Misinformation Machines with Gordon Pennycook – Part 1

Hello and welcome to the Behavioral Design Podcast. This season we're diving into the intersection of behavioral science and AI. We want to make sense of the state of AI, from understanding how humans interact with intelligent systems to using AI to do behavioral design itself. I'm Aline Holsworth, a health tech advisor specializing in AI and product design. Over the past 15 years, I've been crafting human centered products with behavioral science

at the core. At Apple, I LED Behavioral Science for Health AI, designing and launching AI powered features to help users reach their health goals. And I'm Samuel Sultzer, your second Co host. I'm a behavioral strategist specializing in hybrid formation and designing products that drive long term baby change. I work with leading tech organizations integrating AI to scale behavioral design for good.

And I'm also the founder of Baby Bites, a dedicated community on behavioral science and AI. Quick word on Nuance Behavior where we help organizations build impactful digital products using behavioral design. We only take on a few clients at a time to ensure the highest level of quality for our tailored evidence based solutions. If you'd like to become one of our special projects, e-mail us at hello@nuancebehavior.com or we could call directly on our

website, nuancebehavior.com. Hey, I don't see a voting sticker. Have you voted? Well, yes, actually I have already voted. I voted early. We have this wonderful thing in Durham called early voting. There are, there are many, many advantages to it. You can go at any time of day. And like one of many, many places, it's flexible. It's, it's like really the ideal way to vote.

There's no lines. Whereas if you if you vote on actual Election Day, you can only go to a very specific place with all the like very specific things and often have to wait for a very long time time to vote. So highly recommend early voting. Nice. That's awesome. And I had a voting experience this year with voting for the EU elections. And I feel like for me, every time I vote, I feel like good about democracy, like I feel.

Like. It's like positive reinforcement because you know you go to place, people are friendly. So friendly, yeah. Everyone is friendly and you kind of feel like you leave there feeling like you've done something good as well, and you've done your part and you, yeah, you feel thankful. Yeah, that you, you get that, that voter's high after. No, I agree.

I mean, there are really very few situations in life nowadays where you can go and do something, you know, fairly simple and get like 6 thank you's and like smiles the whole time and just all of this positive reinforcement. It's wonderful. And then you even get a sticker at the end. Yeah, I didn't get a sticker. I got, I think a cookie maybe or something. I I got something, but I didn't get a sticker. But. But I'm glad you did and I'm glad you voted. And that sets us up for

obviously this episode and also next episode, we're going to cover a lot of thoughts on how behavioral science can help us think better about misinformation, elections and everything in between. Also with kind of AI as a big

influence as well, potentially. Yeah. And, and overall, we obviously have this big theme of misinformation, especially looking from this side of the Atlantic observing the American election, there seems to be this cloud and worry around misinformation and, and what are things we can trust and what can we trust when it comes to the news leading up to the election? And we've heard about cats and dogs being consumed in some places where I've heard a lot of

different things. But I don't know from my perspective, I obviously see it from a different angle than you. So what has been your experience so far when it comes to encountering misinformation? In this election. Yeah. Well, so first, I think the Haitians eating cats and dogs is maybe one of the more comical form of misinformation. But a lot of the misinformation that's proliferating increasingly in advance of the election is less innocuous than that.

And it's really what I would categorize as disinformation, which is intentional. It's misinformation, but really intentionally trying to deceive people from mail in ballots being destroyed in Pennsylvania and then encouraging or facilitating non citizens to vote in California. These are all examples of disinformation that's already spreading that has been very clearly and unilaterally

debunked. But because it's sensational and because it casts doubt on the election results, it's juicy enough that people are extremely motivated to spread it. And the big fear is that this disinformation will not only have an impact on the election itself, but on whatever happens after the election. Right.

And yeah, maybe Eileen, do you want to introduce our fantastic guest at this important time? Yes, I would love to. We were so lucky to have Gordon Pennycook on the show. He is a psych professor at Cornell, and he studies where we go wrong in reasoning. And these reasoning errors show up in all sorts of domains, from climate change to health issues to politics, like we'll talk about today.

And one fun fact about Gordon. He won the very humorous and highly esteemed IG Nobel Peace Prize for his work on pseudo profound bullshit. This is one of my favorite prizes in all of social science. It's actually, it's not just social science, but many social scientists have been awarded the IG Nobel. And it's also one of my favorite papers that has won the IG Nobel.

So if you're not familiar, the the IG Nobel is this celebration of research that quote UN quote first makes people laugh and then makes them think. And he has some very interesting research on pseudo profound bullshit, which is what he calls seemingly impressive assertions that are presented as true and meaningful but are actually vacuous.

And you can see this in many of the quotes that come from Deepak Chopra. But it's some examples are something like hidden meaning transforms unparalleled abstract beauty, which like has the the syntax of being a statement that could be true and makes sense, but actually is complete nonsense, even though it's it like comes off as something that could be real. Gordon has also done a great deal of research on misinformation, which is exactly

why we invited him here today. And, you know, given that it's election week and the first thing that comes to mind when you think of elections, at least in the US, is misinformation and disinformation. I'm not sure what that says exactly about the state of society, but these two do seem to come hand in hand. And so as a result, who would we be if we weren't tackling this head on?

Yeah. And like, to be fair, I think it's not only true for the US, but I think in every election that happened this year, there's been a lot of elections in the world. And even last year, it's definitely always, always a part of the conversation. And yeah, as I said, we had AI think a really important and valuable discussion when it comes to understanding misinformation from kind of behavioral science perspective,

but in a very fun way. I will say we had a really good conversation and Gordon shared with us a way for us to break down misinformation. It's a big topic, but I think he did a great job of helping us make sense of it in a way. What we know about misinformation, how to think about it, and how we can kind of also think about the causes and ways to measure it and many things. That is interesting. From a behavioral science

perspective, so we cover. That given the system as well, we naturally also cover quite a bit on. A is part. To what degree AI play a role in amplifying and creating misinformation and how to think about that as well? But yeah, so if you're interested to learn what makes misinformation so enticing and spread worthy, and also AI's. Role in all of this Listen to this episode with Gordon pennycook happens to. Murgatroyd. Wow. Perhaps, say, welcome Gordon to the Behavioral, the same

podcast. Thank you, happy to be here. Yeah, you don't sound very happy. Sound very happy to be here. Well, it's 9:30 in the morning, I mean. That is not that early. Because that is, that's true. My kids go to school at 8:00 and I, I'm, they say goodbye to me when I'm in bed still. So I yeah. Yeah. Yeah. Well, we're very happy, yeah. We are excited to have you here. I'm so excited. This is great, Oh my God. That's better.

And yeah, we obviously wanted to talk to you because it's been a lot of elections already this year, but then the Big 1 is coming up. And leading up to this has been

increasingly, I think, questions about how well we understand the role of misinformation, especially in this day and age where AI is having a big role. And so we really want to get your help to make sense of all of this. And I guess to start, it would be really interesting to maybe set the scene a little bit. Yeah, that's good. No, maybe I can give it like a an even broader kind of like

historical thing. Like I started doing research on misinformation in the 2016 election because fake news was this big thing and that and that was a particular case where people, it was like there was a bunch of stories about Macedonian teenagers and like just people trying to make a buck. They'd just make a website that looks like a news website, make up a bunch of stuff. And then that would be like make go viral on Facebook and they'd

make a few bucks off that. And so that was the kind of initial sort of thing that we were investigating. And misinformation has kind of evolved since then. You know, I mean, misinformation has been around since information has been around, like it's just false things.

But there's this, like, unique class of Internet stuff, social media falsehoods and so on. And then as the kind of like bots on social media we got became more advanced, I mean, most of they were pushing misinformation that people are making. It wasn't like, yeah, it was magnifying it. They were just like, a lot of the, it was inauthentic kind of engagement on social media that was helping things go viral. But I mean, a lot of the time it

was purely just people. People were getting their attention grabbed by some insane headline, and then it was just kind of getting spread. But then we like when Trump was the president, it became kind of apparent that there might be bigger problems. And that's misinformation coming directly from political elites in some cases. Like these things work together because I mean, it's just we happened again in the in the debate with Kamala Harris, which was like eating the cats and the dogs.

That was a kind of misinformation that came into Trump's sphere and then he tells everyone, millions of people about this thing, and not that many people would have seen it if it was just a fake news headline. But then it became a gigantic thing. And then it has a certain level of credibility to people who are following Donald Trump that that maybe they would have dismissed otherwise. Yeah.

Yeah, exactly. And I mean, one is just like getting in front of as many people as you can, if you have that size of a platform, but then also adding credibility and people kind of accept what he says and so on. So there's and there's more dynamic, there's different dynamics to that. And so it became a more difficult problem. And then you enter in AI as a

way to kind of like make things easier for the creator side, like, you know, building misinformation kind of that scale. Now, at the same time, I should say, though, it's not difficult to make up crap. It's not, it's not, it's actually pretty trivial for people to just make up stories. And a lot of time it's not even like that. It's like some kernel of something becomes it's like a rumors.

I'll give one explicit example. The people have heard the story about like in schools, teens are identifying as cats and there's litter boxes in bathrooms or whatever. And like the furries, are these the furries? Yeah, well, I mean, but they I mean it's because it's like it's like anti trans kind of trope. People think it's furries, but people think furries are trans people. But it has nothing to do with one or the other, of course.

And so they just think so they think that they're identifying as cats because you could you could identify as anything now and blah, blah, blah, blah, blah. But that rumor came out of like, I think there was actually was Kitty litter in a school, but it was because of shooter scenario. But there was no one obviously like pretending to be a cat and using the Kitty litter in the bathroom. I'm sorry that is lost. I mean what? Why do you need Kitty litter for an active shooter situation?

Well, you might be stuck in the room for a long time. Oh, my uncle who is a literal a teacher in Canada told her he's like, I heard a story about their kind of kid leaders. And I'm like Shane, there's no who's. And by the way, I was, there's a here's a little story. I was my parents were janitors in addition to other things, but I so after school, my after school's chores were to clean

the high school. And if a kid, you know, use the Kitty litter box, there's like they're not going to clean it. Like that's just not, you know what I mean? That's it's not in the cards. So anyways, that's but so like you don't really need AI to make up stuff like that, but it does. But you can. It does help scale things, and then it's. Easier. To scale things. Yeah, exactly. It's not difficult, but it but it does make it easier to make much much more of it. Yeah, exactly.

Yeah. And I'm interested in that sense, like, do you in your research or how you think about this, is that a way that this is in some way separate in terms like we have some things, as mentioned, there's been, you

know, various types of falsehoods and gossip and propaganda in various ways where the objective has been to kind of shape a narrative in some ways. And and we've recently seen some like popular kind of more maybe right wing pundits that has come out that they've been sponsored by Russian television to to push certain things.

And so that it is kind of like this type of misinformation that in some ways maybe has been around, but still obviously still amplified with, you know, Internet existing today. But then you have this maybe in some ways new wave of misinformation that is purely used, accelerated and spewed by used artificial means, like some form of bots or some form of LLM models that are kind of used for for some of this.

Is that something that's actively separated or how would you categorize misinformation in this way? It's. Not separated by us. I don't know, maybe some people are trying to do that. I think it would be very hard to know which was which. And we actually don't generally because I'm a psychologist, we don't spend a lot of time speculating about the source of it because it could who knows really.

And ultimately, and I think actually the, the funding of the influencers by Russian, like by, by Russian interests is a great example because they didn't tell them to do anything. They're just funding them. They're like, they're already saying what we want them to say. So they don't need AI. They don't need they just, we're just giving them money to keep doing what they're doing because they've already adopted the

narrative. And so like, I mean, so it's a case study in the sense that if you think about there's enough stuff out there already to drop on if you want to push false narratives. And then, you know, you can see how AI would help with it, but it's certainly not necessary. And like we had the same. So I don't think that the scope of the problem has really changed that much since AI has come around. It's probably made the lives of people who are trying to actively disinform people a

little bit easier perhaps. But like, I don't think it's really improving the falsehoods or like making them more more catchy or whatever. I think it's just that was already there, people already pretty good at that. So that is, is I think a really good way to frame it.

And so then looking on still technological side of how things have evolved, we you kind of mentioned the 2016 election where there was talk, I think there was Cambridge Analytica was kind of like one of the main things that was talked around. I want to stop on Cambridge Analytica because this, I think, is fake news in itself. I don't know, Gordon, how how familiar you are with what actually happened with Cambridge Analytica. Sure, yeah.

People thought that they were really targeting people, but they weren't. It was, you know, they thought they thought that it was like they had some secret sauce that was really doing persuasion. But it. Yeah, and, and I think a lot of us bought into that as well. I, I certainly did. I used this as a, as a case study to, to test out your debunk bot, which we're going to

talk about in another episode. But you know, I, I was actually, you know, fairly unsure, at least not 100% certain that what happened around this was, you know, certainly a lot of accounts on Facebook were mishandled and information was stolen or whatever you want to call it without consent. But in terms of what was done with that information, I think there's this huge misconception, let's just say, about how what actually happened and then how much that actually affected the

election. So you might ask someone, you know, why? Why did the election tilt in favor of Donald Trump? And someone could reasonably say, oh, you know, that psychological targeting that happened through through Cambridge Analytica's data problem. Yeah, and it probably was. I mean, people say the same thing with fake news as well. I mean, they're, so they're did that did the fake news sway the election? Was it Cambridge Political

people are looking for reasons. I mean, at the time, you can see we were looking for reasons to explain this thing that we could not explain as like, liberal intellectuals, I guess, you know, or like this doesn't make sense. It turns out people actually like Trump and they had different political opinions, you know what I mean? So. And the polling was wrong and they weren't listening to the. They weren't answering the polls. So that was basically what it

was. So you, all of our little explanations that we developed immediately after the election were. Yeah. I guess this for me raises another question, which is how do you quantify the impact of misinformation on something like an election? Can we in any reasonable way actually do that? Yeah, probably not. I mean, it's, I don't know, You can't, I can't quantify it on the like, what people actually believe, you know what I mean? Like how do you know this? Behavior.

Yeah, we want to. We what we really want to understand is what influences voting behavior. What people believe maybe is one thing, not necessarily the same as who they actually vote for. Sure, yeah. But yeah, I guess what I was saying was that it's hard to even pinned down a little of misinformation and what people's beliefs are and like, whether it's, you know, identity or

cultural negatives or whatever. But it's like, yeah, but definitely when it comes to the election, this is it's difficult because and one, respect almost all of like the vast majority of people who vote, you already know who they're going to vote for. That's what it's going to be anyways. And so you have this like weird situation in the States where there's this. Small minority in a particular few states who have a gigantic influence and there's a lot of things that could that's a very

like hard to predict context. And so like small things could have a big influence or not and I don't know. And so when it comes to misinformation, it's just, it's just one of these things where people, one thing that we can say for sure is that people do are walking around with a huge amount of like misunderstanding about how the world is. And that comes from somewhere. And the thing where it comes

from is like misinformation. Now, to the extent with that comes from the sort of thing that we mean when we say misinformation, like the social media viral stuff that everyone kind of worries about or just like simple mistakes or conversations with neighbors or whatever, you know, it's hard to

really tease that apart. Yeah. And maybe the way to tease it apart a little bit is if we then move forward a few years, we have 2020 and we have a strange year with, you know, COVID and many other things happening around there as well. But talking about COVID and and vaccine and misinformation, I know you've you've done research

around that specifically. Yeah. I mean, I think it's a good example because, like, thinking about people not getting vaccinated as a consequence of misinformation. Now, there are lots of reasons why people don't vaccinate. And they, you know, some of it like institutional trusts, you know, how much you trust doctors in the system and all that kind of stuff.

But, I mean, I think there's something to question that the amount of misinformation spreading had some impact on people not deciding not to vaccinate. It's just kind of very obvious, yeah. It's it's primarily attributed to Joe Rogan, right? Yeah, I'm not sure primarily, but like, I think what's it called? Ivermectin? I mean, we've got to be clear on the. I don't want to be too politically biased.

You know, people were saying that Joe Rogan was promoting horse tranquilizer or whatever, which is kind of may be true in a strict sense, but not true. And like, it's not really like what it's for for humans, you know what I mean? So but anyways, so it it goes both ways. But there's been this kind of growing thing in the research on misinformation where people are kind of like denying its importance and consequence. And I think that's kind of just wrong. I mean, I think that is the

case. We can it's it probably is overrated in many respects. Like people think it's the most there was like some like kind of a thing as aun thing or something where it's like misinformation is the most important problem. And like, Oh yeah, it's easy to like make it see it's but it's like proximally speaking, not, you know, like misinformation flourishes because of the structure of our fragmented information environment.

That's a much bigger problem. And there's other like but many other big problems, you know, climate change. But but it still ultimately is like we want people to have accurate beliefs about the world and misinformation is just false things. And so like, just in a strict sense, like, yeah, that's a huge problem. We don't, we want, that's the whole point of education. This is why we have expertise and and This is why we make podcasts where we teach people things.

It ultimately is all about trying to like, persuade people to believe things that are on based on good evidence, and misinformation counteracts that. It's interesting that you put it next to climate change because, well, for example, if you don't believe in climate change, then it doesn't like there's nothing you can even start to do there for getting someone to act. They're just, you know, it doesn't matter how big the actual problem of climate change is.

If they don't believe it, they're not going to do anything. That's that's true. I think that's so perfect example. So that is the case. And people often say in papers about climate change, they'll be like, we really need to kind of convince people that this is a problem because they're not going to do anything. But like at the same time, what are the most effective ways to combat climate change? It's not the everyday behaviors of citizens.

It's governmental policy and, you know, so those are things that cannot be influenced by changing an individual's belief. Then, of course, what people say is, well, if you change what people believe and then the governments will listen to them, well, you know that there's there's nothing there that's not completely wrong, but it's not exactly right either. So yeah. Pretty difficult indirect route for sure. Yeah, exactly.

Yeah. And so if I'm going to try to summarize a little bit where we're come so far. So we started in some ways where misinformation has been around forever and it's kind of it's the twin to information. Like it's kind of hard to avoid in human society. We're always have to deal with that.

We spoke about how a lot of the current misinformation is amplified by credibility through certain voices and like certain people that are kind of spreading this and with help of obviously technology today, like it's making it easier.

But at the same time, it's it's much of that still comes from people doing it. Like a lot of people being involved in, in sharing their maybe false beliefs that we just use masses and then has some kind of consequences, but that it's very hard to understand actually what the consequences

are still. Like it's kind of hard to even with something like vaccine hesitancy or or not taking your vaccine, it's still hard to really know, like to what degree people were vaccinated or not because of misinformation on the sources. Because it's like it's so mixed in to the fragment of like human experience that we talked to people on the like in our lives on online. Like this is very hard to know. Like, OK, this source of misinformation LED people to to not do this thing.

It's like good summary so far. Anything you would like underline? That's a great summary. I think that captures it. We could just use that we don't need. We could just cut out the first part. They've got the gist. Well, that, that's good to hear that sometimes you get it wrong into summarizing, but that's good. And and then I do want to kind of build on kind of moving from COVID in some ways. I would say I've noticed and I think it's been big part of the discourse is that, you know, I

think in 2022 there was some research that was around X or Twitter saying that like 2/3 or at least 1/3 of of all accounts were some form of bots on Twitter. We saw counties rising first, like more edge cases where you're like, hey, you could use deepfakes. Like deepfakes is kind of the thing you could kind of copy someone's voice or or face and that can be used in various ways. But it was kind of like little edge cases and, and more like, hey, be wary as soon it's going

to happen. And then I think it was around last year that was in Slovakia. They had some form of election deep fake thing around, especially with audio I believe where there was some basing up to the election, there was some basically deep fake audio that was spread that had some form of impact in terms of least fact checkers were scrambling to make sense of these fake recordings and how to make sense of it.

Yeah, they they had the Slovakian Liberal Party leader talking about vote rigging and I think it was raising the price of beer. Oh geez, of all the that's the last, That's the final straw. Yeah, we cannot tolerate this. Yeah. So we're moving into, you know, the area now 2024 where it feels a little more real when it comes

to this either. We have also, not only are these bots more like very simple bots that you just jump around and maybe use retweets or post some very generic things, but they're actually powered sometimes by large language models where they can actually have some form of level conversations online and and they can actually engage with people in some form of shack communities and and so on. Maybe the last thing I'll say as an example is recently I've seen this like people being worried

about what we see on images online because someone googled baby Peacock and they realized that of the 20 images on Google there was a baby Peacock, only three of them were actually real baby Peacocks. And all of them else was like AI generated images. So I think more and more of this

fear is sweeping out now. It's like, OK, where are we right now when it comes to this kind of technological influence on misinformation and so on. So you you alluded a little bit to that already in terms of we're not maybe going in orders of magnitudes away from where we're before. Like we're still, we're existing and ordering a pretty, pretty misinformation high levels. But yeah, how do you make sense of the current technology global advancements?

Yeah. I mean, I think what the real concern is essentially is like inauthentic behavior, like what you in a, in a kind of like a democratic society, we can have differences of opinion and like people get things wrong and whatever, and we can learn from each other and all that kind of stuff. But when people are acting inauthentically, then it kind of like disrupts the system.

And what that, what that means is like the obvious case that you're talking about is like literally it's a bot and it's not really a person. You think it's a person. And so like you're taking that to mean that there's all these people are supporting this cause, but actually it's a bunch of Russian bots and it's not actual Americans or whatever. Or it's like this is someone talking, but it's actually not them talking. It's an AI or this is a picture

of something. I think it's not actually that. But there's also like, I mean, those are all concerns, but of course, like inauthentic behaviors. Also people who are influencers who are not generating authentic opinions, they're trying to just say what they think will get them the clicks and the likes and the, and the views and the listens and so on. That's also an authentic

behavior. And So what you end up as a distorted kind of reality that people are entering online and which of course leads offline where it's like, it's hard to know what's going on. Misinformation makes that worse. But also like it's not just about truth and falsehood. It's just about like what, how do we know what other people are like, what's what's actually happening in the country?

And so like our understanding of like in, in the context of like, you know, given the election of like our political others is like way off. People have really, really do not, they totally think that like people on the other side of the aisle are way more extreme than they actually are. There was actually a paper that was published recently where, you know, people think the like way more people are responsible

for trolling than actually are. It's a tiny fraction of people actually engage in kind of like actual users engage in what you might call like an authentic, an authentic sort of behavior. But the, our sense of it is that like 40% of people actually do. It was actually 3% or something. I can't remember what the actual status. So we, our perception of reality is being really distorted by all these things. And then you enter in, these things are just essentially made

to be inauthentic. Yeah, it's not going to make it better, that's for sure. It's a it's an issue. So, so I'm familiar with this research about believing that others are much more extreme than they really are. I actually, I feel very strongly the opposite feeling in in my own experience, which is that I tend to think that other people think like me, even if they're, you know, not like me. And I extend my own belief system to the Republicans of the

world as well. And so when I think about, you know, the tightness of the race, for example, in the 2024 election, and I think, oh, about half the country is going to vote for Donald Trump, that is so inconsistent with my beliefs about what other people could possibly even be capable of. And my liberal bias is very, very strong here.

But, you know, and part of it, I'm sure, could be living physically in a little liberal bubble myself where I mostly interact with other people that are very similar to me. But for me, I am shocked to think about anyone voting for Donald Trump. I think both of those things will be true at once. So like you're kind of intuitive, like sense of you would assume that everyone's more similar to you than they actually are because you have to, you know, you live through your own perspective.

If I ask you, OK, then like think about a Republican and how extreme they are or whatever. Now you think, OK, wait, I think they're similar to me. But then you look and see that they're supporting Donald Trump. And so you make some sort of inference of it must be that, oh, they actually must be like way more extreme or like not different. So like you take. So then you make an inference that they're like, now you overcorrect essentially.

I'll give you I'll give you like another kind of like related examples of this. And this relates to conspiracy beliefs. So like people there's this like common thing for conspiracy believers where they think about the people to, you know, accept mainstream narratives, quote UN quote as like sheeple. You know, they're like there's going along, they're conformists. And like really the conspiracy theorists are the actual kind of critical thinkers, the unique critical free thinkers, they

often say. And So what you might assume from that is so people who believe conspiracies probably,

like, know that they're on the fringe, right? And kind of, they kind of want to be there. And that's the whole point of it. And like, if you ask them that, they probably would agree to it, Yeah. That's a feature, not a bug. Yeah, yeah. But at the same time, we did a study where we asked people to estimate the percent of people that agree with them about conspiracies. OK. And so I'll give an example in. So for the Sandy Hook false flag conspiracy, it's a ludicrous

extreme conspiracy. 8% of people in our sample thought it was true that Sandy Hook didn't actually happen. It was a whole false flag thing, actors and so on. And so we asked the people to both the people who believe and don't believe, but we asked them, estimate the percent of people that you think agree with you. And so 8% thought it was true. If they're calibrated, they'll say 8 to 10%, you know, whatever they said 61%, they thought they

were in the majority. They thought that most people realize that this was a false flag. So like both at the same time, they're like, yeah, I'm unique, I'm critical, freethinker, whatever. But they also kind of think that everyone agrees with them. And that's even when you're totally on the fringe. And so you can see how this there's kind of dueling elements of our psychology that contribute to these issues.

That's great. Yeah, Sam has been calling AI and and and specifically Gen. AI weird humans. And to me, this seems like humans are are really the weird humans. Yeah, exactly. It's it's a reflection of us. We're like, oh, how weird AI is. It's like it's just doing, you know, we created. It right it's. Just it's a little, it's like a weird little mirror, but it's still us in the mirror. Yep, Yep. Yeah, no, but that's super

interesting. So I think we've now spoken quite a bit of, you know, elements to setting the scene in terms of OK, where are we, what is the role of humans? We're humans, we're technology and so on. But used to really make sure we're on the same page in terms

of misinformation. There's a lot of things run misinformation that I think it's kind of interesting to to talk about and that idea in terms of I've seen a lot of versions of this. You know, we talk about a lie spreading X amount of faster than truth. I I've come across this thing called Brandolini's law. Have you come across this idea Brandolini's Law? Yeah. It's it's much easier to create bullshit than to refute it. Yeah, yeah, exactly, exactly.

And yeah, I think it's called some like bullshit, the symmetry principle or something like that as well. I wrote the 1st paper on the psychology of bullshit. You don't think I know about the Brandolini's law? That's a that's a dining criminal. Thing. Thank you. Oh my God, how dare you. Even I've got I've got Frankfurt's own bullshit on my shelf over here. That's good representing and and sticking up for your territory there. So yeah, tell us, tell us about

that. Like what is what is true about how misinformation spreads? Is it true that it spreads X much faster? Yeah. Is it 8 times more? I think so.

I think it is true actually. And So what this is an interesting case study in itself because so there was a paper published, Deb Roy and and others published this paper in Science in the kind of early days of misinformation research or whatever, saying that like misinformation spreads faster than true information, false spreads faster than true on it

was a Twitter analysis. We were really dubious about that because it was like most, it was among things that were fact checked, false things were spread faster than true things. And so we thought that was a weird kind of category of things, like the type of true things that are fact checked is different than just true things in general.

But we've, there's actually been more follow-ups and we've looked at data ourselves and like it does seem to be the case that things that are dubious are have a kind of extra bit of our virality on social media. If you and if you look at all kind of ways of slicing that, it seems to be kind of generally

the case. And there's a there's a kind of simple logic to it, which is that if you have freedom to say whatever you want and you're unconstrained by reality and the truth, you can say something more interesting. I mean, This is why we watch movies and TV. We kind of like say, OK, this is a different world now. It's OK that we can, you know, they're superheroes or whatever it is that's interesting. That's what narratives are, you

know? And so it is the case that like, yeah, in certainly in the context of like kind of news content, if you want to stretch the truth, it's going to get more clicks. And like, it's part of that also, like if you're stretching the truth, then you're probably also being more willing to really hype up the emotional language or like, play on people's fears or whatever. And so, yeah, if you have no scruples, you can you can make a buck on social media, unfortunately.

It's interesting the, the statement that a lie spreads 8 times faster than the truth to me, I, I feel it doesn't have, it doesn't have all of those characteristics, right? It's, there's not as much emotion or creativity to it. But I, I think it has enough of those elements that it makes me wonder, is that true? You know, like it kind of feels like a lie. That's so excessively precise. It's like there's no way you can estimate that because how?

Because think about what the think about what the claim is there. You're like, OK, we've done it. We've captured a representative sent of all true things and all false things. It's eight times we did it. Yeah, yeah. We're we're going to shoot them both out of a rocket and see how quickly they, yeah, they go. It's also a round Number. It's not like 8.35 times, it's eight times, Yeah. Anyways, in general, false things can spread faster than true things.

But it's just, it doesn't mean all false. It's. And who knows how many times more is it's. Yeah. What would you say are some more characteristics of lies? If you're trying to arm someone to maybe spot misinformation out there, this sort of false precision might be part of it. You talked about the the emotional aspect. What are some other features of misinformation? Let's shrink the category to the like, online viral falsehoods because that's easier to kind of wrestle with.

But yes, so I mean, So what? Those sorts of things are common, like using highly emotional language. But really that's what it's about is there's like an angle, a narrow way to think about misinformation is like things strictly speaking being false, which would be like something like if someone were to claim, oh, 10 million people died due to COVID. I actually have no idea how many people I should probably look that up. But actually it was half that number that was strictly

speaking being false. But like ultimately those quantities are not really that different for people to understand. And really the ultimate the, the kind of like actual kind of just you're getting across. There is a lot of people. And so it's in a certain sense, it's kind of true. Yeah, people can't think about those numbers anyway, so it doesn't really. Yeah, exactly. So it's just it's a distinction

without a difference. And so like misinformation often has this kind of quality where it's not just that it's like strictly speaking is false, but there's misleading that you're being deflected over here instead of over there. Haitians are eating their pets, for example.

Yeah, yeah, exactly. Because that what that's about is like, so there's the specific fact, the claim that like they're eating the pets, but what's really being communicated there is that these are dangerous people that are different than you and and so on. And that's also false. You know, that's all misleading like that, you know, exactly. So when when you're kind of engaging online, you have to think about who is the message coming from and what are they trying to get from me?

You have to kind of look behind the actual literal facts. Yeah, what's the point of this message? 1 So 1 issue is that so my oldest brother there is like kind of down the rabbit hole or he more than in the past than he is now. But I had conversations about kind of who he's listening to and what kind of sources he gets because he thinks all mainstream media is fake news and so on. And the, you know, the problem with that is that people who are journalists can actually be fired.

You know what I mean? Like if they would make something up, there's some mechanism by which they could be fired. But he likes listening to some dude on YouTube who like just says a bunch of stuff. That's OK. What stops this guy from lying or like saying like that he can't be fired.

In fact, if anything is incentivized to say extreme things that people will listen to the so they listen to his videos or watch his videos and he says, yeah, but he doesn't lie though, So you know what I mean? But that didn't work. But but I think there's a good old underlying point you have to think about, like, what is this person's incentive to mislead me? And that's that's one way to kind of get around try to figure out what's true or false on social media.

You got to think about what they're trying to do with the sources. Yeah, yeah. And I guess like one specific element we talked a little bit about that, you know, something that's emotionally arousing or appealing in some emotional way can have impact. And a very specific part of that, that honestly, it's always kind of pisses me off in some ways, is it's funny. That funny is one of the few things that can like really amplify a message more than anything.

And I love funny. Like I, I love comedy. I love watching everything from yeah comedy specials to stand up

to TV shows and and done some improv comedy myself. That's so ironic given how unfunny you are. I know it's it's. Wow, way to Crest the man. Anyway, what is unfortunate though is I feel like fun is it's kind of taking hostage by like this really bad actors where you have funny being used. I would say honestly very well by Trump. Most people are on the same page in terms of like part of his appeal and why it's hard to also counteract his bullshit is that he often times veils it in fun

and or funny. Same with, like we mentioned, the idea that certain people have a lot of sway online and and can like amplify things. Elon Musk is probably one of those people. And there's few things that he loves to retweet as much as funny memes, you know, like a fun meme, even if it's lie, it's like that doesn't really bother him. You know, if it's funny, that kind of takes the box and and I'm just interested to hear your thoughts on on funny and it's really misinformation.

Well, I mean, people can use it as a excuse essentially, or like you say something inappropriate, like that was a joke, you know what I mean? Or like, that's what Musk basically does on Twitter where he's like, oh, I was just trying to make people laugh. It just so happens that it was extremely harmful misinformation or whatever, you know what I mean? But but I was thinking, I thought it was funny, You know, if you have to.

So you have to think this is the tension when it comes to comedy, which is like you have to think about what the consequences of the jokes are. And so like, I know that comedians do actually like lament this often, but I saw Jerry Seinfeld, he said he was wrong about it. He said what happens is the culture shifts. You have to know what's funny within the culture. And so that's I think that's a good way to put it.

So I think one more thing about it too, is The Onion was like the original kind of fake news site. Of course, they did it in such a way that it was explicitly and kind of obviously satirical. What happened was used to be it used to. Yeah. And then I was just like reality used. To be an obvious difference. Yeah, yeah, yeah. So Facebook was was started clamping down on this kind of fake news, like where people are, you know, making these websites and this making things up.

And So what those websites started to do is like, oh, we're actually we're satire sites, but they weren't promoting the satirical. And it was like, you could see that they were like the, the jokes were not, they were political. It was like really not meant and also wasn't really funny, unlike the Onion, which is actually quite funny. You know, you can see that they were putting effort into the political side, but not the

funny side. And, and someone who knows comedy knows which, which one you're talking about. And in terms of like Trump being funny, he's usually funny kind of incidentally, you know what I mean? Like he, he doesn't really make jokes. He just says things that are funny. And so that's, you know, that's amusing for people that enjoy

things that are funny. But but I think there's a deeper problem there, which is like there's conduct you'd expect from somebody and said that's a position which is like, it's OK to make jokes, but sometimes you have to be serious or whatever. I think. And it's actually, it's worse than it's different than that, which is Trump.

So Trump's the apparent way that he approaches communication is very different than what you would expect and probably want from somebody in that sort of position, which is it doesn't really matter what I'm saying in a strict sense. I don't like that is the facts don't don't matter. It's the message that matters. That's what the the example was somebody pushed back to on. They asked them like, do you really think that they're eating the cats and the dogs in

Springfield? And he said that was just a report that I heard. He's like, he's like, it doesn't he's like for him, it doesn't matter what I believe. I'm just saying things. I'm just saying things like. Yeah, that's totally fine. It could be true. I'll just say it like it doesn't matter if it's true or not. Like he doesn't. There's no regard for the truth. And so like whether something's funny that was, that's funny, I'll say it does it. Will some people be misled by

it? Will it make people like other people angry? It doesn't matter because like I, you know, it's just something I want to say. So I'm going to say whatever it is that I want to say. And like, that's literally the stream of consciousness thing that he he does. So there's no filtering on like whether something is true, obviously, or like whether it would hurt people or whatever it is.

That's just literally just. And so people, people, what's interesting is that people take that to mean honesty. And it's it's honesty in a certain sense. It's honesty to yourself. Like you're just like, in a certain sense, like a little kids. Yeah. The ones that. Like having any responsibility? Yeah, there's. No, there's no responsibility. There's no like, there's no filter in this. It's like everything that comes in their head comes out of their mouth.

That's one way of being honest. I mean, what we kind of want honesty is like caring about the truth and getting things right. That's what is the important part of being honest. For a president, yeah, yeah. Of the United States. Or someone who's like a science scientist or like, a communicator. These are the things that are important. If you're in the public space and you have a large audience, being honest and being truthful is what important, not just being, like, genuine.

You know, doesn't matter who you are. What matters is that you're doing good in the world. Yeah. All right, it is time for us to jump to our quick fire round. This is a game, little game that we play. It's called to AI or Not to AI, and we're going to give you a bunch of tasks and you're going to tell us whether that use case is well suited to AI or not.

OK. All right, So the first one is about polling to AI or not to AI. Poll individual AI chat bots that have been trained on US demographics, past polling data, and representative media diets in order to predict the outcome of an election. I think the field says unsure. I'm going to say no because I'm a psychologist. OK, to AI or not to AI cast most likely votes for people who can't make it to the polls. No, no. I mean it's from based on the other one. Obviously, it's enough.

OK, translate from brash American English to polite Canadian English. Oh yes, very good. Great use. We'll see if there's some bias to filter saying like it started recommending team. What is it called? Tim horton's. Tim Horton's. Tim horton's. Yeah, exactly. You get a copy. Double double Tim Horton's, which is so much sugar, by the way. You start getting all these hockey ads. Yeah, I would be fine.

I would make me more like AI more appealing if it just like had a tuque on. A tuque is a winter hat, Sorry. I didn't even know that. It's a winter hat. OK, what about this? To AI or not to AI create a new celebrity. No, no, that's what is the Black Mirror. Yeah, let's just not. If it sounds like a Black Mirror episode, don't do it. OK, generate a gullibility profile for every person in the US. What do you mean by goal ability profile? How acceptable is that person to

misinformation, for example? That's part. I think that's something that could be useful and interesting. Create pseudo profile bullshit memes. Oh yeah, yes, no question. Very quick. Did you come across Truth Terminal the the bot that's called Truth Terminal? No. OK, Basically it was a controversial AAI bought that has been able to generate millions in crypto by itself by creating really really obscene memes online. Oh, interesting. Spread fake news on social media. Wait is the answer.

Can it do it or should it do it? Is it a use case well suited to AI? Sure. Oh yeah, making things up. OK then this is the big one then. To AI or not to AI combat the spread of fake news on social media. Yeah, yeah, for sure. TuneIn for the next episode. Perfect. Yes, the battle of the bots. OK. And that's a wrap. You've been listening to the Behavioral Design Podcast, brought to you by Habit Weekly and Nuanced Behavior.

Sam and Aline tell me this season is packed with incredible insights about behavioral design and AI, so be sure to subscribe and share the podcast with your friends. Though you might want to keep it away from your enemies. In case you haven't noticed, I'm an AI voice. Yep, pretty crazy. Quite the improvement since last season's AI outro, don't you think?

And if you'd like to collaborate with us at Nuance Behavior, where we use behavioral design to craft digital products with Nuance, e-mail us at hello@nuancebehavior.com or book a call directly on our website, nuancebehavior.com. A special thanks to the amazing Dave Pizarro for our show music, and to Mei Chen Yap and April English for their help in producing and publishing this episode.

Thanks again for tuning in. We'll be back soon with another exciting conversation where behavioral design and AI intersect. Oh. I still feel burned by that. You said that wasn't funny, but I'll I'll recover it. You know, it was a joke. That's like the irony of that. Like that was me trying to be funny and not being funny. That's that's the deep, deep irony.