1
00:00:05,200 --> 00:00:07,640
Speaker 1: Hey, this is Annie and Smantha and welcome to stuff

2
00:00:07,640 --> 00:00:19,040
Speaker 1: I never told you Production by Heart Radio, and today

3
00:00:19,160 --> 00:00:22,080
Speaker 1: we are so happy to once again be joined by

4
00:00:22,320 --> 00:00:26,240
Speaker 1: the generous, the genuine Bridget Todd, the real deal, as

5
00:00:26,280 --> 00:00:28,640
Speaker 1: they might say, Welcome Bridget.

6
00:00:29,000 --> 00:00:34,280
Speaker 2: Thank you. I love the G theme. I'll take it.

7
00:00:33,240 --> 00:00:36,320
Speaker 1: Yes, I did think about this one beforehand. Sometimes I

8
00:00:36,440 --> 00:00:37,000
Speaker 1: just come a little.

9
00:00:37,159 --> 00:00:42,479
Speaker 3: But I will admit your skin looks amazing. Oh the

10
00:00:42,520 --> 00:00:44,960
Speaker 3: glowing here, That's what I'm saying of the G theme. Hello,

11
00:00:45,080 --> 00:00:52,480
Speaker 3: my God, keep complimenting me. Yes, please, well, Bridget, how

12
00:00:52,560 --> 00:00:53,040
Speaker 3: have you been?

13
00:00:53,159 --> 00:00:54,960
Speaker 1: Happy? Belated birthday?

14
00:00:55,160 --> 00:00:58,280
Speaker 2: Thank you? Thank you. My birthday is on Pie Day,

15
00:00:58,480 --> 00:01:01,520
Speaker 2: which I always love to throw out. When I was

16
00:01:01,520 --> 00:01:04,280
Speaker 2: a kid, I would have pie. Even though I'm not

17
00:01:04,319 --> 00:01:06,399
Speaker 2: a huge pie person, I would have pie instead of

18
00:01:06,400 --> 00:01:07,440
Speaker 2: cake on my birthday.

19
00:01:08,280 --> 00:01:12,320
Speaker 3: I prefer cake. Really, you, Annie, the face.

20
00:01:12,200 --> 00:01:15,800
Speaker 2: Samantha just made when I said I prefer cake, you looked.

21
00:01:15,880 --> 00:01:17,040
Speaker 2: You looked horrified.

22
00:01:17,240 --> 00:01:20,479
Speaker 3: No, I just know Annie's stance on this, and she's

23
00:01:20,520 --> 00:01:22,200
Speaker 3: a pie girl. So oh you knew.

24
00:01:22,319 --> 00:01:26,000
Speaker 1: I was on a podcast that argued pie versus cake

25
00:01:26,040 --> 00:01:26,679
Speaker 1: and I won.

26
00:01:28,040 --> 00:01:29,880
Speaker 2: What were the merits of pie over cake?

27
00:01:31,200 --> 00:01:35,040
Speaker 1: Bridget? I can direct you to the podcast otherwise I

28
00:01:35,080 --> 00:01:37,119
Speaker 1: have a thirty minute fight right now.

29
00:01:38,000 --> 00:01:40,080
Speaker 2: I love that this is I love that you have

30
00:01:40,200 --> 00:01:42,960
Speaker 2: such a like a clarified say, I go, I got

31
00:01:43,040 --> 00:01:45,320
Speaker 2: chuck cake. Here's a resource for you, you know.

32
00:01:46,880 --> 00:01:50,560
Speaker 1: But if that's your preference, then I salute you.

33
00:01:50,880 --> 00:01:53,280
Speaker 3: What kind of cake though? What's your favorite?

34
00:01:53,600 --> 00:01:57,600
Speaker 2: Ooh, what a good question. I'm a pretty traditional cake person.

35
00:01:58,120 --> 00:02:00,560
Speaker 2: First of all, there's no cake that I will not try.

36
00:02:01,040 --> 00:02:04,920
Speaker 2: I like the yellow cake with chocolate icing, like I

37
00:02:05,440 --> 00:02:10,640
Speaker 2: like cream cheese icing. I love. I am not above

38
00:02:11,000 --> 00:02:15,760
Speaker 2: just a Duncan Hines box mix funfetti cake. Honestly all

39
00:02:15,840 --> 00:02:19,360
Speaker 2: cake I am. But you can't see it. But behind

40
00:02:19,400 --> 00:02:22,160
Speaker 2: me there's a chocolate cream cheese cake on my counter.

41
00:02:22,480 --> 00:02:24,840
Speaker 2: And when I'm done with all my calls and interviews

42
00:02:24,840 --> 00:02:26,600
Speaker 2: and stuff, that's gonna be a slice of that cake

43
00:02:26,680 --> 00:02:28,800
Speaker 2: is going to be my reward for that for cheesecake.

44
00:02:28,840 --> 00:02:30,640
Speaker 3: So it is cheesecake? You considered that a cake?

45
00:02:30,960 --> 00:02:32,640
Speaker 2: Or Oh?

46
00:02:32,720 --> 00:02:34,440
Speaker 3: I'm just gonna ask you all the questions. What a

47
00:02:34,480 --> 00:02:38,440
Speaker 3: good question, because I that's how I would celebrate my birthday.

48
00:02:38,440 --> 00:02:42,080
Speaker 3: I traditionally my mother would make me a chocolate mini

49
00:02:42,120 --> 00:02:45,600
Speaker 3: chocolate chip cheesecake for my birthday. It was very specific,

50
00:02:46,000 --> 00:02:47,560
Speaker 3: and then from then on she would get me a

51
00:02:47,639 --> 00:02:50,560
Speaker 3: cheesecake of some sort that she was like testing out.

52
00:02:51,480 --> 00:02:53,600
Speaker 3: But I did like The debate was is this an

53
00:02:53,639 --> 00:02:56,320
Speaker 3: actual cake? Could this be considered cake?

54
00:02:56,720 --> 00:02:59,800
Speaker 2: I don't know that I would call it cake because

55
00:02:59,800 --> 00:03:02,280
Speaker 2: the time that I made a cheesecake, I don't think

56
00:03:02,360 --> 00:03:05,080
Speaker 2: I baked it. I don't, like, really, I think that

57
00:03:05,360 --> 00:03:10,160
Speaker 2: baking might be something that has to happen for it

58
00:03:10,200 --> 00:03:12,040
Speaker 2: to be a cake. I believe the time that I

59
00:03:12,080 --> 00:03:14,919
Speaker 2: made a cheesecake it was no bake. Also, it kind

60
00:03:14,919 --> 00:03:16,480
Speaker 2: of almost put me off cheesecake. I don't know if

61
00:03:16,480 --> 00:03:19,000
Speaker 2: you ever made one from scratch, but the ingredients are

62
00:03:19,040 --> 00:03:23,320
Speaker 2: like eight sticks of butter, eight things of sour cream,

63
00:03:23,400 --> 00:03:25,440
Speaker 2: eight things that cream cheese are like you never think

64
00:03:25,440 --> 00:03:27,919
Speaker 2: about what actually goes into a cheesecake, and then when

65
00:03:27,919 --> 00:03:30,000
Speaker 2: you actually make one, you're like, wow, there's a lot

66
00:03:30,040 --> 00:03:31,120
Speaker 2: of sugar and cream.

67
00:03:31,600 --> 00:03:34,080
Speaker 3: No, it's intense. So the kind of cheesecakes I would

68
00:03:34,080 --> 00:03:37,160
Speaker 3: take I would bake is literal as it would be baked,

69
00:03:37,280 --> 00:03:39,200
Speaker 3: but you had to have the spring form pan and

70
00:03:39,320 --> 00:03:42,880
Speaker 3: the chick time not to let it crack. That was

71
00:03:42,920 --> 00:03:46,440
Speaker 3: the true test. I never did have a perfect cheesecake. No.

72
00:03:46,680 --> 00:03:48,440
Speaker 2: This is why I do the no bake, because I

73
00:03:48,640 --> 00:03:52,480
Speaker 2: that's too much. Your kudos to your mom for figuring

74
00:03:52,520 --> 00:03:53,720
Speaker 2: that out is really good?

75
00:03:54,080 --> 00:03:59,280
Speaker 1: Oh yes, well cake cake talk aside? You earn self

76
00:03:59,280 --> 00:04:01,520
Speaker 1: By Southwest? How was that?

77
00:04:01,920 --> 00:04:06,040
Speaker 2: It was really good? I will say I've been a lot.

78
00:04:06,600 --> 00:04:09,320
Speaker 2: I've been going to south By Southwest on and off

79
00:04:09,360 --> 00:04:11,360
Speaker 2: for a long time, and I really enjoy it. I

80
00:04:11,440 --> 00:04:14,640
Speaker 2: usually spent my birthday there. This time I also spent

81
00:04:14,680 --> 00:04:19,120
Speaker 2: my birthday there. I've seen it go from a kind

82
00:04:19,120 --> 00:04:24,920
Speaker 2: of hip, little arts and music forward festival. I think

83
00:04:24,960 --> 00:04:28,440
Speaker 2: that there was a time where it got really tech

84
00:04:28,480 --> 00:04:32,800
Speaker 2: bro heavy, where they added in tech as its own activation,

85
00:04:32,960 --> 00:04:35,560
Speaker 2: and like everything was tech. I remember going the year

86
00:04:35,560 --> 00:04:39,360
Speaker 2: that everybody was talking about NFTs and it was every

87
00:04:39,400 --> 00:04:42,560
Speaker 2: every session, every talk, every this, every that, every activation

88
00:04:42,720 --> 00:04:45,240
Speaker 2: was NFTs. And then going the next year and it

89
00:04:45,279 --> 00:04:48,000
Speaker 2: was like NFTs, who we don't know her? So this

90
00:04:48,200 --> 00:04:52,200
Speaker 2: year they dialed down the tech a little bit, and

91
00:04:52,240 --> 00:04:54,400
Speaker 2: I think, even as a tech person, I think that

92
00:04:54,520 --> 00:04:55,960
Speaker 2: was good. I think that was the right choice. Just

93
00:04:56,000 --> 00:05:00,880
Speaker 2: little more film, more culture, more art, more music. Tech

94
00:05:01,000 --> 00:05:03,240
Speaker 2: was still there. People like me were still there, but

95
00:05:03,960 --> 00:05:07,320
Speaker 2: I think it felt good having you know, you don't

96
00:05:07,320 --> 00:05:08,560
Speaker 2: want to go to these events and have it to

97
00:05:08,720 --> 00:05:13,360
Speaker 2: be an AI hype machine, and oftentimes that's what it's

98
00:05:13,360 --> 00:05:15,279
Speaker 2: felt like. And it felt a little that was dialed.

99
00:05:15,320 --> 00:05:17,760
Speaker 2: It was still there, but dialed down to the point

100
00:05:17,800 --> 00:05:19,080
Speaker 2: of Okay, I can tolerate this.

101
00:05:19,960 --> 00:05:23,080
Speaker 3: Didn't they do something special with podcast movement? Like they

102
00:05:23,600 --> 00:05:24,640
Speaker 3: joined together? Right?

103
00:05:24,680 --> 00:05:28,839
Speaker 2: They did? Podcast Movement happened at at south By Southwest

104
00:05:28,880 --> 00:05:32,799
Speaker 2: This year they had the iHeartRadio Podcast Awards from south

105
00:05:32,880 --> 00:05:36,080
Speaker 2: By Southwest. This year, I Heart had a whole activation

106
00:05:36,240 --> 00:05:39,800
Speaker 2: the iHeart Podcast Hotel. I got to see some familiar

107
00:05:39,800 --> 00:05:42,720
Speaker 2: faces from the iHeart fam. Yeah. So it was just

108
00:05:42,760 --> 00:05:50,680
Speaker 2: one big amalgamation of south By podcasts, podcast events, podcast awards.

109
00:05:50,760 --> 00:05:54,080
Speaker 2: I met Killer Mike at one of the events. Ramsey,

110
00:05:54,120 --> 00:05:57,360
Speaker 2: who we all know, was there and I was like,

111
00:05:57,520 --> 00:06:00,400
Speaker 2: oh my god. Ramsey shot up for a second. I

112
00:06:00,440 --> 00:06:02,400
Speaker 2: think that's Killer Mike and he was like, not only

113
00:06:02,480 --> 00:06:05,400
Speaker 2: is that Killer Mike, I am working on a podcast

114
00:06:05,400 --> 00:06:07,440
Speaker 2: with Killer Mike, and like, would you like to meet him?

115
00:06:07,880 --> 00:06:10,880
Speaker 2: I'm rarely starstruck. It is rare in my life, and

116
00:06:10,960 --> 00:06:13,960
Speaker 2: I'm like, this person is someone that I really like

117
00:06:14,000 --> 00:06:17,880
Speaker 2: and respect. I think Killer Mike might think I'm nonverbal

118
00:06:18,040 --> 00:06:19,800
Speaker 2: like that. That's how the interaction went.

119
00:06:21,960 --> 00:06:23,920
Speaker 3: Like he is an Atlanta staple, like he is.

120
00:06:23,960 --> 00:06:25,000
Speaker 2: Have you seen him around?

121
00:06:25,760 --> 00:06:27,840
Speaker 3: Yes, so I've seen him around doing a few things

122
00:06:27,839 --> 00:06:28,360
Speaker 3: here and there.

123
00:06:28,480 --> 00:06:31,960
Speaker 1: Yeah, oh my god, you see him about town?

124
00:06:32,360 --> 00:06:34,760
Speaker 3: Yeah, he got his places to go.

125
00:06:35,480 --> 00:06:35,720
Speaker 1: Uh.

126
00:06:35,800 --> 00:06:39,440
Speaker 3: We started watching some YouTube series and they came to

127
00:06:39,480 --> 00:06:42,040
Speaker 3: Atlanta and they highlighted Killer Mike because he's such a

128
00:06:42,279 --> 00:06:47,000
Speaker 3: staple of him of Atlanta. But yes, he is, he's

129
00:06:47,040 --> 00:06:47,960
Speaker 3: a man about town.

130
00:06:49,279 --> 00:06:51,840
Speaker 2: Thinking about it, I guess because you all live in Atlanta,

131
00:06:51,920 --> 00:06:55,479
Speaker 2: it's probably a lot less exciting to see people like

132
00:06:55,480 --> 00:06:57,320
Speaker 2: you've probably seen people like that around all the time,

133
00:06:57,680 --> 00:06:59,840
Speaker 2: and it's probably not the same as it is to

134
00:06:59,880 --> 00:07:03,560
Speaker 2: me where the only living in DC. The only celebrity

135
00:07:03,640 --> 00:07:06,320
Speaker 2: I might see is like Mitch McConnell, which is not

136
00:07:06,360 --> 00:07:07,440
Speaker 2: really the same thing.

137
00:07:08,760 --> 00:07:09,720
Speaker 3: Different reactions.

138
00:07:09,760 --> 00:07:12,280
Speaker 2: Fair, Although, my friends went to a music festival and

139
00:07:12,280 --> 00:07:14,480
Speaker 2: they were and they were standing next to AOC so

140
00:07:14,720 --> 00:07:15,480
Speaker 2: that was exciting.

141
00:07:16,000 --> 00:07:17,440
Speaker 3: Yeah, it's still exciting.

142
00:07:18,440 --> 00:07:22,520
Speaker 1: Yeah, that, yeah, that is exciting. I mean I've never

143
00:07:23,520 --> 00:07:25,720
Speaker 1: just to be clear, I've seen Killer Mike, but I

144
00:07:25,760 --> 00:07:27,680
Speaker 1: don't know what would happen if I tried to talk

145
00:07:27,720 --> 00:07:29,320
Speaker 1: to Killer? So I don't know that would be any

146
00:07:29,400 --> 00:07:32,200
Speaker 1: better than you no rubbing.

147
00:07:31,880 --> 00:07:32,680
Speaker 3: Of the elbows?

148
00:07:33,160 --> 00:07:38,160
Speaker 1: Yeah, no, no, just all oh, there is no yes,

149
00:07:38,200 --> 00:07:41,240
Speaker 1: But I actually was curious. And this brings us to

150
00:07:41,320 --> 00:07:44,200
Speaker 1: our topic, which I'm very excited to talk about. Is

151
00:07:44,360 --> 00:07:47,720
Speaker 1: if you said that they kind of dialed down the

152
00:07:47,800 --> 00:07:52,320
Speaker 1: AI at south By Southwest, But were there conversations or

153
00:07:52,360 --> 00:07:56,760
Speaker 1: anything about what's going on with AI and art there?

154
00:07:57,240 --> 00:08:00,320
Speaker 2: Yes? Oh, good question. It's actually a funny story. Probably

155
00:08:00,360 --> 00:08:03,240
Speaker 2: shouldn't even say this, but like so I was. I

156
00:08:03,400 --> 00:08:06,119
Speaker 2: was there for both south By Southwest and Podcast Movement

157
00:08:06,160 --> 00:08:08,760
Speaker 2: that was happening at the same time, and I was

158
00:08:08,760 --> 00:08:10,760
Speaker 2: there giving a talk. They were doing this series of

159
00:08:10,800 --> 00:08:13,560
Speaker 2: talks throughout the day at Podcast Movement, and so I

160
00:08:13,640 --> 00:08:15,920
Speaker 2: was giving one of the sort of signature talks that

161
00:08:15,960 --> 00:08:19,760
Speaker 2: I do about why audiences are actually kind of skeptical

162
00:08:19,920 --> 00:08:22,600
Speaker 2: of AI and their media and their creative work and

163
00:08:23,120 --> 00:08:25,360
Speaker 2: what that means and like how we should be. Basically,

164
00:08:25,360 --> 00:08:27,000
Speaker 2: the long and short of it is that you know,

165
00:08:27,560 --> 00:08:30,640
Speaker 2: especially in a medium like podcasting, the reason why people

166
00:08:30,640 --> 00:08:33,920
Speaker 2: turn to podcasts is trust. And so if you dial

167
00:08:34,040 --> 00:08:36,400
Speaker 2: up the AI. You are throwing away the thing that

168
00:08:36,440 --> 00:08:39,680
Speaker 2: makes the medium not just good but also lucrative, also

169
00:08:39,679 --> 00:08:43,840
Speaker 2: like a good you know medium. And so I'm deciding

170
00:08:43,920 --> 00:08:47,640
Speaker 2: my slides, and I use a case study of a

171
00:08:47,760 --> 00:08:54,160
Speaker 2: specific AI podcast company, and I have a slide that

172
00:08:54,280 --> 00:08:57,760
Speaker 2: has a picture of their you know, splash landing page.

173
00:08:57,760 --> 00:09:01,080
Speaker 2: It's like, oh, this company has makes the podcasts with

174
00:09:01,160 --> 00:09:04,080
Speaker 2: AI hosts and AI voices, and they say they can

175
00:09:04,120 --> 00:09:07,040
Speaker 2: make you know, however, many podcasts for like a dollar,

176
00:09:07,760 --> 00:09:10,120
Speaker 2: and they're churning them out. And I was using that

177
00:09:10,200 --> 00:09:13,480
Speaker 2: as like a negative in my opinion, negative use case.

178
00:09:14,600 --> 00:09:17,520
Speaker 2: I'm going over the slides with my producer. I'm like, oh,

179
00:09:17,840 --> 00:09:20,719
Speaker 2: maybe I should take this out. Like the person who

180
00:09:20,840 --> 00:09:23,760
Speaker 2: whose company I'm talking about, she could be in the audience,

181
00:09:23,960 --> 00:09:27,040
Speaker 2: make it awkward. Let me remove the slide. I show

182
00:09:27,120 --> 00:09:29,720
Speaker 2: up through my talk. Not only is that woman who

183
00:09:29,840 --> 00:09:33,960
Speaker 2: runs the company there, she's talking right before me about

184
00:09:33,960 --> 00:09:37,800
Speaker 2: her AI podcast company. So sets me up to be like, Okay,

185
00:09:37,840 --> 00:09:40,560
Speaker 2: everything that you just heard, here's why I disagree. But

186
00:09:40,679 --> 00:09:42,880
Speaker 2: I'm so glad that I took out that slide that

187
00:09:43,000 --> 00:09:47,839
Speaker 2: was like specifically screw this AI podcasting company. Am I right?

188
00:09:47,960 --> 00:09:49,400
Speaker 2: I didn't actually say that in the in the in

189
00:09:49,440 --> 00:09:54,400
Speaker 2: the talk, but that would have been the impression given. Yeah, who.

190
00:09:55,800 --> 00:09:56,840
Speaker 1: Bullet dodge for her?

191
00:09:56,960 --> 00:10:00,520
Speaker 2: Yeah, bullet dodged. Seriously, how would that.

192
00:10:00,520 --> 00:10:03,320
Speaker 3: Have been to be like, ohllo, did she stick around

193
00:10:03,360 --> 00:10:03,840
Speaker 3: for yours?

194
00:10:04,520 --> 00:10:06,640
Speaker 2: Oh? I don't know. That's actually a very good question.

195
00:10:07,000 --> 00:10:09,680
Speaker 2: I can't imagine anything that I had to say would

196
00:10:09,679 --> 00:10:14,520
Speaker 2: have moved somebody who is gambling on the idea that

197
00:10:14,559 --> 00:10:19,599
Speaker 2: audiences want more AI in their creative work, in their podcast,

198
00:10:19,760 --> 00:10:22,200
Speaker 2: in their media. When I think if you listen to

199
00:10:22,240 --> 00:10:24,920
Speaker 2: what audiences are telling us, they're telling us the exact opposite.

200
00:10:24,960 --> 00:10:28,840
Speaker 2: They don't want AI in their really anywhere, but especially

201
00:10:28,880 --> 00:10:31,400
Speaker 2: in their creative work. And that really dovetails nicely with

202
00:10:32,160 --> 00:10:33,800
Speaker 2: the situation I want to talk about today.

203
00:10:35,760 --> 00:10:39,959
Speaker 1: Yes, and I we were talking about this before, Samantha

204
00:10:39,960 --> 00:10:45,000
Speaker 1: and I never heard of this specific case, but it

205
00:10:45,160 --> 00:10:47,280
Speaker 1: is a huge conversation right now, and I've heard of

206
00:10:47,320 --> 00:10:50,640
Speaker 1: other cases and there's a lot to unpack and it

207
00:10:50,720 --> 00:10:54,439
Speaker 1: is really interesting. So what specific case are we talking

208
00:10:54,480 --> 00:10:55,120
Speaker 1: about today? Bridge?

209
00:10:55,720 --> 00:11:00,240
Speaker 2: I am talking about the writer drama unfolding around the

210
00:11:00,280 --> 00:11:02,760
Speaker 2: book Shy Girl, and I want to talk about this.

211
00:11:03,240 --> 00:11:05,559
Speaker 2: I wanted to sort of get your thoughts as authors

212
00:11:05,640 --> 00:11:08,200
Speaker 2: and creatives and people who are thinking about AI and

213
00:11:08,240 --> 00:11:11,000
Speaker 2: how it shows up in creative work in general. We

214
00:11:11,160 --> 00:11:14,640
Speaker 2: all had a conversation a while back about why it

215
00:11:14,720 --> 00:11:18,200
Speaker 2: is that women are using AI less in workplaces, and

216
00:11:18,960 --> 00:11:21,640
Speaker 2: the researchers gave lots of reasons, but one of the

217
00:11:21,679 --> 00:11:24,840
Speaker 2: reasons that women are less likely to use AI at

218
00:11:24,880 --> 00:11:27,719
Speaker 2: work is that they are more likely to be accused

219
00:11:27,760 --> 00:11:32,120
Speaker 2: of using AI as a negative thing and judged more

220
00:11:32,160 --> 00:11:35,600
Speaker 2: harshly if there is suspicion that they are using AI,

221
00:11:35,760 --> 00:11:38,160
Speaker 2: and so that was one of the reasons that researchers

222
00:11:38,200 --> 00:11:41,640
Speaker 2: identified as to why women are using AI less in artwork.

223
00:11:42,080 --> 00:11:46,959
Speaker 2: So this drama with the novel Shy Girl, I think,

224
00:11:47,200 --> 00:11:50,680
Speaker 2: really reveals a lot of the anxiety that some of

225
00:11:50,720 --> 00:11:55,320
Speaker 2: us feel around AI, particularly around how AI is being

226
00:11:55,440 --> 00:11:59,559
Speaker 2: used or not used to produce media like writing. So

227
00:12:00,480 --> 00:12:03,080
Speaker 2: I'll just give you the top lines of a situation.

228
00:12:04,080 --> 00:12:07,240
Speaker 2: Mia Ballard is a black woman poet and fiction writer.

229
00:12:07,960 --> 00:12:10,880
Speaker 2: I had actually not heard of her nor read any

230
00:12:10,920 --> 00:12:14,080
Speaker 2: of her writing before the situation hit my social media feed.

231
00:12:14,520 --> 00:12:16,720
Speaker 2: It sounds like she writes a lot of stories with

232
00:12:16,800 --> 00:12:21,360
Speaker 2: themes of feminine rage and it. Actually, I hate at

233
00:12:21,400 --> 00:12:24,679
Speaker 2: the first time that I'm hearing about hers during a scandal,

234
00:12:24,880 --> 00:12:28,319
Speaker 2: because it sounds like her genre of like weird girl

235
00:12:28,400 --> 00:12:30,480
Speaker 2: literature would be very much up my alley.

236
00:12:31,559 --> 00:12:34,280
Speaker 1: Yes, is that agreed?

237
00:12:34,520 --> 00:12:34,720
Speaker 3: Yeah?

238
00:12:34,800 --> 00:12:37,160
Speaker 2: Is that a style of writing you all rock with?

239
00:12:37,760 --> 00:12:41,920
Speaker 1: Oh yeah, oh yeah, oh yeah yeah.

240
00:12:41,960 --> 00:12:45,160
Speaker 2: I'm in a subreddit called weird girl Literature and every

241
00:12:45,200 --> 00:12:48,760
Speaker 2: suggestion I'm like, Ugh, give me a book where a

242
00:12:48,840 --> 00:12:52,360
Speaker 2: twisted woman does some twisted stuff and I'm I'm set.

243
00:12:56,200 --> 00:12:59,400
Speaker 1: That's my stay in best Night I can have.

244
00:13:00,240 --> 00:13:06,080
Speaker 2: Yes. So, Mia initially self published her novel called Shi Girl,

245
00:13:06,200 --> 00:13:09,080
Speaker 2: about a young woman who was forced to live as

246
00:13:09,120 --> 00:13:12,600
Speaker 2: a pet to a man. The book actually had some

247
00:13:12,679 --> 00:13:15,240
Speaker 2: like pretty good reviews on Goodreads, though I would say

248
00:13:15,559 --> 00:13:18,040
Speaker 2: they're kind of some are very good, and some are

249
00:13:18,080 --> 00:13:19,960
Speaker 2: like not so good. I would call them like mixed.

250
00:13:20,360 --> 00:13:22,920
Speaker 2: Some people specifically said that they loved the way that

251
00:13:22,960 --> 00:13:25,120
Speaker 2: Mia writes. Other people were like, oh, I don't like

252
00:13:25,160 --> 00:13:27,920
Speaker 2: this writing style at all. Last year, one of the

253
00:13:27,920 --> 00:13:33,400
Speaker 2: big publishing companies, Hashet, their Orbit division, acquired MIA's book

254
00:13:33,440 --> 00:13:36,560
Speaker 2: Shy Girl for its newest imprint run for It, which

255
00:13:36,600 --> 00:13:41,280
Speaker 2: exclusively publishes horror fiction. So good for Mia. Right, Like,

256
00:13:41,320 --> 00:13:44,640
Speaker 2: if the story ended here, you would be thinking, oh,

257
00:13:45,120 --> 00:13:47,800
Speaker 2: good job, girl, your self published book is going to

258
00:13:47,800 --> 00:13:49,160
Speaker 2: be published by a big publisher.

259
00:13:49,880 --> 00:13:51,600
Speaker 3: You made it, you did it.

260
00:13:52,080 --> 00:13:54,080
Speaker 2: If you don't want to hear like like you know

261
00:13:54,120 --> 00:13:56,680
Speaker 2: when you watch a horror movie and there's a point

262
00:13:56,679 --> 00:13:58,400
Speaker 2: of the movie where you're like, oh, this the family.

263
00:13:59,000 --> 00:14:01,360
Speaker 2: The family moved into the and they had their first

264
00:14:01,360 --> 00:14:03,400
Speaker 2: lovely night together. If you stopped it here, it'd be

265
00:14:03,400 --> 00:14:04,680
Speaker 2: a happy movie.

266
00:14:04,840 --> 00:14:07,520
Speaker 1: It'd be ten minutes long, but it'd be great.

267
00:14:17,400 --> 00:14:20,120
Speaker 2: So here is where things take a term because about

268
00:14:20,120 --> 00:14:24,280
Speaker 2: two months ago, readers who read Shy Girl started to

269
00:14:24,360 --> 00:14:29,360
Speaker 2: publicly ask whether this book was written using AI on

270
00:14:29,440 --> 00:14:33,320
Speaker 2: a Reddit subreddit called horror Lit, One user, who identified

271
00:14:33,320 --> 00:14:36,200
Speaker 2: themselves as a book editor with twelve years of experience

272
00:14:36,280 --> 00:14:41,360
Speaker 2: under their belt, specifically identifying AI pros, asked, does anyone

273
00:14:41,360 --> 00:14:44,960
Speaker 2: else think this was written by chatgebt? The post reads,

274
00:14:45,640 --> 00:14:48,520
Speaker 2: I know not an accusation to make lightly. I'm not

275
00:14:48,560 --> 00:14:50,640
Speaker 2: making it lightly. I have a lot to say and

276
00:14:50,680 --> 00:14:53,320
Speaker 2: I'll try to organize this post as best I can.

277
00:14:53,920 --> 00:14:56,840
Speaker 2: Me book editor of twelve years, I've had people from

278
00:14:56,960 --> 00:14:59,720
Speaker 2: over the last few years send me chatgypt creative writing.

279
00:15:00,480 --> 00:15:02,600
Speaker 2: My job with these AI pieces was to see if

280
00:15:02,600 --> 00:15:04,600
Speaker 2: I could humanize them or get them to a point

281
00:15:04,640 --> 00:15:07,480
Speaker 2: where it was enjoyable to read, or even acceptable. The

282
00:15:07,560 --> 00:15:10,760
Speaker 2: answer was generally no. Chatgept might be able to write

283
00:15:10,800 --> 00:15:13,440
Speaker 2: a passage that sounds good, but there are two problems

284
00:15:13,440 --> 00:15:15,960
Speaker 2: with that. A passage does not a novel make. A

285
00:15:16,040 --> 00:15:19,280
Speaker 2: novel is not a collection of passable passages. It's a

286
00:15:19,320 --> 00:15:21,840
Speaker 2: singular thing and needs to work as a singular thing.

287
00:15:22,360 --> 00:15:25,400
Speaker 2: And it seems good at first glance, on second glance,

288
00:15:25,480 --> 00:15:28,600
Speaker 2: it's not very good. So this was the post that

289
00:15:28,760 --> 00:15:32,720
Speaker 2: kind of got the conversation of whether or not this

290
00:15:32,760 --> 00:15:37,280
Speaker 2: book could be AI generated or chat GPT generated going online.

291
00:15:38,320 --> 00:15:42,080
Speaker 1: Yes, and I know we've talked about some of these

292
00:15:42,120 --> 00:15:44,880
Speaker 1: things before, and I definitely will be talking about fan

293
00:15:44,960 --> 00:15:50,120
Speaker 1: fiction later, but there are some things people look for

294
00:15:50,640 --> 00:15:53,560
Speaker 1: when they're trying to figure out if something is AI generated, right.

295
00:15:53,760 --> 00:15:55,840
Speaker 2: Yes, So this some of this was some of this

296
00:15:55,960 --> 00:15:57,520
Speaker 2: I knew, but a lot of this was stuff I

297
00:15:57,520 --> 00:16:00,560
Speaker 2: had never heard. So this user who said that they

298
00:16:00,560 --> 00:16:03,520
Speaker 2: had all this experience with AI writing, felt that the

299
00:16:03,640 --> 00:16:07,560
Speaker 2: novel had all of these key hallmarks of AI generated writing,

300
00:16:07,840 --> 00:16:13,920
Speaker 2: things like emotional or even emotionally overwrought text, the use

301
00:16:13,960 --> 00:16:18,920
Speaker 2: of adjectives and weather similes, light and dark metaphors, and

302
00:16:19,000 --> 00:16:22,960
Speaker 2: stock words such as quiet, chaos, and violence. They made

303
00:16:23,000 --> 00:16:26,080
Speaker 2: an entire exhaustive list of all of the sort of

304
00:16:26,200 --> 00:16:30,040
Speaker 2: linguistic quirks that they thought made them think like, okay,

305
00:16:30,040 --> 00:16:33,840
Speaker 2: this could be AI. The phrasing of describing a smell

306
00:16:34,000 --> 00:16:38,560
Speaker 2: and describing it as something X something, why things being

307
00:16:38,600 --> 00:16:43,760
Speaker 2: in three a lot, or using the linguistic construction two

308
00:16:44,080 --> 00:16:46,960
Speaker 2: x two y in a lot of different passages. Also,

309
00:16:47,080 --> 00:16:51,360
Speaker 2: they said the book has near perfect grammar, which, according

310
00:16:51,440 --> 00:16:53,400
Speaker 2: to this post is something that is like very rare

311
00:16:53,520 --> 00:16:56,560
Speaker 2: in creative writing, which I had not heard nor even considered.

312
00:16:56,560 --> 00:17:01,360
Speaker 2: But you know, I'm not a book editor. They also

313
00:17:01,400 --> 00:17:05,800
Speaker 2: talked about the tell tell m dashes to separate clauses,

314
00:17:05,800 --> 00:17:08,199
Speaker 2: which is that is actually an AI tell I've heard of,

315
00:17:08,280 --> 00:17:12,200
Speaker 2: although tell that to Emily Dickinson, right, girl loved the

316
00:17:12,320 --> 00:17:13,639
Speaker 2: girl love to dash.

317
00:17:14,280 --> 00:17:16,440
Speaker 1: A lot of people do love a dash, And now

318
00:17:16,480 --> 00:17:19,359
Speaker 1: it's become a like I'm not AI. I promised.

319
00:17:19,960 --> 00:17:21,560
Speaker 3: It threw me off when I first saw that, because

320
00:17:21,560 --> 00:17:22,879
Speaker 3: I used to do that a lot, and I'm like,

321
00:17:22,960 --> 00:17:26,720
Speaker 3: oh damn, I think I'd be like I'd failed classes.

322
00:17:27,440 --> 00:17:31,280
Speaker 2: So I am the person that you've just described Annie

323
00:17:31,320 --> 00:17:35,840
Speaker 2: where I When I was in grad school, I studied

324
00:17:35,840 --> 00:17:39,920
Speaker 2: a lot of Emily Dickinson's writing, and I loved her

325
00:17:40,040 --> 00:17:43,199
Speaker 2: use of dashes. And I was a lot younger, but

326
00:17:44,000 --> 00:17:47,399
Speaker 2: I was like, oh, dashes signal that you are like

327
00:17:47,680 --> 00:17:51,600
Speaker 2: deep and like a real deep writer, deep thinker. Subsequently,

328
00:17:51,640 --> 00:17:55,560
Speaker 2: my work has always been dash heavy, but now I

329
00:17:55,600 --> 00:17:57,240
Speaker 2: take them out because I don't want I don't want

330
00:17:57,240 --> 00:18:00,239
Speaker 2: anyone getting the wrong idea. And I think that that

331
00:18:00,400 --> 00:18:04,959
Speaker 2: the way that you put that is so right, because listen,

332
00:18:05,000 --> 00:18:08,960
Speaker 2: I'm no computer scientist, but I have to imagine that

333
00:18:09,000 --> 00:18:11,680
Speaker 2: since AI is trained on lots and lots and lots

334
00:18:11,680 --> 00:18:16,320
Speaker 2: of human writing, often stolen copyrighted writing that has been

335
00:18:16,400 --> 00:18:20,800
Speaker 2: used to train AI without the author's consent, of course,

336
00:18:20,880 --> 00:18:23,040
Speaker 2: it is going to be harder to tell what is

337
00:18:23,080 --> 00:18:25,520
Speaker 2: written by AI and what is written by a human.

338
00:18:25,560 --> 00:18:28,360
Speaker 2: I read this interesting op ed in New York Times

339
00:18:28,480 --> 00:18:30,359
Speaker 2: by a writer about this whole incident, and she said,

340
00:18:30,760 --> 00:18:34,480
Speaker 2: humans don't write like AI rights like humans. And I

341
00:18:34,520 --> 00:18:37,840
Speaker 2: think that really kind of gets at what you're saying that, Oh,

342
00:18:37,840 --> 00:18:40,120
Speaker 2: a lot of people use these quirks doesn't mean it's

343
00:18:40,119 --> 00:18:41,800
Speaker 2: AI necessarily.

344
00:18:42,119 --> 00:18:44,200
Speaker 1: Yes, And I think that's the heart of the conversation

345
00:18:44,240 --> 00:18:47,400
Speaker 1: that we're going to get into. But because we live

346
00:18:47,440 --> 00:18:51,840
Speaker 1: in such a destabilized like, even before AI had become

347
00:18:51,920 --> 00:18:58,440
Speaker 1: such a presence in our lives, reality just feels so subjective.

348
00:18:58,520 --> 00:19:02,080
Speaker 1: Now it just feels like I don't know what's true

349
00:19:02,160 --> 00:19:06,160
Speaker 1: or not anymore, And now AI is getting involved in that.

350
00:19:07,040 --> 00:19:13,000
Speaker 1: And so you are questioning, did somebody write this or

351
00:19:13,119 --> 00:19:15,760
Speaker 1: is this AI? Where are these things? Is it the

352
00:19:15,960 --> 00:19:18,800
Speaker 1: m dash they're talking about the weather like, looking for

353
00:19:18,880 --> 00:19:22,480
Speaker 1: all of these signs wherein it could be AI or

354
00:19:22,520 --> 00:19:26,040
Speaker 1: it could be somebody that AI learned from.

355
00:19:26,280 --> 00:19:29,840
Speaker 2: Yeah, it's part of this conversation. People were putting in

356
00:19:30,440 --> 00:19:34,120
Speaker 2: works that were clearly written before AI was ever a thing,

357
00:19:34,240 --> 00:19:38,320
Speaker 2: like somebody put in the Declaration of Independence or Frankenstein,

358
00:19:38,800 --> 00:19:41,639
Speaker 2: and a lot of AI detectors settle this is ninety

359
00:19:41,640 --> 00:19:45,320
Speaker 2: eight percent AI. And obviously the founding flowers were not

360
00:19:45,400 --> 00:19:49,439
Speaker 2: using chat shebt right, the Declaration of Independence, But of

361
00:19:49,520 --> 00:19:53,520
Speaker 2: course all of those texts helped train AI. So it

362
00:19:53,600 --> 00:19:58,120
Speaker 2: kind of weigh it is I think particularly perhaps the

363
00:19:58,200 --> 00:20:01,439
Speaker 2: overuse of AI detection might be part of that, but

364
00:20:01,800 --> 00:20:03,720
Speaker 2: I do also just agree. I think that we're in

365
00:20:03,760 --> 00:20:07,280
Speaker 2: a time where we don't know who to trust. I

366
00:20:07,320 --> 00:20:16,320
Speaker 2: think we are really really craving authentic voices, and I

367
00:20:16,359 --> 00:20:19,919
Speaker 2: think media has really gotten to a place where people

368
00:20:19,960 --> 00:20:22,080
Speaker 2: have reason to be really skeptical, and so I think

369
00:20:22,359 --> 00:20:25,399
Speaker 2: everybody's just on a hair trigger right now. In some ways,

370
00:20:25,400 --> 00:20:27,159
Speaker 2: that is how we should be. We should always be

371
00:20:27,280 --> 00:20:29,560
Speaker 2: very skeptical and very critical and like ask questions about

372
00:20:29,560 --> 00:20:32,360
Speaker 2: what we're seeing online and in media. But I think

373
00:20:32,400 --> 00:20:35,240
Speaker 2: this reveals just how kind of fraud it all is

374
00:20:35,320 --> 00:20:36,920
Speaker 2: right now. I think you're totally right.

375
00:20:37,680 --> 00:20:43,600
Speaker 1: Yes, And in this particular case, study does really showcase

376
00:20:43,640 --> 00:20:49,280
Speaker 1: a lot of those anxieties and people looking at this

377
00:20:49,480 --> 00:20:54,480
Speaker 1: author and trying to prove like, oh, they definitely used

378
00:20:54,480 --> 00:20:57,560
Speaker 1: AI and here's other instances and the impact it had

379
00:20:57,600 --> 00:21:00,359
Speaker 1: on her life, but also just why it mattens so

380
00:21:00,440 --> 00:21:03,399
Speaker 1: much to so many people. So yeah, yes, can we

381
00:21:03,440 --> 00:21:04,040
Speaker 1: come back to.

382
00:21:04,160 --> 00:21:08,760
Speaker 2: What totally So some people were even saying that a

383
00:21:08,840 --> 00:21:12,119
Speaker 2: separate interview that Mia did about the book, it seemed

384
00:21:12,160 --> 00:21:14,840
Speaker 2: like maybe she used AI to generate her answer. So

385
00:21:14,960 --> 00:21:17,040
Speaker 2: I said that to say that this is not the

386
00:21:17,040 --> 00:21:22,560
Speaker 2: only project Shy Girl where AI accusations have swirled around

387
00:21:22,560 --> 00:21:27,639
Speaker 2: this particular writer. So this popular book YouTube channel Frankie's Shelf,

388
00:21:27,640 --> 00:21:31,240
Speaker 2: which has over two hundred thousand followers, made a two

389
00:21:31,280 --> 00:21:36,200
Speaker 2: and a half hour long video titled quote I think

390
00:21:36,200 --> 00:21:39,320
Speaker 2: this is AI slop that over a million people watched.

391
00:21:39,720 --> 00:21:44,040
Speaker 2: I clicked in honestly two and a half hours. I

392
00:21:44,080 --> 00:21:48,040
Speaker 2: don't got the time, So I don't got the time.

393
00:21:48,720 --> 00:21:50,840
Speaker 2: I did not. I did not watch the video myself.

394
00:21:50,840 --> 00:21:56,720
Speaker 2: That's what I'm saying. But yeah, that's like Avengers Endgame,

395
00:21:56,880 --> 00:22:03,400
Speaker 2: that's that's that's the length of a real movie, long one.

396
00:22:03,840 --> 00:22:07,080
Speaker 2: So John Wick Chapter four is the longest movie in

397
00:22:07,119 --> 00:22:08,800
Speaker 2: the franchise. And I use that as an example because

398
00:22:08,840 --> 00:22:10,359
Speaker 2: I saw John Wick in the theater and I was like,

399
00:22:10,359 --> 00:22:13,960
Speaker 2: how longest movie? It's two hours and forty nine minutes long,

400
00:22:14,320 --> 00:22:18,240
Speaker 2: So about as long as John Wick. The longest John

401
00:22:18,280 --> 00:22:21,120
Speaker 2: Wick movie that I saw in theaters and was counting

402
00:22:21,240 --> 00:22:23,960
Speaker 2: down the minutes until the movie was over, Like what

403
00:22:24,119 --> 00:22:26,320
Speaker 2: is it illegal to make a movie that's ninety minutes anymore?

404
00:22:26,440 --> 00:22:32,879
Speaker 2: But I digress. So in this video, Frankie points out

405
00:22:32,960 --> 00:22:35,719
Speaker 2: that the word sharp appeared one hundred and fifty nine

406
00:22:35,800 --> 00:22:39,720
Speaker 2: times across the two hundred and fourteen pages and had

407
00:22:39,760 --> 00:22:44,119
Speaker 2: been used to describe abstract concepts like guilt or a

408
00:22:44,200 --> 00:22:47,600
Speaker 2: silence in a room, like sharp is not really a

409
00:22:47,640 --> 00:22:50,879
Speaker 2: word that you would associate with describing abstract concepts like

410
00:22:50,960 --> 00:22:52,800
Speaker 2: usually it's like a physical thing, And so he was saying, well,

411
00:22:52,840 --> 00:22:56,720
Speaker 2: that's some weird quirk. He also found other overused words

412
00:22:56,800 --> 00:22:59,480
Speaker 2: like weight, which was used ninety four times, edge eighty

413
00:22:59,480 --> 00:23:04,840
Speaker 2: four times, hum or humming used as a descriptor for atmosphere,

414
00:23:04,920 --> 00:23:08,960
Speaker 2: like a humming atmosphere, and rhythm or rhythmic forty two times.

415
00:23:08,960 --> 00:23:11,639
Speaker 2: So all of these words that Frankie says, oh, like,

416
00:23:11,840 --> 00:23:15,400
Speaker 2: nobody would use these words to this amount in one

417
00:23:15,440 --> 00:23:18,840
Speaker 2: project if they weren't using AI. Now, Frankie is careful

418
00:23:18,880 --> 00:23:21,760
Speaker 2: to say that this doesn't mean the book was AI

419
00:23:21,840 --> 00:23:25,080
Speaker 2: generated just because it has repetitive words and words use

420
00:23:25,119 --> 00:23:29,000
Speaker 2: a particular way, because humans also write in this way,

421
00:23:29,040 --> 00:23:32,399
Speaker 2: like humans can be formulaic, humans can be repetitive, And

422
00:23:32,520 --> 00:23:35,679
Speaker 2: just to be super clear, as far as I know,

423
00:23:35,760 --> 00:23:39,159
Speaker 2: there is no definitive proof that AI was used to

424
00:23:39,200 --> 00:23:41,920
Speaker 2: write this book. So I don't know, and that's not

425
00:23:41,960 --> 00:23:43,720
Speaker 2: a question I'm able to answer that I'm going to

426
00:23:43,760 --> 00:23:46,520
Speaker 2: be trying to answer in this episode, but more on

427
00:23:46,560 --> 00:23:49,359
Speaker 2: that in a moment. I guess what I would say

428
00:23:49,560 --> 00:23:53,840
Speaker 2: is that this whole thing, I'm sorry to say, it

429
00:23:53,880 --> 00:23:56,400
Speaker 2: really showed me that I don't necessarily have a good

430
00:23:56,440 --> 00:23:59,960
Speaker 2: sense of when text is or is not AI generated

431
00:24:00,200 --> 00:24:03,639
Speaker 2: like her writing. I read the first few pages of

432
00:24:03,680 --> 00:24:06,280
Speaker 2: the book. It did not jump out to me as

433
00:24:06,359 --> 00:24:08,399
Speaker 2: being AI generated. I thought it was. I thought it

434
00:24:08,440 --> 00:24:12,280
Speaker 2: was a bit unusually written in a style perspective, but

435
00:24:12,359 --> 00:24:14,160
Speaker 2: it didn't. It did. It would not have jumped out

436
00:24:14,160 --> 00:24:16,040
Speaker 2: to me as like, oh, that's a telltale AI. I

437
00:24:16,040 --> 00:24:18,399
Speaker 2: feel like I have a sense of AI writing, and

438
00:24:18,480 --> 00:24:22,240
Speaker 2: this book to me I was. It didn't jump out

439
00:24:22,240 --> 00:24:23,280
Speaker 2: to me as AI generated.

440
00:24:24,520 --> 00:24:28,840
Speaker 1: Yeah. I haven't read it obviously, but it is kind

441
00:24:28,880 --> 00:24:32,639
Speaker 1: of scary to think that, like, because I feel like

442
00:24:32,680 --> 00:24:35,720
Speaker 1: I can recognize AI writing too, But what if I

443
00:24:35,760 --> 00:24:38,199
Speaker 1: can anymore? What if it's gotten beyond that point?

444
00:24:38,480 --> 00:24:42,080
Speaker 2: So yeah, that's my point, And I guess researching this

445
00:24:42,800 --> 00:24:45,520
Speaker 2: for me personally, I think I am at that point

446
00:24:45,560 --> 00:24:50,840
Speaker 2: where AI has gotten so good at sounding like a

447
00:24:50,920 --> 00:24:56,399
Speaker 2: human writer that what I thought I knew, because you know,

448
00:24:56,480 --> 00:24:59,880
Speaker 2: when you read something AI generated like that's not blank,

449
00:25:00,160 --> 00:25:03,600
Speaker 2: it's blank. That's just blank blank blank, like you have

450
00:25:03,680 --> 00:25:05,840
Speaker 2: a I have a sense of it in my mind.

451
00:25:06,480 --> 00:25:09,480
Speaker 2: And researching this, they also as part of researching at

452
00:25:09,520 --> 00:25:11,760
Speaker 2: the New York Times, had a thing that was like, oh,

453
00:25:11,760 --> 00:25:13,720
Speaker 2: can you tell which of these was a I generated?

454
00:25:14,000 --> 00:25:16,680
Speaker 2: I bombed that test, So I I clear I clearly

455
00:25:17,760 --> 00:25:20,600
Speaker 2: have lost the ability to tell the way that I

456
00:25:20,600 --> 00:25:25,360
Speaker 2: always And I'm actually curious how you all determine whether

457
00:25:25,440 --> 00:25:28,760
Speaker 2: or not something's AI generated. There's just a flatness to it.

458
00:25:29,240 --> 00:25:31,760
Speaker 2: There's just like it's like they're they're just a like

459
00:25:32,720 --> 00:25:36,240
Speaker 2: hollowness to the writing that is kind of it's like

460
00:25:36,320 --> 00:25:37,800
Speaker 2: kinda know it when I see it kind of thing.

461
00:25:37,840 --> 00:25:39,600
Speaker 2: What about what's what's your benchmark?

462
00:25:40,720 --> 00:25:46,640
Speaker 1: Well, I I think again, this is something that we're

463
00:25:46,680 --> 00:25:49,240
Speaker 1: going to get into more towards the like conclusion part

464
00:25:50,400 --> 00:25:54,280
Speaker 1: is that I'm really struggling with there. Usually when I

465
00:25:54,320 --> 00:25:57,680
Speaker 1: think something is AI, it does have a flatness to it.

466
00:25:57,680 --> 00:26:02,760
Speaker 1: It does have kind of a either a strange bluntness

467
00:26:03,400 --> 00:26:07,280
Speaker 1: or a strange like way too much like way way way,

468
00:26:07,320 --> 00:26:15,040
Speaker 1: way way too much and very repetitive. But uh, it

469
00:26:15,200 --> 00:26:19,800
Speaker 1: is true that someone can have a style or someone

470
00:26:19,800 --> 00:26:24,439
Speaker 1: could be a bad writer, or someone might be in

471
00:26:24,480 --> 00:26:28,000
Speaker 1: the world of fan fiction, a non English speaker and

472
00:26:28,040 --> 00:26:32,760
Speaker 1: they're writing in English, and so it's I have these

473
00:26:32,920 --> 00:26:37,359
Speaker 1: things that I associate with AI, but I still cannot

474
00:26:37,840 --> 00:26:41,639
Speaker 1: definitively say that it is AI.

475
00:26:42,720 --> 00:26:45,719
Speaker 3: Yeah, I honestly I couldn't tell you. I really couldn't,

476
00:26:45,760 --> 00:26:49,240
Speaker 3: mainly because there are so many bad writers in general,

477
00:26:51,960 --> 00:26:53,440
Speaker 3: not that I know anybody who's a bad writer. I

478
00:26:53,440 --> 00:26:55,520
Speaker 3: don't know anybody who's a bad writer. Co sure personally

479
00:26:55,600 --> 00:27:00,639
Speaker 3: personally that when you see some bad write that's like

480
00:27:00,720 --> 00:27:03,000
Speaker 3: so off the mark that I'm like, this could be AI.

481
00:27:05,800 --> 00:27:09,159
Speaker 3: That's kind of what I expect sometimes, And that's the

482
00:27:09,200 --> 00:27:12,600
Speaker 3: one that is too little, too much, like where it

483
00:27:12,720 --> 00:27:16,280
Speaker 3: sounds like it's a bad infrovercial almost and I'm like, yeah,

484
00:27:16,480 --> 00:27:20,200
Speaker 3: this also could be AI. So I really have bad

485
00:27:20,359 --> 00:27:22,560
Speaker 3: like marked when it comes to writing. I feel like

486
00:27:23,119 --> 00:27:26,720
Speaker 3: videos and pictures and such are little more telltale signs.

487
00:27:27,400 --> 00:27:30,679
Speaker 3: Still I've gotten I've still gotten got by some of

488
00:27:30,680 --> 00:27:33,199
Speaker 3: these videos. I'm like, somen of a bescuit, You mean

489
00:27:33,280 --> 00:27:36,360
Speaker 3: that random animal? It's not doing this really cute thing.

490
00:27:37,800 --> 00:27:40,679
Speaker 3: Videos get me a while. Why do they do this

491
00:27:40,840 --> 00:27:43,560
Speaker 3: that plays with my emotion. I needed to be real anyway.

492
00:27:43,600 --> 00:27:46,360
Speaker 3: But like, yeah, I think I'm just gonna be real

493
00:27:46,400 --> 00:27:48,040
Speaker 3: honest to be like, I'm not sure if I would

494
00:27:48,080 --> 00:27:48,480
Speaker 3: catch it.

495
00:27:50,520 --> 00:27:55,840
Speaker 2: Somebody listening is thinking, these three idiots, how can they

496
00:27:55,920 --> 00:27:59,920
Speaker 2: not catch it? To that person, I say, take that.

497
00:28:00,040 --> 00:28:03,080
Speaker 2: At New York Times AI writing tests, it is harder

498
00:28:03,240 --> 00:28:05,520
Speaker 2: than you think. And I'm someone who spends a lot

499
00:28:05,520 --> 00:28:08,560
Speaker 2: of time thinking about AI, and I was surprised by

500
00:28:09,040 --> 00:28:12,520
Speaker 2: how difficult it was to tell with writing at this point.

501
00:28:12,600 --> 00:28:14,960
Speaker 2: With video, I honestly I have almost kind of looked

502
00:28:14,960 --> 00:28:17,919
Speaker 2: divested from short form video because so much of it

503
00:28:18,000 --> 00:28:22,639
Speaker 2: is AI, and I just kept there. It started with

504
00:28:22,640 --> 00:28:25,840
Speaker 2: a video of bunnies on a trampoline. This was like

505
00:28:25,920 --> 00:28:27,960
Speaker 2: months and months and months ago, and I sent that

506
00:28:28,000 --> 00:28:30,040
Speaker 2: to a bunch of my friends. I put it in

507
00:28:30,080 --> 00:28:32,479
Speaker 2: the group chat. My friend's roasted me. They were like,

508
00:28:32,680 --> 00:28:37,640
Speaker 2: what are you someone's mema putting AI generated video slop

509
00:28:37,720 --> 00:28:39,760
Speaker 2: in the group chat. What's the matter with you? I

510
00:28:40,080 --> 00:28:42,800
Speaker 2: got such a like. I I was like, Okay, well

511
00:28:42,800 --> 00:28:44,400
Speaker 2: I'll never post another video in the group.

512
00:28:46,440 --> 00:28:48,920
Speaker 3: The telltale sign you have to go into comments, because

513
00:28:48,960 --> 00:28:51,920
Speaker 3: you got the comments will tell you someone is out

514
00:28:51,920 --> 00:28:54,680
Speaker 3: there being sleuth because I almost felt for that one too.

515
00:28:54,720 --> 00:28:56,480
Speaker 3: I was like oh, and then up pulled up the comments.

516
00:28:56,520 --> 00:29:00,560
Speaker 3: I was like, damn it, AI. Yeah, you're talking about people.

517
00:29:00,600 --> 00:29:03,040
Speaker 2: I would say, like, when I do the trainings around

518
00:29:03,040 --> 00:29:07,800
Speaker 2: recognizing AI generated content, like people should know what their

519
00:29:07,880 --> 00:29:10,520
Speaker 2: triggers are, because I clearly have some sort of a

520
00:29:10,600 --> 00:29:14,240
Speaker 2: trigger for like cute animal videos. Clearly that is a

521
00:29:15,600 --> 00:29:18,160
Speaker 2: place where I am totally willing to let my guard

522
00:29:18,240 --> 00:29:21,920
Speaker 2: down and like suspend my disbelief and I need to

523
00:29:21,960 --> 00:29:24,360
Speaker 2: stay when I see a cute animal video. Now I'm

524
00:29:24,400 --> 00:29:26,840
Speaker 2: like extra alert. So get a sense of what your

525
00:29:27,240 --> 00:29:31,960
Speaker 2: specific triggers are around like content online. Although we were

526
00:29:32,120 --> 00:29:36,000
Speaker 2: talking about this skit that the comedian Drew Ski made

527
00:29:36,080 --> 00:29:39,160
Speaker 2: where he's like dressed up as a conservative woman and

528
00:29:39,520 --> 00:29:41,320
Speaker 2: a lot of people were like, oh, that's Erica Kirk

529
00:29:41,680 --> 00:29:45,840
Speaker 2: and so that happened, And then there was this almost

530
00:29:46,000 --> 00:29:51,320
Speaker 2: entirely AI generated, fabricated claim that Erica Kirk was suing him,

531
00:29:51,320 --> 00:29:54,160
Speaker 2: which is not happening. And I saw this video that

532
00:29:54,200 --> 00:29:56,960
Speaker 2: purported to show Drewski, who I know what he looks like,

533
00:29:57,280 --> 00:30:00,680
Speaker 2: talking about how Erica Kirk was suing him, and I

534
00:30:00,720 --> 00:30:02,800
Speaker 2: had to research this, so I knew it wasn't the case.

535
00:30:02,880 --> 00:30:05,320
Speaker 2: But I would never have looked at that video and

536
00:30:05,360 --> 00:30:07,719
Speaker 2: thought like, it looks real Like I would never if

537
00:30:07,760 --> 00:30:10,280
Speaker 2: I didn't already know from researching it that that was

538
00:30:10,760 --> 00:30:13,240
Speaker 2: not true. I would have never, in a million years

539
00:30:13,280 --> 00:30:15,800
Speaker 2: thought this was not a video of Drewski talking about

540
00:30:15,840 --> 00:30:17,080
Speaker 2: a situation happening to him.

541
00:30:17,520 --> 00:30:19,680
Speaker 3: Okay, you just alerted me to that, because I just

542
00:30:19,760 --> 00:30:21,480
Speaker 3: heard that video and it kind of went on with it,

543
00:30:21,480 --> 00:30:22,960
Speaker 3: But I didn't think either way. I just went, yeah,

544
00:30:22,960 --> 00:30:24,240
Speaker 3: that sounds about right and moved on.

545
00:30:24,520 --> 00:30:27,440
Speaker 2: Yeah, as far as we know, as of our recording,

546
00:30:27,520 --> 00:30:29,840
Speaker 2: Erica Kirk has had no response. So if you see

547
00:30:30,640 --> 00:30:35,640
Speaker 2: a tweet or a video purporting to say otherwise, it's

548
00:30:36,040 --> 00:30:38,760
Speaker 2: don't get don't don't get got there that AI tool

549
00:30:38,880 --> 00:30:41,720
Speaker 2: that they're so good that one would if I didn't

550
00:30:41,720 --> 00:30:43,400
Speaker 2: know already that one would have got me.

551
00:30:43,920 --> 00:30:47,520
Speaker 3: They're trying to get me. We are both mem Yes.

552
00:30:51,000 --> 00:30:56,320
Speaker 1: Well, so there's kind of another interesting side to this too, though,

553
00:30:56,560 --> 00:31:00,840
Speaker 1: is that you know, when we're talking about Shy Girl,

554
00:31:01,240 --> 00:31:05,640
Speaker 1: there was this uproar about oh is there AI? And

555
00:31:05,680 --> 00:31:09,400
Speaker 1: it went to the publisher. It went like all over

556
00:31:09,440 --> 00:31:12,920
Speaker 1: the place with people demanding, like, is this say I

557
00:31:13,000 --> 00:31:13,800
Speaker 1: do something about it?

558
00:31:14,120 --> 00:31:18,960
Speaker 2: Yes, So people were pissed. They were reaching out to me.

559
00:31:19,080 --> 00:31:22,360
Speaker 2: A's publisher hashet asking whether or not this book was

560
00:31:22,360 --> 00:31:25,880
Speaker 2: AI generated, and so there was a little bit of

561
00:31:25,920 --> 00:31:28,120
Speaker 2: a witch hunt element to this. I don't want to

562
00:31:28,120 --> 00:31:30,400
Speaker 2: say that like with like a negative connotation necessarily, but

563
00:31:30,480 --> 00:31:33,080
Speaker 2: like I just got to say, I think that just

564
00:31:33,160 --> 00:31:37,600
Speaker 2: shows how much people do not want AI and their writing.

565
00:31:38,200 --> 00:31:41,760
Speaker 2: I personally can ever see myself taking it upon myself

566
00:31:41,760 --> 00:31:44,920
Speaker 2: to like email a stranger's publisher to ask if AI

567
00:31:45,120 --> 00:31:50,000
Speaker 2: was used. I understand why people want to know. Maybe

568
00:31:50,040 --> 00:31:52,440
Speaker 2: you bought the book or read the book and feel misled.

569
00:31:52,480 --> 00:31:55,240
Speaker 2: Now it taps into exactly the kind of distrust that

570
00:31:55,280 --> 00:31:58,640
Speaker 2: you were talking about earlier. Annie, I personally find the

571
00:31:58,680 --> 00:32:03,400
Speaker 2: witch hunt angle to be, like, so it surprises me

572
00:32:04,120 --> 00:32:07,080
Speaker 2: that people are that invested in it that they will

573
00:32:07,080 --> 00:32:08,720
Speaker 2: take it take it upon themselves to be like I'm

574
00:32:08,720 --> 00:32:10,600
Speaker 2: going to watch a three hour long video about it

575
00:32:10,600 --> 00:32:12,960
Speaker 2: and then email of a publisher like that surprises me.

576
00:32:13,000 --> 00:32:14,160
Speaker 2: But I just think it goes to show how much

577
00:32:14,200 --> 00:32:16,920
Speaker 2: people don't like and don't want and don't trust AI.

578
00:32:17,880 --> 00:32:22,240
Speaker 2: After The New York Times got involved. The publisher dropped

579
00:32:22,280 --> 00:32:24,880
Speaker 2: the book. So the book is being pulled. It is

580
00:32:24,880 --> 00:32:27,560
Speaker 2: no more so. The New York Times they did an

581
00:32:27,560 --> 00:32:29,880
Speaker 2: investigation and they said that they concluded the book was

582
00:32:29,920 --> 00:32:33,040
Speaker 2: written by AI. This is from the Times. The Times

583
00:32:33,080 --> 00:32:37,080
Speaker 2: analyzed passages from the novel using several AI detection tools

584
00:32:37,080 --> 00:32:40,840
Speaker 2: and found recurring patterns characteristic of AI generated text, like

585
00:32:40,920 --> 00:32:44,840
Speaker 2: Gabson logic, excessive use of melodramatic adjectives, and an over

586
00:32:44,920 --> 00:32:48,640
Speaker 2: reliance on the rule of three. Max Spiro, who is

587
00:32:48,680 --> 00:32:51,160
Speaker 2: the founder and chief executive of Pangram, which is like

588
00:32:51,200 --> 00:32:55,360
Speaker 2: an AI detection program, says that he heard about this controversy,

589
00:32:55,760 --> 00:32:57,680
Speaker 2: he got a copy of the book and ran his

590
00:32:57,800 --> 00:33:01,480
Speaker 2: own test, and that test indicated the book was seventy

591
00:33:01,480 --> 00:33:06,480
Speaker 2: eight percent AI generated. He published those findings on social media,

592
00:33:06,520 --> 00:33:10,400
Speaker 2: saying I'm confident that this is largely AI generated or

593
00:33:10,440 --> 00:33:16,920
Speaker 2: heavily AI assisted, and so whether or not this definitively

594
00:33:17,120 --> 00:33:19,320
Speaker 2: proves it, I cannot say that is like above my

595
00:33:19,360 --> 00:33:22,880
Speaker 2: pay grade, but that all of that was enough, plus

596
00:33:22,920 --> 00:33:27,200
Speaker 2: the public pressure for the publisher to drop this book.

597
00:33:28,240 --> 00:33:36,480
Speaker 1: Yes, And one thing that I find interesting about this

598
00:33:36,600 --> 00:33:44,720
Speaker 1: whole conversation is the Winch hunt aspect, because I think

599
00:33:44,800 --> 00:33:49,880
Speaker 1: there is sometimes you just don't like an ending perhaps

600
00:33:50,080 --> 00:33:53,000
Speaker 1: and then you're like it was AI. You don't have

601
00:33:53,200 --> 00:33:57,360
Speaker 1: maybe the proof, you might have something some evidence, but

602
00:33:57,480 --> 00:34:01,840
Speaker 1: also you know, really relying on these AI detection tools

603
00:34:02,440 --> 00:34:05,200
Speaker 1: when they are not perfect.

604
00:34:05,960 --> 00:34:08,040
Speaker 3: Yeah, I was gonna ask about that because I saw

605
00:34:08,600 --> 00:34:12,200
Speaker 3: pretty early on when people are using detection site for

606
00:34:12,440 --> 00:34:15,520
Speaker 3: papers for young students, and a lot of them were dained,

607
00:34:15,640 --> 00:34:17,920
Speaker 3: not because they were actually using AI, but they were

608
00:34:17,960 --> 00:34:21,480
Speaker 3: using things like autocorrect or they use it after like

609
00:34:21,640 --> 00:34:24,759
Speaker 3: after the fact, which was specific for the program of

610
00:34:24,800 --> 00:34:28,640
Speaker 3: correcting your grammar like grammarly and that's specific. You can

611
00:34:28,800 --> 00:34:31,480
Speaker 3: use that, you are encouraged to use that. But that

612
00:34:31,680 --> 00:34:35,160
Speaker 3: actually brought up the detector and saying they used AI,

613
00:34:35,400 --> 00:34:37,160
Speaker 3: like is it? How does this work?

614
00:34:37,560 --> 00:34:40,080
Speaker 2: Yeah, I'm glad that you used that example. You know,

615
00:34:40,120 --> 00:34:44,600
Speaker 2: I got my start in education and I was teaching

616
00:34:45,200 --> 00:34:48,960
Speaker 2: right when a lot of these AI plagiarism checker and

617
00:34:49,000 --> 00:34:51,759
Speaker 2: detector tools were coming up. And the long and short

618
00:34:51,760 --> 00:34:56,840
Speaker 2: of it is that AI detection platforms they just really

619
00:34:56,840 --> 00:35:01,040
Speaker 2: cannot be trusted. The time spoke to hit Chaprabardi, who

620
00:35:01,080 --> 00:35:04,000
Speaker 2: is a professor of computer science at Stonybrook University, and

621
00:35:04,400 --> 00:35:07,239
Speaker 2: he used that guy I told you about earlier, Max Biro.

622
00:35:07,480 --> 00:35:11,440
Speaker 2: He used Max Biro's program Pangram to check more than

623
00:35:11,480 --> 00:35:15,480
Speaker 2: fourteen thousand self published novels on Amazon for AI writing,

624
00:35:15,920 --> 00:35:18,799
Speaker 2: and so that program found that nearly twenty percent of

625
00:35:18,920 --> 00:35:23,359
Speaker 2: novels had been essentially written by AI. And so he

626
00:35:23,840 --> 00:35:29,480
Speaker 2: admits that AI detectors sometimes mistakenly flag human writing as

627
00:35:29,560 --> 00:35:32,600
Speaker 2: AI generated, But still he told The Times that he

628
00:35:32,719 --> 00:35:36,760
Speaker 2: was confident that Pangram was indeed picking up chapbot language.

629
00:35:37,160 --> 00:35:39,880
Speaker 2: He says the program was built to detect linguistic patterns

630
00:35:40,320 --> 00:35:43,200
Speaker 2: that are frequently used by large language models like chatgept

631
00:35:43,320 --> 00:35:46,400
Speaker 2: and Gemini, and has a FoST positive rate of around

632
00:35:46,440 --> 00:35:49,640
Speaker 2: one and ten thousand, and that it's also designed to

633
00:35:49,760 --> 00:35:53,880
Speaker 2: catch human efforts to cover up aius through editing. So,

634
00:35:54,880 --> 00:35:57,799
Speaker 2: I again, I am no computer scientist, but I do

635
00:35:57,920 --> 00:36:00,439
Speaker 2: think that we should talk about the fact that these

636
00:36:00,480 --> 00:36:05,680
Speaker 2: tools are not just randomly mistaken. Sometimes they are biased.

637
00:36:06,080 --> 00:36:09,640
Speaker 2: They more frequently flagged text written by non native English

638
00:36:09,680 --> 00:36:12,200
Speaker 2: speakers as being AI generated. But it's not just like

639
00:36:12,239 --> 00:36:15,400
Speaker 2: Anny said that you see in the fan fiction community.

640
00:36:15,560 --> 00:36:20,640
Speaker 2: Research has consistently shown that these detectors misclassify non native

641
00:36:20,680 --> 00:36:25,279
Speaker 2: English as AI generated, while Native English is often accurately identified.

642
00:36:25,600 --> 00:36:28,280
Speaker 2: The numbers are to me pretty damning. In one study

643
00:36:28,320 --> 00:36:32,200
Speaker 2: in patterns called GPT, detectors are biased against non native

644
00:36:32,239 --> 00:36:35,040
Speaker 2: English writers, more than half of the essays written by

645
00:36:35,080 --> 00:36:38,440
Speaker 2: non native English speakers were flagged as AI generated, with

646
00:36:38,520 --> 00:36:41,720
Speaker 2: a false positive rate of sixty one point three percent.

647
00:36:42,280 --> 00:36:43,640
Speaker 2: So that's pretty significant.

648
00:36:44,600 --> 00:36:48,879
Speaker 1: Yeah, I wouldn't feel confident at all if I knew

649
00:36:48,880 --> 00:36:51,560
Speaker 1: that the percentage was that high that it would be incorrect.

650
00:36:52,160 --> 00:36:55,600
Speaker 1: So why is this happening?

651
00:36:57,000 --> 00:37:00,160
Speaker 2: Yeah, that's a great question. It sounds like these tools

652
00:37:00,200 --> 00:37:03,960
Speaker 2: are trained on style guides and writing examples that are

653
00:37:04,000 --> 00:37:08,440
Speaker 2: dominated by native English syntax and vocabulary, so writing that

654
00:37:08,520 --> 00:37:12,120
Speaker 2: deviates from those norms are more often to get flagged

655
00:37:12,120 --> 00:37:15,040
Speaker 2: as machine like. So in other words, they're basically just

656
00:37:15,080 --> 00:37:19,120
Speaker 2: treating standard American or British English as the baseline for

657
00:37:19,239 --> 00:37:22,799
Speaker 2: what human writing looks like, and then people can get

658
00:37:22,880 --> 00:37:26,040
Speaker 2: punished if they deviate from that. And as I said,

659
00:37:26,080 --> 00:37:28,840
Speaker 2: like in educational settings, this was already a big problem

660
00:37:28,880 --> 00:37:32,000
Speaker 2: that educators were dealing with. When I was teaching, I

661
00:37:32,120 --> 00:37:35,560
Speaker 2: was told to use these detectors that we now know

662
00:37:35,640 --> 00:37:39,319
Speaker 2: are biased and it ever happened in any of my

663
00:37:39,360 --> 00:37:42,320
Speaker 2: classes specifically, but like students could be accused of cheating

664
00:37:42,800 --> 00:37:46,319
Speaker 2: and have serious consequences like failing a course or being

665
00:37:46,320 --> 00:37:49,480
Speaker 2: expelled as a result. There have already been documented cases

666
00:37:49,480 --> 00:37:53,000
Speaker 2: of students being penalized for work that they insist they

667
00:37:53,040 --> 00:37:56,640
Speaker 2: wrote themselves simply because their writing style did not match

668
00:37:56,680 --> 00:37:59,880
Speaker 2: what an algorithm expected human writing to look like. And

669
00:38:00,120 --> 00:38:04,680
Speaker 2: so whether or not that I can't speak to the

670
00:38:04,719 --> 00:38:09,000
Speaker 2: New York Times, is AI detection rigor or like how

671
00:38:09,040 --> 00:38:11,480
Speaker 2: they did it, They don't go into that, but I

672
00:38:11,520 --> 00:38:14,440
Speaker 2: wanted to mention that because, like, as we're talking about

673
00:38:14,600 --> 00:38:18,560
Speaker 2: the Shy Girl situation and everyone rushing to cite AI

674
00:38:18,719 --> 00:38:22,320
Speaker 2: detection scores as proof that this book was AI generated,

675
00:38:22,719 --> 00:38:25,880
Speaker 2: I do think it's worth asking who do these tools

676
00:38:26,080 --> 00:38:29,640
Speaker 2: actually protect and who do they kind of keep vulnerable

677
00:38:29,719 --> 00:38:33,200
Speaker 2: or keep at risk for accusations around AI use that

678
00:38:33,200 --> 00:38:34,400
Speaker 2: may or may not be unfounded.

679
00:38:35,320 --> 00:38:38,920
Speaker 1: Yes, especially when, as you mentioned earlier in previous episodes

680
00:38:38,960 --> 00:38:42,840
Speaker 1: we were talking about, women are more likely to be

681
00:38:42,960 --> 00:38:46,719
Speaker 1: penalized for using AI if they are found out that

682
00:38:46,760 --> 00:38:51,040
Speaker 1: they are, whereas for men it's more likely to be like, oh,

683
00:38:51,280 --> 00:38:53,680
Speaker 1: you're using the system. Great, you know what you're doing.

684
00:38:55,760 --> 00:39:01,359
Speaker 1: And then to put this faith into these detect that

685
00:39:01,560 --> 00:39:05,960
Speaker 1: do have bias in them does leave a lot of

686
00:39:06,040 --> 00:39:11,319
Speaker 1: room for error and for marginalized people specifically to be

687
00:39:11,480 --> 00:39:12,040
Speaker 1: hurt more.

688
00:39:13,160 --> 00:39:15,840
Speaker 2: Yeah, I mean, I can't not mention the fact that

689
00:39:15,880 --> 00:39:20,200
Speaker 2: Mia as a black woman. I have to wonder if

690
00:39:20,280 --> 00:39:23,279
Speaker 2: that is part of the conversation of why people are

691
00:39:24,040 --> 00:39:26,880
Speaker 2: kind of going so hard for her and like calling

692
00:39:26,920 --> 00:39:30,000
Speaker 2: the publisher and all of that like that, that is

693
00:39:30,040 --> 00:39:32,120
Speaker 2: a reaction that I am surprised by, and I have

694
00:39:32,160 --> 00:39:34,279
Speaker 2: to wonder if her race and gender is part of it.

695
00:39:34,960 --> 00:39:36,920
Speaker 2: And I did want to make a little comparison because

696
00:39:36,960 --> 00:39:39,600
Speaker 2: do y'all remember the writer James Fray. He wrote that

697
00:39:39,640 --> 00:39:41,359
Speaker 2: book One Million Little Pieces.

698
00:39:41,640 --> 00:39:45,479
Speaker 3: Yeah, he was found out for a lying exactly about

699
00:39:45,480 --> 00:39:45,799
Speaker 3: his book.

700
00:39:45,880 --> 00:39:49,560
Speaker 2: Yes, yeah. His memoir was like, I mean this this

701
00:39:49,680 --> 00:39:53,000
Speaker 2: incident was it like burned in my brain. I was

702
00:39:53,040 --> 00:39:55,480
Speaker 2: on an episode of Behind the Bastards about Oprah and

703
00:39:55,840 --> 00:39:57,160
Speaker 2: when we got to this part, I was like, I

704
00:39:57,160 --> 00:39:59,640
Speaker 2: could tell you everything. I watched the episode in real time.

705
00:40:00,040 --> 00:40:02,920
Speaker 2: What do you want to know? But yeah, Oprah had

706
00:40:02,960 --> 00:40:05,239
Speaker 2: made it one of her book club picks and then

707
00:40:05,680 --> 00:40:08,120
Speaker 2: when it was revealed to be largely fabricated, she had

708
00:40:08,200 --> 00:40:10,239
Speaker 2: him on the show to like scold him, and I

709
00:40:10,280 --> 00:40:14,640
Speaker 2: remember he had I think he had an outbreak of

710
00:40:15,239 --> 00:40:17,640
Speaker 2: what is it called when adults get chicken pox? It

711
00:40:17,680 --> 00:40:21,080
Speaker 2: has a name when its shingles. He had like face

712
00:40:21,160 --> 00:40:25,000
Speaker 2: shingles at the time. And this, this image of him

713
00:40:25,200 --> 00:40:29,800
Speaker 2: with face shingles getting scolded by Oprah is just burned

714
00:40:29,880 --> 00:40:33,279
Speaker 2: in my memory. But so you might. You might think

715
00:40:33,400 --> 00:40:35,799
Speaker 2: that guy when it was revealed that he basically made

716
00:40:35,880 --> 00:40:38,640
Speaker 2: up his memoir an Embarrassed Oprah's Book Club, you might

717
00:40:38,680 --> 00:40:42,240
Speaker 2: think that he has been like shunned by the literary world.

718
00:40:42,360 --> 00:40:44,360
Speaker 2: You would be wrong, because he has a new book

719
00:40:44,400 --> 00:40:49,360
Speaker 2: out and he's talked about, albeit inconsistently, using AI in

720
00:40:49,400 --> 00:40:52,120
Speaker 2: his writing. He said that AI helped write parts of

721
00:40:52,120 --> 00:40:56,520
Speaker 2: his book, that he trained AI to match his writing style.

722
00:40:56,960 --> 00:40:58,800
Speaker 2: He told readers that they would not be able to

723
00:40:58,880 --> 00:41:02,040
Speaker 2: tell what was AI and what he wrote. He did

724
00:41:02,200 --> 00:41:04,640
Speaker 2: walk some of this back later when people were like,

725
00:41:04,880 --> 00:41:07,520
Speaker 2: m I'm not so sure about that. So I'm not

726
00:41:07,560 --> 00:41:10,400
Speaker 2: sure what the truth is. But it is not like

727
00:41:10,480 --> 00:41:14,000
Speaker 2: he was pillared for this. And in fact, his newest book,

728
00:41:14,000 --> 00:41:18,080
Speaker 2: which just came out included a very splashy piece in

729
00:41:18,120 --> 00:41:22,959
Speaker 2: the British Times under the headline America's literary bad boy

730
00:41:23,040 --> 00:41:29,640
Speaker 2: is back. So there's you know, there's no accompanying which

731
00:41:29,719 --> 00:41:33,279
Speaker 2: hunt to find out if this book used AI like

732
00:41:33,320 --> 00:41:35,919
Speaker 2: there is with me is and Yeah, I just I'm

733
00:41:35,960 --> 00:41:39,160
Speaker 2: just curious, like who gets to just be called like, yeah,

734
00:41:39,200 --> 00:41:42,200
Speaker 2: he's the literary bad boy, and who gets like call

735
00:41:42,320 --> 00:41:43,920
Speaker 2: angry calls to their publisher.

736
00:41:44,719 --> 00:41:46,719
Speaker 3: That's such a cliche phrase, though, I wonder if they

737
00:41:46,719 --> 00:41:48,239
Speaker 3: were trying to make fun of him a little bit

738
00:41:48,680 --> 00:42:02,319
Speaker 3: with that literary bad boy. Yeah, all right, I'm just

739
00:42:02,360 --> 00:42:06,480
Speaker 3: gonna ask this question in ernest because there's a little

740
00:42:06,520 --> 00:42:10,000
Speaker 3: part of me in this that people who loved her book,

741
00:42:10,360 --> 00:42:13,880
Speaker 3: who really enjoyed this book and really seem to it,

742
00:42:13,920 --> 00:42:18,000
Speaker 3: is it so awful if she did use AI? And

743
00:42:18,040 --> 00:42:20,240
Speaker 3: I say this in this context of like, we don't

744
00:42:20,280 --> 00:42:22,680
Speaker 3: want this to be taken Yeah, of course not. We

745
00:42:22,760 --> 00:42:26,640
Speaker 3: want good storytelling in any ways, in most all that

746
00:42:26,760 --> 00:42:30,080
Speaker 3: we want genuine writing. But at the same time, we

747
00:42:30,160 --> 00:42:32,840
Speaker 3: see a lot of smut out there that's just randomly

748
00:42:32,880 --> 00:42:34,440
Speaker 3: thrown at us, like you know what I mean, Like

749
00:42:34,520 --> 00:42:38,359
Speaker 3: it's not sincere oftentimes you see series that are picked

750
00:42:38,400 --> 00:42:40,720
Speaker 3: up by pen writers who are no longer the writers.

751
00:42:41,560 --> 00:42:44,439
Speaker 3: Is that such a bad in the scope of things?

752
00:42:44,440 --> 00:42:46,240
Speaker 3: And I say this, I know this is a slippery slope,

753
00:42:46,239 --> 00:42:49,000
Speaker 3: but yeah, just a little bit antagonistic care ask It's

754
00:42:49,000 --> 00:42:50,240
Speaker 3: a fascinating question.

755
00:42:50,360 --> 00:42:52,759
Speaker 2: And Annie, I'm curious to get your thoughts too, as

756
00:42:52,760 --> 00:42:56,160
Speaker 2: somebody who does a lot of fan fiction. So, I

757
00:42:56,200 --> 00:43:01,440
Speaker 2: do think that publishers have not really set a firm

758
00:43:01,560 --> 00:43:05,040
Speaker 2: boundary around what is or is not acceptable when it

759
00:43:05,080 --> 00:43:08,239
Speaker 2: comes to AIU so and so I think perhaps there

760
00:43:08,280 --> 00:43:12,319
Speaker 2: is a spectrum where you just have CHATGETPT spit out

761
00:43:12,320 --> 00:43:13,920
Speaker 2: of text and then you copy and paste out and

762
00:43:14,000 --> 00:43:17,680
Speaker 2: that's the book. And did you use chat Gypt or

763
00:43:17,719 --> 00:43:22,359
Speaker 2: AI to like brainstorm or you know, clean up your

764
00:43:22,400 --> 00:43:26,040
Speaker 2: pros And if you did that, can you still call

765
00:43:26,080 --> 00:43:30,000
Speaker 2: it genuine human work? That's not a question I can answer.

766
00:43:30,160 --> 00:43:32,640
Speaker 2: And the issue is that publishers have not made it

767
00:43:32,680 --> 00:43:34,640
Speaker 2: clear where they stand. They have not made it clear

768
00:43:35,000 --> 00:43:37,400
Speaker 2: what is acceptable and what is a breach of contract

769
00:43:37,480 --> 00:43:40,600
Speaker 2: and what is okay. And I think that they're really

770
00:43:40,640 --> 00:43:44,319
Speaker 2: allowing public sentiment and public outrage to be the thing

771
00:43:44,400 --> 00:43:48,120
Speaker 2: that drives what should be hard and fast, clear rules

772
00:43:48,160 --> 00:43:49,520
Speaker 2: and regulations, right.

773
00:43:49,719 --> 00:43:52,560
Speaker 3: I mean in all this to say when people steal content,

774
00:43:52,640 --> 00:43:55,839
Speaker 3: like taking other people's writings as their inspiration and then

775
00:43:55,920 --> 00:43:59,320
Speaker 3: using that for their chat gypt writings, that's stealing, like

776
00:43:59,680 --> 00:44:01,600
Speaker 3: we're going to but that as like an od yeahs

777
00:44:01,760 --> 00:44:05,600
Speaker 3: statement there, Like obviously Mia was already an established writer

778
00:44:06,080 --> 00:44:11,360
Speaker 3: according to previous history. She had been doing writing previously. Again,

779
00:44:11,400 --> 00:44:13,920
Speaker 3: they said maybe a couple of her other works had

780
00:44:13,960 --> 00:44:18,120
Speaker 3: been AI. We don't know. We don't know with all

781
00:44:18,160 --> 00:44:21,680
Speaker 3: of this, Yeah, what does she say? So?

782
00:44:21,880 --> 00:44:24,680
Speaker 2: In an email to The Times, Balor denied using AI

783
00:44:24,719 --> 00:44:28,640
Speaker 2: to write Shy Girl. She did contend that an acquaintance

784
00:44:28,680 --> 00:44:32,160
Speaker 2: that she hired to edit the self published version of

785
00:44:32,200 --> 00:44:35,800
Speaker 2: the book might have used AI in the editing process.

786
00:44:36,239 --> 00:44:39,680
Speaker 2: She said, the controversy has changed my life in many ways,

787
00:44:39,719 --> 00:44:41,600
Speaker 2: and my mental health is at an all time low,

788
00:44:41,920 --> 00:44:43,960
Speaker 2: and my name is ruined for something I didn't even

789
00:44:44,000 --> 00:44:47,560
Speaker 2: personally do, noting that she could not elaborate on how

790
00:44:47,560 --> 00:44:49,719
Speaker 2: the book had been edited with AI because she was

791
00:44:49,760 --> 00:44:53,279
Speaker 2: pursuing legal action, and so that's what she says. She says.

792
00:44:53,520 --> 00:44:56,200
Speaker 2: She says, I didn't use AI. If AI was used,

793
00:44:56,239 --> 00:44:58,680
Speaker 2: it was somebody that helped me edit this, the self

794
00:44:58,719 --> 00:45:01,560
Speaker 2: published version of this book. I've never self published a

795
00:45:01,560 --> 00:45:04,239
Speaker 2: book before. But my understanding is, when do you self

796
00:45:04,239 --> 00:45:07,480
Speaker 2: publish something, you are kind of just on the line

797
00:45:07,719 --> 00:45:11,080
Speaker 2: for any of the issues that arise, or just sort

798
00:45:11,080 --> 00:45:13,479
Speaker 2: of like on the hook for all of that. Because

799
00:45:13,480 --> 00:45:15,120
Speaker 2: I don't know if that's but again, I've never done

800
00:45:15,120 --> 00:45:17,239
Speaker 2: that before myself. So don't you know, take that for

801
00:45:17,280 --> 00:45:17,799
Speaker 2: what it's worth.

802
00:45:18,160 --> 00:45:21,279
Speaker 3: Yeah, I think you can do it on Amazon and such,

803
00:45:21,320 --> 00:45:25,440
Speaker 3: like it's pretty easy to access those types of places.

804
00:45:25,520 --> 00:45:28,280
Speaker 3: I have a couple of old coworkers who loved writing

805
00:45:28,360 --> 00:45:32,520
Speaker 3: Mormon romance novels. Oh yeah, if you want to look

806
00:45:32,520 --> 00:45:36,920
Speaker 3: that up, I'll tell you later. But like, it's interesting

807
00:45:36,920 --> 00:45:40,000
Speaker 3: because could we come back and see much like some

808
00:45:40,080 --> 00:45:42,680
Speaker 3: of the students that we were talking about previously who

809
00:45:42,719 --> 00:45:46,440
Speaker 3: came back trying to fight because they were like expelled

810
00:45:46,800 --> 00:45:50,240
Speaker 3: completely kicked out of school, being accused of using AI,

811
00:45:50,360 --> 00:45:52,839
Speaker 3: but it was like grammarly and things that you came

812
00:45:52,880 --> 00:45:56,520
Speaker 3: back to check their grammar. Could it be something like that,

813
00:45:56,680 --> 00:45:59,840
Speaker 3: like did she use a system that helped check her

814
00:46:00,480 --> 00:46:04,160
Speaker 3: or whatnot? And in this moment, all these people went

815
00:46:04,200 --> 00:46:06,440
Speaker 3: after her as an example.

816
00:46:07,960 --> 00:46:11,120
Speaker 2: Yeah, I feel the same way. I guess we don't.

817
00:46:11,160 --> 00:46:14,399
Speaker 2: We don't know, and we may never know in what

818
00:46:14,440 --> 00:46:19,160
Speaker 2: capacity this was AI generated, if at all, but something

819
00:46:19,200 --> 00:46:25,319
Speaker 2: about the vibe of the entire Internet saying you used

820
00:46:25,360 --> 00:46:29,560
Speaker 2: AI And I'm not even sure how like if like

821
00:46:30,040 --> 00:46:32,399
Speaker 2: if if that happened to me, I'm not even sure

822
00:46:32,480 --> 00:46:36,879
Speaker 2: what I would do to illustrate that I hadn't other

823
00:46:37,000 --> 00:46:40,000
Speaker 2: than I've just written a book and it's been so

824
00:46:40,160 --> 00:46:44,439
Speaker 2: heavily track changes, edited by me and the editor going

825
00:46:44,480 --> 00:46:47,440
Speaker 2: back and forth and back and forth that if somebody

826
00:46:47,440 --> 00:46:49,080
Speaker 2: wanted to pour through all of that, I guess that

827
00:46:49,120 --> 00:46:51,840
Speaker 2: would that would illustrate it. But like, it does feel

828
00:46:52,880 --> 00:46:57,520
Speaker 2: strange that you could write a book and people could

829
00:46:57,560 --> 00:47:00,440
Speaker 2: just say, oh, this was AI generated and that might

830
00:47:00,480 --> 00:47:02,839
Speaker 2: not even be true. And it's just it just it's

831
00:47:02,880 --> 00:47:06,560
Speaker 2: just we're in a very weird dynamic where the public's

832
00:47:06,560 --> 00:47:10,120
Speaker 2: feelings around this have not caught up with our ability

833
00:47:10,160 --> 00:47:12,839
Speaker 2: to say, for serve if something is AI generated or not,

834
00:47:13,000 --> 00:47:17,480
Speaker 2: and publishing companies their own guidelines like they're like they're

835
00:47:17,560 --> 00:47:21,160
Speaker 2: really behind on the whole AI conversation when it comes

836
00:47:21,160 --> 00:47:22,399
Speaker 2: to what they will and will not.

837
00:47:22,400 --> 00:47:25,800
Speaker 3: Accept any need to go to look at our book.

838
00:47:26,680 --> 00:47:30,760
Speaker 1: You know what I learned thanks to fan fiction. Okay,

839
00:47:30,800 --> 00:47:36,080
Speaker 1: the fan fiction community is very, very into this whole thing.

840
00:47:36,560 --> 00:47:38,239
Speaker 1: There's a lot I could talk about. But one thing

841
00:47:38,280 --> 00:47:40,760
Speaker 1: I learned recently is if you have a Google Doc,

842
00:47:41,600 --> 00:47:46,040
Speaker 1: AI can scrape your Google Doc. So they fan fiction

843
00:47:46,080 --> 00:47:50,120
Speaker 1: writers recommend you do not use Google Docs, and what

844
00:47:50,160 --> 00:47:50,480
Speaker 1: we use.

845
00:47:51,360 --> 00:47:53,640
Speaker 2: All I use is Google Docs.

846
00:47:54,520 --> 00:47:57,239
Speaker 1: Yeah, but guess what for the fan fiction I've written,

847
00:47:57,320 --> 00:47:59,920
Speaker 1: I've never used Google Docs. Even before I knew that,

848
00:48:01,920 --> 00:48:07,439
Speaker 1: I just write on a doc and then copulate doc. Yeah. Yeah,

849
00:48:08,000 --> 00:48:11,520
Speaker 1: but I did get called out for being somebody. Brown

850
00:48:11,600 --> 00:48:13,719
Speaker 1: In was like your AI clearly, and I was like,

851
00:48:13,800 --> 00:48:18,400
Speaker 1: I can't prove you're wrong, but I'm not, so it

852
00:48:18,520 --> 00:48:19,440
Speaker 1: is strange.

853
00:48:20,040 --> 00:48:20,279
Speaker 2: Yeah.

854
00:48:20,320 --> 00:48:22,560
Speaker 1: But also you get AI reviews and I'm like, no,

855
00:48:22,680 --> 00:48:24,320
Speaker 1: this is an AI for sure.

856
00:48:25,239 --> 00:48:28,040
Speaker 3: So an AI review called out you being AI.

857
00:48:28,680 --> 00:48:32,360
Speaker 1: No, Okay, there's like a spade of You'll get reviews

858
00:48:32,360 --> 00:48:35,520
Speaker 1: that are just like, what a good story? Did you know?

859
00:48:35,680 --> 00:48:39,560
Speaker 1: Kyber is from the planet Ilium, And I'm like, yeah,

860
00:48:39,560 --> 00:48:42,000
Speaker 1: that's a Star Wars fact that's unrelated to anything.

861
00:48:43,239 --> 00:48:44,560
Speaker 2: Yeah.

862
00:48:44,640 --> 00:48:49,239
Speaker 1: Yeah, but anyway, Google Docs, I don't know the specifics,

863
00:48:49,280 --> 00:48:51,520
Speaker 1: but I just know that in the fan fiction community,

864
00:48:51,520 --> 00:48:52,680
Speaker 1: they're like, don't do it.

865
00:48:54,000 --> 00:48:59,480
Speaker 2: Always trust fan fiction folks, and like other niche nerdy

866
00:48:59,800 --> 00:49:03,319
Speaker 2: in interest people who are gathered by interest in some

867
00:49:03,480 --> 00:49:07,480
Speaker 2: niche nerdy thing to be ahead that they're always got

868
00:49:07,520 --> 00:49:10,319
Speaker 2: their finger on the pulse. They're ahead like whatever it is.

869
00:49:10,320 --> 00:49:12,960
Speaker 2: It's like, oh, we're three steps ahead of y'all. Whole read.

870
00:49:13,120 --> 00:49:15,440
Speaker 2: We've already got a system worked out to avoid this.

871
00:49:15,760 --> 00:49:18,319
Speaker 1: We know about the M dashes, we know that non

872
00:49:18,400 --> 00:49:21,960
Speaker 1: native English speakers get penalized where we go.

873
00:49:24,040 --> 00:49:24,720
Speaker 3: On the ready.

874
00:49:25,480 --> 00:49:30,000
Speaker 1: Yes, yes, but anyway, writing the book, we did use.

875
00:49:29,840 --> 00:49:33,440
Speaker 3: Google Docs, so so it would say that we were AI.

876
00:49:35,160 --> 00:49:37,759
Speaker 1: No, it's just going to be if it if it

877
00:49:37,880 --> 00:49:41,120
Speaker 1: scrapes Google Docs and it's going to take that and

878
00:49:42,440 --> 00:49:43,000
Speaker 1: run with it.

879
00:49:43,840 --> 00:49:47,600
Speaker 2: Yeah. I have used Google Docs, like it's all I use.

880
00:49:47,880 --> 00:49:51,280
Speaker 2: So everything I've written is probably being used to train AI.

881
00:49:52,000 --> 00:49:54,560
Speaker 2: You're welcome open AI, sam all, and then you can

882
00:49:54,560 --> 00:49:59,520
Speaker 2: get me a check anytime, I guess Sundar Pachai, head

883
00:49:59,560 --> 00:50:01,840
Speaker 2: of Google, Well, you can cut me a check anytime.

884
00:50:03,440 --> 00:50:05,120
Speaker 1: I don't know the specifics. We should come back and

885
00:50:05,120 --> 00:50:05,799
Speaker 1: talk about that though.

886
00:50:05,880 --> 00:50:07,240
Speaker 2: Yeah, I'll do some research.

887
00:50:07,840 --> 00:50:10,280
Speaker 1: Yes, there's a lot.

888
00:50:10,239 --> 00:50:12,719
Speaker 4: For you to be talking about right now, Bridget, that's

889
00:50:12,800 --> 00:50:18,879
Speaker 4: for sure, But in this specific conversation, so you've been

890
00:50:18,920 --> 00:50:20,279
Speaker 4: talking about the publishers.

891
00:50:20,600 --> 00:50:25,279
Speaker 1: That's kind of the issue is that there aren't clear guidelines,

892
00:50:26,320 --> 00:50:29,080
Speaker 1: and I think that is I do think that is

893
00:50:29,120 --> 00:50:32,920
Speaker 1: really important because this actually happens at our local conference,

894
00:50:33,000 --> 00:50:38,120
Speaker 1: dragon Con, where there's an art room, and they were like,

895
00:50:39,160 --> 00:50:42,160
Speaker 1: you have to tell people you used AI and if

896
00:50:42,200 --> 00:50:45,080
Speaker 1: you don't get out, like there has to be that

897
00:50:45,160 --> 00:50:48,440
Speaker 1: trust you were talking about earlier, Bridget, wherein maybe you

898
00:50:48,480 --> 00:50:52,000
Speaker 1: do use it for ideas, but even then people might

899
00:50:52,120 --> 00:50:54,080
Speaker 1: like to know that, and they might be okay with that,

900
00:50:54,560 --> 00:50:58,839
Speaker 1: but it's sort of that wanting to know, yeah, feeling misled.

901
00:51:00,040 --> 00:51:02,680
Speaker 1: So if there aren't these guidelines and this is a

902
00:51:02,840 --> 00:51:09,160
Speaker 1: very rapidly changing internet world that we live in, then

903
00:51:09,160 --> 00:51:13,359
Speaker 1: I think people feel betrayed if they don't know what

904
00:51:13,400 --> 00:51:14,319
Speaker 1: they're getting.

905
00:51:14,840 --> 00:51:18,880
Speaker 2: Exactly, and we just don't have a framework for that

906
00:51:19,040 --> 00:51:21,480
Speaker 2: right now. You would think that publishers would be setting

907
00:51:21,480 --> 00:51:25,839
Speaker 2: that framework, and I guess there is initiatives where they'll

908
00:51:25,880 --> 00:51:27,920
Speaker 2: say like, oh, this is a human led work, but

909
00:51:27,920 --> 00:51:30,400
Speaker 2: there's not really a way. That's all kind of like

910
00:51:30,440 --> 00:51:33,719
Speaker 2: an honor system right now, and so we're all just

911
00:51:33,760 --> 00:51:38,120
Speaker 2: sort of waiting for these institutions and mechanisms to catch

912
00:51:38,200 --> 00:51:40,640
Speaker 2: up with the reality that we're in. And I'll just

913
00:51:40,680 --> 00:51:43,280
Speaker 2: say it doesn't sound like publishing companies have moved very quickly.

914
00:51:43,840 --> 00:51:47,319
Speaker 2: The Times noted that while publishers have maintained a firm

915
00:51:47,400 --> 00:51:50,680
Speaker 2: line against AI generated texts and images and require authors

916
00:51:50,719 --> 00:51:53,640
Speaker 2: to attest that their work is original and their publishing contracts,

917
00:51:54,160 --> 00:51:57,719
Speaker 2: but few publishing companies have clear policies or measures to

918
00:51:57,760 --> 00:52:01,120
Speaker 2: prevent users from writing with AI. And with the Shy

919
00:52:01,200 --> 00:52:07,440
Speaker 2: Girls situation, I will say I was surprised that, and

920
00:52:07,719 --> 00:52:11,000
Speaker 2: maybe it has to do with her book being self

921
00:52:11,040 --> 00:52:15,359
Speaker 2: published and then being acquired. I was surprised that a

922
00:52:15,400 --> 00:52:19,680
Speaker 2: big publishing company like Kshet would not have editors. I

923
00:52:19,680 --> 00:52:21,520
Speaker 2: feel like so many people have to read a book

924
00:52:21,520 --> 00:52:23,640
Speaker 2: before it's published, and I was surprised that this was

925
00:52:23,680 --> 00:52:25,560
Speaker 2: something that would only be being addressed the first time

926
00:52:25,600 --> 00:52:29,000
Speaker 2: by social media commenters. Like that surprised me, And I

927
00:52:29,040 --> 00:52:32,440
Speaker 2: think like, even if contracts say that writers need to

928
00:52:32,480 --> 00:52:35,840
Speaker 2: affirm that their writing is original, what does that mean exactly.

929
00:52:36,440 --> 00:52:40,680
Speaker 2: The Times reports that publishers are skeptical of AI generated content,

930
00:52:41,040 --> 00:52:44,640
Speaker 2: partly because it can't be copyrighted, but that since AI

931
00:52:44,920 --> 00:52:48,080
Speaker 2: is so commonly used in things like outlining and research

932
00:52:48,160 --> 00:52:51,560
Speaker 2: and other early stages of writing, there's no real consensus,

933
00:52:51,640 --> 00:52:55,040
Speaker 2: according to The Times, in the publishing industry about where

934
00:52:55,040 --> 00:52:58,000
Speaker 2: that line should be drawn, and so many people feel

935
00:52:58,080 --> 00:53:02,719
Speaker 2: like this ambiguity just opens the door to bad actors

936
00:53:03,239 --> 00:53:06,719
Speaker 2: or even well meaning writers who don't realize that they've

937
00:53:06,719 --> 00:53:08,719
Speaker 2: maybe gone too far when it comes to AI use

938
00:53:08,760 --> 00:53:09,279
Speaker 2: in their work.

939
00:53:09,960 --> 00:53:17,719
Speaker 1: And we also don't really know the scope of AI writing, right.

940
00:53:18,160 --> 00:53:21,839
Speaker 2: Yeah. That I found to be so interesting because if

941
00:53:21,840 --> 00:53:24,960
Speaker 2: you just open up Amazon, you'll find so much obvious

942
00:53:25,080 --> 00:53:28,080
Speaker 2: AI slop that's like easy to determine, but when it

943
00:53:28,080 --> 00:53:31,040
Speaker 2: comes to big publishing companies, it seems harder to tell.

944
00:53:31,280 --> 00:53:34,160
Speaker 2: And The Times reported that AI is just leading to

945
00:53:34,200 --> 00:53:37,680
Speaker 2: more books being published. In general, it's nearly impossible to

946
00:53:37,719 --> 00:53:40,279
Speaker 2: gauge how much AI writing is getting published, but there's

947
00:53:40,320 --> 00:53:42,720
Speaker 2: evidence that the technology has led to a surge of books.

948
00:53:43,080 --> 00:53:45,480
Speaker 2: Last year, more than three point five million books for

949
00:53:45,560 --> 00:53:48,640
Speaker 2: self published, up from two point five million in twenty

950
00:53:48,640 --> 00:53:52,279
Speaker 2: twenty four, according to industry data, And so it does

951
00:53:52,320 --> 00:53:56,520
Speaker 2: seem like more and more books are being traditionally published,

952
00:53:56,520 --> 00:54:00,000
Speaker 2: and some of those books are probably in part written

953
00:54:00,239 --> 00:54:04,319
Speaker 2: with or buy AI, and so a publishing company not

954
00:54:04,480 --> 00:54:08,520
Speaker 2: having clear rules around it is really a problem. And

955
00:54:08,560 --> 00:54:10,680
Speaker 2: another thing they pointed out in that piece is that

956
00:54:11,360 --> 00:54:16,560
Speaker 2: one reason why publishing companies might be a little less

957
00:54:16,560 --> 00:54:20,839
Speaker 2: willing to set a firm boundary is because they might

958
00:54:20,920 --> 00:54:24,200
Speaker 2: want to be able to use AI in house for

959
00:54:24,239 --> 00:54:27,440
Speaker 2: things like marketing or translation or narration. Right, so if

960
00:54:27,640 --> 00:54:29,480
Speaker 2: they can't, really they're in a little bit of a

961
00:54:29,560 --> 00:54:32,680
Speaker 2: rub where they can't really say we have banned AI

962
00:54:32,800 --> 00:54:37,000
Speaker 2: usage altogether only for our writers in house. We'll use

963
00:54:37,040 --> 00:54:38,680
Speaker 2: AI however we see it, and we don't have to

964
00:54:38,680 --> 00:54:41,480
Speaker 2: tell you. Like, I understand the sort of rub that

965
00:54:41,480 --> 00:54:45,040
Speaker 2: they're in where they want to use it in some capacities,

966
00:54:45,400 --> 00:54:47,799
Speaker 2: but they want to control when and how and who.

967
00:54:51,280 --> 00:54:54,799
Speaker 1: Yes, and I do think it's really important that this

968
00:54:54,840 --> 00:54:59,040
Speaker 1: whole public perspective on it as well, of how the

969
00:54:59,040 --> 00:55:02,680
Speaker 1: public is viewing the usage of AI in art and

970
00:55:02,719 --> 00:55:06,239
Speaker 1: how companies are reacting to it, because it's clear that

971
00:55:06,280 --> 00:55:08,759
Speaker 1: companies do they're like, yes, I would like to use

972
00:55:08,800 --> 00:55:10,880
Speaker 1: it for this, this and this. Let us please not

973
00:55:10,960 --> 00:55:15,560
Speaker 1: have this conversation and move on but something like this

974
00:55:15,800 --> 00:55:20,680
Speaker 1: with what happened with Shy Girl, kind of forced the

975
00:55:20,760 --> 00:55:23,560
Speaker 1: company to come out make a statement.

976
00:55:24,560 --> 00:55:27,280
Speaker 2: So Hashet did come out with a public statement. I

977
00:55:27,320 --> 00:55:29,960
Speaker 2: really think that the heart of their statement is like,

978
00:55:30,480 --> 00:55:33,399
Speaker 2: do whatever, just don't embarrass us. I genuinely think that's

979
00:55:33,440 --> 00:55:36,280
Speaker 2: what it is. So they only put out this statement

980
00:55:36,520 --> 00:55:39,439
Speaker 2: when the Shy Girl controversy blew up on social media.

981
00:55:39,480 --> 00:55:43,440
Speaker 2: Their statement says, we encourage responsible experimentation with AI for

982
00:55:43,520 --> 00:55:47,360
Speaker 2: operational uses and recognize the benefits of remaining curious and

983
00:55:47,440 --> 00:55:51,160
Speaker 2: embracing technology. Our industry relies on the creative talent of

984
00:55:51,239 --> 00:55:53,840
Speaker 2: humans and would not exist without it. The value of

985
00:55:53,880 --> 00:55:57,200
Speaker 2: our partnerships with our creative contributors cannot be overstated. We

986
00:55:57,320 --> 00:56:01,440
Speaker 2: draw a distinction between operational uses uses that help us

987
00:56:01,440 --> 00:56:03,880
Speaker 2: fulfill our mission and make it easy for more people

988
00:56:04,080 --> 00:56:06,560
Speaker 2: to have access to our books and other products, and

989
00:56:06,760 --> 00:56:10,960
Speaker 2: creative uses IE uses that harness AI to replace the

990
00:56:11,000 --> 00:56:14,960
Speaker 2: creative work of a human author, designer, illustrator, translator, or photographer.

991
00:56:15,480 --> 00:56:18,640
Speaker 2: For this reason, we are opposed to machine creativity in

992
00:56:18,760 --> 00:56:22,239
Speaker 2: order to protect original creative content produced by humans. We

993
00:56:22,320 --> 00:56:24,719
Speaker 2: recognize that this is a fast moving area and will

994
00:56:24,719 --> 00:56:27,840
Speaker 2: continue to be guided by industry standards and the needs

995
00:56:27,880 --> 00:56:32,759
Speaker 2: of key creative rights holders. So to me, that is

996
00:56:32,840 --> 00:56:37,240
Speaker 2: that is exist. I'm no lawyer, but reading that statement,

997
00:56:37,280 --> 00:56:41,880
Speaker 2: to me, it sounds like we use AI when we

998
00:56:41,920 --> 00:56:44,200
Speaker 2: want to. What are you crazy? We're not gonna use AI,

999
00:56:44,680 --> 00:56:46,799
Speaker 2: but we don't think our writers should use it, and

1000
00:56:47,160 --> 00:56:50,359
Speaker 2: we'll continue to update that as we see fit.

1001
00:56:50,560 --> 00:56:58,840
Speaker 3: Thanks that phrase machine creativity. That's a whole like oxymoron.

1002
00:56:59,320 --> 00:57:03,880
Speaker 2: Yeah, yes, And I guess that's my whole kind of

1003
00:57:03,920 --> 00:57:05,960
Speaker 2: So what And this is just my opinion. I think

1004
00:57:05,960 --> 00:57:09,279
Speaker 2: that big publishing companies are kind of hanging writers out

1005
00:57:09,280 --> 00:57:12,960
Speaker 2: to dry by not having clear policies. It's like they're

1006
00:57:13,040 --> 00:57:17,280
Speaker 2: waiting for a big embarrassing dust up that will put

1007
00:57:17,320 --> 00:57:20,720
Speaker 2: the burden on the individual writer to deal with, rather

1008
00:57:20,760 --> 00:57:24,480
Speaker 2: than proactively come up with a clear policy because they

1009
00:57:24,520 --> 00:57:26,600
Speaker 2: want to make sure that they have carveouts for how

1010
00:57:26,800 --> 00:57:28,720
Speaker 2: for how they want to use AI or how they

1011
00:57:28,760 --> 00:57:30,600
Speaker 2: might want to use AI going down the line. So

1012
00:57:31,200 --> 00:57:33,640
Speaker 2: and I think in twenty twenty six, it's just common

1013
00:57:33,680 --> 00:57:35,680
Speaker 2: sense that you need to have a policy around AI.

1014
00:57:37,040 --> 00:57:39,440
Speaker 3: It sounds like they're opening themselves up to a lawsuit.

1015
00:57:40,240 --> 00:57:41,840
Speaker 3: That's what I think yeah, I mean it feels like

1016
00:57:41,920 --> 00:57:42,880
Speaker 3: very discriminatory.

1017
00:57:43,680 --> 00:57:47,479
Speaker 2: Well, so again, I don't know what's going to happen

1018
00:57:47,520 --> 00:57:50,920
Speaker 2: with the Mia Ballard shy girl situation, but let's say

1019
00:57:51,640 --> 00:57:55,720
Speaker 2: that for the sake of argument. She says, Okay, well,

1020
00:57:55,840 --> 00:57:59,240
Speaker 2: my editor used AI in this editing process, but the

1021
00:57:59,280 --> 00:58:02,960
Speaker 2: only stipulation in my contract is that AI will not

1022
00:58:03,040 --> 00:58:08,080
Speaker 2: be used to create the first pass or something. Is

1023
00:58:08,120 --> 00:58:10,080
Speaker 2: she able to sue for breach of contract because they

1024
00:58:10,120 --> 00:58:14,440
Speaker 2: dropped her book like not having I'm just a big

1025
00:58:14,960 --> 00:58:17,680
Speaker 2: believer in the power of contracts. I'm like, I can't

1026
00:58:17,680 --> 00:58:20,760
Speaker 2: tell you how many times having a good contract negotiated

1027
00:58:20,800 --> 00:58:24,200
Speaker 2: by the lawyer has saved my butt. And I think

1028
00:58:24,320 --> 00:58:26,439
Speaker 2: having just a just a And again I've not seen

1029
00:58:26,480 --> 00:58:27,880
Speaker 2: her contracts, so I don't know what it says. But

1030
00:58:28,240 --> 00:58:30,600
Speaker 2: just having a line that says, oh, authors will attest

1031
00:58:30,640 --> 00:58:34,400
Speaker 2: that the work is original is not enough. And I want,

1032
00:58:34,440 --> 00:58:37,360
Speaker 2: I just wonder what kind of legal precedent that setting

1033
00:58:37,400 --> 00:58:39,560
Speaker 2: or what kind of holes are being left open, because

1034
00:58:39,560 --> 00:58:42,920
Speaker 2: it's just it's just so ambiguous what that actually means.

1035
00:58:43,440 --> 00:58:49,200
Speaker 3: I mean, it sounds original. Yeah, the idea, yeah.

1036
00:58:59,400 --> 00:59:01,760
Speaker 2: I just read is that right before we came on

1037
00:59:02,080 --> 00:59:05,160
Speaker 2: to to talk. I just read that. You know, the

1038
00:59:05,240 --> 00:59:09,320
Speaker 2: romance publisher Harlequin. So just today it was announced that

1039
00:59:09,360 --> 00:59:12,880
Speaker 2: they struck a multi year deal with dash Verse, which

1040
00:59:12,880 --> 00:59:16,640
Speaker 2: is an AI entertainment company, to make forty animated micro

1041
00:59:16,760 --> 00:59:21,440
Speaker 2: dramas based on Harlequin titles using dash versus AI tool Framio,

1042
00:59:21,840 --> 00:59:25,720
Speaker 2: which generates complete short films from text prompts in about

1043
00:59:25,760 --> 00:59:29,600
Speaker 2: three weeks each. And so the details of this are

1044
00:59:29,640 --> 00:59:34,240
Speaker 2: that authors will get royalties through monetized AD then subscriptions,

1045
00:59:35,120 --> 00:59:37,520
Speaker 2: though Harlequin did not give deal specifics on what that

1046
00:59:37,560 --> 00:59:42,520
Speaker 2: looks like, and they say that the videos are illustrator assisted.

1047
00:59:42,800 --> 00:59:45,320
Speaker 2: I'm not one hundred percent sure what they mean by that,

1048
00:59:45,360 --> 00:59:48,160
Speaker 2: but I think that they're positioning it as a human

1049
00:59:48,400 --> 00:59:52,960
Speaker 2: AI creative collaboration rather than fully automated. I think I

1050
00:59:52,960 --> 00:59:55,560
Speaker 2: think that's how they are trying to frame it. I

1051
00:59:55,560 --> 00:59:59,800
Speaker 2: cannot speak to whether or not that is actually the case,

1052
01:00:00,240 --> 01:00:02,520
Speaker 2: but it's just goes to show that some big publishers

1053
01:00:02,520 --> 01:00:03,680
Speaker 2: are clearly fine with AI.

1054
01:00:05,920 --> 01:00:08,440
Speaker 3: That's an interesting take though, to take the old school

1055
01:00:08,480 --> 01:00:13,040
Speaker 3: Harlequin romance, cause there's numerous amouths believe. Yeah, as a child,

1056
01:00:13,080 --> 01:00:16,520
Speaker 3: when I was taking to my you know, not my

1057
01:00:16,560 --> 01:00:19,000
Speaker 3: mom for some mod reason, but for other moms that

1058
01:00:19,080 --> 01:00:22,240
Speaker 3: had the closet full of all of the romances. I'm like, hey,

1059
01:00:22,400 --> 01:00:24,760
Speaker 3: let me go find that I was a weird teenager.

1060
01:00:25,520 --> 01:00:28,440
Speaker 3: But like there's tons of that. So that's an interesting

1061
01:00:28,600 --> 01:00:31,800
Speaker 3: idea to go back to taking some of those which

1062
01:00:31,960 --> 01:00:36,760
Speaker 3: is still beloved obviously in ways, and then making these

1063
01:00:36,920 --> 01:00:40,720
Speaker 3: types of creations and also noting that they're going to

1064
01:00:40,760 --> 01:00:43,760
Speaker 3: get royalties, but is it like fifty cents like they

1065
01:00:43,800 --> 01:00:47,280
Speaker 3: get the residual weirdly checks like from streaming type of thing.

1066
01:00:48,080 --> 01:00:50,360
Speaker 3: That's an interesting guy, but it's an interesting idea.

1067
01:00:51,000 --> 01:00:53,920
Speaker 2: Yeah. I mean the head of dash Verse, their CEO,

1068
01:00:54,280 --> 01:00:56,760
Speaker 2: said exactly what you just said, that they're excited that

1069
01:00:56,800 --> 01:01:02,880
Speaker 2: this is bringing global entertainment franchises existing IP powered by

1070
01:01:02,920 --> 01:01:05,480
Speaker 2: AI and so, and it kind of like I kind

1071
01:01:05,480 --> 01:01:06,680
Speaker 2: of get what he's saying that, like if you have

1072
01:01:06,720 --> 01:01:10,080
Speaker 2: a bunch of existing titles in IP that they're like, oh, well,

1073
01:01:10,080 --> 01:01:12,280
Speaker 2: why wouldn't we use AI to just like get to

1074
01:01:12,280 --> 01:01:14,880
Speaker 2: squeeze more revenue out of that. I get what they

1075
01:01:14,880 --> 01:01:19,600
Speaker 2: are saying, but I think that there is no one

1076
01:01:19,920 --> 01:01:26,280
Speaker 2: clear standard around AI in publishing. In some cases, it's fine,

1077
01:01:26,320 --> 01:01:29,560
Speaker 2: and it's celebrated in this you know, splashy press release

1078
01:01:29,600 --> 01:01:31,400
Speaker 2: about how this is going to use AI to get

1079
01:01:31,400 --> 01:01:34,360
Speaker 2: more revenue and value out of your IP. In some

1080
01:01:34,440 --> 01:01:37,120
Speaker 2: cases people are losing their book deals, right, And so

1081
01:01:37,200 --> 01:01:41,160
Speaker 2: I guess I guess it's we're just in a we're

1082
01:01:41,200 --> 01:01:43,680
Speaker 2: just in a new territory, and I think we need

1083
01:01:43,720 --> 01:01:47,840
Speaker 2: more guidance, more guard rails, more clarity around it because

1084
01:01:47,880 --> 01:01:49,720
Speaker 2: we're all sort of navigating it in real time, and

1085
01:01:49,760 --> 01:01:51,040
Speaker 2: I don't think it has to be that way.

1086
01:01:51,760 --> 01:01:53,960
Speaker 3: Is Fabio still alive because I need to know his

1087
01:01:54,000 --> 01:01:55,840
Speaker 3: take on these harlequin romance.

1088
01:01:55,720 --> 01:01:58,200
Speaker 2: Is still alive because he got hit by a what

1089
01:01:58,280 --> 01:02:01,360
Speaker 2: if he get hiphie, he was by a bird that

1090
01:02:01,440 --> 01:02:03,600
Speaker 2: was forever. I guess that was nineteen ninety nine.

1091
01:02:03,840 --> 01:02:05,480
Speaker 1: Forever, go, bridget.

1092
01:02:09,520 --> 01:02:10,000
Speaker 3: Everything.

1093
01:02:12,200 --> 01:02:14,600
Speaker 1: He was in Sharknado. Come on, okay.

1094
01:02:14,880 --> 01:02:17,600
Speaker 2: I was like, that was nineteen ninety nine. That was

1095
01:02:17,600 --> 01:02:23,080
Speaker 2: like what ten years ago? Someone say like it was

1096
01:02:23,120 --> 01:02:24,919
Speaker 2: ten years ago, and I was like, you're talking about

1097
01:02:24,920 --> 01:02:27,600
Speaker 2: it was like two years ago. Like tell me, no,

1098
01:02:27,640 --> 01:02:31,320
Speaker 2: don't tell me about how long. Like in my mind

1099
01:02:32,520 --> 01:02:35,320
Speaker 2: the year two thousand was ten years ago. I won't

1100
01:02:35,360 --> 01:02:37,120
Speaker 2: hear anything else and anything else that you say I

1101
01:02:37,160 --> 01:02:38,800
Speaker 2: had to say about that to me is wrong.

1102
01:02:39,040 --> 01:02:41,640
Speaker 3: It's wrong, it's true. But I need to know if

1103
01:02:41,680 --> 01:02:44,160
Speaker 3: he's to get his second wind, that he's about to

1104
01:02:44,200 --> 01:02:48,880
Speaker 3: come to relevance again because of this, specifically, Parlican romance

1105
01:02:48,920 --> 01:02:53,280
Speaker 3: is coming back. Dig Yeah, but they better, they better.

1106
01:02:53,360 --> 01:02:55,160
Speaker 2: If they're gonna do this, they better have some like

1107
01:02:55,280 --> 01:02:59,280
Speaker 2: blowing long hair and some heaving bosoms.

1108
01:02:58,440 --> 01:03:01,640
Speaker 3: On the on the uh on the pirate ship first.

1109
01:03:01,720 --> 01:03:06,000
Speaker 1: Yes, yes, well see this is also the issue though,

1110
01:03:06,040 --> 01:03:08,560
Speaker 1: are they going to pay Are they going to pay Fabio?

1111
01:03:09,040 --> 01:03:11,560
Speaker 3: Are they going to stand ins that you would literally

1112
01:03:11,600 --> 01:03:16,240
Speaker 3: take art from real life people to make these covers,

1113
01:03:16,280 --> 01:03:18,160
Speaker 3: So I'll wonder exactly.

1114
01:03:18,280 --> 01:03:22,200
Speaker 1: So that's like if you're bringing this AI element to it,

1115
01:03:22,600 --> 01:03:25,240
Speaker 1: because contracts back then didn't know about AI, they didn't

1116
01:03:25,320 --> 01:03:29,400
Speaker 1: know about any of that, so probably not, probably not

1117
01:03:29,440 --> 01:03:31,200
Speaker 1: going to get anything. And then you're watching yourself in

1118
01:03:31,240 --> 01:03:34,720
Speaker 1: an AI video like, oh that is a whole other

1119
01:03:36,000 --> 01:03:37,880
Speaker 1: I'll come back and well that's a whole like that

1120
01:03:38,000 --> 01:03:38,600
Speaker 1: is happening.

1121
01:03:38,920 --> 01:03:40,960
Speaker 2: It sounds black mirror, but that is happening.

1122
01:03:41,480 --> 01:03:43,680
Speaker 3: I wonder how successful this will be because part of

1123
01:03:43,720 --> 01:03:47,040
Speaker 3: the Harlequinn wrote romance, the books, the muddy books that

1124
01:03:47,040 --> 01:03:50,280
Speaker 3: we love is because you can imagine it for yourself.

1125
01:03:50,440 --> 01:03:54,480
Speaker 3: So you create these pictures. So and then what happens

1126
01:03:54,560 --> 01:03:57,280
Speaker 3: in when movies. When books are made into movies, people

1127
01:03:57,280 --> 01:03:59,320
Speaker 3: have a lot of feels because you've picked the wrong

1128
01:03:59,360 --> 01:04:01,800
Speaker 3: people in their minds, you know, the people they pictured.

1129
01:04:02,080 --> 01:04:06,959
Speaker 3: So now that they're taking this from these very descriptive

1130
01:04:07,320 --> 01:04:10,200
Speaker 3: books that really play into your imagination and then playing

1131
01:04:10,240 --> 01:04:13,000
Speaker 3: in into this, I wonder how successful it will be.

1132
01:04:14,080 --> 01:04:17,280
Speaker 2: That's a good question. And I mean part of it

1133
01:04:17,360 --> 01:04:21,680
Speaker 2: is like I cannot imagine that this is takes a

1134
01:04:21,720 --> 01:04:26,160
Speaker 2: lot of human labor to feed the whatever book prompts

1135
01:04:26,240 --> 01:04:29,120
Speaker 2: into their emit their video generator. I'm sure it takes

1136
01:04:29,240 --> 01:04:33,200
Speaker 2: not not zero human labor, but you know, I wonder

1137
01:04:33,240 --> 01:04:36,280
Speaker 2: if they've just made like a cost benefit analysis of

1138
01:04:36,480 --> 01:04:39,360
Speaker 2: even if hardly any I mean, it kind of goes

1139
01:04:39,360 --> 01:04:41,600
Speaker 2: back to what I was talking about with AI podcasts.

1140
01:04:42,320 --> 01:04:44,920
Speaker 2: I don't think, I mean, AI podcasts exist. If you

1141
01:04:44,960 --> 01:04:47,160
Speaker 2: wanted to listen to them, you absolutely could. There's plenty

1142
01:04:47,240 --> 01:04:49,479
Speaker 2: of them to choose from. People don't want to listen

1143
01:04:49,480 --> 01:04:52,400
Speaker 2: to them, And I think that the mass that these

1144
01:04:52,440 --> 01:04:54,760
Speaker 2: people are doing are that like, oh, well, they're cheap

1145
01:04:54,800 --> 01:04:57,480
Speaker 2: to make. So even if not a lot of people

1146
01:04:57,760 --> 01:04:59,760
Speaker 2: want this kind of content, and boy do they not

1147
01:04:59,800 --> 01:05:02,320
Speaker 2: want that's kind of AI generated content. Even if a

1148
01:05:02,320 --> 01:05:04,800
Speaker 2: few people engage with it, that might make it worth it.

1149
01:05:05,000 --> 01:05:07,000
Speaker 2: That's what I kind of That's the sort of math

1150
01:05:07,040 --> 01:05:07,840
Speaker 2: I bet they're doing.

1151
01:05:08,920 --> 01:05:11,240
Speaker 1: It's interesting to see how many cases of this are

1152
01:05:11,280 --> 01:05:16,960
Speaker 1: playing out right now of maybe companies seeing oh there

1153
01:05:17,120 --> 01:05:20,000
Speaker 1: is something more to this math that is not just

1154
01:05:21,320 --> 01:05:27,800
Speaker 1: the monetary because it has become like you as a creative,

1155
01:05:27,880 --> 01:05:31,360
Speaker 1: you do not want to get the AI thing attached

1156
01:05:31,400 --> 01:05:36,080
Speaker 1: to you. No, that is not good. It is something

1157
01:05:36,200 --> 01:05:40,600
Speaker 1: that you try to avoid. And I do think that

1158
01:05:40,840 --> 01:05:44,840
Speaker 1: it touches on some anxieties we have, but also what

1159
01:05:44,960 --> 01:05:48,840
Speaker 1: people want, like what you were talking about earlier, like

1160
01:05:48,880 --> 01:05:53,800
Speaker 1: that trust, which I think personally, as someone in this

1161
01:05:53,920 --> 01:05:56,080
Speaker 1: industry a company should pay attention to.

1162
01:05:57,240 --> 01:06:00,840
Speaker 2: Yeah, I mean in my talk about AIU, that was

1163
01:06:00,880 --> 01:06:03,920
Speaker 2: my entire kind of so what. Yes, it might be

1164
01:06:04,200 --> 01:06:07,960
Speaker 2: cheap and quick and easy to produce AI generative content,

1165
01:06:08,800 --> 01:06:12,880
Speaker 2: but in doing so, it's not just the warm, fuzzy,

1166
01:06:12,920 --> 01:06:17,080
Speaker 2: squishy reasons why somebody would want to prioritize human voices

1167
01:06:17,600 --> 01:06:20,920
Speaker 2: in content because we all trust humans, yah yah, YadA.

1168
01:06:20,960 --> 01:06:23,880
Speaker 2: It's also a business question do you want your brand

1169
01:06:23,960 --> 01:06:28,240
Speaker 2: associated with quick, cheap, lazy slop. I know that I

1170
01:06:28,280 --> 01:06:30,880
Speaker 2: do not want my brand associated with quick, cheap, lazy

1171
01:06:30,920 --> 01:06:33,200
Speaker 2: slop that might not even be accurate, might not even

1172
01:06:33,240 --> 01:06:36,480
Speaker 2: be true. And so that's not an at that's not

1173
01:06:36,600 --> 01:06:40,280
Speaker 2: just like an ethical, squishy question. It's also a dollars

1174
01:06:40,320 --> 01:06:44,120
Speaker 2: and cents business question of would that be good for

1175
01:06:44,200 --> 01:06:47,480
Speaker 2: your brand? Would that be a positive association with your

1176
01:06:47,480 --> 01:06:50,320
Speaker 2: brand to have your brand be associated with AI content

1177
01:06:50,360 --> 01:06:53,920
Speaker 2: that we know can be, let's face it, just trash,

1178
01:06:53,960 --> 01:06:56,040
Speaker 2: And so I think that we have kind of I

1179
01:06:56,280 --> 01:06:58,800
Speaker 2: would like to see the conversation reframed in that way

1180
01:06:58,840 --> 01:07:02,720
Speaker 2: so that you don't associate people who are talking about

1181
01:07:02,840 --> 01:07:05,560
Speaker 2: why it's so important to champion human voices and creative work.

1182
01:07:06,000 --> 01:07:09,920
Speaker 2: We're not sort of being cast aside of like you know, warm,

1183
01:07:09,960 --> 01:07:14,760
Speaker 2: fuzzy do gooders. We're actually talking hardline dollars and cents,

1184
01:07:14,800 --> 01:07:17,560
Speaker 2: the kind of stuff that people who are financial decision

1185
01:07:17,560 --> 01:07:20,000
Speaker 2: makers that media companies ought to be paying attention to.

1186
01:07:21,280 --> 01:07:22,760
Speaker 3: Well, we're guaranteed humans, so.

1187
01:07:23,040 --> 01:07:28,480
Speaker 2: Ah, that's right, guaranteed human I'm not saying it even worse,

1188
01:07:31,280 --> 01:07:33,240
Speaker 2: And I guess I want to end on this because

1189
01:07:33,760 --> 01:07:37,080
Speaker 2: I think the big takeaway here is how much audiences

1190
01:07:37,120 --> 01:07:39,560
Speaker 2: are telling us they do not want AI generated content

1191
01:07:39,560 --> 01:07:43,080
Speaker 2: in their writing. And for all the stuff I've talked about,

1192
01:07:43,640 --> 01:07:46,640
Speaker 2: it actually kind of makes me feel excited about being

1193
01:07:46,680 --> 01:07:50,040
Speaker 2: a creative professional. I really liked this hopeful advice from

1194
01:07:50,240 --> 01:07:52,840
Speaker 2: writer Andrea Bart's in The New York Times. She writes,

1195
01:07:53,560 --> 01:07:55,480
Speaker 2: I fear it could follow on us authors to prove

1196
01:07:55,520 --> 01:07:57,920
Speaker 2: our writing as our own. The advice I often hear

1197
01:07:57,960 --> 01:08:01,000
Speaker 2: from fellow authors is hopeful, if a bit hippy write

1198
01:08:01,040 --> 01:08:03,880
Speaker 2: something weird, break all the rules, pour your heart and

1199
01:08:03,920 --> 01:08:06,400
Speaker 2: soul into it, because nobody can tell a story quite

1200
01:08:06,440 --> 01:08:10,760
Speaker 2: like you, baby. And yeah, this whole kerfuffle has really

1201
01:08:10,760 --> 01:08:14,280
Speaker 2: got me appreciating the fact that I have terrible grammar.

1202
01:08:15,040 --> 01:08:18,760
Speaker 2: I really can't spell you know, I you know it's

1203
01:08:19,040 --> 01:08:21,880
Speaker 2: you know, and AI didn't write it. Look how badly

1204
01:08:21,920 --> 01:08:22,599
Speaker 2: written it is.

1205
01:08:25,400 --> 01:08:27,000
Speaker 3: All the astericks are mine, damn it.

1206
01:08:27,360 --> 01:08:33,559
Speaker 1: Yeah that's quite a silver learning. Yes, but it is

1207
01:08:33,640 --> 01:08:40,479
Speaker 1: true because I think when you see that, like if

1208
01:08:40,560 --> 01:08:43,400
Speaker 1: we made a mistake on the podcast, we are people,

1209
01:08:43,760 --> 01:08:45,840
Speaker 1: and people will write in and be like, you made

1210
01:08:45,880 --> 01:08:49,960
Speaker 1: this mistake and then we'll respond to it, and that

1211
01:08:50,200 --> 01:08:52,960
Speaker 1: having that kind of connection, but also having that moment

1212
01:08:53,000 --> 01:08:55,439
Speaker 1: of like, oh, yeah, it wasn't perfect because I am

1213
01:08:55,800 --> 01:08:58,800
Speaker 1: a human person who does make mistakes.

1214
01:08:59,640 --> 01:09:02,680
Speaker 2: Yeah. I was reading there was some celebrity that did

1215
01:09:02,760 --> 01:09:04,799
Speaker 2: I can't remember who it was, but that did something

1216
01:09:05,680 --> 01:09:08,400
Speaker 2: bad for which they had to publish like a notes

1217
01:09:08,439 --> 01:09:11,360
Speaker 2: app apology. And I remember the top comment and it

1218
01:09:11,400 --> 01:09:14,200
Speaker 2: was a bad apology, and the top comment was, well,

1219
01:09:14,200 --> 01:09:15,960
Speaker 2: at least we know they didn't use AI to write

1220
01:09:15,960 --> 01:09:20,479
Speaker 2: this apology. At least it came from the heart.

1221
01:09:22,400 --> 01:09:25,960
Speaker 1: I can't imagine going to AI being like, write this

1222
01:09:26,040 --> 01:09:26,960
Speaker 1: apology for me.

1223
01:09:27,040 --> 01:09:30,400
Speaker 3: Please, here's what they used to use their publicists.

1224
01:09:30,520 --> 01:09:32,800
Speaker 2: So yeah, I mean, is that any better?

1225
01:09:34,080 --> 01:09:41,600
Speaker 1: That's true? That's true. Well, once again, I have so

1226
01:09:41,640 --> 01:09:44,040
Speaker 1: many thoughts. We could keep going and going and going.

1227
01:09:44,120 --> 01:09:46,759
Speaker 1: But thank you so much Bridget for bringing this to us.

1228
01:09:47,200 --> 01:09:50,160
Speaker 1: I had not heard of it, and I learned a lot.

1229
01:09:50,720 --> 01:09:52,080
Speaker 3: Try to follow up on this case.

1230
01:09:52,680 --> 01:09:57,840
Speaker 1: Yes, yes, we should check back in. Well before we

1231
01:09:57,920 --> 01:10:00,679
Speaker 1: do that, where can the good listeners find you? Bridget?

1232
01:10:01,120 --> 01:10:04,960
Speaker 2: You can pre order my audio book Love at First Prompt,

1233
01:10:05,040 --> 01:10:07,960
Speaker 2: all about Intimacy and AI at Love at First prompt

1234
01:10:08,000 --> 01:10:10,439
Speaker 2: dot ai. You can follow my podcast there art articles

1235
01:10:10,439 --> 01:10:12,280
Speaker 2: on the internet or check me out on Instagram at

1236
01:10:12,320 --> 01:10:13,160
Speaker 2: bridget Marie and DC.

1237
01:10:13,720 --> 01:10:17,240
Speaker 1: Yes and listeners, go and do all of that. If

1238
01:10:17,280 --> 01:10:19,880
Speaker 1: you haven't already, If you would like to email us,

1239
01:10:19,960 --> 01:10:23,280
Speaker 1: you can. You can email Hello, It's stuffonnevertold you dot com.

1240
01:10:23,360 --> 01:10:24,960
Speaker 1: You can find us on Blue Sky I'm Mom Stuff

1241
01:10:25,000 --> 01:10:27,360
Speaker 1: podcast or on Instagram and TikTok and Stuff I've Never

1242
01:10:27,400 --> 01:10:29,519
Speaker 1: Told You for also on YouTube. We have some merchandise

1243
01:10:29,560 --> 01:10:31,320
Speaker 1: to Cottmuau and we have a book and can get

1244
01:10:31,320 --> 01:10:34,000
Speaker 1: wherever you get Brooks. Thanks as always to our superduce

1245
01:10:34,040 --> 01:10:36,800
Speaker 1: Christine or executive producer, my anti contributor Joey. Thank you

1246
01:10:36,920 --> 01:10:38,920
Speaker 1: and thanks to you for listening Stuff Won't Never Told

1247
01:10:38,960 --> 01:10:40,680
Speaker 1: You to protection by heart Radio. For more podcasts from

1248
01:10:40,680 --> 01:10:42,360
Speaker 1: my heart Radio, you can check out the heart Radio app,

1249
01:10:42,360 --> 01:10:44,760
Speaker 1: Apple podcast or if you listen to your favorite shows,