshua Rothman
Right now, we are living through some of the most tumultuous political times our.
- [34:44]
Tyler Foggit
Country has ever known.
- [34:45]
Joshua Rothman
I'm David Remnick, and each week on the New Yorker Radio Hour, I'll try to make sense of what's happening alongside politicians and thinkers like Cory Booker, Nancy Pelosi, Liz Cheney, Tim Waltz, Ketanji Brown Jackson, Newt Gingrich, Robert F. Kennedy Jr. Charlemagne, tha God, and so many more that's all in the New Yorker Radio Hour, wherever you listen to podcasts from prx.