1 00:00:00,120 --> 00:00:03,239 Speaker 1: When you look at teen surveys, two thirds of kids 2 00:00:03,279 --> 00:00:06,440 Speaker 1: say that they have a joyful experience on social media. 3 00:00:06,600 --> 00:00:09,560 Speaker 1: They like using the products because they give them a 4 00:00:09,600 --> 00:00:12,520 Speaker 1: sense of connection among other teens and they're sending them 5 00:00:12,520 --> 00:00:15,040 Speaker 1: content that they want to watch. But on the flip 6 00:00:15,040 --> 00:00:17,640 Speaker 1: side of that, you also have more than fifty percent 7 00:00:17,760 --> 00:00:20,880 Speaker 1: of teenagers who engage with social media saying that they 8 00:00:20,920 --> 00:00:24,040 Speaker 1: feel overwhelmed by all of the drama on these apps, 9 00:00:24,600 --> 00:00:27,000 Speaker 1: and more than a quarter of these kids saying that 10 00:00:27,280 --> 00:00:33,680 Speaker 1: they feel worse about themselves after using social media. 11 00:00:34,800 --> 00:00:40,599 Speaker 2: From Bloomberg News and iHeartRadio, it's the big take. I'm 12 00:00:40,640 --> 00:00:46,960 Speaker 2: Weiscasova today the downside of the algorithm that powers TikTok's popularity. 13 00:00:54,880 --> 00:00:58,560 Speaker 2: Bloomberg investigative correspondent Olivia Carvel has been on the show 14 00:00:58,560 --> 00:01:02,600 Speaker 2: twice now to report about the enormous influence of TikTok, 15 00:01:03,120 --> 00:01:06,920 Speaker 2: especially on young people. She's back today to tell us 16 00:01:06,959 --> 00:01:10,400 Speaker 2: about something the company doesn't talk much about, the secret 17 00:01:10,520 --> 00:01:15,200 Speaker 2: algorithm that keeps many of TikTok's billion plus users glued 18 00:01:15,240 --> 00:01:18,559 Speaker 2: to the app for hours every day. And just a note, 19 00:01:18,600 --> 00:01:22,760 Speaker 2: We'll be talking about some difficult subjects here, including suicide 20 00:01:23,160 --> 00:01:25,520 Speaker 2: and some of it is not easy to listen to. 21 00:01:25,680 --> 00:01:28,840 Speaker 2: If there are kids nearby, you might want to use headphones. 22 00:01:30,959 --> 00:01:33,319 Speaker 2: Can we start just by really defining what is an 23 00:01:33,360 --> 00:01:35,919 Speaker 2: algorithm and how does it work in this context? 24 00:01:36,120 --> 00:01:39,760 Speaker 1: An algorithm is a mathematical equation that instructs a computer 25 00:01:39,840 --> 00:01:42,560 Speaker 1: program how to behave. It is not good, it is 26 00:01:42,600 --> 00:01:46,120 Speaker 1: not bad. It's not benivolent, it's not malevolent. It is 27 00:01:46,319 --> 00:01:49,400 Speaker 1: just a mathematical equation. And when you think about an 28 00:01:49,440 --> 00:01:53,440 Speaker 1: algorithm that is powering a social media product like TikTok, 29 00:01:54,040 --> 00:01:57,080 Speaker 1: its main goal is to keep a user engaged with 30 00:01:57,160 --> 00:02:00,240 Speaker 1: the product as long as possible, and the algorithm is 31 00:02:00,240 --> 00:02:01,280 Speaker 1: trained to do just that. 32 00:02:01,840 --> 00:02:04,680 Speaker 2: And so if we're talking about TikTok, which is videos, 33 00:02:05,480 --> 00:02:08,760 Speaker 2: it's trying to show you videos that people have made 34 00:02:08,919 --> 00:02:11,240 Speaker 2: that you're going to be interested in and watch it 35 00:02:11,560 --> 00:02:13,359 Speaker 2: and then watch the next one, and the next one 36 00:02:13,400 --> 00:02:14,120 Speaker 2: and the next one. 37 00:02:14,320 --> 00:02:14,720 Speaker 3: That's right. 38 00:02:14,760 --> 00:02:18,400 Speaker 1: The algorithm is learning from you as you use the product. 39 00:02:18,840 --> 00:02:22,760 Speaker 1: So every movement your finger makes on that app, every like, 40 00:02:22,919 --> 00:02:27,799 Speaker 1: every rewatch, every follow, every comment, every time you make 41 00:02:27,880 --> 00:02:31,960 Speaker 1: an action on TikTok, the algorithm is tracking you to 42 00:02:32,120 --> 00:02:35,640 Speaker 1: then deliver you a steady stream of content that it 43 00:02:35,720 --> 00:02:38,440 Speaker 1: now knows you want to see. And a number of 44 00:02:38,600 --> 00:02:41,680 Speaker 1: users of TikTok actually believe the algorithm can read their 45 00:02:41,760 --> 00:02:45,040 Speaker 1: minds because they feel like the videos they are sent 46 00:02:45,120 --> 00:02:47,680 Speaker 1: through TikTok are exactly what they want to see. But 47 00:02:47,720 --> 00:02:51,079 Speaker 1: they don't understand how TikTok does this and what's powering 48 00:02:51,160 --> 00:02:52,560 Speaker 1: that is the algorithm. 49 00:02:52,880 --> 00:02:55,040 Speaker 2: And I think we've all had that experience on social 50 00:02:55,040 --> 00:02:57,680 Speaker 2: media where you see something that appears in your feed 51 00:02:57,840 --> 00:03:01,279 Speaker 2: and it's not anything that you're in interested in necessarily 52 00:03:01,360 --> 00:03:03,880 Speaker 2: or anything you asked for, but you do wind up 53 00:03:03,919 --> 00:03:06,720 Speaker 2: clicking on it and you do find it interesting, and 54 00:03:06,720 --> 00:03:09,600 Speaker 2: then suddenly there's a lot more of that kind of 55 00:03:09,639 --> 00:03:12,160 Speaker 2: content sitting there waiting for you the next time you 56 00:03:12,280 --> 00:03:12,600 Speaker 2: use it. 57 00:03:12,960 --> 00:03:16,520 Speaker 1: That's right. Yeah, we can relay this to targeted advertising 58 00:03:16,639 --> 00:03:20,280 Speaker 1: when maybe you've been looking to buy a workblazer online 59 00:03:20,320 --> 00:03:22,480 Speaker 1: and then you open up Instagram and it's right there. 60 00:03:22,600 --> 00:03:25,480 Speaker 1: A number of women have talked about thinking about egg 61 00:03:25,520 --> 00:03:28,880 Speaker 1: freezing or getting married and then seeing very targeted advertising 62 00:03:28,960 --> 00:03:32,240 Speaker 1: relating to those things popping up in their feeds, and 63 00:03:32,400 --> 00:03:37,320 Speaker 1: algorithms are actually really effective and very useful in some cases. 64 00:03:37,920 --> 00:03:41,240 Speaker 1: You know, in this infinite world of information. They're delivering 65 00:03:41,240 --> 00:03:44,680 Speaker 1: you content that you want to see. If algorithms didn't exist, 66 00:03:44,720 --> 00:03:48,240 Speaker 1: we'd just be wading through endless forms of content that 67 00:03:48,240 --> 00:03:51,200 Speaker 1: are largely irrelevant to us. So they are useful in 68 00:03:51,200 --> 00:03:54,200 Speaker 1: that respect, but on the flip side of that, they 69 00:03:54,200 --> 00:03:55,440 Speaker 1: can also be dangerous. 70 00:03:56,280 --> 00:04:00,840 Speaker 2: And you write that TikTok's algorithm, among social media companies 71 00:04:00,840 --> 00:04:04,040 Speaker 2: in particular, is really really effective. 72 00:04:04,840 --> 00:04:09,480 Speaker 1: Yeah, TikTok's algorithm is so successful that we have seen 73 00:04:09,600 --> 00:04:13,120 Speaker 1: other social media platforms in Silicon Valley try to copy it. 74 00:04:13,800 --> 00:04:18,200 Speaker 1: TikTok's algorithm is, you know, sending users the vast majority 75 00:04:18,320 --> 00:04:20,680 Speaker 1: of what they see on the app, and it has 76 00:04:20,680 --> 00:04:23,960 Speaker 1: become so good at what it does that people are 77 00:04:24,040 --> 00:04:26,919 Speaker 1: trying to understand it. They want to know what drives it, 78 00:04:27,000 --> 00:04:30,680 Speaker 1: what power is it, and also how does it influence 79 00:04:30,800 --> 00:04:34,080 Speaker 1: what a user sees. At the heart of the algorithm, 80 00:04:34,480 --> 00:04:38,200 Speaker 1: what makes the platform so addictive is the for you page. 81 00:04:38,480 --> 00:04:40,919 Speaker 1: As a user on TikTok, when you open up the app, 82 00:04:41,120 --> 00:04:44,200 Speaker 1: the first thing you see is this for you page, 83 00:04:44,240 --> 00:04:47,080 Speaker 1: which is delivering a steady stream of content that you 84 00:04:47,120 --> 00:04:50,320 Speaker 1: don't want to look away from because it's so hyper personalized. 85 00:04:50,600 --> 00:04:51,600 Speaker 1: To what you want to. 86 00:04:51,560 --> 00:04:56,000 Speaker 2: See, and it's really what the algorithm has decided what 87 00:04:56,040 --> 00:04:56,599 Speaker 2: you want to see. 88 00:04:56,640 --> 00:04:59,240 Speaker 1: Is that right, That's a great catch. It's not what 89 00:04:59,320 --> 00:05:01,919 Speaker 1: you want to see, it is what the algorithm thinks 90 00:05:02,120 --> 00:05:05,000 Speaker 1: you want to see. Social media apps have really become 91 00:05:05,080 --> 00:05:09,360 Speaker 1: ubiquitous in our lives now. Everyone uses them, particularly kids. 92 00:05:09,440 --> 00:05:13,080 Speaker 1: They have a lot of positives, like they enable connection 93 00:05:13,279 --> 00:05:17,960 Speaker 1: among teens, and particularly through the pandemic, was really crucial. 94 00:05:18,000 --> 00:05:21,000 Speaker 1: It allowed teens to connect with one another, to talk 95 00:05:21,000 --> 00:05:24,400 Speaker 1: among their friendship groups. But with all social media products, 96 00:05:24,800 --> 00:05:28,080 Speaker 1: they have different pros and coms. When you look at 97 00:05:28,320 --> 00:05:31,440 Speaker 1: Facebook or Instagram, a lot of what you see is 98 00:05:31,520 --> 00:05:34,320 Speaker 1: driven by who you follow, It's driven by your friendship groups. 99 00:05:34,800 --> 00:05:38,640 Speaker 1: What distinguishes TikTok from other social media products is that 100 00:05:38,680 --> 00:05:41,880 Speaker 1: it's for you. Feed is really a steady stream of 101 00:05:41,960 --> 00:05:46,160 Speaker 1: content sent in from random users around the world. It 102 00:05:46,200 --> 00:05:49,840 Speaker 1: is not driven by who you follow, who you're friends with. 103 00:05:50,520 --> 00:05:53,799 Speaker 1: It is driven by what the algorithm thinks you want 104 00:05:53,839 --> 00:05:55,920 Speaker 1: to see. So what the algorithm can send you to 105 00:05:56,000 --> 00:05:58,440 Speaker 1: keep you watching for even longer. So it's a totally 106 00:05:58,520 --> 00:06:03,080 Speaker 1: different business than the social media apps before it, and 107 00:06:03,080 --> 00:06:05,599 Speaker 1: that's where it's really interesting when you're seeing Facebook and 108 00:06:05,680 --> 00:06:08,719 Speaker 1: Instagram and Snapchat start to try and tweak their own 109 00:06:08,720 --> 00:06:11,680 Speaker 1: business models to reflect more of that of TikTok, where 110 00:06:11,680 --> 00:06:15,160 Speaker 1: they're sending random content to users rather than sending them 111 00:06:15,200 --> 00:06:18,440 Speaker 1: content from people that they already follow or are already 112 00:06:18,440 --> 00:06:22,280 Speaker 1: friends with. When you look at TikTok's algorithm, it does 113 00:06:22,360 --> 00:06:25,239 Speaker 1: a lot of great things for users, feeding them content 114 00:06:25,320 --> 00:06:27,120 Speaker 1: they do want to see, that they do want to 115 00:06:27,160 --> 00:06:31,320 Speaker 1: be engaging with. But it can also send content to 116 00:06:31,400 --> 00:06:35,159 Speaker 1: teenagers that can target their vulnerabilities and in some cases 117 00:06:35,279 --> 00:06:41,119 Speaker 1: send repetitive videos about very distressing, dangerous, or harmful types 118 00:06:41,160 --> 00:06:44,320 Speaker 1: of issues. And in the reporting in this story, I've 119 00:06:44,360 --> 00:06:47,039 Speaker 1: talked to a number of young people and families of 120 00:06:47,080 --> 00:06:50,680 Speaker 1: young people who are aware that the TikTok algorithm have 121 00:06:50,839 --> 00:06:54,440 Speaker 1: been sending them never ending streams of content about anorexia, 122 00:06:55,080 --> 00:07:00,120 Speaker 1: are the eating disordered related content, depression, self harm, and suicide? 123 00:07:00,360 --> 00:07:04,640 Speaker 2: How did they describe the content the algorithm delivered to 124 00:07:04,760 --> 00:07:06,720 Speaker 2: their for you page? 125 00:07:06,760 --> 00:07:08,480 Speaker 1: A number of the young people that I talked to 126 00:07:08,600 --> 00:07:11,600 Speaker 1: said they started using the platform and they loved TikTok. 127 00:07:11,720 --> 00:07:15,480 Speaker 1: They loved the silly, funny dancing videos, they loved the 128 00:07:15,480 --> 00:07:17,880 Speaker 1: content they were seeing, and they liked engaging with it 129 00:07:17,920 --> 00:07:20,400 Speaker 1: and trying to make their own content. But then they 130 00:07:20,480 --> 00:07:23,200 Speaker 1: noticed that what their full you feed or what the 131 00:07:23,240 --> 00:07:28,360 Speaker 1: algorithm was sending them started to change. It switched from funny, 132 00:07:28,400 --> 00:07:32,320 Speaker 1: silly dancing videos to much darker topics. You know, a 133 00:07:32,320 --> 00:07:34,280 Speaker 1: lot of the young people who were watching this kind 134 00:07:34,320 --> 00:07:37,880 Speaker 1: of content pointed out that it wasn't just videos to 135 00:07:38,040 --> 00:07:41,480 Speaker 1: raise awareness of eating disorders and mental health issues. It 136 00:07:41,560 --> 00:07:45,400 Speaker 1: was videos that actually glorified eating disorders, that embraced it, 137 00:07:45,480 --> 00:07:49,160 Speaker 1: that encouraged it, that told these young girls to eat 138 00:07:49,240 --> 00:07:52,120 Speaker 1: less than three hundred calories a day, or try this 139 00:07:52,280 --> 00:07:56,560 Speaker 1: exercise to make your legs look skinnier. It promoted videos 140 00:07:56,560 --> 00:08:00,440 Speaker 1: and images of other young women with eating disorder who 141 00:08:00,480 --> 00:08:04,400 Speaker 1: suffer from anorexia and encourage them to look the same way. 142 00:08:05,520 --> 00:08:09,720 Speaker 2: Olivia, your most recent story about TikTok focus is on 143 00:08:09,880 --> 00:08:13,520 Speaker 2: a teenager named Chase Naska. Can you tell us about him? 144 00:08:14,080 --> 00:08:18,000 Speaker 1: Chase was a sixteen year old boy who grew up 145 00:08:18,440 --> 00:08:21,800 Speaker 1: on Long Island in Bayport. He went to Bayport Blue 146 00:08:21,840 --> 00:08:25,680 Speaker 1: Point High School. He loved playing soccer. He was on 147 00:08:25,720 --> 00:08:28,920 Speaker 1: his high school soccer and swimming teams. He was a 148 00:08:29,000 --> 00:08:32,240 Speaker 1: very competitive athlete. He was an honors student, so he 149 00:08:32,360 --> 00:08:35,400 Speaker 1: was excelling academically at school as well. He had a 150 00:08:35,400 --> 00:08:39,160 Speaker 1: big group of friends in the Bayport area, and in 151 00:08:39,200 --> 00:08:43,680 Speaker 1: twenty twenty, when the COVID nineteen pandemic shut down his school, 152 00:08:44,040 --> 00:08:47,720 Speaker 1: Chase started to use social media more like most other 153 00:08:47,760 --> 00:08:51,400 Speaker 1: teens in America did. He would use TikTok in particular 154 00:08:51,480 --> 00:08:55,160 Speaker 1: because that's the most popular app among teenagers today, and 155 00:08:55,200 --> 00:08:58,920 Speaker 1: he would send his friends' videos and often they were funny, 156 00:08:59,160 --> 00:09:02,640 Speaker 1: kind of silly video about sports or weightlifting or batman 157 00:09:03,320 --> 00:09:07,200 Speaker 1: or basketball. And then towards the end of twenty twenty one, 158 00:09:07,400 --> 00:09:10,199 Speaker 1: a number of his friends noticed the content that Chase 159 00:09:10,280 --> 00:09:14,120 Speaker 1: was sending them through TikTok took a darker turn. It 160 00:09:14,240 --> 00:09:20,120 Speaker 1: was about suicide. It was posts glorifying suicide, talking about 161 00:09:20,240 --> 00:09:23,720 Speaker 1: not wanting to live anymore, it's not worth it anymore. 162 00:09:24,480 --> 00:09:26,760 Speaker 1: Some of these posts really concerned his friends and they 163 00:09:26,760 --> 00:09:30,040 Speaker 1: didn't understand what was going on. In February of twenty 164 00:09:30,240 --> 00:09:34,360 Speaker 1: twenty two, Chase took his own life, and afterwards his 165 00:09:34,640 --> 00:09:40,280 Speaker 1: parents just they didn't understand. They obviously felt blindsided, and 166 00:09:40,880 --> 00:09:43,960 Speaker 1: they started searching for an answer, looking for a clue. 167 00:09:44,400 --> 00:09:47,720 Speaker 1: He didn't leave a note, He didn't mention anything to 168 00:09:47,840 --> 00:09:50,240 Speaker 1: either of them. He was keeping up with his grades, 169 00:09:50,920 --> 00:09:55,840 Speaker 1: with his sports schedule, and his mum wanted access to 170 00:09:55,880 --> 00:09:57,840 Speaker 1: his cell phone to get a sense of what he 171 00:09:57,960 --> 00:10:00,400 Speaker 1: was doing online in the days leading up to his death. 172 00:10:00,760 --> 00:10:04,040 Speaker 1: So when the police returned his phone, she tried to 173 00:10:04,040 --> 00:10:07,040 Speaker 1: open it, but like all phones today, it had a 174 00:10:07,040 --> 00:10:08,920 Speaker 1: pass code and she couldn't break into it. So she 175 00:10:09,080 --> 00:10:12,240 Speaker 1: ended up figuring out a way to reset his passwords 176 00:10:12,240 --> 00:10:16,120 Speaker 1: on different social media apps so she could access them. 177 00:10:16,600 --> 00:10:19,680 Speaker 1: She ended up getting access to both Snapchat and TikTok. 178 00:10:20,920 --> 00:10:23,920 Speaker 1: Once she gained access to his TikTok app and she 179 00:10:24,040 --> 00:10:29,280 Speaker 1: opened his for you page, she was horrified at what 180 00:10:29,559 --> 00:10:33,920 Speaker 1: her son had been sent by TikTok's algorithm. She watched 181 00:10:33,960 --> 00:10:36,400 Speaker 1: the for you feed for more than an hour and 182 00:10:36,440 --> 00:10:39,760 Speaker 1: said there were no happy videos, There were no funny videos. 183 00:10:40,200 --> 00:10:45,120 Speaker 1: It was a non stop stream of content about heartbreak, 184 00:10:45,600 --> 00:10:52,480 Speaker 1: unrequited love, depression, anxiety, self harm, and suicide. His mother, 185 00:10:52,600 --> 00:10:56,720 Speaker 1: Michelle Naske, when she opened his TikTok account, she didn't 186 00:10:56,720 --> 00:10:58,760 Speaker 1: really know what she was looking for. She'd never used 187 00:10:58,760 --> 00:11:01,080 Speaker 1: the app before. She didn't underst stand how it worked. 188 00:11:01,400 --> 00:11:03,600 Speaker 1: She just wanted to be closer to Chase. She wanted 189 00:11:03,600 --> 00:11:05,160 Speaker 1: to get a sense of what he was doing on 190 00:11:05,200 --> 00:11:07,439 Speaker 1: these apps, what he was seeing, who he was talking to. 191 00:11:08,240 --> 00:11:11,839 Speaker 1: And she told me that once she opened TikTok, she 192 00:11:12,000 --> 00:11:14,120 Speaker 1: realized what she was looking for and she found it 193 00:11:14,200 --> 00:11:14,760 Speaker 1: right there. 194 00:11:15,600 --> 00:11:17,040 Speaker 3: Then I'm gonna put a shotgun in my mouth and 195 00:11:17,040 --> 00:11:20,720 Speaker 3: blow the brains at the back of my head. Cool, 196 00:11:24,520 --> 00:11:27,040 Speaker 3: you gotta kill yourself a word. 197 00:11:29,440 --> 00:11:37,160 Speaker 2: Like right now? If my dad would you listening everything 198 00:11:37,200 --> 00:11:41,280 Speaker 2: we just heard. There is content that Chase's mother found 199 00:11:41,400 --> 00:11:44,520 Speaker 2: on his TikTok account, and they were shown to the 200 00:11:44,640 --> 00:11:49,280 Speaker 2: public at a recent televised congressional hearing where TikTok's CEO 201 00:11:49,400 --> 00:11:52,680 Speaker 2: showed you testified and we're going to hear more about 202 00:11:52,679 --> 00:11:53,760 Speaker 2: that in a minute. 203 00:11:54,160 --> 00:11:57,160 Speaker 1: Not only was it the for you feed showing her 204 00:11:57,200 --> 00:12:01,000 Speaker 1: a steady stream of content about depression and suicide, she 205 00:12:01,080 --> 00:12:04,560 Speaker 1: actually managed to go deeper into his account to see 206 00:12:05,120 --> 00:12:10,080 Speaker 1: the videos that he had liked, saved, favorited, bookmarked. It 207 00:12:10,160 --> 00:12:14,000 Speaker 1: became a library of what Chase had seen in the 208 00:12:14,120 --> 00:12:16,640 Speaker 1: months leading up to his death. There were over three 209 00:12:16,760 --> 00:12:21,240 Speaker 1: thousand videos saved within this account, preserved within this account 210 00:12:21,320 --> 00:12:24,080 Speaker 1: almost like a digital fingerprint of Chase that he had 211 00:12:24,160 --> 00:12:29,640 Speaker 1: left behind, and all of these videos were glorifying suicide 212 00:12:29,640 --> 00:12:32,520 Speaker 1: in some way. I've seen a lot of this content 213 00:12:32,600 --> 00:12:36,000 Speaker 1: and it is really distressing to see it. And I'm 214 00:12:36,000 --> 00:12:38,800 Speaker 1: not a teenager, you know, my brain is fully developed, 215 00:12:38,840 --> 00:12:41,800 Speaker 1: and even I struggled watching some of this content and 216 00:12:41,920 --> 00:12:45,520 Speaker 1: particularly thinking about how a teenage boy would have felt 217 00:12:45,880 --> 00:12:48,959 Speaker 1: watching this kind of content. During my reporting, I spoke 218 00:12:49,000 --> 00:12:52,839 Speaker 1: to one child psychologist, Professor Jennifer Harriger, who has done 219 00:12:52,880 --> 00:12:55,920 Speaker 1: a lot of work on social media and how it 220 00:12:56,000 --> 00:12:56,839 Speaker 1: impacts kids. 221 00:12:56,880 --> 00:13:00,920 Speaker 2: Today. Our producer Rebecca Shassan asked profefd us your Harger 222 00:13:01,280 --> 00:13:05,559 Speaker 2: about her research on social media in young people's brains. 223 00:13:06,200 --> 00:13:10,600 Speaker 3: A teenage brain has a less developed prefrontal cortex, that's 224 00:13:10,640 --> 00:13:13,120 Speaker 3: the part of the brain that's associated with judgment and 225 00:13:13,160 --> 00:13:17,320 Speaker 3: planning and impulse control. And a teenage brain also is 226 00:13:17,400 --> 00:13:21,240 Speaker 3: more sensitive to rewards and social feedback, and we know 227 00:13:21,360 --> 00:13:24,640 Speaker 3: that from numerous studies. And so if you think about 228 00:13:25,000 --> 00:13:28,320 Speaker 3: a teen viewing this type of content, they would have 229 00:13:28,559 --> 00:13:35,320 Speaker 3: a harder time recognizing that these messages are not necessarily normalized. 230 00:13:35,480 --> 00:13:40,040 Speaker 3: They would have difficulty critically evaluating the content of the messages, 231 00:13:40,440 --> 00:13:43,920 Speaker 3: and they would also be more sensitive to the emotional 232 00:13:43,960 --> 00:13:47,240 Speaker 3: component of the messages. And then when you combine that 233 00:13:47,360 --> 00:13:51,000 Speaker 3: with the prefrontal cortex, that would also lead to difficulties 234 00:13:51,040 --> 00:13:53,200 Speaker 3: with them being able to regulate the time that they're 235 00:13:53,240 --> 00:13:55,559 Speaker 3: spending on social media, so they might be viewing more 236 00:13:55,600 --> 00:13:59,520 Speaker 3: and more images or content compared to people that might 237 00:13:59,559 --> 00:14:01,440 Speaker 3: be able to recognize that, oh, it's time for me 238 00:14:01,520 --> 00:14:02,120 Speaker 3: to get off. 239 00:14:04,440 --> 00:14:07,160 Speaker 1: When his mum started reviewing all of this content and 240 00:14:07,200 --> 00:14:09,880 Speaker 1: all of these videos that Chase had saved, she said, 241 00:14:09,920 --> 00:14:13,080 Speaker 1: the one that devastated her the most, that caused her 242 00:14:13,120 --> 00:14:16,680 Speaker 1: to actually physically gasp, was a video sent to his 243 00:14:16,760 --> 00:14:19,920 Speaker 1: account three days before his death. It was a picture 244 00:14:19,960 --> 00:14:23,000 Speaker 1: of a young man smiling on train tracks with an 245 00:14:23,040 --> 00:14:26,680 Speaker 1: oncoming train in the background, and the caption said, went 246 00:14:26,760 --> 00:14:30,000 Speaker 1: for a quick little walk to clear my head. Chase 247 00:14:30,080 --> 00:14:32,520 Speaker 1: died by stepping in front of an oncoming. 248 00:14:32,200 --> 00:14:38,080 Speaker 2: Train, Olivia, When you asked TikTok about Chase, where did 249 00:14:38,080 --> 00:14:38,560 Speaker 2: they say. 250 00:14:38,840 --> 00:14:42,720 Speaker 1: Chase's parents have filed a wrongful death lawsuit against TikTok, 251 00:14:43,120 --> 00:14:47,080 Speaker 1: And when a lawsuits filed, that really removes the ability 252 00:14:47,200 --> 00:14:49,920 Speaker 1: for any corporation to be able to comment on the case, 253 00:14:50,200 --> 00:14:53,080 Speaker 1: so it's fairly typical for TikTok to come back to 254 00:14:53,200 --> 00:14:56,520 Speaker 1: us saying we can't comment on pending litigation. But the 255 00:14:56,600 --> 00:14:59,880 Speaker 1: company did send a statement through saying that it is 256 00:15:00,040 --> 00:15:02,360 Speaker 1: committed to the safety and well being of its users, 257 00:15:02,600 --> 00:15:06,160 Speaker 1: especially teens, and a spokeswoman from the company said, our 258 00:15:06,200 --> 00:15:09,680 Speaker 1: hearts break for any family that experience is a tragic loss. 259 00:15:10,040 --> 00:15:13,600 Speaker 1: We strive to provide a positive and enriching experience and 260 00:15:13,640 --> 00:15:17,960 Speaker 1: will continue our significant investment in safeguarding our platform, and 261 00:15:18,040 --> 00:15:20,680 Speaker 1: TikTok has done a lot of work in this area, 262 00:15:20,760 --> 00:15:24,000 Speaker 1: particularly in the last few years, to try and make 263 00:15:24,080 --> 00:15:28,120 Speaker 1: the app a safer space for kids. The challenges that 264 00:15:28,200 --> 00:15:32,200 Speaker 1: when you find specific accounts like Chase's account, and you 265 00:15:32,240 --> 00:15:35,120 Speaker 1: can see that the for you feed is continuously sending 266 00:15:35,160 --> 00:15:39,520 Speaker 1: him dark, depressing or distressing content, it shows us just 267 00:15:39,640 --> 00:15:43,080 Speaker 1: how challenging this task is, and that even with all 268 00:15:43,120 --> 00:15:46,120 Speaker 1: of these efforts, all of these safeguards, or all of 269 00:15:46,160 --> 00:15:50,560 Speaker 1: these guard rails to protect kids, ultimately TikTok hasn't been 270 00:15:50,600 --> 00:15:54,280 Speaker 1: able to fully solve this problem. In the March congressional hearing, 271 00:15:54,400 --> 00:15:58,720 Speaker 1: when TikTok CEO Shaou Chu had to testify before Congress. 272 00:15:59,080 --> 00:16:03,760 Speaker 1: Chase NASCAR's account was actually broadcast before the entire congressional hearing. 273 00:16:04,440 --> 00:16:09,680 Speaker 4: The content and chases for You page was not a 274 00:16:09,760 --> 00:16:14,200 Speaker 4: window to discovery, as you boldly claimed in your testimony. 275 00:16:14,840 --> 00:16:19,120 Speaker 4: It wasn't content from a creator that you invited to 276 00:16:19,200 --> 00:16:23,640 Speaker 4: roam the hill today or stam education content that children 277 00:16:24,040 --> 00:16:29,680 Speaker 4: in China see. Instead, his for you page was sadly 278 00:16:29,760 --> 00:16:36,440 Speaker 4: a window to discover suicide. It is unacceptable, sir, that 279 00:16:36,600 --> 00:16:41,000 Speaker 4: even after knowing all these dangers, you still claim TikTok 280 00:16:41,080 --> 00:16:43,239 Speaker 4: is something grand to behold. 281 00:16:43,880 --> 00:16:46,880 Speaker 1: Chu was asked if he would allow his own children 282 00:16:46,920 --> 00:16:50,280 Speaker 1: to watch that type of content, and then a representative 283 00:16:50,400 --> 00:16:53,160 Speaker 1: displayed a number of videos that had been sent to 284 00:16:53,240 --> 00:16:55,440 Speaker 1: Chase that were glorifying suicide. 285 00:16:55,680 --> 00:17:00,440 Speaker 4: Would you share this content with your children? With your 286 00:17:00,480 --> 00:17:02,880 Speaker 4: two children, would you want them to see this? 287 00:17:04,280 --> 00:17:08,360 Speaker 2: That was Florida Republican Congressman Gus Bill A. Racketts, who 288 00:17:08,400 --> 00:17:12,040 Speaker 2: was questioning TikTok CEO show Chew and Chew is not 289 00:17:12,119 --> 00:17:15,720 Speaker 2: really given a chance to answer the question he was asked. 290 00:17:16,080 --> 00:17:19,679 Speaker 1: And during the hearing, Chase Nescar's parents were in the 291 00:17:19,720 --> 00:17:23,040 Speaker 1: audience and they were asked to stand up while his 292 00:17:23,160 --> 00:17:27,320 Speaker 1: account was displayed before all the representatives in the room, 293 00:17:27,680 --> 00:17:30,720 Speaker 1: and it resulted in probably one of the most emotive 294 00:17:30,840 --> 00:17:34,200 Speaker 1: scenes from the entire five hour testimony, when you saw 295 00:17:34,280 --> 00:17:37,119 Speaker 1: Dean and Michelle Nescar stand up in the audience and 296 00:17:37,280 --> 00:17:40,919 Speaker 1: break down in tears as Chase's death was explained to 297 00:17:41,000 --> 00:17:41,520 Speaker 1: the CEO. 298 00:17:42,680 --> 00:17:46,480 Speaker 2: After the break, the frustration of some people inside TikTok 299 00:17:46,520 --> 00:17:58,000 Speaker 2: whose job is trying to protect users from dark content, Olivia. 300 00:17:58,040 --> 00:18:00,719 Speaker 2: Before the break, you had said that tik talk is 301 00:18:01,200 --> 00:18:04,040 Speaker 2: trying to reduce the amount of this sort of content 302 00:18:04,080 --> 00:18:06,920 Speaker 2: on the platform. What exactly are they doing, What sort 303 00:18:06,960 --> 00:18:10,160 Speaker 2: of guardrails, as you put it, are they putting in play? 304 00:18:10,600 --> 00:18:13,800 Speaker 1: So TikTok has a trust and safety team. It has 305 00:18:13,920 --> 00:18:17,320 Speaker 1: more than forty thousand people working on this team to 306 00:18:17,520 --> 00:18:21,119 Speaker 1: try and make the platform a safer space, particularly for kids, 307 00:18:21,680 --> 00:18:24,480 Speaker 1: and they have done many things over the years to 308 00:18:24,600 --> 00:18:28,320 Speaker 1: try and enhance the protection of the app. They've tried 309 00:18:28,359 --> 00:18:31,920 Speaker 1: things like not allowing users under the age of sixteen 310 00:18:32,040 --> 00:18:36,480 Speaker 1: to send private messages. They have created a family pairing 311 00:18:36,520 --> 00:18:39,600 Speaker 1: tool which allows parents to kind of get a mirror 312 00:18:39,640 --> 00:18:42,280 Speaker 1: image of what their kids are seeing on TikTok to 313 00:18:42,400 --> 00:18:44,800 Speaker 1: check in on what they're doing on the app. They 314 00:18:44,960 --> 00:18:48,879 Speaker 1: have also more recently, they've opened up tools that allow 315 00:18:48,960 --> 00:18:52,639 Speaker 1: researchers to study some of the content that is displayed 316 00:18:52,640 --> 00:18:56,600 Speaker 1: on TikTok, although in this case these researchers have to 317 00:18:56,640 --> 00:18:59,240 Speaker 1: actually get it cleared by TikTok before it's published, so 318 00:18:59,280 --> 00:19:02,679 Speaker 1: they've received a bit criticism around that move. But you know, 319 00:19:02,760 --> 00:19:04,760 Speaker 1: not only are they trying to be a safer space 320 00:19:04,800 --> 00:19:07,240 Speaker 1: for kids, they're also trying to be more transparent with 321 00:19:07,320 --> 00:19:10,879 Speaker 1: how the app works by opening a transparency center in 322 00:19:10,960 --> 00:19:14,639 Speaker 1: Los Angeles and you know, sharing more information about what 323 00:19:14,800 --> 00:19:18,360 Speaker 1: it is that powers TikTok. And the most recent thing 324 00:19:18,560 --> 00:19:21,960 Speaker 1: the app announced was actually in early March, they said 325 00:19:21,960 --> 00:19:24,840 Speaker 1: that they were pushing out a feature that allowed users 326 00:19:24,880 --> 00:19:28,639 Speaker 1: to refresh their for you page. So this means if 327 00:19:28,680 --> 00:19:30,800 Speaker 1: you didn't like what the app was sending you, you 328 00:19:30,800 --> 00:19:34,520 Speaker 1: could push a button that effectively resets the whole account 329 00:19:34,560 --> 00:19:36,879 Speaker 1: as if you've opened a new one, and then you 330 00:19:36,920 --> 00:19:39,720 Speaker 1: can start using TikTok as a completely new user, and 331 00:19:40,000 --> 00:19:42,159 Speaker 1: it pushes away all of the content that you may 332 00:19:42,200 --> 00:19:44,280 Speaker 1: have been sending in the past that the algorithm thinks 333 00:19:44,320 --> 00:19:45,200 Speaker 1: you want to see. 334 00:19:45,640 --> 00:19:49,040 Speaker 2: You also write that the Trust and Safety team pulls 335 00:19:49,119 --> 00:19:51,399 Speaker 2: down videos when they can find them, but it's not 336 00:19:51,440 --> 00:19:52,240 Speaker 2: really a fair fight. 337 00:19:52,680 --> 00:19:55,480 Speaker 1: Yeah that's right. I mean billions of pieces of content 338 00:19:55,520 --> 00:19:57,960 Speaker 1: are uploaded onto TikTok, you know, in a daily basis, 339 00:19:58,040 --> 00:20:01,119 Speaker 1: and the Trust and Safety tea is tasked with trying 340 00:20:01,119 --> 00:20:06,320 Speaker 1: to moderate this content. This is a really complicated nuance 341 00:20:06,520 --> 00:20:10,040 Speaker 1: complex area. If we just talk about depression for a moment, 342 00:20:10,359 --> 00:20:13,480 Speaker 1: think about how heart it would be for a human 343 00:20:13,560 --> 00:20:16,280 Speaker 1: moderator to decide what kind of content it should leave 344 00:20:16,359 --> 00:20:18,080 Speaker 1: up on the platform and what kind of content it 345 00:20:18,080 --> 00:20:21,320 Speaker 1: should pull down. So, for example, say we talk about 346 00:20:21,320 --> 00:20:24,560 Speaker 1: a video where a user is saying I'm feeling really 347 00:20:24,600 --> 00:20:27,560 Speaker 1: sad today. Should that be allowed on TikTok or should 348 00:20:27,600 --> 00:20:31,240 Speaker 1: it not? And the human moderators have to ultimately decide this. 349 00:20:31,240 --> 00:20:35,160 Speaker 1: This gets harder as the topics get, you know, more 350 00:20:35,200 --> 00:20:38,840 Speaker 1: and more complex. So when it comes to suicide, for example, 351 00:20:39,400 --> 00:20:42,399 Speaker 1: if someone writes a caption like I don't want to 352 00:20:42,440 --> 00:20:46,040 Speaker 1: be here tomorrow. Are they joking? Are they being serious? 353 00:20:46,240 --> 00:20:48,920 Speaker 1: Is this a crif for help, should we send law 354 00:20:49,000 --> 00:20:52,320 Speaker 1: enforcement to the adore. It's very hard for human moderators 355 00:20:52,359 --> 00:20:56,280 Speaker 1: to determine what a user means when they send out 356 00:20:56,320 --> 00:20:59,439 Speaker 1: a video onto the app, and the human moderators have 357 00:20:59,480 --> 00:21:01,919 Speaker 1: to ultimate decide what to leave up and what to 358 00:21:01,960 --> 00:21:02,440 Speaker 1: take down. 359 00:21:02,840 --> 00:21:05,680 Speaker 2: You spoke to quite a few current and former members 360 00:21:05,680 --> 00:21:08,879 Speaker 2: of TikTok's trust in safety team, and you say that 361 00:21:08,960 --> 00:21:12,640 Speaker 2: they were frustrated because they felt they couldn't even get 362 00:21:12,640 --> 00:21:15,280 Speaker 2: the information they needed to do their jobs. 363 00:21:15,760 --> 00:21:18,280 Speaker 1: I spoke to more than a dozen trust in safety 364 00:21:18,320 --> 00:21:21,639 Speaker 1: representatives for this article. Many of them are former employees 365 00:21:21,680 --> 00:21:24,720 Speaker 1: of TikTok, some who left as recently as last year, 366 00:21:25,320 --> 00:21:28,480 Speaker 1: and what they were telling me is that not only 367 00:21:28,560 --> 00:21:32,760 Speaker 1: is this a really difficult issue for the platform to moderate, 368 00:21:32,960 --> 00:21:37,440 Speaker 1: but they felt as though they were limited in what 369 00:21:37,480 --> 00:21:39,760 Speaker 1: they could achieve in their role. And what I mean 370 00:21:39,800 --> 00:21:42,439 Speaker 1: by that is their task with trying to make the 371 00:21:42,480 --> 00:21:46,240 Speaker 1: algorithm safer, but they don't know how it works. Many 372 00:21:46,280 --> 00:21:49,000 Speaker 1: of them described the algorithm as a black box. They 373 00:21:49,080 --> 00:21:52,760 Speaker 1: said that all decisions pertaining to the algorithm were made 374 00:21:52,880 --> 00:21:56,240 Speaker 1: in Beijing, where byte Dance, the parent company of TikTok, 375 00:21:56,440 --> 00:22:00,520 Speaker 1: is based, and even though these forty thousand people who 376 00:22:00,600 --> 00:22:03,280 Speaker 1: work on trust and safety, who are largely based outside 377 00:22:03,280 --> 00:22:05,760 Speaker 1: of China, what they could do is come up with 378 00:22:05,920 --> 00:22:09,639 Speaker 1: tools to protect users. That really put the onus on 379 00:22:09,760 --> 00:22:12,840 Speaker 1: the user, So you, as the user, have to decide 380 00:22:12,880 --> 00:22:15,919 Speaker 1: what content you're comfortable seeing. If you, as a user 381 00:22:16,119 --> 00:22:19,120 Speaker 1: don't want to see content about anorexia, you can put 382 00:22:19,119 --> 00:22:20,879 Speaker 1: that in as a hashtag and say you want to 383 00:22:20,880 --> 00:22:24,760 Speaker 1: filter out any content related to that. But this does 384 00:22:24,840 --> 00:22:27,119 Speaker 1: not change the underlying algorithm. 385 00:22:27,160 --> 00:22:27,840 Speaker 2: It does not. 386 00:22:27,960 --> 00:22:30,879 Speaker 1: Change the way the for you feed works. The only 387 00:22:31,000 --> 00:22:34,320 Speaker 1: way to change what a user sees what content is 388 00:22:34,359 --> 00:22:38,040 Speaker 1: being sent out broadcasts to the billion people who use 389 00:22:38,080 --> 00:22:42,280 Speaker 1: TikTok on a daily basis is to tweak the underlying algorithm. 390 00:22:42,400 --> 00:22:45,200 Speaker 1: And the Trust and Safety team had no access to it. 391 00:22:45,600 --> 00:22:48,400 Speaker 2: And you write. The members of the Trust and Safety 392 00:22:48,440 --> 00:22:53,879 Speaker 2: team wrote to the higher upside TikTok asking questions about 393 00:22:53,880 --> 00:22:56,200 Speaker 2: how the algorithm works so that they would be able 394 00:22:56,280 --> 00:22:59,120 Speaker 2: to make suggestions about how to make it safer. And 395 00:22:59,200 --> 00:23:01,359 Speaker 2: what did the company say to those requests. 396 00:23:01,680 --> 00:23:04,679 Speaker 1: Many of those requests were just flat out ignored. They 397 00:23:04,720 --> 00:23:07,960 Speaker 1: said that it was a one way information flow. Trust 398 00:23:07,960 --> 00:23:11,280 Speaker 1: and Safety was expected to provide information to the engineering 399 00:23:11,320 --> 00:23:13,720 Speaker 1: team that was largely based in Beijing, But when they 400 00:23:13,800 --> 00:23:17,879 Speaker 1: asked follow up questions for the engineers who programmed the 401 00:23:17,960 --> 00:23:21,000 Speaker 1: algorithm to talk to trust and safety, they were stonewalled. 402 00:23:21,040 --> 00:23:24,439 Speaker 1: They couldn't get any answers. They said that the secrecy 403 00:23:24,480 --> 00:23:29,439 Speaker 1: pertaining to the algorithm isn't just an external thing, it 404 00:23:29,560 --> 00:23:31,560 Speaker 1: is a deep internal thing as well. 405 00:23:32,200 --> 00:23:34,480 Speaker 2: And what does TikTok say about that. 406 00:23:35,160 --> 00:23:37,800 Speaker 1: In a response to a number of questions I asked 407 00:23:37,840 --> 00:23:40,520 Speaker 1: the company, TikTok said that it takes these concerns that 408 00:23:40,560 --> 00:23:44,000 Speaker 1: have been voiced by employees seriously. It says that members 409 00:23:44,040 --> 00:23:47,440 Speaker 1: of its trust and safety team do work directly with engineering, 410 00:23:48,040 --> 00:23:51,440 Speaker 1: and that it's made significant changes in the past few 411 00:23:51,520 --> 00:23:54,040 Speaker 1: years to try and make the app a safer space. 412 00:23:54,200 --> 00:23:57,880 Speaker 1: It's also been trying to make the app more transparent. 413 00:23:58,160 --> 00:24:02,440 Speaker 1: It's been releasing transparency reports, telling the public how many 414 00:24:02,520 --> 00:24:06,840 Speaker 1: videos it removes every quarter, and opening this transparency center 415 00:24:06,840 --> 00:24:11,440 Speaker 1: in LA to try and allow researchers, third party members, 416 00:24:11,640 --> 00:24:15,600 Speaker 1: journalists to get a better sense of how the algorithm works. 417 00:24:15,880 --> 00:24:18,960 Speaker 1: So the company is definitely working to try and address 418 00:24:19,000 --> 00:24:20,440 Speaker 1: this issue. 419 00:24:20,680 --> 00:24:24,840 Speaker 2: You right about one person in particular who shows just 420 00:24:24,920 --> 00:24:27,560 Speaker 2: how difficult this is. Can you tell us about him? 421 00:24:28,040 --> 00:24:32,840 Speaker 1: Charles Barr was an early employee on TikTok's sales team 422 00:24:33,080 --> 00:24:36,159 Speaker 1: and Germany. What he told me was that before he 423 00:24:36,200 --> 00:24:39,439 Speaker 1: started working at TikTok, he would see a lot of fun, silly, 424 00:24:39,560 --> 00:24:42,480 Speaker 1: kind of joking videos and he loved the product. He 425 00:24:42,520 --> 00:24:44,600 Speaker 1: thought it was amazing and way better than any other 426 00:24:44,640 --> 00:24:47,840 Speaker 1: social media product out there. But once he joined TikTok 427 00:24:47,920 --> 00:24:50,600 Speaker 1: and started to post about working as an employee at 428 00:24:50,600 --> 00:24:53,680 Speaker 1: the company, he would get a lot of videos sent 429 00:24:53,720 --> 00:24:57,920 Speaker 1: to him from users who felt like specific content should 430 00:24:57,920 --> 00:25:00,960 Speaker 1: come down, was dangerous, was harmful. One of the first 431 00:25:01,000 --> 00:25:04,359 Speaker 1: really scary videos he remembers being sent was of a 432 00:25:04,400 --> 00:25:08,000 Speaker 1: man shooting himself in the head. And when Charles would 433 00:25:08,040 --> 00:25:10,639 Speaker 1: engage with this content, what I mean by engage is 434 00:25:10,720 --> 00:25:14,360 Speaker 1: just open it, just watch it, just foughd it on 435 00:25:14,400 --> 00:25:16,320 Speaker 1: maybe to the trust and safety team, which he did 436 00:25:16,359 --> 00:25:20,439 Speaker 1: on many occasions. The algorithm was learning from him that 437 00:25:20,600 --> 00:25:24,639 Speaker 1: he was opening, engaging and watching dark, depressing content, so 438 00:25:24,680 --> 00:25:27,159 Speaker 1: it started to send it to him. It became so 439 00:25:27,440 --> 00:25:30,840 Speaker 1: dark that he said, on many occasions it led him 440 00:25:30,840 --> 00:25:33,639 Speaker 1: to cry. And you know, that just goes to show 441 00:25:33,680 --> 00:25:38,159 Speaker 1: that even TikTok's own employees didn't understand the algorithm and 442 00:25:38,200 --> 00:25:41,000 Speaker 1: didn't understand how to change what it was sending them, 443 00:25:41,560 --> 00:25:44,600 Speaker 1: so he raised these internal concerns, and he says a 444 00:25:44,640 --> 00:25:48,960 Speaker 1: lot of his questions were not answered. After Charles raised 445 00:25:49,000 --> 00:25:52,040 Speaker 1: these concerns internally, he was actually fired from TikTok in 446 00:25:52,160 --> 00:25:56,479 Speaker 1: late twenty twenty one. He was accused of expense fraud 447 00:25:57,080 --> 00:26:01,560 Speaker 1: and misusing company tools. TikTok has said to me in 448 00:26:01,600 --> 00:26:04,280 Speaker 1: an on record statement that it can't really comment on 449 00:26:04,359 --> 00:26:07,880 Speaker 1: Charles's criticisms of the company, and it can't validate any 450 00:26:07,960 --> 00:26:10,760 Speaker 1: of the concerns that he raised internally because they can't 451 00:26:10,760 --> 00:26:14,040 Speaker 1: find those messages. He actually still maintains his innocence to 452 00:26:14,080 --> 00:26:17,199 Speaker 1: this day. He sued TikTok for wrongful dismissal and they 453 00:26:17,200 --> 00:26:18,920 Speaker 1: settled outside of court. 454 00:26:19,240 --> 00:26:22,480 Speaker 2: When we come back, the pressure builds on social media 455 00:26:22,520 --> 00:26:34,040 Speaker 2: companies to do something to have fixed this problem, Olivia. 456 00:26:34,080 --> 00:26:37,480 Speaker 2: We talked about the congressional hearing where TikTok CEO is 457 00:26:37,640 --> 00:26:41,800 Speaker 2: asked a lot of tough questions, and here's there's growing 458 00:26:41,840 --> 00:26:45,120 Speaker 2: pressure on the company that do something about this kind 459 00:26:45,119 --> 00:26:49,120 Speaker 2: of content that the algorithm is pushing to users. 460 00:26:49,760 --> 00:26:52,439 Speaker 1: I mean, the winds of change are coming for social media. 461 00:26:52,520 --> 00:26:55,880 Speaker 1: We're seeing movements in so many different areas. Now we're 462 00:26:55,880 --> 00:27:00,520 Speaker 1: seeing dozens of lawsuits filed accusing these companies of wrongfol death, 463 00:27:00,640 --> 00:27:04,879 Speaker 1: or product harm liability. Many of these cases are ongoing 464 00:27:04,920 --> 00:27:08,960 Speaker 1: before the courts today. We're seeing Section two thirty, which 465 00:27:09,160 --> 00:27:12,720 Speaker 1: is a section of the Communications Decency Act that essentially 466 00:27:12,840 --> 00:27:17,280 Speaker 1: provides tech companies with a liability shield saying they cannot 467 00:27:17,359 --> 00:27:20,160 Speaker 1: be sued for the content that's posted on their platforms. 468 00:27:20,640 --> 00:27:24,320 Speaker 1: Section two thirty cases before the Supreme Court right now. 469 00:27:24,440 --> 00:27:27,960 Speaker 1: We've also seen more and more cases like the Chase 470 00:27:28,040 --> 00:27:31,000 Speaker 1: Nascar case and parents involved in those cases willing to 471 00:27:31,040 --> 00:27:34,960 Speaker 1: speak out. We're seeing researchers and academics also push back 472 00:27:35,000 --> 00:27:39,720 Speaker 1: against big tech, calling out for access to these algorithms 473 00:27:39,760 --> 00:27:42,360 Speaker 1: to understand how to study them, to see the impact 474 00:27:42,359 --> 00:27:44,919 Speaker 1: that they have on kids and on society in general. 475 00:27:45,560 --> 00:27:48,920 Speaker 2: As are practical matter, what can be done to fix 476 00:27:48,960 --> 00:27:49,679 Speaker 2: this problem? 477 00:27:49,880 --> 00:27:51,640 Speaker 1: Well, in terms of what can be done, I mean 478 00:27:51,680 --> 00:27:55,840 Speaker 1: a lot of these academics are calling for more transparency 479 00:27:55,920 --> 00:27:58,440 Speaker 1: around how these algorithms work. 480 00:27:58,680 --> 00:28:02,600 Speaker 2: And Professor Jennifer, who we've heard from earlier, said something 481 00:28:03,000 --> 00:28:06,679 Speaker 2: similar that research is hard to do if you don't 482 00:28:06,720 --> 00:28:08,120 Speaker 2: have access to all the information. 483 00:28:08,680 --> 00:28:10,480 Speaker 3: We have not been able to find a way to 484 00:28:10,520 --> 00:28:16,240 Speaker 3: actually test the algorithm itself and whether that is increasing risk, 485 00:28:16,600 --> 00:28:18,560 Speaker 3: And the reason for that is that the companies are 486 00:28:18,560 --> 00:28:21,760 Speaker 3: not very transparent about how the algorithm works, so there's 487 00:28:21,800 --> 00:28:25,639 Speaker 3: no way for a researcher to test whether or not 488 00:28:25,720 --> 00:28:30,480 Speaker 3: the algorithm is making this experience more dangerous for the 489 00:28:30,560 --> 00:28:34,200 Speaker 3: child without knowing how the algorithm is actually working. If 490 00:28:34,400 --> 00:28:38,240 Speaker 3: someone is experiencing negative effects based on the content that 491 00:28:38,280 --> 00:28:41,440 Speaker 3: they're viewing on a social media platform, it would be 492 00:28:41,480 --> 00:28:45,120 Speaker 3: difficult to know if it's because they themselves have sought 493 00:28:45,160 --> 00:28:48,480 Speaker 3: out the content, or if the algorithm is pushing more 494 00:28:48,520 --> 00:28:51,120 Speaker 3: content on them that they would have not normally been 495 00:28:51,120 --> 00:28:51,680 Speaker 3: exposed to. 496 00:28:52,200 --> 00:28:56,440 Speaker 2: We asked Jennifer Harrager for some practical advice for parents 497 00:28:56,560 --> 00:28:59,800 Speaker 2: who are here in all this and wondering how they 498 00:28:59,800 --> 00:29:04,600 Speaker 2: can help their kids stay emotionally safe on social media. 499 00:29:04,920 --> 00:29:08,400 Speaker 3: If a child's already on social media, then I would 500 00:29:08,440 --> 00:29:12,920 Speaker 3: recommend that the parents help create a plan to first 501 00:29:13,000 --> 00:29:15,080 Speaker 3: of all, limit the time that the child is on 502 00:29:15,120 --> 00:29:18,680 Speaker 3: social media, because they probably will have a very difficult 503 00:29:18,760 --> 00:29:22,400 Speaker 3: time getting off of it themselves without some sort of 504 00:29:22,720 --> 00:29:27,920 Speaker 3: prompt And I would also recommend that parents talk with 505 00:29:28,000 --> 00:29:31,959 Speaker 3: their children and help them with some media literacy. So 506 00:29:32,360 --> 00:29:36,440 Speaker 3: how do you critically evaluate the messages that you're being 507 00:29:36,480 --> 00:29:39,520 Speaker 3: exposed to? What kinds of things are you seeing on 508 00:29:39,560 --> 00:29:42,800 Speaker 3: social media? Let's talk about it. How are you feeling 509 00:29:43,000 --> 00:29:45,280 Speaker 3: when you're on social media? If you feel a lot 510 00:29:45,400 --> 00:29:48,120 Speaker 3: worse about yourself when you're on it, you know, is 511 00:29:48,160 --> 00:29:49,760 Speaker 3: this the best thing for you right now? 512 00:29:52,240 --> 00:29:56,400 Speaker 2: Oliba? Given all the pressure lately for more transparency, do 513 00:29:56,480 --> 00:30:00,040 Speaker 2: you think social media companies will begin to allow some 514 00:30:00,200 --> 00:30:03,040 Speaker 2: visibility into the way their algorithms work. 515 00:30:03,840 --> 00:30:06,520 Speaker 1: One of the biggest difficulties we've seen is that tech 516 00:30:06,560 --> 00:30:11,520 Speaker 1: companies are notoriously secretive about this data and about their algorithms. 517 00:30:12,000 --> 00:30:17,560 Speaker 1: They protect that intellectual property fiercely. And this isn't just TikTok, 518 00:30:17,680 --> 00:30:20,480 Speaker 1: this is, you know, a big tech social media thing. 519 00:30:20,800 --> 00:30:23,840 Speaker 1: None of these companies want third party researchers coming in 520 00:30:23,920 --> 00:30:27,480 Speaker 1: and studying their algorithm. That's their secret source, that's what 521 00:30:27,600 --> 00:30:31,000 Speaker 1: makes them money, that's their whole business model. So now 522 00:30:31,040 --> 00:30:34,719 Speaker 1: we're getting to a point where researchers are saying, well, look, 523 00:30:34,960 --> 00:30:40,880 Speaker 1: we've seen rising rates of mental health issues, hopelessness, suicide, 524 00:30:40,920 --> 00:30:44,920 Speaker 1: and self harm among teens. That has tracked precisely with 525 00:30:45,480 --> 00:30:49,760 Speaker 1: the explosive use of social media. So you know, correlation 526 00:30:50,280 --> 00:30:53,600 Speaker 1: is very easy to prove here, but causation is very 527 00:30:53,600 --> 00:30:56,640 Speaker 1: difficult to prove. The reason why that's difficult to prove 528 00:30:56,720 --> 00:31:00,400 Speaker 1: is because these researchers cannot gain access to what happening 529 00:31:00,400 --> 00:31:03,640 Speaker 1: inside these companies and how these algorithms work and how 530 00:31:03,680 --> 00:31:06,680 Speaker 1: they're affecting their users. So one of the things that 531 00:31:06,720 --> 00:31:08,959 Speaker 1: a lot of people are calling for, one of the 532 00:31:08,960 --> 00:31:12,040 Speaker 1: things that we should be talking about is ways to 533 00:31:12,040 --> 00:31:15,400 Speaker 1: put pressure on these companies to be more transparent, to 534 00:31:15,520 --> 00:31:19,400 Speaker 1: not only allow people to study the public content that 535 00:31:19,520 --> 00:31:21,800 Speaker 1: might be on their platforms, which is what TikTok has 536 00:31:21,840 --> 00:31:26,120 Speaker 1: recently done, but to actually study the underlying algorithm, the 537 00:31:26,200 --> 00:31:30,000 Speaker 1: recommendation engine, how it's weighted, what it thinks you know 538 00:31:30,160 --> 00:31:32,680 Speaker 1: you want to see, and how it makes those decisions. 539 00:31:33,240 --> 00:31:35,560 Speaker 1: And that's what I've heard time and time again from 540 00:31:35,560 --> 00:31:38,240 Speaker 1: the researchers I've talked to, is we need access to 541 00:31:38,320 --> 00:31:42,640 Speaker 1: study these because these algorithms and these companies are having 542 00:31:42,720 --> 00:31:45,600 Speaker 1: a huge impact on the youth of today and we 543 00:31:45,760 --> 00:31:48,280 Speaker 1: do not know what the long term consequences are going 544 00:31:48,280 --> 00:31:51,480 Speaker 1: to be. The vast majority of kids do really like 545 00:31:51,560 --> 00:31:55,680 Speaker 1: social media, but it has this darker side and the 546 00:31:55,800 --> 00:31:59,080 Speaker 1: darker side we don't understand because we can't study it. 547 00:31:59,120 --> 00:32:00,480 Speaker 2: Olivia, thanks so. 548 00:32:00,520 --> 00:32:02,720 Speaker 1: Much for being here, Thanks for hearing me. 549 00:32:03,880 --> 00:32:05,800 Speaker 2: Thanks for listening to us here at The Big Take. 550 00:32:06,000 --> 00:32:09,480 Speaker 2: It's a daily podcast from Bloomberg and iHeartRadio. For more 551 00:32:09,520 --> 00:32:13,760 Speaker 2: shows from iHeartRadio, visit the iHeartRadio app, Apple Podcasts, or 552 00:32:13,840 --> 00:32:16,400 Speaker 2: wherever you listen, and we'd love to hear from you. 553 00:32:16,760 --> 00:32:20,400 Speaker 2: Email us questions or comments to Big Take at Bloomberg 554 00:32:20,440 --> 00:32:24,000 Speaker 2: dot net. The supervising producer of The Big Take is 555 00:32:24,120 --> 00:32:28,760 Speaker 2: Vicky Virgalina. Our senior producer is Catherine Fink. Rebecca Shasson 556 00:32:28,920 --> 00:32:33,160 Speaker 2: is our producer. Our associate producer is Sam Goobauer. Philde 557 00:32:33,200 --> 00:32:36,960 Speaker 2: Garcia is our engineer. Our original music was composed by 558 00:32:37,080 --> 00:32:40,640 Speaker 2: Leo Sidrin. I'm Wes Kasova. We'll be back on Monday 559 00:32:40,720 --> 00:32:43,320 Speaker 2: with another Big Take. Have a great weekend.