WEBVTT - Why is misinformation so powerful? 0:00:00.009 --> 0:00:02.990 This podcast was made on the lands of the Wurundjeri people, 0:00:03.240 --> 0:00:06.109 the Woi Wurrung and the Bunurong. We'd like to pay 0:00:06.110 --> 0:00:09.238 respects to their elders, past and present, and emerging. 0:00:14.550 --> 0:00:17.870 From the Melbourne School of Psychological Sciences at the University 0:00:17.870 --> 0:00:20.350 of Melbourne, this is PsychTalks. 0:00:22.850 --> 0:00:25.250 Welcome to PsychTalks, the podcast where we dive into the 0:00:25.250 --> 0:00:28.370 fascinating research happening here at the University of Melbourne's School 0:00:28.370 --> 0:00:31.889 of Psychological Sciences. I'm Professor Nick Haslam, and I'm joined 0:00:31.889 --> 0:00:34.810 by my amazing co-host, Associate Professor Cassie Hayward. 0:00:34.970 --> 0:00:37.889 Hi, Nick. Great to be here. Today, we're tackling an 0:00:37.889 --> 0:00:42.720 incredibly important issue, the rise of misinformation and disinformation. Now, 0:00:42.729 --> 0:00:45.409 this podcast has previously looked at why we can fall 0:00:45.409 --> 0:00:49.209 victim to conspiracy theories and misinformation way back in season one. 0:00:49.630 --> 0:00:51.778 But we wanted to zoom in on the developments in 0:00:51.779 --> 0:00:54.979 the research since then. We've seen how the spread and 0:00:54.979 --> 0:00:58.939 influence of misinformation and disinformation has only increased, posing real 0:00:58.939 --> 0:01:02.450 threats to democratic processes, public health, and social trust. 0:01:02.939 --> 0:01:05.139 And joining us to unpack it all is Associate Professor 0:01:05.139 --> 0:01:08.738 Piers Howe, who leads research at the Information and Influence Lab. 0:01:13.080 --> 0:01:16.720 Piers, welcome to Psych Talks. Um, could you start by 0:01:16.720 --> 0:01:19.910 telling us a bit about your background, your current research, 0:01:20.080 --> 0:01:24.370 and I guess what drew you to study misinformation and disinformation? 0:01:24.800 --> 0:01:27.440 Yes, indeed, and thank you for having me on PsychTalks. 0:01:27.680 --> 0:01:32.080 So by training, I am a cognitive scientist, so I 0:01:32.080 --> 0:01:36.080 am very interested in studying, um, information processing and how 0:01:36.080 --> 0:01:39.440 people make decisions. So naturally I got drawn to this topic. 0:01:39.849 --> 0:01:43.860 But obviously, what's been going on in the news and 0:01:44.029 --> 0:01:46.629 um what has been going on elsewhere. For example, last 0:01:46.629 --> 0:01:51.269 year the World Economic Forum listed mis- and disinformation as the 0:01:51.269 --> 0:01:54.230 number one short term global threats, that kind of gets 0:01:54.230 --> 0:01:56.910 your attention. So there's been lots of reasons why I've 0:01:56.910 --> 0:01:59.260 been focusing on this more and more. 0:01:59.809 --> 0:02:03.129 In my lab, the um Information and Influence lab, we 0:02:03.129 --> 0:02:08.970 focus on creating interventions both to reduce the spread of 0:02:08.970 --> 0:02:13.279 false information primarily on social media, but also to try 0:02:13.279 --> 0:02:16.888 and mitigate the harmful effects of mis- and disinformation. 0:02:17.229 --> 0:02:21.520 So perhaps to reduce uh support for undemocratic um practises 0:02:21.520 --> 0:02:23.830 or um even partisan violence. 0:02:24.240 --> 0:02:28.070 So Piers, you're using terms like misinformation and disinformation and 0:02:28.070 --> 0:02:30.399 you also hear about uh fake news. 0:02:30.880 --> 0:02:32.829 Uh, terms like that thrown around all the time. Can 0:02:32.830 --> 0:02:34.970 you break down what they mean and how they differ? 0:02:35.220 --> 0:02:40.179 Yes, sure. So, um, misinformation is false information that is 0:02:40.179 --> 0:02:44.740 communicated without the intent to deceive, though perhaps by people 0:02:44.740 --> 0:02:48.979 who haven't bothered to check whether it's true or not. Conversely, 0:02:49.100 --> 0:02:53.258 disinformation is communicated with the specific intent to deceive and 0:02:53.258 --> 0:02:57.619 the person communicating it knows it's false or at least misleading. 0:02:58.389 --> 0:03:03.339 Um, fake news is not, not a particularly scientific term, um, 0:03:03.350 --> 0:03:07.309 popularised by a certain person and it generally means something 0:03:07.309 --> 0:03:10.050 which is put in a news format but hasn't undergone 0:03:10.050 --> 0:03:14.389 the editorial standards of a mainstream newspaper. So the attempt 0:03:14.389 --> 0:03:16.710 is to try and give something which hasn't gone through 0:03:16.710 --> 0:03:20.070 the editorial process, the appearance that it has to give 0:03:20.070 --> 0:03:22.660 it credibility when it shouldn't receive as much credibility. 0:03:22.869 --> 0:03:24.740 Uh, I thought it was just news you didn't agree with. 0:03:26.460 --> 0:03:29.740 [Laughs] There has been an alternative definition put forward by 0:03:29.740 --> 0:03:33.449 a certain president, but, um, no, that's not its definition. 0:03:34.100 --> 0:03:37.770 With all of those with disinformation, misinformation, and fake news, 0:03:38.179 --> 0:03:41.820 one impact that's been talked about is the impact they 0:03:41.820 --> 0:03:45.589 can have on elections. Well, we're talking about certain politicians. Um, 0:03:45.699 --> 0:03:50.300 is there actual evidence that election outcomes have been manipulated 0:03:50.300 --> 0:03:53.899 or influenced, or is it kind of a more subtle 0:03:54.309 --> 0:03:56.429 impact on population beliefs? 0:03:56.669 --> 0:04:03.500 So there's increasing evidence that coordinated disinformation campaigns can change 0:04:03.779 --> 0:04:08.309 voting behaviour by a few percentage points. So there is 0:04:08.309 --> 0:04:12.529 theoretically the possibility that they could swing a close election. 0:04:12.880 --> 0:04:15.110 Proving that, of course, would be hard because it would 0:04:15.110 --> 0:04:17.070 only occur for a close election. 0:04:17.700 --> 0:04:22.420 However, the bigger threat is perhaps the um decrease in 0:04:22.420 --> 0:04:26.459 trust regarding election results. And there was a particularly pertinent 0:04:26.459 --> 0:04:31.700 example of this in the 2024 Romanian presidential election where 0:04:31.700 --> 0:04:34.738 the um the first round was nullified. 0:04:35.339 --> 0:04:39.320 Um, in part because of fears of the influence of 0:04:39.320 --> 0:04:43.589 a Russian disinformation campaign. So there's an example of, um, 0:04:43.600 --> 0:04:47.720 false information, disinformation having a very real and tangible effect. 0:04:47.880 --> 0:04:50.640 Just on, on close elections, I mean, we've just had 0:04:50.640 --> 0:04:53.160 our Australian election and a lot of the seats were 0:04:53.160 --> 0:04:58.678 very close, coupled with the ability to specifically target certain 0:04:58.678 --> 0:05:02.428 electorates with social media messages, it can't be too far 0:05:02.428 --> 0:05:06.260 off the radar that it's possible to have those, um, 0:05:06.630 --> 0:05:10.479 impacts in those close seats through targeted messaging from either 0:05:10.829 --> 0:05:14.669 real or, or fake news outlets or information outlets. 0:05:15.109 --> 0:05:17.670 I would have to agree. So there was a lot 0:05:17.670 --> 0:05:21.470 of um misinformation and disinformation going around. I should say 0:05:21.470 --> 0:05:25.269 from both sides. Um, maybe it roughly balanced out, maybe 0:05:25.269 --> 0:05:27.950 it didn't quite balance out. It would be surprising if 0:05:27.950 --> 0:05:30.829 at least some of those results weren't swung by mis 0:05:30.829 --> 0:05:36.029 and disinformation, especially due to coordinated campaigns, but we can't 0:05:36.029 --> 0:05:37.869 prove that, so we have to, as scientists be a 0:05:37.869 --> 0:05:39.750 little bit cautious about claiming that. 0:05:40.440 --> 0:05:45.220 So it could potentially swing closely contested elections, um, and 0:05:45.220 --> 0:05:46.839 that's why we care about it because it might have 0:05:46.839 --> 0:05:49.399 all sorts of influence for who's in power. But is 0:05:49.399 --> 0:05:52.160 there any evidence it's actually caused other harms outside the 0:05:52.160 --> 0:05:52.910 political system? 0:05:53.200 --> 0:05:57.149 So I would regard mis and disinformation as a "meta risk". 0:05:57.399 --> 0:06:01.558 The principal problem is that it reduces our ability to 0:06:01.559 --> 0:06:03.679 respond to other risks. 0:06:04.309 --> 0:06:07.880 Um, the classic example is our ability or slow response 0:06:07.880 --> 0:06:12.040 to healthcare risks. So for example, in the 1950s, um, 0:06:12.049 --> 0:06:14.719 in the USA but also in the UK there was 0:06:14.720 --> 0:06:19.799 growing evidence linking cigarette smoking to cancer. And in response 0:06:19.799 --> 0:06:24.828 to this, the top six American tobacco companies, um, um, 0:06:24.839 --> 0:06:28.200 came together and they funded the PR firm Hill and 0:06:28.200 --> 0:06:28.719 Norton 0:06:29.040 --> 0:06:34.440 to create a decades-long disinformation campaign, which was perhaps one 0:06:34.440 --> 0:06:39.790 of the most successful um disinformation campaigns in history. Um, historically, 0:06:39.920 --> 0:06:43.269 looking back, it seems to have delayed effective legislation by 0:06:43.269 --> 0:06:46.200 at least 10 years, and the legislation that did come 0:06:46.200 --> 0:06:49.678 in was very much watered down. Um, it really boosted 0:06:49.678 --> 0:06:53.149 their profits and depending on who's modelling you look at, 0:06:53.359 --> 0:06:55.660 this campaign alone, and it wasn't just the 0:06:55.734 --> 0:06:58.484 single event that was um it started off on the 0:06:58.484 --> 0:07:04.204 4th of January 1954 with this um with this advert, 0:07:04.375 --> 0:07:07.855 a frank statement to cigarette smokers which was actually viewed 0:07:07.855 --> 0:07:11.375 by 43 million Americans. So it, it was a very 0:07:11.375 --> 0:07:12.894 well funded, very big campaign. 0:07:13.320 --> 0:07:17.049 But um it probably cost the lives of somewhere between 0:07:17.049 --> 0:07:21.859 3 and 7 million Americans just by reducing the legislation 0:07:21.859 --> 0:07:25.779 which would have hampered cigarette sales. So that's one incredible example. 0:07:25.859 --> 0:07:28.299 The reason why I picked on that example is it 0:07:28.299 --> 0:07:33.220 literally became a playbook that other commercial companies used because 0:07:33.220 --> 0:07:36.890 it had the forgive me for saying this, the genius 0:07:36.890 --> 0:07:37.779 insight 0:07:38.059 --> 0:07:41.890 that you don't have to persuade something that someone's not true. 0:07:42.200 --> 0:07:45.670 You just have to cast enough doubt that it is true, 0:07:45.880 --> 0:07:49.519 and then people will delay their actions, especially if they 0:07:49.519 --> 0:07:51.950 don't want to act. So for example, people are smoking, 0:07:52.079 --> 0:07:54.359 they're addicted, they don't want to give up smoking, and 0:07:54.359 --> 0:07:56.359 so you all you have to do is give them 0:07:56.359 --> 0:07:59.600 enough doubt and then they won't give up smoking. Legislators 0:07:59.600 --> 0:08:02.950 won't impose laws because they're getting money from taxes from this. 0:08:03.119 --> 0:08:04.440 So they don't want to do that. 0:08:04.940 --> 0:08:08.299 So this literally created a playbook on how to most 0:08:08.299 --> 0:08:12.250 effectively create doubt and I'm talking about things like, fund 0:08:12.250 --> 0:08:16.570 scientific research that just throws doubt on the issue, um, 0:08:16.739 --> 0:08:21.010 hire fake experts to testify that the evidence isn't yet conclusive. 0:08:21.220 --> 0:08:24.500 Things like that, very effective, and other companies, um, the 0:08:24.500 --> 0:08:29.940 petrochemical industry, for example, various polluting industries, even pharmaceutical industry 0:08:29.940 --> 0:08:33.150 trying to deny that certain medications were addictive, literally 0:08:33.205 --> 0:08:35.544 use the playbook going forward. 0:08:35.875 --> 0:08:40.075 Um, a more modern example would be, um, COVID-19 in 0:08:40.075 --> 0:08:46.084 the US. So between May of 2021 and September 2022, 0:08:46.354 --> 0:08:52.863 over 230,000 Americans died unnecessarily due to COVID. So these 0:08:52.864 --> 0:08:54.755 were people who were offered the vaccine, 0:08:55.250 --> 0:08:58.750 refused the vaccine, went on to die, and this takes 0:08:58.750 --> 0:09:01.099 into account that not all of them would have lived 0:09:01.099 --> 0:09:05.500 had they got the vaccine. Now, misinformation wasn't the only 0:09:05.500 --> 0:09:07.539 reason why they didn't take the vaccine. A lot of 0:09:07.539 --> 0:09:10.500 these people didn't want to take vaccines anyway, but it 0:09:10.500 --> 0:09:14.380 was an enabling factor. It cast enough doubt that it 0:09:14.380 --> 0:09:18.750 helped them decide not to take the vaccine and ultimately die. So, 0:09:19.539 --> 0:09:23.500 we often think of misinformation's primary aim as trying to 0:09:23.500 --> 0:09:27.140 convince something of something false. That's actually not how it's 0:09:27.140 --> 0:09:31.210 generally used. It's generally used to create doubt which either 0:09:31.210 --> 0:09:34.140 delays actions or allows people to act in a way 0:09:34.140 --> 0:09:37.468 that they want to act anyway. That's a second function 0:09:37.469 --> 0:09:40.940 of mis and disinformation. The third one is actually by foreign 0:09:40.940 --> 0:09:43.830 disinformation campaigns, is to purposely designed 0:09:44.244 --> 0:09:47.835 to actually increase social divisions and weaken a country, and again, 0:09:48.005 --> 0:09:51.195 to reduce the ability of that country to take action 0:09:51.484 --> 0:09:52.955 against a foreign threat. 0:09:53.164 --> 0:09:55.965 Just on your playbook there, you can see that happening 0:09:55.965 --> 0:09:58.844 in the US at the moment with funding of studies 0:09:58.844 --> 0:10:02.085 to look at the link between autism and vaccines, and 0:10:02.085 --> 0:10:05.364 just the fact that they're funding those studies throws doubt 0:10:05.364 --> 0:10:08.325 on what is pretty much established as no link. 0:10:08.780 --> 0:10:11.099 And then you see the, the vaping companies doing pretty 0:10:11.099 --> 0:10:13.949 much the same playbook as the tobacco companies, often the 0:10:13.950 --> 0:10:16.939 same companies, doing the same thing with vaping that you see. So, 0:10:16.989 --> 0:10:19.390 you must see a lot of these examples just every 0:10:19.390 --> 0:10:20.270 day at the moment. 0:10:21.090 --> 0:10:24.359 Well, yes, that's exactly right. And of course, what else 0:10:24.359 --> 0:10:26.718 you're seeing in the US is not only are these 0:10:26.969 --> 0:10:27.729 um 0:10:29.159 --> 0:10:32.359 dubious, not only is this dubious research being funded, but 0:10:32.359 --> 0:10:36.270 the proper research is being defunded and you're getting a 0:10:36.270 --> 0:10:41.150 lot of statements being um made to support um unproven 0:10:41.150 --> 0:10:44.549 scientific techniques and you're getting a lot of true scientific 0:10:44.549 --> 0:10:51.799 information removed from US government websites, again, to create confusion. 0:10:52.130 --> 0:10:54.250 It's primarily about creating confusion. 0:10:54.580 --> 0:10:56.949 Which presumably makes it harder to fight because it's not 0:10:56.950 --> 0:11:00.700 just telling lies, right? It's, uh, increasing doubt and there's 0:11:00.700 --> 0:11:04.809 always some doubt about pretty much any proposition, uh, I guess, uh, 0:11:04.820 --> 0:11:08.580 or removing truth rather than spreading lies. So that presumably 0:11:08.580 --> 0:11:09.809 makes it a lot harder to fight? 0:11:10.020 --> 0:11:13.450 Yes, so that was why it was the genius insight 0:11:13.780 --> 0:11:17.505 of John W. Hill back in the 1950s that it's, 0:11:17.544 --> 0:11:20.815 it's such an easier battle to fight, not to, not to, 0:11:20.945 --> 0:11:24.664 it's impossible to prove that vaccines cause autism because they don't. 0:11:24.905 --> 0:11:28.294 But it's much easier to pour doubt on the fact 0:11:28.544 --> 0:11:31.505 that they've been proved not to cause autism and that's, 0:11:31.604 --> 0:11:32.905 that's what's going on. 0:11:33.770 --> 0:11:36.968 You've mentioned in your work that misinformation can shift societal 0:11:36.969 --> 0:11:40.609 narratives like the illusory truth effect or normalising fringe ideas. 0:11:41.080 --> 0:11:43.869 Can you explain how this works and why it's so dangerous? 0:11:44.200 --> 0:11:47.760 So, the illusionary truth effects is the psychological phenomenon that 0:11:47.760 --> 0:11:50.479 the more you hear something, even if it's not true 0:11:50.479 --> 0:11:53.119 and even if it's not supported by evidence, the more 0:11:53.119 --> 0:11:56.080 likely you are to believe it, or at least to 0:11:56.080 --> 0:11:59.440 begin to believe it's plausible even if you can't believe 0:11:59.440 --> 0:12:01.760 it entirely. So that's coming back to this doubt theme. 0:12:01.960 --> 0:12:04.159 You begin to doubt that it's impossible. 0:12:04.659 --> 0:12:06.890 And the reason why this is an issue is this 0:12:06.900 --> 0:12:10.169 this concept of the Overton window, which is the range 0:12:10.460 --> 0:12:14.650 of um issues that the public think is plausible and 0:12:14.650 --> 0:12:18.700 so are willing to discuss. And using the illusionary truth effects, 0:12:18.710 --> 0:12:19.409 if you 0:12:19.849 --> 0:12:23.608 just keep on repeating extreme views, you can alter that 0:12:23.609 --> 0:12:28.570 Overton window to include um more and more extreme content 0:12:28.570 --> 0:12:33.169 and thus shift the narrative of the discussions and that's 0:12:33.169 --> 0:12:35.809 by shifting the narrative, by shifting what people are willing 0:12:35.809 --> 0:12:40.239 to discuss and entertain, um, you can often shift their 0:12:40.239 --> 0:12:44.489 decisions because one of the heuristics that people use is 0:12:44.489 --> 0:12:46.000 they tend to try and 0:12:46.440 --> 0:12:49.150 stick on the middle decision. So they hear a range 0:12:49.150 --> 0:12:52.520 of ideas and they try and pick the central one 0:12:52.520 --> 0:12:56.959 because as a heuristic, that's generally quite a um a 0:12:56.960 --> 0:12:58.590 good um thing to do. 0:12:59.000 --> 0:13:02.159 So a heuristic is um a mental shortcut. 0:13:02.729 --> 0:13:07.219 It's a way of making an educated guess. In this case, 0:13:07.309 --> 0:13:10.340 an educated guess of whether something is true or not, 0:13:10.710 --> 0:13:14.070 but it is not a reliable method. It's a quick 0:13:14.070 --> 0:13:16.739 and dirty method which usually gives you the right answer, 0:13:16.789 --> 0:13:17.760 but not always. 0:13:18.570 --> 0:13:22.729 People who are pushing disinformation campaigns know that um as 0:13:22.729 --> 0:13:25.619 humans use heuristics and so if they can shift one end, 0:13:25.729 --> 0:13:29.369 they can shift what reasonable people trying to be reasonable 0:13:29.369 --> 0:13:31.969 but being time poor do. And that's one of the 0:13:31.969 --> 0:13:36.049 reasons why um humans fall for disinformation. We're all busy, 0:13:36.210 --> 0:13:37.130 we all have lives. 0:13:37.744 --> 0:13:42.224 We can't actually research every topic we hear about and 0:13:42.224 --> 0:13:45.064 so we need to fall back on heuristics simply to 0:13:45.065 --> 0:13:46.464 get through our lives. 0:13:46.825 --> 0:13:49.385 When I first started in this field, there were media 0:13:49.385 --> 0:13:53.135 literacy studies which were actually trying to teach people how 0:13:53.135 --> 0:13:56.015 to spot false information and they were saying things like, 0:13:56.344 --> 0:13:59.184 you can spot a false advert by looking for misspellings, 0:13:59.265 --> 0:14:03.705 looking for unprofessional formatting. Those are heuristics, and nowadays a 0:14:03.705 --> 0:14:06.905 lot of the false information is very, very professionally done. 0:14:07.489 --> 0:14:08.479 Um, so 0:14:09.659 --> 0:14:13.340 this, humans aren't mean, they aren't out to, um, deceive 0:14:13.340 --> 0:14:16.820 other people, mostly. Some are, but most aren't, but they're 0:14:16.820 --> 0:14:19.789 time poor and forced to use heuristics, which can be exploited. 0:14:19.940 --> 0:14:21.940 Just on the heuristics thing, and I don't want to 0:14:21.940 --> 0:14:24.940 get too much into the anti-vax weeds because we'll get 0:14:24.940 --> 0:14:29.570 hate mail. But, um, when we're talking about how anti-vax 0:14:29.570 --> 0:14:33.409 campaigners talk about the issues, they will be very direct 0:14:33.409 --> 0:14:36.340 and assertive about their claims, and they will say, vaccines 0:14:36.340 --> 0:14:37.140 cause autism. 0:14:37.479 --> 0:14:39.969 And then you look on the other side, and scientists 0:14:39.969 --> 0:14:43.619 aren't trained to talk like that. They talk in words like, 0:14:43.650 --> 0:14:47.080 there's no evidence that vaccines cause autism. So it's a very, 0:14:47.250 --> 0:14:50.799 from a listener's point of view, one seems very sure 0:14:50.799 --> 0:14:54.489 and one seems very kind of weak, and you can 0:14:54.489 --> 0:14:57.570 kind of see how people get led down certain paths 0:14:58.159 --> 0:15:00.650 from those heuristics just in terms of how people talk 0:15:00.650 --> 0:15:01.210 about things. 0:15:01.799 --> 0:15:07.039 Would you suggest that scientists start to talk more certainly 0:15:07.039 --> 0:15:10.760 about their findings? Or just in terms of the, the 0:15:10.760 --> 0:15:13.679 damage that it does when we have the other side 0:15:13.679 --> 0:15:18.000 of a lot of these issues talking very, kind of 0:15:18.000 --> 0:15:20.080 with a lot of assurance in their claims. 0:15:20.280 --> 0:15:22.799 So what you say is very true. It's been very 0:15:22.799 --> 0:15:25.559 well studied in the literature that just by giving facts 0:15:25.559 --> 0:15:29.039 and figures, even when you can't justify them, makes the 0:15:29.039 --> 0:15:30.710 message sound more plausible. 0:15:31.479 --> 0:15:36.260 But as scientists, we can't do that. We have to 0:15:36.260 --> 0:15:40.739 stick with the truth. We can't become um what would 0:15:40.739 --> 0:15:44.140 amount to being um partisan, wouldn't it? Because then we'd 0:15:44.140 --> 0:15:46.619 be trying to push a political message because once we 0:15:46.619 --> 0:15:50.250 go beyond what the science is, then we're getting political. 0:15:50.659 --> 0:15:53.859 And that's probably going to backfire. So no, I wouldn't 0:15:53.859 --> 0:15:55.260 recommend scientists do that. 0:15:55.609 --> 0:15:58.619 But what I would recommend science, and this is actually 0:15:58.619 --> 0:16:00.609 one of the things I'm, we're doing in our lab, 0:16:00.780 --> 0:16:04.619 is explaining what the truth looks like and trying to 0:16:04.619 --> 0:16:08.020 get people not to use heuristics and not to assume 0:16:08.020 --> 0:16:10.219 that every time they see a number. On my way 0:16:10.219 --> 0:16:13.580 back when on my commute home, I pass I I 0:16:13.580 --> 0:16:17.830 go past the barbershop which um says that um, 0:16:18.679 --> 0:16:26.119 That character is, um, 84 84% innate and 16, 16% 0:16:26.119 --> 0:16:29.729 the haircut, and they fix the 16%. Now that's a 0:16:29.729 --> 0:16:32.789 joke and we can all see it's a joke, but, um, 0:16:33.119 --> 0:16:34.840 we need to train people, and this is one of 0:16:34.840 --> 0:16:35.900 the things my lab is trying 0:16:35.955 --> 0:16:39.744 to do. So they don't fall for these simple tricks 0:16:39.744 --> 0:16:41.895 that a number means you're credible. 0:16:42.544 --> 0:16:44.465 What are some other things for our listeners who want 0:16:44.465 --> 0:16:48.784 to avoid being pulled by disinformation campaigns? What practical steps 0:16:48.784 --> 0:16:50.825 can they take or what things can they do to 0:16:50.825 --> 0:16:53.335 avoid falling down these rabbit holes? 0:16:53.744 --> 0:16:58.729 So, the best advice is perhaps the most boring, which 0:16:58.729 --> 0:17:03.450 is watch what information sources you use. So the majority 0:17:03.450 --> 0:17:08.640 of Australians now um get the majority of the um 0:17:08.640 --> 0:17:12.810 news from social media. It's just popping up in their feed. 0:17:13.089 --> 0:17:18.040 Social media is a very unreliable source, consciously deciding not 0:17:18.040 --> 0:17:21.050 to get your information from an unreliable source because we 0:17:21.050 --> 0:17:23.369 all suffer from the illusionary truth effects. 0:17:23.780 --> 0:17:26.989 And another thing we all suffer from is believing we 0:17:26.989 --> 0:17:30.349 don't suffer from the illusionary truth effects. If you just 0:17:30.349 --> 0:17:33.390 see it more and more often, you will shift what 0:17:33.390 --> 0:17:36.709 you believe is plausible. That's just how humans are. So 0:17:36.709 --> 0:17:40.150 use reliable news sources, go out of your way to 0:17:40.150 --> 0:17:44.229 subscribe to these reliable mainstream news sources, ABC, BBC, New 0:17:44.229 --> 0:17:48.988 York Times, The Age, The Australian. Distinguish between when these 0:17:48.989 --> 0:17:53.218 reliable sources are reporting factually and when you have opinions. 0:17:53.859 --> 0:17:56.849 Um, a lot of people don't make that distinction. In fact, 0:17:56.939 --> 0:18:01.020 a surprisingly large amount of people cannot distinguish between a fact, 0:18:01.060 --> 0:18:04.420 a newspaper report and an advert put in this in 0:18:04.420 --> 0:18:07.020 this news newspaper with the word advert at the top, 0:18:07.170 --> 0:18:10.300 but put in the format of a newspaper article. So 0:18:10.300 --> 0:18:13.020 be clear which bit of the source you're reading and 0:18:13.020 --> 0:18:14.739 that bit is actually reliable. 0:18:15.189 --> 0:18:17.989 But also, even the reliable news sources are now beginning 0:18:17.989 --> 0:18:21.790 to use clickbaity headlines. What pops up on social media 0:18:21.790 --> 0:18:26.229 is typically just the headline. Most people share that without 0:18:26.229 --> 0:18:28.020 even clicking on the link. 0:18:28.479 --> 0:18:31.079 Um, so they don't even know what they're sharing because 0:18:31.079 --> 0:18:34.349 the actual article, sometimes these, these um links go to 0:18:34.349 --> 0:18:38.319 reputable articles, but the actual article, even it's reputable, might 0:18:38.319 --> 0:18:42.609 not quite encapsulate that headline. That headline is short. Um, 0:18:42.880 --> 0:18:45.489 and in some cases, some social media companies actually changed 0:18:45.489 --> 0:18:47.839 the headline as well, which is, uh, which, which can 0:18:47.839 --> 0:18:49.438 be a problem. Um. 0:18:50.060 --> 0:18:54.140 Don't just put things into search engines because the AI 0:18:54.140 --> 0:18:57.939 will spit out an amalgamated result which may not correspond 0:18:57.939 --> 0:18:59.180 to reality in any way. 0:18:59.750 --> 0:19:06.290 Um, reliable sources also include things like government websites, but also, um, Wikipedia. 0:19:06.530 --> 0:19:08.369 There's a bit of a backlash at the moment in 0:19:08.369 --> 0:19:12.640 our schools against Wikipedia. Wikipedia can be a very, very 0:19:12.640 --> 0:19:16.810 valuable source, but it sometimes misses the mark. And at 0:19:16.810 --> 0:19:19.530 the top of the Wikipedia page, there's a talk tab 0:19:19.530 --> 0:19:22.170 and if you go there, you can get a rating 0:19:22.170 --> 0:19:25.169 of how reliable that particular Wikipedia page is. 0:19:25.729 --> 0:19:29.420 And although the article itself may be unreliable, that rating 0:19:29.420 --> 0:19:32.469 is usually very good indeed. And I would argue because 0:19:32.469 --> 0:19:36.060 of the illusionary truth effect, if that rating is low, don't, 0:19:36.219 --> 0:19:39.010 don't read, don't think, oh, OK, this, this isn't reliable, 0:19:39.020 --> 0:19:41.099 but I'm gonna read it anyway and I'll just, you know, 0:19:41.189 --> 0:19:44.489 be sceptic because you will end up fooling yourself. 0:19:45.020 --> 0:19:48.780 A lot of those tactics of critical thinking are tactics 0:19:48.780 --> 0:19:52.698 that the misinformation side will often say that they're doing, right? 0:19:52.810 --> 0:19:54.239 They'll say, don't fall for 0:19:54.689 --> 0:19:56.890 the, you know, don't be a sheep, don't follow this, 0:19:56.979 --> 0:19:59.409 this is the real truth. And they kind of believe 0:19:59.410 --> 0:20:03.050 that they're critical thinkers and doing all those things. They're 0:20:03.050 --> 0:20:07.359 just going to the real source rather than the fake news. So, 0:20:07.609 --> 0:20:10.770 how do you get around people thinking that they are 0:20:10.770 --> 0:20:13.810 doing all those things? It's just that they're being led 0:20:13.810 --> 0:20:15.719 to the wrong sources. 0:20:16.050 --> 0:20:17.930 So that's the million dollar question. 0:20:18.290 --> 0:20:21.949 And I'll just be straight, I don't have a good answer, 0:20:21.979 --> 0:20:24.540 and the field doesn't have a good answer. Our answer 0:20:24.540 --> 0:20:28.329 used to be go to third party fact checkers. Now 0:20:28.329 --> 0:20:32.550 we have Breitbart doing fact checking and those fact checks 0:20:32.550 --> 0:20:35.500 bear no resemblance at all to any of the other 0:20:35.500 --> 0:20:36.219 fact checkers. 0:20:37.140 --> 0:20:42.930 Um, but people are quoting these Breitbart fact checks. They're quoting, um, um, 0:20:42.939 --> 0:20:48.530 Fox News. So you have an alternative information ecosystem and 0:20:48.599 --> 0:20:51.649 we're not really sure how to deal with it. We 0:20:51.650 --> 0:20:54.760 thought that you could go to the ultimate source of truth, 0:20:54.829 --> 0:20:58.349 which would be science. Well, hasn't that been proved wrong? 0:20:58.560 --> 0:21:01.319 I remember a junior member of faculty coming up to 0:21:01.319 --> 0:21:03.479 me being terribly, terribly excited. 0:21:03.949 --> 0:21:06.510 Because he knew I I studied misinformation and he had 0:21:06.510 --> 0:21:10.260 the solution. The solution was very simple for him. Australia 0:21:10.260 --> 0:21:16.069 needed an independent scientific body which would research um topics 0:21:16.069 --> 0:21:19.829 which were important to elections and give the facts to 0:21:19.829 --> 0:21:23.910 the Australian people. So of course I mentioned CSIRO, I 0:21:23.910 --> 0:21:27.510 mentioned the nuclear report, and I mentioned that it was 0:21:27.510 --> 0:21:30.369 pretty much ignored by a certain party. 0:21:30.824 --> 0:21:33.094 We don't have a good answer. So my lab is 0:21:33.094 --> 0:21:36.535 trying other ways. That, that's what you, what you're talking 0:21:36.535 --> 0:21:38.734 about is too hard for us to tackle. But there's 0:21:38.734 --> 0:21:40.574 other ways you can do. There's the foot in the 0:21:40.574 --> 0:21:43.974 door technique. You can talk about other values and you 0:21:43.974 --> 0:21:48.094 can build up getting people more and more reasonable, and 0:21:48.094 --> 0:21:52.334 then eventually they can perhaps realise that perhaps Fox News 0:21:52.334 --> 0:21:53.974 isn't always so factually accurate. 0:21:54.640 --> 0:21:57.520 So looking ahead, do you think the fight against misinformation, 0:21:57.599 --> 0:22:01.198 disinformation will be getting easier or harder with AI and 0:22:01.199 --> 0:22:03.069 other technological advances? 0:22:03.160 --> 0:22:06.319 It's gonna change, but I don't think it's gonna get 0:22:06.319 --> 0:22:09.500 either easier or harder. So AI is having a huge, 0:22:09.599 --> 0:22:14.109 huge impact. So we're now getting far more sophisticated botnets, 0:22:14.560 --> 0:22:20.390 which are coordinating in far more human-like manners, um, and, um, 0:22:20.640 --> 0:22:24.040 and also even in some cases acting as sleepers. 0:22:24.359 --> 0:22:27.489 So in the past, you would be befriended by what 0:22:27.489 --> 0:22:29.129 who you thought was a human but was actually a 0:22:29.130 --> 0:22:32.369 bot and would immediately try to persuade you to do something, 0:22:32.489 --> 0:22:34.760 vote for this person or don't vote for that person. 0:22:35.050 --> 0:22:37.959 Now they can act like your friends for maybe 6 0:22:37.959 --> 0:22:40.409 months and then try and persuade you to do something. 0:22:40.489 --> 0:22:42.010 So it's getting pretty sophisticated. 0:22:42.290 --> 0:22:46.189 Um, sending out personalised messages is getting increasingly easy with 0:22:46.189 --> 0:22:49.739 these large language models. The game is changing very much, 0:22:49.989 --> 0:22:55.030 but also we're developing techniques using AI to spot these botnets, 0:22:55.189 --> 0:22:59.910 to spot this inauthentic activity, and also to, um, there's 0:22:59.910 --> 0:23:03.410 quite a lot of work done by, um, um, Costello 0:23:03.410 --> 0:23:07.959