1 00:00:00,400 --> 00:00:02,719 Speaker 1: You can kind of think of Lindsa as the fast 2 00:00:02,759 --> 00:00:10,160 Speaker 1: fashion of self portraits. There Are No Girls on the 3 00:00:10,160 --> 00:00:13,000 Speaker 1: Internet as a production of I Heart Radio and Unbossed Creative. 4 00:00:17,400 --> 00:00:19,479 Speaker 1: I'm Bridget Todd, and this is there are No Girls 5 00:00:19,480 --> 00:00:24,560 Speaker 1: on the Internet. So it's your social media feeds are 6 00:00:24,680 --> 00:00:27,280 Speaker 1: anything like mine, that you probably have been seeing these 7 00:00:27,320 --> 00:00:30,880 Speaker 1: AI images of your friends all over Facebook. And it 8 00:00:30,920 --> 00:00:34,400 Speaker 1: really reminded me of a few other early Facebook picture trends, 9 00:00:34,400 --> 00:00:37,640 Speaker 1: like when we all picked cartoon characters or famous actors 10 00:00:37,640 --> 00:00:39,519 Speaker 1: that we thought we looked like, and I'll changed our 11 00:00:39,560 --> 00:00:42,920 Speaker 1: profile pictures to reflect that. I'm probably dating myself a 12 00:00:42,960 --> 00:00:46,280 Speaker 1: little bit here, but early Facebook was full of earnest 13 00:00:46,280 --> 00:00:50,160 Speaker 1: stuff like that, and this AI image generation trend kind 14 00:00:50,159 --> 00:00:52,360 Speaker 1: of seems like a call back to those days. So 15 00:00:52,400 --> 00:00:55,920 Speaker 1: at first I was confused, but then I started to 16 00:00:55,960 --> 00:00:59,320 Speaker 1: get intrigued. I mean, who doesn't want to see a 17 00:00:59,440 --> 00:01:05,440 Speaker 1: representationation of themselves reimagined and an otherworldly, futuristic, cyberpunk aesthetic. 18 00:01:05,800 --> 00:01:07,800 Speaker 1: So I tracked down where they were all coming from 19 00:01:07,840 --> 00:01:10,960 Speaker 1: and found lens Us, the app behind these viral images. 20 00:01:11,440 --> 00:01:14,680 Speaker 1: So there I was finding myself happily clicking around the 21 00:01:14,720 --> 00:01:17,440 Speaker 1: app and scrolling around looking for the ten to twenty 22 00:01:17,440 --> 00:01:19,880 Speaker 1: pictures of myself that they said they needed to generate 23 00:01:19,959 --> 00:01:22,959 Speaker 1: AI images. And it wasn't until I was prompted to 24 00:01:22,959 --> 00:01:26,000 Speaker 1: grab my credit card that I pumped the brakes. First 25 00:01:26,000 --> 00:01:28,039 Speaker 1: of all, isn't it funny how easy it is, at 26 00:01:28,120 --> 00:01:30,959 Speaker 1: least for me to set aside the inclination to ask 27 00:01:31,040 --> 00:01:33,760 Speaker 1: questions if everyone else is doing something on social media 28 00:01:33,920 --> 00:01:36,280 Speaker 1: and it looks kind of fun or cool. I am 29 00:01:36,319 --> 00:01:39,520 Speaker 1: embarrassed about how quickly I went from this is stupid too, 30 00:01:39,680 --> 00:01:42,120 Speaker 1: I should probably try it? And I feel like this 31 00:01:42,200 --> 00:01:45,160 Speaker 1: is exactly the feeling that tech companies are so good 32 00:01:45,200 --> 00:01:48,080 Speaker 1: at exploiting in all of us. So what exactly are 33 00:01:48,120 --> 00:01:52,360 Speaker 1: the ethical and privacy implications of using an AI image 34 00:01:52,360 --> 00:01:57,120 Speaker 1: generation apps like lens up Engineer and AI ethicis Azaria 35 00:01:57,200 --> 00:02:00,640 Speaker 1: Cole Shephard wrote a popular Twitter thread break it all down, 36 00:02:01,000 --> 00:02:03,880 Speaker 1: starting with the tweet I am here on my soapbox 37 00:02:03,920 --> 00:02:06,120 Speaker 1: to ruin your day by talking about why you should 38 00:02:06,160 --> 00:02:09,240 Speaker 1: delete length to AI and stop blindly using image based 39 00:02:09,280 --> 00:02:13,280 Speaker 1: AI generator apps. Hi, my name is Azariah Coole Ship, 40 00:02:13,280 --> 00:02:15,359 Speaker 1: but I don't really have a fancy title. I had 41 00:02:15,400 --> 00:02:17,639 Speaker 1: just graduated with my degree in electrical engineering with the 42 00:02:17,760 --> 00:02:21,520 Speaker 1: emphasis and AI and implicit advice. You're an electrical engineer, 43 00:02:21,680 --> 00:02:24,519 Speaker 1: you're a real passion for ethics and AI, UH and 44 00:02:24,600 --> 00:02:26,960 Speaker 1: implicit bias and a I. How did this become something 45 00:02:27,000 --> 00:02:30,959 Speaker 1: that you were interested in pursuing professionally and really thinking about. 46 00:02:31,800 --> 00:02:34,560 Speaker 1: Early in my college career, I became really really interested 47 00:02:34,720 --> 00:02:38,440 Speaker 1: in AI, and we talked about like brave, Brave New 48 00:02:38,480 --> 00:02:42,040 Speaker 1: World and all of the different robotic developments and stuff 49 00:02:42,040 --> 00:02:43,920 Speaker 1: in some of my classes, and I started to write 50 00:02:44,040 --> 00:02:49,919 Speaker 1: essays criticizing basically how technology perpetuates violences against certain communities, 51 00:02:49,919 --> 00:02:54,880 Speaker 1: how lack of regulation also perpetuated by violence, and so 52 00:02:54,919 --> 00:02:57,239 Speaker 1: on and so forth. And so when it came time 53 00:02:57,280 --> 00:02:59,560 Speaker 1: to pick my senior research project as well as my 54 00:02:59,639 --> 00:03:03,320 Speaker 1: elect is, all of them were centered around algorithmic development 55 00:03:03,480 --> 00:03:07,079 Speaker 1: and addressing implicit bias and AI and also making a 56 00:03:09,120 --> 00:03:11,560 Speaker 1: more accessible if it's going to be used despite the 57 00:03:11,560 --> 00:03:14,760 Speaker 1: fact that it has extreme and what I feel at 58 00:03:14,840 --> 00:03:19,239 Speaker 1: this point in time, very irreparable flaws. What first went 59 00:03:19,240 --> 00:03:21,239 Speaker 1: through your mind when you wake up in the morning 60 00:03:21,320 --> 00:03:24,519 Speaker 1: and you see all these people on social media have 61 00:03:24,720 --> 00:03:28,800 Speaker 1: fed their images to this app where they are getting 62 00:03:28,840 --> 00:03:33,920 Speaker 1: these AI generated cartoon renderings of themselves like what were you? Like? 63 00:03:34,800 --> 00:03:37,720 Speaker 1: Hate to be a debut downer, guys, But this has 64 00:03:37,720 --> 00:03:41,520 Speaker 1: some issues. So it was kind of multifaceted because one, 65 00:03:41,720 --> 00:03:43,240 Speaker 1: like I said, a lot of my research is about 66 00:03:43,440 --> 00:03:48,200 Speaker 1: addressing implicit bias and so like increasing access to images 67 00:03:48,200 --> 00:03:50,280 Speaker 1: of diverse groups in order to enhance the ways in 68 00:03:50,320 --> 00:03:53,320 Speaker 1: which people engage with facial recognition technology and that kind 69 00:03:53,320 --> 00:03:57,280 Speaker 1: of stuff, to increase accuracy, to decrease frimalization and so 70 00:03:57,440 --> 00:04:01,240 Speaker 1: on and so forth. However, um, I think that there, 71 00:04:02,720 --> 00:04:04,920 Speaker 1: like I said, there's a lack of regulation as is, 72 00:04:04,960 --> 00:04:06,560 Speaker 1: and so I don't even think that AI is in 73 00:04:06,600 --> 00:04:10,120 Speaker 1: a place where it should be allowed to function the 74 00:04:10,120 --> 00:04:12,520 Speaker 1: way that it does. And so when I saw people 75 00:04:12,760 --> 00:04:14,880 Speaker 1: using the app, at first I was kind of like, Okay, 76 00:04:14,920 --> 00:04:17,719 Speaker 1: this is kind of crazy, very trendy. I asked somebody, 77 00:04:17,720 --> 00:04:19,599 Speaker 1: I was like, what app did you use? They told me, 78 00:04:20,080 --> 00:04:22,080 Speaker 1: and I just kind of said okay and left it there. 79 00:04:22,320 --> 00:04:24,880 Speaker 1: And then over the course the next couple of days, 80 00:04:24,880 --> 00:04:27,880 Speaker 1: I just started doing research and like really digging into 81 00:04:28,800 --> 00:04:32,320 Speaker 1: what people were actually branding. And I belonged to a 82 00:04:32,320 --> 00:04:35,480 Speaker 1: couple of different AI development groups, even the AI art 83 00:04:35,520 --> 00:04:39,600 Speaker 1: development group where people are generating art using prompts, and 84 00:04:39,640 --> 00:04:42,640 Speaker 1: I think that that's very different than using your photos 85 00:04:42,680 --> 00:04:46,280 Speaker 1: to generate images on apps like this one. And so 86 00:04:46,960 --> 00:04:50,960 Speaker 1: as a result, um, I started talking about it on 87 00:04:51,000 --> 00:04:52,560 Speaker 1: like my close friends on Instagram, and I was just like, 88 00:04:52,560 --> 00:04:54,359 Speaker 1: this is pretty crazy. I think that people just stop 89 00:04:54,400 --> 00:04:56,720 Speaker 1: using this app, y'all, do me a favorite, don't use it. 90 00:04:56,839 --> 00:04:58,800 Speaker 1: And people are like, oh, what happened? And like some 91 00:04:58,880 --> 00:05:02,120 Speaker 1: people ask questions, and so I started like expanding on 92 00:05:02,240 --> 00:05:05,240 Speaker 1: my opinion and explaining my findings to people, and I 93 00:05:05,320 --> 00:05:07,640 Speaker 1: kind of turned into, oh, where can I repost this? 94 00:05:07,680 --> 00:05:09,919 Speaker 1: So then I went to Twitter and I posted on 95 00:05:09,960 --> 00:05:14,120 Speaker 1: Twitter and yeah, that's how we got here. And I 96 00:05:14,160 --> 00:05:18,599 Speaker 1: still think that it's very chaotic. Um, there's a lot 97 00:05:18,640 --> 00:05:21,000 Speaker 1: of people who are mad or this Twitter thread, and 98 00:05:21,040 --> 00:05:22,560 Speaker 1: I don't even think it's that deep. I think it's 99 00:05:22,600 --> 00:05:26,040 Speaker 1: literally just me sharing information that you can find doing 100 00:05:26,040 --> 00:05:28,159 Speaker 1: basic research and if you decide to google things, like 101 00:05:28,200 --> 00:05:31,640 Speaker 1: people oftentimes jump on the bandwagon and don't do any reading. 102 00:05:32,360 --> 00:05:35,080 Speaker 1: And I feel like thirty minutes of reading in this case, 103 00:05:35,120 --> 00:05:37,760 Speaker 1: which takes people so for so far or so much 104 00:05:37,760 --> 00:05:40,760 Speaker 1: further than they are because they've taken different face value. 105 00:05:40,800 --> 00:05:43,280 Speaker 1: Most people who are interacting with this tweet didn't even 106 00:05:43,320 --> 00:05:45,760 Speaker 1: read the terms of service, and so that creates a 107 00:05:45,800 --> 00:05:48,960 Speaker 1: whole another question as to how seriously people take their 108 00:05:49,000 --> 00:05:53,520 Speaker 1: data privacy and how people interact with different apps on 109 00:05:53,560 --> 00:05:56,720 Speaker 1: social media, and which one for different apps and which 110 00:05:56,720 --> 00:05:59,480 Speaker 1: ones are worth engaging in and which ones aren't versus 111 00:05:59,680 --> 00:06:03,200 Speaker 1: what's returning an investment or is there returning investment overall 112 00:06:03,240 --> 00:06:05,920 Speaker 1: when you're using these apps? Oh god, that's such an 113 00:06:05,960 --> 00:06:08,599 Speaker 1: interesting point. Like, I first of all, we are big 114 00:06:08,640 --> 00:06:12,080 Speaker 1: proponents of reading those terms terms and service. If they 115 00:06:12,120 --> 00:06:15,400 Speaker 1: are if they if they are difficult for you know, 116 00:06:15,440 --> 00:06:17,719 Speaker 1: your average person to understand, if they're not a lawyer, 117 00:06:18,040 --> 00:06:20,080 Speaker 1: there might be like might there be some reason for that? 118 00:06:20,160 --> 00:06:23,560 Speaker 1: You know? And also, I think your point about return 119 00:06:23,680 --> 00:06:25,880 Speaker 1: on an investment, I have this theory. This is just 120 00:06:25,920 --> 00:06:30,920 Speaker 1: my opinion that when everybody is posting something on social media, 121 00:06:32,400 --> 00:06:38,320 Speaker 1: apps like Lensa are banking on people wanting the return 122 00:06:38,360 --> 00:06:41,400 Speaker 1: of investment, being like being able to be part of that, 123 00:06:41,520 --> 00:06:43,360 Speaker 1: being able to do what their friends are doing, and 124 00:06:43,400 --> 00:06:45,479 Speaker 1: like being part of the trend. And so I think 125 00:06:45,880 --> 00:06:48,800 Speaker 1: even the inclination to just stop and be like is 126 00:06:48,880 --> 00:06:51,400 Speaker 1: this it is being is doing what everybody else is 127 00:06:51,440 --> 00:06:53,520 Speaker 1: doing on social media and wanting to be part of 128 00:06:53,520 --> 00:06:56,080 Speaker 1: the you know, part of the trend, which I understand, 129 00:06:56,160 --> 00:06:58,320 Speaker 1: Like I get caught up in that too. Is that 130 00:06:58,400 --> 00:07:01,120 Speaker 1: worth the potential risk for what I'm doing? And like, 131 00:07:01,400 --> 00:07:04,040 Speaker 1: I think that things move so quickly on the Internet 132 00:07:04,080 --> 00:07:06,880 Speaker 1: that we really have created a situation where we're inviting 133 00:07:06,920 --> 00:07:09,559 Speaker 1: people to not take that minute to actually think about 134 00:07:09,560 --> 00:07:11,680 Speaker 1: if the return is going to be worth it. I 135 00:07:11,720 --> 00:07:14,840 Speaker 1: definitely agree. I don't think that people take time to 136 00:07:14,880 --> 00:07:16,680 Speaker 1: contemplate that at all. And I think that that was 137 00:07:16,720 --> 00:07:18,800 Speaker 1: Like one of the biggest questions that I got throughout 138 00:07:18,880 --> 00:07:21,400 Speaker 1: the day today was like, well, you use apps like 139 00:07:21,440 --> 00:07:23,760 Speaker 1: Twitter and Instagram. It's like me, using these apps is 140 00:07:23,760 --> 00:07:25,720 Speaker 1: not an endorsement of these apps, it's not an endorsement 141 00:07:25,720 --> 00:07:29,360 Speaker 1: of their politics, it's not an endorsement of their owners, 142 00:07:29,760 --> 00:07:32,440 Speaker 1: so on and so forth. However, I can say we 143 00:07:32,600 --> 00:07:36,000 Speaker 1: use apps like Twitter and Instagram to communicate. When you're 144 00:07:36,000 --> 00:07:39,280 Speaker 1: looking at apps like Linsa and you look at the 145 00:07:39,320 --> 00:07:41,920 Speaker 1: fact that people are just data dumping, basically just dumping 146 00:07:41,920 --> 00:07:44,800 Speaker 1: their faces in here, paying somebody to process them and 147 00:07:44,840 --> 00:07:47,800 Speaker 1: then regurgitate an image that's either stealing from an artist 148 00:07:48,360 --> 00:07:53,000 Speaker 1: that actually painted whatever the images, or is using it 149 00:07:53,080 --> 00:07:57,200 Speaker 1: to basically broaden and train their AI and as a 150 00:07:57,240 --> 00:08:01,040 Speaker 1: result eventually sexualized or violate the person whose images it is. 151 00:08:01,280 --> 00:08:03,920 Speaker 1: I just think that the returnal investment is very different. Like, 152 00:08:04,120 --> 00:08:06,200 Speaker 1: I don't see what the purpose of paying somebody to 153 00:08:06,280 --> 00:08:09,440 Speaker 1: violate you or to take your intelectric property is personally, 154 00:08:09,720 --> 00:08:11,560 Speaker 1: And I know that we give up plenty of information 155 00:08:11,600 --> 00:08:13,560 Speaker 1: by being on like I said, these other apps, So 156 00:08:13,600 --> 00:08:15,840 Speaker 1: this is not an endorsement of them at all, but 157 00:08:15,880 --> 00:08:17,320 Speaker 1: I think that if you're going to use something, at 158 00:08:17,360 --> 00:08:21,720 Speaker 1: least have a reason besides it looked cool. Yeah, So 159 00:08:21,840 --> 00:08:26,080 Speaker 1: what are some of the ethical issues involving using apps 160 00:08:26,120 --> 00:08:29,680 Speaker 1: like lensa to do this kind of AI generation? Um? So, 161 00:08:29,720 --> 00:08:33,079 Speaker 1: some of the biggest things that I noticed, like particularly, 162 00:08:33,160 --> 00:08:37,920 Speaker 1: were one the fact that when we're talking about in 163 00:08:37,960 --> 00:08:40,400 Speaker 1: the world of like me too, or in the world 164 00:08:40,520 --> 00:08:43,760 Speaker 1: of protecting ourselves against sexual assault and violation and that 165 00:08:43,800 --> 00:08:46,840 Speaker 1: kind of stuff, the conversation around revenge point is a 166 00:08:46,880 --> 00:08:49,240 Speaker 1: really important one in this case. When you think about 167 00:08:49,280 --> 00:08:52,160 Speaker 1: the lack of legislation around a I think about how 168 00:08:52,200 --> 00:08:55,520 Speaker 1: somebody can upload any image because there's no filter on 169 00:08:55,559 --> 00:08:59,200 Speaker 1: there that says, hey, you can't upload this. Um. It 170 00:08:59,280 --> 00:09:02,960 Speaker 1: makes it easy for AI renderings of naked bodies to happen, 171 00:09:03,280 --> 00:09:09,040 Speaker 1: and somebody could absolutely use that to blackmail or harass 172 00:09:09,120 --> 00:09:11,400 Speaker 1: somebody who they've been involved with in the past. And 173 00:09:11,440 --> 00:09:13,400 Speaker 1: I mean, this isn't something that's new. People will do 174 00:09:13,400 --> 00:09:16,240 Speaker 1: this with deep fick technology already. And so when you 175 00:09:16,280 --> 00:09:19,320 Speaker 1: think about an app that's readily available at everybody's finger 176 00:09:19,360 --> 00:09:21,560 Speaker 1: tips that doesn't require you to have special knowledge like 177 00:09:21,640 --> 00:09:24,320 Speaker 1: deep fike does, UM, you think about the danger that 178 00:09:24,320 --> 00:09:27,920 Speaker 1: that perpetuates in the door that that opens for people 179 00:09:27,960 --> 00:09:31,920 Speaker 1: to be violated and to be exposing with if they 180 00:09:31,960 --> 00:09:33,960 Speaker 1: didn't consent to And so I think consent is a 181 00:09:33,960 --> 00:09:37,600 Speaker 1: really big portion of this. Also, the sexualization of people's 182 00:09:37,640 --> 00:09:40,360 Speaker 1: images at the hands of LINDSA itself, like the fact 183 00:09:40,360 --> 00:09:44,480 Speaker 1: that people can submit headshots and it returned images with 184 00:09:44,960 --> 00:09:49,439 Speaker 1: boobs or with no shirt on, or that kind of 185 00:09:49,480 --> 00:09:52,960 Speaker 1: assumes somebody's proportions or whatever the case may be, creates 186 00:09:52,960 --> 00:09:55,320 Speaker 1: a whole different problem because nobody consents to do that, 187 00:09:55,360 --> 00:09:58,520 Speaker 1: and since that's the property of LINDSA based on their 188 00:09:58,600 --> 00:10:01,760 Speaker 1: terms and conditions, those images can be misused however they 189 00:10:01,760 --> 00:10:04,720 Speaker 1: see fit and so we think about how this infringes 190 00:10:04,720 --> 00:10:07,800 Speaker 1: on people's personal autonomy, and we discourage people from using 191 00:10:07,840 --> 00:10:12,400 Speaker 1: this because people talk a lot about agency, all without 192 00:10:12,480 --> 00:10:16,480 Speaker 1: reading into how they're stripping themselves of their agency when 193 00:10:16,480 --> 00:10:19,760 Speaker 1: they participate in different trends that our present online. Another 194 00:10:19,800 --> 00:10:22,520 Speaker 1: thing is like when you look at the augmentation of 195 00:10:22,520 --> 00:10:24,840 Speaker 1: faces and so on and so forth, it will create 196 00:10:24,880 --> 00:10:27,240 Speaker 1: a lot of insecurity and so some people I had 197 00:10:27,240 --> 00:10:29,880 Speaker 1: saw it throat. I was talking about how some people 198 00:10:29,920 --> 00:10:32,840 Speaker 1: felt insecure about the results that they got from this 199 00:10:32,880 --> 00:10:36,240 Speaker 1: app in both directions. Some felt that it overly enhances 200 00:10:36,240 --> 00:10:38,080 Speaker 1: their beauty and they couldn't match it, and others feel 201 00:10:38,080 --> 00:10:41,200 Speaker 1: like it made them ugly. Body dysphoria and body dysmorphia 202 00:10:41,600 --> 00:10:44,560 Speaker 1: may increase as well, because also there's no way for 203 00:10:44,600 --> 00:10:47,480 Speaker 1: this app to know your gender, and so what happens 204 00:10:47,480 --> 00:10:49,960 Speaker 1: if somebody's gender fluid and they get misgendered using the app. 205 00:10:50,200 --> 00:10:52,320 Speaker 1: A lot of these different violences that people talk about 206 00:10:52,360 --> 00:10:55,800 Speaker 1: they want to mitigate, they've opened the door for AI 207 00:10:55,880 --> 00:10:58,680 Speaker 1: to perpetuate them, which creates a whole different problem that 208 00:10:58,720 --> 00:11:00,319 Speaker 1: I don't think a lot of people to the time 209 00:11:00,360 --> 00:11:02,960 Speaker 1: to contemplate or think about. Yeah, I was reading a 210 00:11:03,000 --> 00:11:06,479 Speaker 1: piece and Wired by Olivia Snow who wrote really compellingly 211 00:11:06,480 --> 00:11:10,160 Speaker 1: about the fact that apps like Lensa essentially made a 212 00:11:10,480 --> 00:11:13,880 Speaker 1: sexualized image of her as a child, and sort of 213 00:11:13,920 --> 00:11:18,160 Speaker 1: like the kind of problems they're in that that an 214 00:11:18,160 --> 00:11:22,800 Speaker 1: app would do that definitely, Um, I think that that 215 00:11:22,920 --> 00:11:24,640 Speaker 1: was the other thing that was really alarming to me 216 00:11:24,679 --> 00:11:26,599 Speaker 1: is when you look in like their terms of conditions 217 00:11:26,679 --> 00:11:28,640 Speaker 1: or their terms of conditions and their terms of service, 218 00:11:28,679 --> 00:11:31,320 Speaker 1: it talks about how you shouldn't upload images of miners 219 00:11:31,800 --> 00:11:34,360 Speaker 1: because there's risk of them being sexualized. And I think 220 00:11:34,480 --> 00:11:38,640 Speaker 1: that's really interesting because why is your algorithm program to 221 00:11:38,679 --> 00:11:40,480 Speaker 1: do this? What kind of training did you give your 222 00:11:40,520 --> 00:11:43,840 Speaker 1: algorithm where it sees a child's face and automatically sexualizes 223 00:11:43,920 --> 00:11:49,959 Speaker 1: the image. Additionally, um, why does the protections for sexualization 224 00:11:50,040 --> 00:11:52,240 Speaker 1: stop at miners? Why is that where your concern is? 225 00:11:52,360 --> 00:11:54,319 Speaker 1: And I think that that's also something that should be 226 00:11:54,360 --> 00:12:11,960 Speaker 1: really alarming for people. Let's take a quick break at her. 227 00:12:12,000 --> 00:12:17,360 Speaker 1: Back back in Microsoft introduced an AI chat thought called 228 00:12:17,440 --> 00:12:20,760 Speaker 1: Ta to Twitter. Kay's engineers trained Tay to have a 229 00:12:20,800 --> 00:12:23,400 Speaker 1: basic grasp of the English language. It was meant to 230 00:12:23,520 --> 00:12:26,600 Speaker 1: learn from the Internet, and eventually the plan was for 231 00:12:26,679 --> 00:12:29,600 Speaker 1: TAY to sound like the Internet, using quick memes and 232 00:12:29,679 --> 00:12:34,000 Speaker 1: jokes and Internet references, to be, as Microsoft said, the 233 00:12:34,120 --> 00:12:38,840 Speaker 1: AI with zero chill. But just a short sixteen hours 234 00:12:38,880 --> 00:12:42,679 Speaker 1: after TAY was introduced to Twitter, k started tweeting inflammatory 235 00:12:42,800 --> 00:12:47,959 Speaker 1: things like, according to the Engineering and Tech magazine I e. E. Spectrum, 236 00:12:48,000 --> 00:12:50,040 Speaker 1: I fucking hate feminist and they should all die in 237 00:12:50,080 --> 00:12:54,000 Speaker 1: Burnen Hell or Bush did nine eleven and Hitler would 238 00:12:54,000 --> 00:12:56,719 Speaker 1: have done a better job. Azaria says that this is 239 00:12:56,720 --> 00:12:59,640 Speaker 1: a good example of what happens when everybody can have 240 00:12:59,679 --> 00:13:03,560 Speaker 1: access as to certain AI tools. You mentioned early on 241 00:13:03,679 --> 00:13:06,920 Speaker 1: that you actually don't feel like certain AI technologies are 242 00:13:06,920 --> 00:13:09,400 Speaker 1: in a place where just anybody should be able to 243 00:13:09,440 --> 00:13:11,280 Speaker 1: be able to play with them like this necessarily. Do 244 00:13:11,320 --> 00:13:13,480 Speaker 1: I have that right? That's correct. I mean you can 245 00:13:13,600 --> 00:13:16,880 Speaker 1: look back at just the stability of AI itself all 246 00:13:16,880 --> 00:13:19,520 Speaker 1: the way back to Microsoft's Ta and how TA got 247 00:13:19,559 --> 00:13:21,880 Speaker 1: on Twitter and went on a racist and homophobic rant 248 00:13:22,520 --> 00:13:26,600 Speaker 1: without prompting and was like using slurs and able ism 249 00:13:26,600 --> 00:13:29,040 Speaker 1: and a whole bunch of other things several years ago, 250 00:13:29,360 --> 00:13:30,760 Speaker 1: and there was no way to check that, And I 251 00:13:30,800 --> 00:13:32,280 Speaker 1: mean they had to pull it, but they had to 252 00:13:32,320 --> 00:13:36,640 Speaker 1: do a lot of clean up for the rhetoric that 253 00:13:37,080 --> 00:13:39,920 Speaker 1: was spewing. You also have to look at things like 254 00:13:40,040 --> 00:13:45,120 Speaker 1: when Google Photos was categorizing black women as monkeys and 255 00:13:45,440 --> 00:13:48,199 Speaker 1: how that perpetuated violence, and all of these different studies 256 00:13:48,200 --> 00:13:50,920 Speaker 1: that have come out that talk about the lack of 257 00:13:51,000 --> 00:13:55,880 Speaker 1: performance of AI and the inability to be ethical or 258 00:13:56,080 --> 00:13:58,160 Speaker 1: to not be biased in the way that it functions. 259 00:13:58,160 --> 00:14:00,240 Speaker 1: And when you think about that, then you have to 260 00:14:00,280 --> 00:14:03,520 Speaker 1: ask yourself, why should we allow people who aren't trained 261 00:14:03,600 --> 00:14:08,559 Speaker 1: in ethics to freely have access to something that perpetuates 262 00:14:08,600 --> 00:14:12,080 Speaker 1: harms like this. I think that one of the most 263 00:14:12,120 --> 00:14:15,560 Speaker 1: interesting things about this whole Twitter threat is probably the 264 00:14:15,600 --> 00:14:18,640 Speaker 1: fact that when I talked about how somebody could upload 265 00:14:18,679 --> 00:14:21,440 Speaker 1: your nudes, somebody said, you're focused on the wrong thing. 266 00:14:21,600 --> 00:14:24,840 Speaker 1: That's free nudes for all of us, and I like, 267 00:14:25,520 --> 00:14:28,440 Speaker 1: and I'm like what, And so that was also like 268 00:14:28,480 --> 00:14:30,920 Speaker 1: one of my really big precautions about bringing this conversation 269 00:14:31,000 --> 00:14:32,440 Speaker 1: up is like, Yeah, there are going to be people 270 00:14:32,440 --> 00:14:35,240 Speaker 1: who gain education from this, and then they're the morally 271 00:14:35,320 --> 00:14:38,600 Speaker 1: unsound who see this as an opportunity and as inspiration 272 00:14:38,920 --> 00:14:41,560 Speaker 1: to kind of explore what that avenue is and to 273 00:14:41,640 --> 00:14:43,840 Speaker 1: figure out how they can manipulate it in order to 274 00:14:43,880 --> 00:14:46,320 Speaker 1: create these sick images. And I mean we already see 275 00:14:46,320 --> 00:14:49,960 Speaker 1: with deep fake the influx and like deep fake porn 276 00:14:50,040 --> 00:14:51,800 Speaker 1: and that kind of stuff, and how that's been used 277 00:14:51,840 --> 00:14:55,320 Speaker 1: to blackmail people, and so thinking about how now that 278 00:14:55,400 --> 00:14:58,800 Speaker 1: it's on everybody's phones, it becomes really easy for people 279 00:14:58,880 --> 00:15:01,200 Speaker 1: to violate the terms of use and there's no filter, 280 00:15:01,320 --> 00:15:03,720 Speaker 1: like I said, on this app. I also think that 281 00:15:03,760 --> 00:15:06,360 Speaker 1: them saying that this is not acceptable behavior on their 282 00:15:06,360 --> 00:15:10,520 Speaker 1: app is hilarious when their app literally does it without 283 00:15:11,240 --> 00:15:14,040 Speaker 1: without the consent of users, Like all of the women 284 00:15:14,080 --> 00:15:16,120 Speaker 1: that are posted talking about how they didn't consent to 285 00:15:16,160 --> 00:15:19,640 Speaker 1: being naked and that's what they got back from LINDSA 286 00:15:19,920 --> 00:15:23,640 Speaker 1: should be alarming to people. It should be it should 287 00:15:23,680 --> 00:15:26,880 Speaker 1: be completely alarming, But instead there are people who are 288 00:15:26,880 --> 00:15:29,720 Speaker 1: dismissing it as what's just imagination, is just a I 289 00:15:30,280 --> 00:15:32,280 Speaker 1: and then we're not even going to talk or I mean, 290 00:15:32,320 --> 00:15:35,080 Speaker 1: we can't talk about the intellectual property right violations that 291 00:15:35,120 --> 00:15:37,720 Speaker 1: comes in when we're talking about how there's so many 292 00:15:37,720 --> 00:15:41,360 Speaker 1: different artists who would be willing to do portraits for 293 00:15:41,360 --> 00:15:46,400 Speaker 1: people that are accurate and ethical for reasonable prices. But 294 00:15:46,520 --> 00:15:49,400 Speaker 1: instead you can kind of think of LINDSA as the 295 00:15:49,440 --> 00:15:53,800 Speaker 1: fast fashion of self portraits. People want fast and accessible 296 00:15:53,840 --> 00:15:56,640 Speaker 1: without thinking about the implications of the harms, and so 297 00:15:56,720 --> 00:15:59,200 Speaker 1: it directly benefits them, So don't they don't think about 298 00:15:59,280 --> 00:16:01,520 Speaker 1: how it impacts anybody else down the line, or how 299 00:16:01,560 --> 00:16:05,320 Speaker 1: to even impact them down the line, because immediate gratification 300 00:16:05,360 --> 00:16:08,800 Speaker 1: is what they're seeking. Yeah, one I saw somebody tweeting 301 00:16:08,840 --> 00:16:12,840 Speaker 1: about how when they used the lensa app it generated 302 00:16:13,280 --> 00:16:17,360 Speaker 1: very realistic imaginings of her breasts and she was like, 303 00:16:17,400 --> 00:16:20,360 Speaker 1: I didn't consent to have my like to have this 304 00:16:20,440 --> 00:16:23,600 Speaker 1: be like what's happening? And you know, I think even 305 00:16:23,640 --> 00:16:26,360 Speaker 1: beyond things like nudes, something that I noticed is that 306 00:16:26,520 --> 00:16:32,480 Speaker 1: the AI really created these very these images that adhered 307 00:16:32,800 --> 00:16:35,640 Speaker 1: very traditionally to sort of like standard beauty practice or 308 00:16:35,720 --> 00:16:39,840 Speaker 1: beauty standards. So like they lightened skin tones, they slim noses, 309 00:16:40,120 --> 00:16:43,200 Speaker 1: they make they make the women like much more sort 310 00:16:43,200 --> 00:16:46,920 Speaker 1: of traditionally like sexualized. And I think, like, what does 311 00:16:46,960 --> 00:16:49,360 Speaker 1: that tell you about the limits of AI as it's 312 00:16:49,360 --> 00:16:52,360 Speaker 1: currently being used, Because I would imagine that if you have, 313 00:16:53,160 --> 00:16:57,960 Speaker 1: you know, limitless ability to render things, having it be 314 00:16:58,200 --> 00:17:02,160 Speaker 1: so so having the things that are rendered adhere to 315 00:17:02,280 --> 00:17:07,000 Speaker 1: such traditional understandings of the limits of gender, the limits 316 00:17:07,000 --> 00:17:09,760 Speaker 1: of beauty, whatever, these very conventional attitudes. It is so 317 00:17:09,880 --> 00:17:13,520 Speaker 1: disappointing and such a like, such a it says, so 318 00:17:13,720 --> 00:17:15,760 Speaker 1: I feel like it might say something about how this 319 00:17:15,840 --> 00:17:20,439 Speaker 1: technology is being used. So I think that we should 320 00:17:20,480 --> 00:17:22,919 Speaker 1: we should definitely talk about the implicit bias aspect of it. 321 00:17:22,960 --> 00:17:25,520 Speaker 1: And this is something that I'm really passionate about. Is 322 00:17:26,160 --> 00:17:29,040 Speaker 1: technology holds the bi or algorithms hold the bias of 323 00:17:29,040 --> 00:17:31,600 Speaker 1: the programmer. So whoever's training them has the biases that 324 00:17:31,640 --> 00:17:35,119 Speaker 1: are being reproduced and regurgitated in these images. So the 325 00:17:35,200 --> 00:17:38,760 Speaker 1: hyper sexualization of women, that's because most likely the person 326 00:17:38,800 --> 00:17:43,000 Speaker 1: training these algorithms is somebody who's looking for hyper sexuality, 327 00:17:43,400 --> 00:17:48,199 Speaker 1: like like if they are if they're inputting images of 328 00:17:48,240 --> 00:17:51,199 Speaker 1: women that look promiscuous, of course it's going to regurgitate that, 329 00:17:51,760 --> 00:17:54,879 Speaker 1: Like if they're inputting images of only white women, or 330 00:17:54,920 --> 00:17:57,600 Speaker 1: only Asian women, or only women that look a certain way. 331 00:17:58,080 --> 00:18:00,760 Speaker 1: Whenever it's faced with an image that does not look 332 00:18:00,800 --> 00:18:02,440 Speaker 1: like that, it's going to try to do its best 333 00:18:02,480 --> 00:18:06,719 Speaker 1: to emulate those images while also trying to stay true. Allegedly, 334 00:18:07,119 --> 00:18:09,439 Speaker 1: to what the original image look like. But most of 335 00:18:09,480 --> 00:18:11,480 Speaker 1: the time when you see black people who have used it, 336 00:18:11,600 --> 00:18:15,560 Speaker 1: their skin is significantly lightened, or their hair texture is changed, 337 00:18:15,800 --> 00:18:18,280 Speaker 1: or a lot of different things, and so it's not 338 00:18:18,359 --> 00:18:22,040 Speaker 1: in alignment with people's actual characteristics, which I think is 339 00:18:22,080 --> 00:18:23,800 Speaker 1: also really weird, because why would you want to pay 340 00:18:23,800 --> 00:18:26,320 Speaker 1: for something that people have repeatedly told you doesn't give 341 00:18:26,320 --> 00:18:29,280 Speaker 1: accurate renderings of yourself and you're wasting your money? And 342 00:18:29,480 --> 00:18:33,000 Speaker 1: I just I don't get that one either. But more importantly, 343 00:18:33,000 --> 00:18:36,800 Speaker 1: I think that if we're going to talk about how 344 00:18:36,880 --> 00:18:41,960 Speaker 1: this is regulated and how this app itself perpetuates like 345 00:18:42,000 --> 00:18:44,879 Speaker 1: all of the modern stereotypes of what people look like, 346 00:18:44,880 --> 00:18:47,639 Speaker 1: then we have to talk about implacit bias and how 347 00:18:47,720 --> 00:18:52,600 Speaker 1: it's inherently anti black and ablest and anti trans and 348 00:18:52,640 --> 00:18:56,720 Speaker 1: a bunch of other things all within like the conversation 349 00:18:56,840 --> 00:18:59,640 Speaker 1: of what the biases of these algorithms are. I think 350 00:18:59,640 --> 00:19:01,959 Speaker 1: that's why the simplest way to put it is that 351 00:19:02,480 --> 00:19:06,320 Speaker 1: this is a much deeper black hole than just like 352 00:19:06,359 --> 00:19:09,480 Speaker 1: the Twitter thread. It's so much deeper than that, and 353 00:19:09,520 --> 00:19:14,040 Speaker 1: it requires so much more conversation and critical understanding. And 354 00:19:14,040 --> 00:19:15,919 Speaker 1: I think the other thing is people don't believe that 355 00:19:15,960 --> 00:19:18,600 Speaker 1: reading is fundamental. Like if people picked up a book 356 00:19:19,200 --> 00:19:22,680 Speaker 1: or or googled for five minutes, they could find out 357 00:19:22,680 --> 00:19:24,800 Speaker 1: all of the information they needed to tell them why 358 00:19:24,840 --> 00:19:29,520 Speaker 1: this was a bad idea. But they don't. They see 359 00:19:29,520 --> 00:19:31,880 Speaker 1: a trend and they have on the bandwagon and don't 360 00:19:31,880 --> 00:19:34,760 Speaker 1: think about what it means for them, and then they 361 00:19:34,800 --> 00:19:37,919 Speaker 1: act shocked or surprised afterwards like, oh, nobody told me, 362 00:19:38,600 --> 00:19:41,640 Speaker 1: Or they label people who bring up these conversations as 363 00:19:42,080 --> 00:19:46,280 Speaker 1: lead eyes who are anti technological development and that kind 364 00:19:46,280 --> 00:19:48,200 Speaker 1: of stuff. And I think that that's ridiculous as well, 365 00:19:48,240 --> 00:19:50,160 Speaker 1: because most of the people who are critiguing this are 366 00:19:50,160 --> 00:20:04,439 Speaker 1: people who are in the industry more. After a quick break, 367 00:20:07,160 --> 00:20:10,080 Speaker 1: let's get right back into it. Something that we talk 368 00:20:10,119 --> 00:20:12,600 Speaker 1: about a lot on the show are that it really 369 00:20:12,640 --> 00:20:16,520 Speaker 1: seems to be marginalized people, in this case, black women 370 00:20:16,560 --> 00:20:19,760 Speaker 1: who are doing a lot of the I don't know 371 00:20:20,520 --> 00:20:23,639 Speaker 1: questioning of technology, like like AI and the way that 372 00:20:23,680 --> 00:20:25,760 Speaker 1: it is being used to cause harm. Do you find 373 00:20:25,760 --> 00:20:27,440 Speaker 1: that to be the case, like in the field, the 374 00:20:27,480 --> 00:20:30,879 Speaker 1: people who are sort of asking the poking questions about 375 00:20:30,960 --> 00:20:33,359 Speaker 1: ethics in AI, do you find that that tends to 376 00:20:33,359 --> 00:20:36,280 Speaker 1: be people who are marginalized or even more specifically like 377 00:20:36,280 --> 00:20:39,120 Speaker 1: black women. I definitely do. I think that oftentimes we're 378 00:20:39,160 --> 00:20:41,479 Speaker 1: expected to be the ones who bear the burden of 379 00:20:41,600 --> 00:20:45,320 Speaker 1: accountability when it comes to how we engage with these technologies. 380 00:20:45,400 --> 00:20:49,679 Speaker 1: I think that in a field like electrical engineering, for example, 381 00:20:49,720 --> 00:20:53,040 Speaker 1: when there's less than three percent of us in the field, 382 00:20:53,400 --> 00:20:57,360 Speaker 1: oftentimes we are we are doing a liquork. We are 383 00:20:57,400 --> 00:21:00,359 Speaker 1: talking about the anti blackness, we are talking about the 384 00:21:00,440 --> 00:21:03,040 Speaker 1: hyper sexualization of black women. We are talking about this 385 00:21:03,200 --> 00:21:06,080 Speaker 1: categorization and the other biases that are present. And I 386 00:21:06,119 --> 00:21:08,480 Speaker 1: don't think that there's anybody else that really advocates for 387 00:21:08,560 --> 00:21:11,840 Speaker 1: these things because it's not a part of their interest, right. 388 00:21:12,160 --> 00:21:14,119 Speaker 1: So these are the same people who are programming and 389 00:21:14,160 --> 00:21:17,720 Speaker 1: developing these algorithms. Of course they don't see a flaw 390 00:21:17,800 --> 00:21:19,760 Speaker 1: in the ways in which they function. And I think 391 00:21:19,800 --> 00:21:21,879 Speaker 1: that this goes all the way back to just like 392 00:21:21,920 --> 00:21:24,359 Speaker 1: my senior research when I was in college, when I 393 00:21:24,400 --> 00:21:26,240 Speaker 1: was working on my project and I had a professor 394 00:21:26,240 --> 00:21:28,840 Speaker 1: who told me that implicit bias didn't actually exist and 395 00:21:28,840 --> 00:21:31,439 Speaker 1: that I was creating a problem in my head. And 396 00:21:31,480 --> 00:21:33,480 Speaker 1: then when I provided evidence of this, he tried to 397 00:21:33,480 --> 00:21:35,720 Speaker 1: gas like me and told me that it was made 398 00:21:35,760 --> 00:21:38,320 Speaker 1: up evidence, until all of a sudden he decided it 399 00:21:38,359 --> 00:21:39,960 Speaker 1: was good research. And then he said that he was 400 00:21:40,000 --> 00:21:42,800 Speaker 1: going to publish his own study on this using my 401 00:21:42,880 --> 00:21:45,399 Speaker 1: research and my results, which I thought was really wild. 402 00:21:45,840 --> 00:21:50,000 Speaker 1: And so I think that oftentimes Black women are expected 403 00:21:50,040 --> 00:21:52,080 Speaker 1: to be like the mules. We're supposed to do all 404 00:21:52,119 --> 00:21:54,320 Speaker 1: of the heavy work, and then somebody else gets the gratification, 405 00:21:54,400 --> 00:21:56,840 Speaker 1: somebody else gets the credit. And I think that's true 406 00:21:56,840 --> 00:22:00,719 Speaker 1: across all different disciplines in life. Just like a lot 407 00:22:00,760 --> 00:22:03,080 Speaker 1: of times black women are the trailblazers, especially as the 408 00:22:03,080 --> 00:22:07,280 Speaker 1: most educated or college educated demographic in the nation, we 409 00:22:08,119 --> 00:22:12,160 Speaker 1: are oftentimes sought out to provide the knowledge, but then 410 00:22:12,200 --> 00:22:16,080 Speaker 1: not credited or given the resources in order to expand 411 00:22:16,119 --> 00:22:19,440 Speaker 1: on our knowledge sets because we've given people what they 412 00:22:19,440 --> 00:22:21,240 Speaker 1: want and they've milked us for what we have to offer, 413 00:22:21,280 --> 00:22:23,679 Speaker 1: and then they think that they can discard us. I 414 00:22:23,720 --> 00:22:26,919 Speaker 1: think that that's so. I mean, who gets the credit 415 00:22:27,200 --> 00:22:29,320 Speaker 1: is also who has the power, right, And so you're 416 00:22:29,640 --> 00:22:32,439 Speaker 1: your professor being like, no, no, no, that doesn't exist, 417 00:22:32,520 --> 00:22:35,880 Speaker 1: and then when he feels sufficiently like it does exist, Oh, 418 00:22:35,880 --> 00:22:37,840 Speaker 1: now I'm going to write my own paper about it, 419 00:22:37,880 --> 00:22:39,960 Speaker 1: and so that I'll get the credit for pointing out 420 00:22:40,000 --> 00:22:41,640 Speaker 1: something that has spent the last couple of weeks gas 421 00:22:41,760 --> 00:22:44,040 Speaker 1: lighting you about saying doesn't exist. You know, It's like 422 00:22:44,440 --> 00:22:49,720 Speaker 1: that is clearly a a a push pull about who 423 00:22:49,760 --> 00:22:52,640 Speaker 1: has power, who has legitimacy, and that matters so much 424 00:22:52,720 --> 00:22:55,639 Speaker 1: in these fields. It definitely does, especially when you have 425 00:22:55,720 --> 00:22:58,159 Speaker 1: to lobby for a seat at the table in the 426 00:22:58,160 --> 00:23:00,960 Speaker 1: first place. It's like AI was, in itself is already 427 00:23:01,040 --> 00:23:04,840 Speaker 1: something that's, like I said, very biased, and it's not 428 00:23:05,000 --> 00:23:08,200 Speaker 1: very inclusionary, even in terms of who's developing the algorithms. 429 00:23:08,240 --> 00:23:11,840 Speaker 1: And so you have organizations like the Algorithm Justice League 430 00:23:11,840 --> 00:23:14,240 Speaker 1: that are working tirelessly, which was also founded by a 431 00:23:14,280 --> 00:23:19,520 Speaker 1: black woman. Um you have UM what else? You have 432 00:23:19,600 --> 00:23:22,840 Speaker 1: a j O, You have black and AI and a 433 00:23:22,840 --> 00:23:24,840 Speaker 1: whole bunch of other different words that are trying to 434 00:23:24,880 --> 00:23:28,360 Speaker 1: bring attention to the ethical imbalances that are present within 435 00:23:28,440 --> 00:23:32,359 Speaker 1: AI and development. And the work is taken and people 436 00:23:32,400 --> 00:23:36,280 Speaker 1: celebrate it. However, oftentimes the people who are behind the 437 00:23:36,280 --> 00:23:39,320 Speaker 1: work are overlooked. And I don't think that there's really 438 00:23:39,600 --> 00:23:43,320 Speaker 1: ever space for conversations around this unless we're creating the 439 00:23:43,320 --> 00:23:48,520 Speaker 1: conversations ourselves, which is pretty terrible and not all not 440 00:23:48,640 --> 00:23:52,400 Speaker 1: all that encouraging, And so I think that now more 441 00:23:52,440 --> 00:23:55,800 Speaker 1: than ever, this is when the black women who are 442 00:23:55,800 --> 00:23:57,560 Speaker 1: in the field I really have to rally around each 443 00:23:57,560 --> 00:23:59,840 Speaker 1: other and rally around our youth in order to develop 444 00:24:00,320 --> 00:24:04,000 Speaker 1: a wave of people who are equally as passionate and 445 00:24:04,600 --> 00:24:09,760 Speaker 1: equally as qualified to create the ripples in this industry, 446 00:24:09,960 --> 00:24:11,720 Speaker 1: because nobody else is going to do it if we're 447 00:24:11,720 --> 00:24:18,160 Speaker 1: not facilitating the conversations A fucking men. Are there alternatives 448 00:24:18,160 --> 00:24:21,480 Speaker 1: for folks who might be interested in checking out AI 449 00:24:21,560 --> 00:24:23,920 Speaker 1: and like AI arts, but doing someone a way that 450 00:24:24,000 --> 00:24:26,960 Speaker 1: might be safer or less harmful, or might have less 451 00:24:26,960 --> 00:24:34,840 Speaker 1: ethical implications. I personally will not list any other apps 452 00:24:34,880 --> 00:24:39,200 Speaker 1: because I feel that there's risks associated with all of them, 453 00:24:39,240 --> 00:24:41,879 Speaker 1: and like everything, I mean, every app that we use 454 00:24:41,920 --> 00:24:45,440 Speaker 1: has a risk. But I'm not necessarily here to endorse 455 00:24:45,960 --> 00:24:50,040 Speaker 1: any AI based art generator. I would encourage people to 456 00:24:50,080 --> 00:24:53,760 Speaker 1: learn about building algorithms themselves, learn about how to input 457 00:24:53,880 --> 00:24:56,600 Speaker 1: something and get an output on your own, as opposed 458 00:24:56,640 --> 00:24:59,879 Speaker 1: to trusting a system and handing over your data to somebody. 459 00:25:00,080 --> 00:25:07,160 Speaker 1: There's plenty of beginner algorithm development works workshops and different 460 00:25:07,480 --> 00:25:09,760 Speaker 1: tutorials that you can access online that you can do 461 00:25:09,840 --> 00:25:13,440 Speaker 1: on your own. Um And I encourage people to learn 462 00:25:13,480 --> 00:25:16,160 Speaker 1: the fundamentals of it actually before trying to get results 463 00:25:16,160 --> 00:25:18,560 Speaker 1: that they can post on social media. That's just me personally. 464 00:25:19,000 --> 00:25:21,320 Speaker 1: I know that people in the thread did list some 465 00:25:21,359 --> 00:25:24,480 Speaker 1: of their like ideal creations. I think that acts like 466 00:25:24,600 --> 00:25:27,240 Speaker 1: mid Journey that people are using for prop based image 467 00:25:27,240 --> 00:25:30,919 Speaker 1: generation have their own risks and have their own harms. 468 00:25:31,000 --> 00:25:32,600 Speaker 1: I also think that if you look at some of 469 00:25:32,680 --> 00:25:35,399 Speaker 1: the art, it's really creepy, and you're better off just 470 00:25:35,480 --> 00:25:38,440 Speaker 1: paying people who paint for real to do these images 471 00:25:38,880 --> 00:25:41,240 Speaker 1: like some of the have you seen AI hands and 472 00:25:41,359 --> 00:25:46,439 Speaker 1: fingers creepy and they are very creepy. And so I 473 00:25:46,480 --> 00:25:49,199 Speaker 1: think that, like I said, people should just explore like 474 00:25:49,200 --> 00:25:54,000 Speaker 1: the fundamentals of algorithms, understanding layers, understanding outputs, understanding bounding 475 00:25:54,040 --> 00:25:57,760 Speaker 1: boxes simplistically, um And I think that knowledge about that 476 00:25:58,440 --> 00:26:04,119 Speaker 1: would one make them more intentional when it comes to 477 00:26:04,160 --> 00:26:07,040 Speaker 1: how they choose to interact with AI, but then also 478 00:26:07,440 --> 00:26:09,280 Speaker 1: give them the skills that that they need in order 479 00:26:09,320 --> 00:26:11,199 Speaker 1: to create the art that they're seeking to create on 480 00:26:11,200 --> 00:26:14,240 Speaker 1: their own without infringing on the rights of other creatives 481 00:26:14,280 --> 00:26:16,840 Speaker 1: that are working tirelessly to paint real images or to 482 00:26:16,920 --> 00:26:20,879 Speaker 1: create real digital art without using AI generators. I just 483 00:26:20,880 --> 00:26:23,439 Speaker 1: think that there's a way that we can do this 484 00:26:23,560 --> 00:26:27,320 Speaker 1: that doesn't violate people, or there's ways that people can 485 00:26:27,359 --> 00:26:30,960 Speaker 1: participate without violating other people. And if you have a 486 00:26:31,040 --> 00:26:33,600 Speaker 1: genuine interest in learning it, then I encourage people to 487 00:26:33,640 --> 00:26:35,919 Speaker 1: seek that out as opposed to trying to hop on 488 00:26:35,960 --> 00:26:39,480 Speaker 1: the fast wave and using apps like Lensa or whatever 489 00:26:39,560 --> 00:26:42,800 Speaker 1: other AI generators are really popular right now. Yeah, I 490 00:26:42,800 --> 00:26:44,600 Speaker 1: will say about the you know, you brought up the 491 00:26:44,640 --> 00:26:48,159 Speaker 1: intellectual property rights issues. I did see a couple of 492 00:26:48,320 --> 00:26:51,719 Speaker 1: the Lensa generated images where if you looked real close 493 00:26:51,840 --> 00:26:54,720 Speaker 1: in the bottom corner, you would see a watermark or 494 00:26:54,800 --> 00:26:57,960 Speaker 1: like a like a signature. And so we have this 495 00:26:58,080 --> 00:27:01,200 Speaker 1: idea that, oh, this is just being generated I uh, 496 00:27:01,440 --> 00:27:04,480 Speaker 1: computer artist robots somewhere, and it's like, no, if it 497 00:27:04,520 --> 00:27:08,400 Speaker 1: has a signature, they've just lifted this image from another 498 00:27:08,840 --> 00:27:12,399 Speaker 1: human artist and maybe changed a little something about it, 499 00:27:12,440 --> 00:27:14,600 Speaker 1: and now they're giving it back to you and calling 500 00:27:14,640 --> 00:27:17,440 Speaker 1: it a high art. Absolutely, And I think that that's 501 00:27:17,480 --> 00:27:19,280 Speaker 1: another big thing right now is that a lot of 502 00:27:19,359 --> 00:27:22,840 Speaker 1: artists are fighting for their rights to their own images 503 00:27:22,920 --> 00:27:25,959 Speaker 1: after their artwork has been fed into the AI generators, 504 00:27:25,960 --> 00:27:28,480 Speaker 1: and the generators are producing something that's a little bit 505 00:27:28,520 --> 00:27:32,520 Speaker 1: more enhanced, and then people are publishing it and selling 506 00:27:32,520 --> 00:27:34,600 Speaker 1: it as their own and AI R I mean, the 507 00:27:34,600 --> 00:27:37,520 Speaker 1: original artist isn't getting any of their credits and being 508 00:27:37,600 --> 00:27:41,440 Speaker 1: robbed of their i P and their copyright because well, 509 00:27:41,560 --> 00:27:45,680 Speaker 1: AI generated eight and so they're allowed to evade responsibility 510 00:27:45,680 --> 00:27:48,760 Speaker 1: for infringing on that person's work. That's another problem that 511 00:27:50,040 --> 00:27:52,399 Speaker 1: I think correlates to all of this is that the 512 00:27:52,440 --> 00:27:57,600 Speaker 1: output images technically don't have owners, like, yes, the app 513 00:27:57,680 --> 00:28:01,280 Speaker 1: owns it, but there's no individual person that gains rights 514 00:28:01,320 --> 00:28:04,800 Speaker 1: over these images. And that means that there's no identity 515 00:28:04,840 --> 00:28:11,119 Speaker 1: associated with who maybe train the AI to turn people's 516 00:28:11,119 --> 00:28:15,200 Speaker 1: images into nudes, or who input somebody's nudes into these 517 00:28:15,240 --> 00:28:18,959 Speaker 1: images or publish them accordingly. I think that it allows 518 00:28:18,960 --> 00:28:23,840 Speaker 1: people to dodge responsibility a lot, and it violates other 519 00:28:23,880 --> 00:28:28,840 Speaker 1: people who are actively trying to establish themselves because they're 520 00:28:28,840 --> 00:28:33,119 Speaker 1: passionate about real art. Because people want to participate in 521 00:28:33,160 --> 00:28:36,360 Speaker 1: a fad that doesn't really have any net benefit besides 522 00:28:37,119 --> 00:28:40,360 Speaker 1: fast images. But I don't understand why people would want 523 00:28:40,400 --> 00:28:47,160 Speaker 1: images that violate the rights and violate ethics. How can 524 00:28:47,200 --> 00:28:50,400 Speaker 1: folks get a better understanding of AI. I just encourage 525 00:28:50,440 --> 00:28:53,800 Speaker 1: people to read books and and if you if you 526 00:28:53,840 --> 00:28:56,880 Speaker 1: want to learn more, read read. I think that there's 527 00:28:56,920 --> 00:28:58,840 Speaker 1: so many different studies that are out right now. So 528 00:28:58,960 --> 00:29:01,600 Speaker 1: many different people have been screaming at the top of 529 00:29:01,600 --> 00:29:05,520 Speaker 1: their lungs how we have to be cognizant of the 530 00:29:05,520 --> 00:29:08,760 Speaker 1: ways in which AI is used and how we propagated 531 00:29:08,800 --> 00:29:11,160 Speaker 1: in society, and a lot of people have not done 532 00:29:11,200 --> 00:29:15,240 Speaker 1: the reading and our mindlessly participating in different social media 533 00:29:15,280 --> 00:29:18,280 Speaker 1: trends without understanding how they perpetuate the harms that they 534 00:29:18,280 --> 00:29:22,080 Speaker 1: claim that they want to mitigate, whether it be child 535 00:29:22,080 --> 00:29:26,080 Speaker 1: pornography or different industries that infringe upon the rights and 536 00:29:26,160 --> 00:29:28,120 Speaker 1: the autonomy of other people. I don't think that people 537 00:29:28,760 --> 00:29:33,760 Speaker 1: are paying acute attention to how they are exacerbating these problems. 538 00:29:33,840 --> 00:29:36,600 Speaker 1: And so when we facilitate a conversation around that and 539 00:29:36,840 --> 00:29:39,560 Speaker 1: we are intentional about learning about these things, I think 540 00:29:39,600 --> 00:29:44,600 Speaker 1: that it creates a more safe interaction with AI. I 541 00:29:44,640 --> 00:29:46,640 Speaker 1: won't say it makes AI a safe place, because I 542 00:29:46,680 --> 00:29:49,080 Speaker 1: don't think that that exists at this time, but it 543 00:29:49,120 --> 00:29:53,040 Speaker 1: does make our interactions with AI a little bit safer. Well, 544 00:29:53,040 --> 00:29:57,560 Speaker 1: I'm so glad that you're out there like poking everyone 545 00:29:57,600 --> 00:29:59,560 Speaker 1: to be a little bit more informed and asking the 546 00:29:59,640 --> 00:30:03,600 Speaker 1: questions to make this very very powerful technology a little 547 00:30:03,640 --> 00:30:07,480 Speaker 1: bit more ethical. Is there a place where folks can 548 00:30:07,640 --> 00:30:10,720 Speaker 1: follow all the work that you're up to. So I 549 00:30:10,920 --> 00:30:14,040 Speaker 1: have a tech Instagram that I'll be posting kind of 550 00:30:14,080 --> 00:30:17,080 Speaker 1: like this good and continuing this conversation on It's called 551 00:30:17,120 --> 00:30:22,280 Speaker 1: public phocology. P O L Y s A c O 552 00:30:24,080 --> 00:30:28,960 Speaker 1: l O g Y. Sorry I can't still. And then 553 00:30:29,680 --> 00:30:33,640 Speaker 1: my regular at is at melan and elevating M E 554 00:30:33,800 --> 00:30:36,040 Speaker 1: l A N I N E L E v A 555 00:30:36,160 --> 00:30:43,560 Speaker 1: T I N got a story about an interesting thing 556 00:30:43,560 --> 00:30:45,600 Speaker 1: in tech. I just want to say Hi. You can 557 00:30:45,680 --> 00:30:47,720 Speaker 1: reach just at Hello at tang godi dot com. You 558 00:30:47,760 --> 00:30:50,960 Speaker 1: can also find transcripts for today's episode at tangdi dot com. 559 00:30:50,960 --> 00:30:52,640 Speaker 1: There Are No Girls on the Internet was created by 560 00:30:52,640 --> 00:30:55,200 Speaker 1: me bridgetad. It's a production of I Heart Radio and 561 00:30:55,280 --> 00:30:59,200 Speaker 1: Unboss creative Jonathan Strickland as our executive producer. Tara Harrison 562 00:30:59,240 --> 00:31:02,920 Speaker 1: is our producer. Sound engineer Michael Amato is our contributing producer. 563 00:31:03,200 --> 00:31:05,880 Speaker 1: I'm your host, Bridget Todd If you want to help 564 00:31:05,960 --> 00:31:08,880 Speaker 1: us grow, rate and review us on Apple Podcasts. For 565 00:31:09,000 --> 00:31:11,160 Speaker 1: more podcasts from i heeart Radio, check out the iHeart 566 00:31:11,240 --> 00:31:13,760 Speaker 1: Radio app, Apple podcast or wherever you get your podcasts.