WEBVTT - Sumer series: How bots are distorting the truth 0:00:00.520 --> 0:00:04.080 Already, and this is the daily This is the Daily 0:00:04.160 --> 0:00:06.840 os oh. Now it makes sense. 0:00:14.840 --> 0:00:16.920 Good morning and welcome to the Daly Ohs. 0:00:17.200 --> 0:00:19.920 I'm Zara and over the next little while, we're going 0:00:19.960 --> 0:00:23.079 to be bringing you a bonus series featuring our favorite 0:00:23.120 --> 0:00:26.079 deep dives from twenty twenty four. We've put together the 0:00:26.120 --> 0:00:28.800 best deep dives to listen to on the beach, road, 0:00:28.800 --> 0:00:30.600 tripping when you don't want to talk to the person 0:00:30.680 --> 0:00:33.280 next to you, or just reflecting on the year that was. 0:00:33.720 --> 0:00:40.559 Welcome to tda's summer series. Earlier this year, we did 0:00:40.560 --> 0:00:43.519 a deep dive into the new digital player helping to 0:00:43.560 --> 0:00:48.000 shape public discourse online. Now the TLDR is the bots 0:00:48.040 --> 0:00:51.760 are targeting your social media algorithms and the content that 0:00:51.800 --> 0:00:55.520 you get served online. In today's deep dive, Chloe explores 0:00:55.560 --> 0:00:59.560 how these digital armies are being mobilized to influence public opinion. 0:01:00.080 --> 0:01:03.680 It's a fascinating element of our online world. So without 0:01:03.680 --> 0:01:07.240 further ado, let's get into the episode. Now, Chloe, you've 0:01:07.319 --> 0:01:08.920 become something of a bot. 0:01:08.720 --> 0:01:11.160 Expert in the last couple of weeks. 0:01:11.280 --> 0:01:13.680 But for anyone that hasn't spent as much time in 0:01:13.720 --> 0:01:16.479 the weeds of bot mania as you have, Yes, can 0:01:16.520 --> 0:01:19.319 you just explain what exactly is a bot? 0:01:19.600 --> 0:01:22.119 I have been in the weeds of the botmania. It's 0:01:22.280 --> 0:01:25.080 quite a place to be, and it can sound really 0:01:25.120 --> 0:01:28.360 techy when we talk about this stuff, but it's pretty 0:01:28.440 --> 0:01:30.560 much every time you jump on social media and you 0:01:30.600 --> 0:01:34.400 see somewhat of a spammy comment on a popular Instagram account. 0:01:34.760 --> 0:01:36.800 Just look at the Daily Yours's account, you'll see plenty 0:01:36.800 --> 0:01:37.120 of them. 0:01:37.280 --> 0:01:39.240 Excellent plug, excellent plug. 0:01:39.840 --> 0:01:42.520 You'll see them most often in the comment section within 0:01:42.560 --> 0:01:44.319 seconds of making a post on Instagram. 0:01:44.560 --> 0:01:46.600 Yeah, and I feel like, you know, even if you 0:01:46.640 --> 0:01:49.120 just look at our Instagram, you're seeing things like how 0:01:49.160 --> 0:01:51.320 coach Sarah changed my life. They changed the name of 0:01:51.360 --> 0:01:53.160 Sarah to multiple different women. 0:01:53.240 --> 0:01:54.000 But there you go. 0:01:54.440 --> 0:01:57.320 There's some like semi nude women that have come up 0:01:57.360 --> 0:01:59.120 in gifts in our comment section. 0:01:59.360 --> 0:02:01.800 Click on my io you'll get rich fast. 0:02:02.160 --> 0:02:04.559 So those sorts of comments you're saying, those are coming 0:02:04.560 --> 0:02:05.800 from bot accounts, right. 0:02:05.800 --> 0:02:08.000 Yeap, So that is a bot account. But when we 0:02:08.040 --> 0:02:12.519 look at bots more broadly, bots are programmed to perform automated, 0:02:12.800 --> 0:02:16.880 repetitive tasks over a network. Okay, so they're deliberately designed 0:02:16.919 --> 0:02:20.000 to mimic human behavior. They look like humans, they sound 0:02:20.080 --> 0:02:22.960 like real users. Only bots can generate content at a 0:02:23.040 --> 0:02:26.640 speed and a scale that we humans simply couldn't. So 0:02:26.680 --> 0:02:30.440 that's thousands upon thousands of comments a day. There are 0:02:30.440 --> 0:02:34.840 some helpful bots like chat gpt is, a chatbot, Siri Alexa, 0:02:35.040 --> 0:02:37.160 but others, as we know, are less helpful. 0:02:37.800 --> 0:02:41.560 They are less helpful, but they are absolutely everywhere. Like 0:02:41.639 --> 0:02:44.919 it takes, you know, one second of looking, as you said, 0:02:44.919 --> 0:02:47.200 at a popular account and you just get flooded with 0:02:47.280 --> 0:02:50.560 this stuff. How significant I mean, I'm just using anecdotal 0:02:50.639 --> 0:02:53.560 evidence here, So how significant are the presence of these 0:02:53.600 --> 0:02:54.840 bots in the online world. 0:02:55.280 --> 0:02:57.280 Yeah, it's wild. And to give you a sense of 0:02:57.320 --> 0:03:01.400 the scale of things, nearly half of all online traffic 0:03:01.480 --> 0:03:04.760 in twenty twenty three came from fake users. 0:03:05.000 --> 0:03:06.040 That is crazy. 0:03:06.120 --> 0:03:08.640 So bots, yeah, now, that's according to a new report 0:03:08.680 --> 0:03:12.680 by IT security firm Imperva, which also found that bad 0:03:12.760 --> 0:03:16.680 bots programmed to defraud and scam users accounted for nearly 0:03:16.720 --> 0:03:18.640 one third of all of that traffic. 0:03:18.919 --> 0:03:19.239 Wow. 0:03:19.600 --> 0:03:21.760 I mean, I knew they were everywhere, but that's really 0:03:22.000 --> 0:03:22.760 there everywhere. 0:03:22.800 --> 0:03:23.720 They're everywhere. 0:03:24.000 --> 0:03:26.320 And so when we're thinking about bots, you know, I 0:03:26.320 --> 0:03:28.440 think that anyone that has spent time on the internet 0:03:28.520 --> 0:03:31.919 knows not to click on you know, come earn money 0:03:31.919 --> 0:03:34.280 with me with your crypto dollars, you know in our 0:03:34.280 --> 0:03:37.600 comment section. But what are some of the other ways 0:03:37.640 --> 0:03:39.520 the bots show up or the other ways that they 0:03:39.520 --> 0:03:43.000 can I guess influence behaviors or conversations. 0:03:43.280 --> 0:03:46.120 Well, that's a whole different podcast about how people are 0:03:46.120 --> 0:03:49.080 scammed and frauded into clicking on things that they shouldn't. 0:03:49.520 --> 0:03:52.000 But what I was really interested in was the way 0:03:52.120 --> 0:03:55.120 bots can be used to make an opinion seem like 0:03:55.160 --> 0:03:57.440 a fact, or to make it feel and zeem as 0:03:57.440 --> 0:04:01.080 though it has either widespread support or widespace opposition for something. 0:04:01.120 --> 0:04:04.400 What does that do for real users online? Because if 0:04:04.400 --> 0:04:07.320 you can gear thousands upon thousands of accounts to push 0:04:07.360 --> 0:04:10.120 a certain narrative, that can be really dangerous if we 0:04:10.160 --> 0:04:14.680 think about elections or democracy or just public discourse at large. 0:04:14.960 --> 0:04:18.279 It's a really interesting topic, but it does still seem 0:04:18.480 --> 0:04:20.159 quite up there, and I would love to bring it 0:04:20.200 --> 0:04:22.960 down to down here. Is there an example that you 0:04:23.000 --> 0:04:25.560 can just provide, I guess to give some orientation as 0:04:25.560 --> 0:04:26.920 to what you're actually talking about here. 0:04:27.240 --> 0:04:30.120 So let's just say you're scrolling on X and you 0:04:30.200 --> 0:04:33.960 read a bunch of different tweets, retweets shares comments of 0:04:33.960 --> 0:04:37.440 those tweets saying apples are really really bad for you 0:04:37.920 --> 0:04:41.559 and instead you should be buying oranges. This is really basic, 0:04:41.600 --> 0:04:44.159 but just still wrapping heads around it. You're probably going 0:04:44.240 --> 0:04:46.840 to start feeling a little bit suss about apples, even 0:04:46.880 --> 0:04:49.159 though you might really like them. You just sort of 0:04:49.160 --> 0:04:51.279 have a bit of questions about that. You might even 0:04:51.360 --> 0:04:55.560 consider buying more oranges. And bots can create these spaces 0:04:55.560 --> 0:04:58.359 that feel like communities are sharing ideas that it's just 0:04:58.440 --> 0:05:02.080 normal people talking about how we all don't really like apples, 0:05:02.240 --> 0:05:04.719 but there is an agenda at play. Now when you 0:05:04.760 --> 0:05:06.920 consider the influence that, as I said, this could have 0:05:07.000 --> 0:05:09.480 on politics or elections. 0:05:08.839 --> 0:05:10.560 Not just apples and oranges, Not just. 0:05:10.600 --> 0:05:12.919 Apples and oranges, it can be really scary. 0:05:13.520 --> 0:05:15.880 What are the experts say about bots? 0:05:15.920 --> 0:05:18.640 Because you know, they are a fairly new phenomenon, but 0:05:18.880 --> 0:05:20.880 there must be a body of research out there. 0:05:21.279 --> 0:05:23.119 I reached out to a bot doc. 0:05:23.520 --> 0:05:26.239 Of course, a bot doc has a dock of bot. 0:05:26.400 --> 0:05:28.239 She does have a real name. Her name is doctor 0:05:28.279 --> 0:05:32.880 Sophia Mellinson Richard done from Canada's McMaster University. Now She 0:05:33.040 --> 0:05:37.880 did a PhD on Bota gander when bot armies generate 0:05:38.120 --> 0:05:42.320 mass content to saturate social media feeds and then manipulate audiences. 0:05:42.920 --> 0:05:46.720 So doctor Richardoni told me about something called hashtag flooding, 0:05:47.080 --> 0:05:51.520 which is essentially a tweet containing nothing but popularized keywords 0:05:51.560 --> 0:05:54.360 and catchphrases in the form of hashtag. 0:05:53.839 --> 0:05:56.880 After hashtag up to hashtag so annoying, so annoying. 0:05:57.200 --> 0:06:01.039 Now, this tactic, as well as the rapid resharing of 0:06:01.120 --> 0:06:05.520 human posts so twenty five thousand in ten minutes, can create, 0:06:05.680 --> 0:06:08.359 like we were saying before, this illusion of widespread support 0:06:08.839 --> 0:06:13.000 or widespread opposition for specific viewpoints. She says that this 0:06:13.040 --> 0:06:16.080 happens as though the idea embedded in the tweed came 0:06:16.120 --> 0:06:20.839 from grassroots popularization. So when we're talking about mimicking human behavior, 0:06:20.880 --> 0:06:23.320 it's this feeling of oh, everyone thinks. 0:06:23.080 --> 0:06:25.400 This, and you want to be a part of the everyone. 0:06:25.520 --> 0:06:27.600 Like that's the human condition, right exactly. 0:06:27.680 --> 0:06:30.240 It's the concept of herd mentality, which is the idea 0:06:30.279 --> 0:06:34.280 that individuals naturally want to conform to the dominant view 0:06:34.400 --> 0:06:35.120 of the community. 0:06:35.480 --> 0:06:38.880 Yeah, I mean, it's so fascinating, and the reason that 0:06:38.960 --> 0:06:41.920 you came to I don't want to say, become obsessed 0:06:41.920 --> 0:06:45.160 with this topic, but I mean had started your fascination 0:06:45.640 --> 0:06:49.359 was around one example that we've seen in the last 0:06:49.360 --> 0:06:50.159 couple of years. 0:06:50.600 --> 0:06:53.000 Talk me through the Johnny Depp Amber Herd story. 0:06:53.600 --> 0:06:56.799 I don't know if many people remember what the internet 0:06:56.839 --> 0:06:59.560 looked like in twenty twenty two during the Johnny Depp 0:06:59.600 --> 0:07:02.520 amber her defamation trial, but I recently listened to a 0:07:02.560 --> 0:07:06.200 six part podcast that brought me right back there. So 0:07:06.400 --> 0:07:09.080 for people who need a refresher on the story. Essentially, 0:07:09.160 --> 0:07:12.400 Johnny Depp filed a lawsuit against his ex wife Amber 0:07:12.440 --> 0:07:15.840 heard over an opinion article Herd wrote for The Washington Post. 0:07:16.600 --> 0:07:20.560 She alleged she had experienced domestic abuse. Her didn't name Debt, 0:07:20.640 --> 0:07:24.120 but he launched defamation proceedings against her, arguing he was 0:07:24.240 --> 0:07:27.760 identifiable from the article and depth denies claims that he 0:07:27.800 --> 0:07:32.239 physically abused her. Now, a jury in Fairfax County, Virginia 0:07:32.400 --> 0:07:35.960 ultimately sided with Depp, and Herd was sued for defaming him. 0:07:36.000 --> 0:07:38.200 But I think you might remember that the court proceedings 0:07:38.240 --> 0:07:39.640 were streamed online. 0:07:39.760 --> 0:07:42.280 Yeah, so I was going to say, can you, in 0:07:42.320 --> 0:07:44.800 the minds of our listeners connect what we've just been 0:07:44.840 --> 0:07:46.800 talking about which is you know all of this stuff 0:07:46.840 --> 0:07:49.880 about bots and this case, what's the connection here? 0:07:50.360 --> 0:07:53.320 So as the case played out in the courtroom, the 0:07:53.440 --> 0:07:58.120 internet mounted its own unofficial trial of Amber Heard. This 0:07:58.200 --> 0:08:00.400 is the premise of the podcast I was talking about earlier, 0:08:00.440 --> 0:08:04.720 Who Trolled Amber? It's by UK media organization Tortoise. The 0:08:04.800 --> 0:08:07.800 podcast found that a large part of the online hate 0:08:07.840 --> 0:08:13.520 campaign against her was actually manufactured and executed by bot accounts. 0:08:13.720 --> 0:08:16.520 Now I spoke to Xavia Greenwood, he produced the Who 0:08:16.560 --> 0:08:17.920 Trolled Amber podcast. 0:08:18.080 --> 0:08:21.200 Some of the main themes were that Amber Heerd deserves prison, 0:08:21.280 --> 0:08:24.240 Amber Heard as a gold digger, Amber Heard as a fraud, 0:08:24.320 --> 0:08:26.520 Amber Heard as a liar. To give you a sense 0:08:26.520 --> 0:08:29.560 of the scale of things, there was a hashtag justice 0:08:29.560 --> 0:08:33.080 for Johnny Depp that was viewed fifteen point seven billion 0:08:33.160 --> 0:08:36.520 times hashtag just as Amber Heerd viewed a fraction of that, 0:08:36.960 --> 0:08:40.240 and my friends were suddenly saying things that didn't really 0:08:40.280 --> 0:08:42.640 sound like them. They were saying, well, you know what 0:08:42.760 --> 0:08:45.040 if this time she was the abuser. And these were 0:08:45.040 --> 0:08:47.520 people who typically in the Me too movement would be 0:08:47.520 --> 0:08:50.120 a bit more cynical or maybe a bit more reserved 0:08:50.120 --> 0:08:53.680 in making a judgment like that. So yeah, from the beginning, 0:08:54.120 --> 0:08:56.720 we sort of saw that this was quite suspicious. 0:08:57.000 --> 0:08:59.200 What else did this investigation uncover? 0:08:59.840 --> 0:09:04.000 The team Tortoise brought together a database of over one 0:09:04.120 --> 0:09:07.439 million anti amber Heard tweets, and they brought it together 0:09:07.520 --> 0:09:11.200 and they found that more than half of them were inauthentic. 0:09:11.760 --> 0:09:14.000 So what that means is either they were posted from 0:09:14.080 --> 0:09:17.200 spam accounts with three followers and they were built in 0:09:17.240 --> 0:09:20.640 the last two months, or they were amplified in an 0:09:20.679 --> 0:09:25.200 inauthentic way so reshared and reshared thousands upon thousands of times. 0:09:25.840 --> 0:09:29.320 Okay, I think the thing that I wonder when I 0:09:29.360 --> 0:09:32.400 hear about this story is like, who is behind this 0:09:32.480 --> 0:09:35.200 trolling campaign? It's very clear who benefits from it, but 0:09:35.280 --> 0:09:36.520 who's behind it. 0:09:36.760 --> 0:09:39.800 We don't know. According to the investigation, the team could 0:09:39.920 --> 0:09:44.120 hypothesize multiple different scenarios and their likely were multiple different 0:09:44.280 --> 0:09:45.120 agendas at play. 0:09:45.200 --> 0:09:47.359 Yeah, it's crazy, So this could. 0:09:47.080 --> 0:09:50.840 Be genuine Johnny Depp fans who can acquire bots for 0:09:50.840 --> 0:09:54.360 pretty cheap online and push support for their favorite actor 0:09:54.760 --> 0:09:57.720 in other scenarios. Xavier floats what he calls a more 0:09:57.760 --> 0:09:58.680 abstract theory. 0:09:59.040 --> 0:10:01.760 I think it's sort of fairly widely known now that 0:10:01.960 --> 0:10:06.440 authoritarian regimes who want to sow discouse or want to 0:10:06.440 --> 0:10:10.480 cause confusion in the West that they seek out wedge issues. 0:10:10.520 --> 0:10:14.720 So they seek out issues which divide Britain's, divide Americans, 0:10:14.760 --> 0:10:18.160 divide Australians. They try to sort of drive a deeper wedge. 0:10:18.440 --> 0:10:21.160 But something that really stuck with me from my conversation 0:10:21.240 --> 0:10:23.959 with Xavier was a point he made about the future 0:10:23.960 --> 0:10:26.600 of political discourse on social media. 0:10:26.679 --> 0:10:30.880 If you can attack a celebrity who has enormous an 0:10:31.000 --> 0:10:34.400 enormous amount of resources, as Amberhard did, was to stop 0:10:34.600 --> 0:10:37.480 someone doing the same thing when it comes to attacking 0:10:37.520 --> 0:10:40.120 a politician or when it comes to trying to sway 0:10:40.240 --> 0:10:40.800 an election. 0:10:41.440 --> 0:10:44.720 I mean, we're talking there about elections, and we of 0:10:44.760 --> 0:10:47.160 course know that there is one coming up just around 0:10:47.160 --> 0:10:50.280 the corner in the US. Are we expecting to see 0:10:50.320 --> 0:10:53.200 this sort of interference from bots in the US election 0:10:53.679 --> 0:10:54.160 this year? 0:10:54.600 --> 0:10:59.400 We already have. Last month, the Biden administration charged Russian 0:10:59.600 --> 0:11:03.960 media executives over an alleged targeted online campaign to influence 0:11:04.200 --> 0:11:08.240 voters in the US and push hidden Russian government messaging. 0:11:08.559 --> 0:11:12.120 Now it comes after US officials seized thirty two Internet 0:11:12.120 --> 0:11:16.760 domain names that were covertly targeting specific demographics on social 0:11:16.760 --> 0:11:20.360 media and promoting AI generated false narratives and pushing that 0:11:20.559 --> 0:11:21.439 to those groups. 0:11:21.800 --> 0:11:24.640 Okay, so there are ready allegations that we've seen this 0:11:24.880 --> 0:11:27.600 at play when it comes to the US election. Ye, 0:11:28.240 --> 0:11:31.280 I mean, I think that a natural endpoint here is 0:11:31.320 --> 0:11:33.120 that if we're talking about the fact that even the 0:11:33.120 --> 0:11:37.160 most powerful institutions, people, whatever, in the world can be 0:11:37.200 --> 0:11:40.480 susceptible to this kind of bodagander, as we'll call it, 0:11:41.240 --> 0:11:43.200 what is like an average dough like you or I 0:11:43.280 --> 0:11:46.360 meant to do when it comes to protecting yourself against something. 0:11:46.200 --> 0:11:48.160 Like this, Well, that was what came back in so 0:11:48.200 --> 0:11:50.640 many of the conversations I was having. For the average 0:11:50.679 --> 0:11:52.520 dough like you and me, A lot of it comes 0:11:52.559 --> 0:11:56.839 down to awareness, staying vigilant, staying curious, question what you're 0:11:56.840 --> 0:11:59.559 getting served, who's pushing it. And that's generally for anything 0:11:59.600 --> 0:12:02.559 that you've seen online, but particularly when you're thinking something 0:12:02.600 --> 0:12:06.440 looks a bit spammy, it likely might be it's important 0:12:06.480 --> 0:12:08.440 to be critical about the content you're seeing before you 0:12:08.640 --> 0:12:12.440 form an opinion, especially ahead of an election. Now, as 0:12:12.480 --> 0:12:15.640 for what comes next, Xavier Greenwood says, we're largely in 0:12:15.800 --> 0:12:16.840 uncharted waters. 0:12:17.400 --> 0:12:19.800 What we saw with Amber heard trial, we may well 0:12:20.040 --> 0:12:22.400 continue to see in an even more intense way in 0:12:22.440 --> 0:12:24.640 the future. To some extent, the genie is out at 0:12:24.640 --> 0:12:25.040 the box. 0:12:25.920 --> 0:12:29.840 Thanks for listening to today's episode of tda's summer series. 0:12:30.120 --> 0:12:32.640 We'll be back again tomorrow with another of our favorite 0:12:32.679 --> 0:12:39.680 deep dives, but until then, have a wonderfully warm day. 0:12:39.960 --> 0:12:42.280 My name is Lily Maddon and I'm a proud Arunda 0:12:42.520 --> 0:12:47.280 Bungelung Caalcuttin woman from Gadighl Country. The Daily oz acknowledges 0:12:47.360 --> 0:12:49.560 that this podcast is recorded on the lands of the 0:12:49.559 --> 0:12:53.120 Gadighl people and pays respect to all Aboriginal and Torres 0:12:53.120 --> 0:12:56.040 Strait Island and nations. We pay our respects to the 0:12:56.080 --> 0:12:58.840 first peoples of these countries, both past and present.