1
00:00:00,240 --> 00:00:04,680
Speaker 1: From UFOs to psychic powers and government conspiracies. History is

2
00:00:04,760 --> 00:00:09,080
Speaker 1: riddled with unexplained events. You can turn back now or

3
00:00:09,200 --> 00:00:12,119
Speaker 1: learn the stuff they don't want you to know. A

4
00:00:12,240 --> 00:00:24,880
Speaker 1: production of I Heart Radios How Stuff Works. Welcome back

5
00:00:24,920 --> 00:00:27,760
Speaker 1: to the show. My name is Matt, my name is

6
00:00:27,840 --> 00:00:30,680
Speaker 1: no They call me Ben. We are joined as always

7
00:00:30,680 --> 00:00:34,480
Speaker 1: with our super producer Paul Michigantrol dec and most importantly,

8
00:00:34,560 --> 00:00:37,640
Speaker 1: you are you. You are here and that makes this

9
00:00:38,080 --> 00:00:40,800
Speaker 1: stuff they don't want you to know as we are

10
00:00:40,880 --> 00:00:44,239
Speaker 1: barreling toward the end of the year. If that is

11
00:00:44,360 --> 00:00:49,680
Speaker 1: you believe in the current calendar? Yes, right? And then

12
00:00:50,240 --> 00:00:55,680
Speaker 1: is this the real life? Is this just fantasy? We're

13
00:00:55,680 --> 00:00:58,600
Speaker 1: gonna do a three part harmony now, okay here, let's

14
00:00:58,600 --> 00:01:02,240
Speaker 1: start this way. What's the what's the very last thing

15
00:01:02,560 --> 00:01:07,040
Speaker 1: you watched on video before we recorded today? Like vhs

16
00:01:07,400 --> 00:01:12,080
Speaker 1: you just video, just any video at any screen. I

17
00:01:12,120 --> 00:01:15,240
Speaker 1: can't really remember, but right now I'm feeling lost in

18
00:01:15,280 --> 00:01:19,880
Speaker 1: the woods? Alright? Is that another queen reference? That's a

19
00:01:19,920 --> 00:01:22,480
Speaker 1: reference for all the parents out there who've watched for

20
00:01:22,959 --> 00:01:27,080
Speaker 1: two Was that the song that actually held up I liked?

21
00:01:27,200 --> 00:01:30,199
Speaker 1: People were saying that the music was a little lackluster. Um,

22
00:01:30,240 --> 00:01:34,959
Speaker 1: I'm sorry, um when when when I get older, everything

23
00:01:35,160 --> 00:01:37,800
Speaker 1: is going to make sense. That is performed by Josh

24
00:01:37,840 --> 00:01:40,240
Speaker 1: Gad is one of the best songs written. Who's here?

25
00:01:40,360 --> 00:01:46,480
Speaker 1: Who's Josh Gad? He plays the Little Snowman. So so

26
00:01:46,560 --> 00:01:50,640
Speaker 1: he saw that, uh, which is entirely a deep fake. Basically,

27
00:01:50,680 --> 00:01:54,120
Speaker 1: it's you know, it's ethan real going on there. I

28
00:01:54,160 --> 00:01:58,160
Speaker 1: watched The Irishman, which is an interesting thing to bring

29
00:01:58,280 --> 00:02:01,000
Speaker 1: up when when we're talking about Day's topic because it

30
00:02:01,160 --> 00:02:06,200
Speaker 1: uses and anti aging d aging technology, which sort of

31
00:02:06,240 --> 00:02:08,840
Speaker 1: a hybrid of like video and c g I to

32
00:02:09,000 --> 00:02:12,520
Speaker 1: make the actors look younger in the earlier parts of

33
00:02:12,520 --> 00:02:14,840
Speaker 1: the film where they're you know, they're younger selves. The

34
00:02:14,880 --> 00:02:18,040
Speaker 1: movie takes place over decades, uh, and it still has

35
00:02:18,080 --> 00:02:20,280
Speaker 1: a little bit of that Uncanny Valley kind of vibe.

36
00:02:20,440 --> 00:02:22,720
Speaker 1: It looks a little just too perfect, almost like a

37
00:02:22,760 --> 00:02:26,160
Speaker 1: cut scene from a final Fantasy game or something like that. Um,

38
00:02:26,160 --> 00:02:28,920
Speaker 1: But you eventually stopped noticing it. Um. So that was

39
00:02:28,960 --> 00:02:31,120
Speaker 1: the last I've been watching that I had to watch

40
00:02:31,120 --> 00:02:33,120
Speaker 1: in two sittings. It's quite long, and I just scored

41
00:02:33,120 --> 00:02:36,400
Speaker 1: Casey would not approve. I watched The Irishman on the

42
00:02:36,440 --> 00:02:40,359
Speaker 1: recommendation of our own Poul Michigantrol Deccand, who has fantastic,

43
00:02:40,440 --> 00:02:43,720
Speaker 1: impeccable taste in film and also has a film of

44
00:02:43,880 --> 00:02:47,239
Speaker 1: his own out of It now on Amazon Prime. You

45
00:02:47,360 --> 00:02:49,720
Speaker 1: okay with me plaguing that poll. It's called Annie Get

46
00:02:49,800 --> 00:02:52,400
Speaker 1: your Gun. There are some people here at the apartment

47
00:02:52,480 --> 00:02:56,560
Speaker 1: complex and they mean business. Uh. It does take place

48
00:02:56,600 --> 00:02:59,880
Speaker 1: in the city and he is involved. It is h

49
00:03:00,360 --> 00:03:03,120
Speaker 1: Annie in the City. Do check it out. Your astute

50
00:03:03,200 --> 00:03:08,760
Speaker 1: listeners will notice some cameos from some of our podcast cohort,

51
00:03:10,080 --> 00:03:14,520
Speaker 1: especially been Well. I think my favorite turn is probably

52
00:03:14,680 --> 00:03:18,040
Speaker 1: one of our super producers, Chandler Mayze. Uh He's He's

53
00:03:18,480 --> 00:03:21,720
Speaker 1: really the scene stealer for me. Uh. So check it out,

54
00:03:22,320 --> 00:03:26,760
Speaker 1: tell us what you think, and we can assure you

55
00:03:26,960 --> 00:03:28,880
Speaker 1: to the best of our ability that the people who

56
00:03:28,919 --> 00:03:31,639
Speaker 1: appear to be on screen saying and doing things are

57
00:03:31,720 --> 00:03:35,080
Speaker 1: actually on screen saying and doing things. This is not

58
00:03:35,200 --> 00:03:40,240
Speaker 1: a revolutionary thought. For centuries, people used to say scene

59
00:03:40,520 --> 00:03:43,760
Speaker 1: is believing, and this meant that while people can make

60
00:03:43,840 --> 00:03:49,320
Speaker 1: up anything in conversation or writing, actually witnessing something with

61
00:03:49,480 --> 00:03:55,480
Speaker 1: your own eyes presented inarguable proof of an event. This

62
00:03:56,000 --> 00:04:01,200
Speaker 1: began to change with the rise of photography. Uh, photographs

63
00:04:01,280 --> 00:04:04,680
Speaker 1: could be faked. During the heyday of spiritualism in the West,

64
00:04:05,120 --> 00:04:09,120
Speaker 1: it was very common for mediums or people purporting to

65
00:04:09,200 --> 00:04:13,040
Speaker 1: be mediums to fake photographs right ye, old double exposure

66
00:04:13,320 --> 00:04:16,640
Speaker 1: right right, which was an unfamiliar alien technology to the

67
00:04:16,720 --> 00:04:20,520
Speaker 1: casual observer. We cannot blame people for believing that they

68
00:04:20,560 --> 00:04:22,680
Speaker 1: saw it with their own eyes. They were unaware of

69
00:04:22,760 --> 00:04:25,840
Speaker 1: the trickery that could or could not be involved. And

70
00:04:25,960 --> 00:04:29,120
Speaker 1: then things escalate further with the dawn of moving pictures.

71
00:04:29,240 --> 00:04:34,480
Speaker 1: That's when directors, filmmakers another industry professionals begin working assiduously

72
00:04:34,640 --> 00:04:38,640
Speaker 1: to make the unreal seem real for fantasy's sake. For

73
00:04:38,800 --> 00:04:42,280
Speaker 1: fantasy's sake, yes exactly, Matt, to bring the fantasies of

74
00:04:42,360 --> 00:04:46,120
Speaker 1: the human mind to life of a sort on screen.

75
00:04:46,480 --> 00:04:50,279
Speaker 1: Our species also quickly realized the power inherent in film,

76
00:04:50,680 --> 00:04:54,840
Speaker 1: and at times history has hinged on specific images or

77
00:04:54,880 --> 00:04:58,520
Speaker 1: even just bits of video. So today's question what happens

78
00:04:58,560 --> 00:05:01,880
Speaker 1: in a world where see is no longer believing? What

79
00:05:02,080 --> 00:05:06,400
Speaker 1: happens when the line between film fiction and film fact blur.

80
00:05:07,240 --> 00:05:10,600
Speaker 1: Let's let's start with some of the most immediately important

81
00:05:10,760 --> 00:05:15,280
Speaker 1: video in the world, the news. Here are the facts. So,

82
00:05:15,360 --> 00:05:19,640
Speaker 1: according to a survey from the Pew Research Center, of

83
00:05:19,640 --> 00:05:23,960
Speaker 1: Americans prefer watching the news um rather than reading or

84
00:05:24,120 --> 00:05:27,719
Speaker 1: listening to it. On the other hand, prefer to read

85
00:05:27,760 --> 00:05:31,160
Speaker 1: the news and nineteen percent to listen to it. And uh,

86
00:05:31,200 --> 00:05:33,960
Speaker 1: and know, those people who prefer their news on video,

87
00:05:34,680 --> 00:05:37,040
Speaker 1: most of them still really want to watch it on

88
00:05:37,200 --> 00:05:40,320
Speaker 1: a big screen on a maybe not a huge screen,

89
00:05:40,640 --> 00:05:43,880
Speaker 1: but a bigger screen like a television, um, rather than

90
00:05:44,160 --> 00:05:46,560
Speaker 1: internet video, you know, maybe on your phone or on

91
00:05:46,640 --> 00:05:49,440
Speaker 1: your computer. And that may be surprising to a lot

92
00:05:49,440 --> 00:05:52,200
Speaker 1: of people listening. I know it's a little surprising to me, right.

93
00:05:52,279 --> 00:05:55,760
Speaker 1: I think it's because, look, this is my understanding of it.

94
00:05:55,880 --> 00:05:57,880
Speaker 1: But I think I think a lot of the people

95
00:05:57,880 --> 00:06:01,000
Speaker 1: who end up responding to a Pew survey maybe are

96
00:06:01,200 --> 00:06:04,720
Speaker 1: leaning more towards people who would be watching their television's

97
00:06:05,160 --> 00:06:08,040
Speaker 1: or maybe a little skewing skewing a little older maybe,

98
00:06:08,320 --> 00:06:11,440
Speaker 1: or the kind of people who say, sure, I'll take

99
00:06:11,480 --> 00:06:15,240
Speaker 1: a survey. Yeah, right, that's just my read of it.

100
00:06:15,320 --> 00:06:17,719
Speaker 1: I think it's a good point. And then Pew found

101
00:06:17,920 --> 00:06:21,520
Speaker 1: just over four in ten, so four out of ten

102
00:06:21,760 --> 00:06:25,760
Speaker 1: United States adults prefer TV, compared with about a third

103
00:06:25,800 --> 00:06:29,320
Speaker 1: who prefer the web, and then four who prefer radio

104
00:06:29,880 --> 00:06:33,760
Speaker 1: and seven only seven percent who actually want to read

105
00:06:33,960 --> 00:06:37,120
Speaker 1: the words of the news right for one reason or another.

106
00:06:37,240 --> 00:06:40,960
Speaker 1: That's the other thing. We don't have a solid methodological

107
00:06:41,440 --> 00:06:47,120
Speaker 1: grasp of how these questions were accurate, like how how

108
00:06:47,200 --> 00:06:49,120
Speaker 1: they were built, how they were framed, and there's so

109
00:06:49,320 --> 00:06:53,160
Speaker 1: much that goes into there. But these numbers feel pretty solid,

110
00:06:53,360 --> 00:06:57,960
Speaker 1: if surprising, I think it's safe to say that, Uh,

111
00:06:58,200 --> 00:07:02,240
Speaker 1: the four of us recording obviously long time listeners. You

112
00:07:02,320 --> 00:07:08,080
Speaker 1: know this. We don't just go home and turn on CNN,

113
00:07:08,520 --> 00:07:12,200
Speaker 1: Fox or MSNBC or something. We we get to the

114
00:07:12,360 --> 00:07:15,720
Speaker 1: edges of stuff. We look into the weird things. Generally,

115
00:07:16,040 --> 00:07:19,240
Speaker 1: I just really quickly. Maybe we do a quick poll here. Generally,

116
00:07:19,280 --> 00:07:22,720
Speaker 1: if I'm if I'm consuming the news, it is through

117
00:07:22,920 --> 00:07:26,640
Speaker 1: a written article. What about you, guys? Yeah, I typically

118
00:07:26,760 --> 00:07:28,480
Speaker 1: read stuff because of the way we research for the

119
00:07:28,560 --> 00:07:31,120
Speaker 1: show and other shows. UM. It tends to be a

120
00:07:31,200 --> 00:07:34,400
Speaker 1: kind of our bread and butter. Occasionally watch documentary, UM

121
00:07:34,680 --> 00:07:36,840
Speaker 1: read parts of books. We don't always have the leg

122
00:07:36,960 --> 00:07:40,160
Speaker 1: luxury of reading, you know, entire book for one podcast episode,

123
00:07:40,200 --> 00:07:43,440
Speaker 1: but definitely excerpts in chapters and then occasionally and stuff

124
00:07:43,480 --> 00:07:45,680
Speaker 1: like YouTube videos, which tend to sort of show you

125
00:07:45,800 --> 00:07:49,360
Speaker 1: where the kind of internet culture, the zeitgeist is is, uh,

126
00:07:49,600 --> 00:07:51,120
Speaker 1: you know, coming down on one side or the other

127
00:07:51,200 --> 00:07:54,640
Speaker 1: of an issue. I think I would watch television more

128
00:07:54,840 --> 00:07:59,480
Speaker 1: often if there were more if there was a larger

129
00:07:59,560 --> 00:08:02,800
Speaker 1: degree of variants in the narratives and stories presented in

130
00:08:02,920 --> 00:08:05,800
Speaker 1: the West, at least, so people like me naturally end

131
00:08:05,880 --> 00:08:08,679
Speaker 1: up watching things on the Internet because that's the easiest

132
00:08:08,760 --> 00:08:11,520
Speaker 1: way to get opposing viewpoints. Right. You want to see

133
00:08:11,640 --> 00:08:16,040
Speaker 1: tin Hua or Al Jazeera or RT, all of which

134
00:08:16,080 --> 00:08:19,120
Speaker 1: are imperfect, then you you probably are not going to

135
00:08:19,280 --> 00:08:22,680
Speaker 1: see that in your basic Comcast package, right, So the

136
00:08:22,760 --> 00:08:26,720
Speaker 1: majority of Americans don't prefer online video just yet, even

137
00:08:26,760 --> 00:08:29,520
Speaker 1: though you know, I think a lot of us listening

138
00:08:29,600 --> 00:08:33,120
Speaker 1: because we listen to podcasts, are already all up on

139
00:08:33,160 --> 00:08:37,240
Speaker 1: the internet for video content for news, right and for

140
00:08:37,600 --> 00:08:41,960
Speaker 1: honestly audio podcasts about the news, like The Daily or Guys,

141
00:08:42,520 --> 00:08:45,320
Speaker 1: the Daily Site Guys. There's all kinds of shows out

142
00:08:45,360 --> 00:08:48,040
Speaker 1: there that will give you what you need in an

143
00:08:48,080 --> 00:08:53,800
Speaker 1: audio way. The Economist, BBC, Global News, etcetera. And this

144
00:08:54,240 --> 00:08:57,520
Speaker 1: this makes sense for US. But even if the majority

145
00:08:57,720 --> 00:09:03,199
Speaker 1: of Americans or the largest suave of Americans don't prefer

146
00:09:03,320 --> 00:09:06,600
Speaker 1: online video just yet, at least for the news, there's

147
00:09:06,640 --> 00:09:08,880
Speaker 1: no arguing that the Internet has made a massive impact

148
00:09:08,960 --> 00:09:12,880
Speaker 1: on video technology and filming in general. And we're in

149
00:09:13,040 --> 00:09:16,719
Speaker 1: the midst of this evolution as we speak. It is

150
00:09:16,840 --> 00:09:21,839
Speaker 1: an evolution at a very fast cadence. The earlier the

151
00:09:21,880 --> 00:09:25,079
Speaker 1: earliest in the earlier film technology, and you guys and

152
00:09:25,480 --> 00:09:28,760
Speaker 1: Paul know this very very well. Like the earliest stuff

153
00:09:29,280 --> 00:09:33,240
Speaker 1: was super expensive, highly specialized, it was cantankerous, and it

154
00:09:33,360 --> 00:09:37,360
Speaker 1: broke a lot. Add to that, distribution channels for anything

155
00:09:37,520 --> 00:09:41,319
Speaker 1: filmed were owned by a fairly small number of corporate

156
00:09:41,400 --> 00:09:44,520
Speaker 1: or state interests, and this meant that whether your film

157
00:09:44,679 --> 00:09:47,559
Speaker 1: was a sci fi blockbuster or whether it was just

158
00:09:47,800 --> 00:09:51,160
Speaker 1: some shameless World War two propaganda, it would go through

159
00:09:51,240 --> 00:09:55,120
Speaker 1: predictable channels. The same people would shake the same hands

160
00:09:55,240 --> 00:09:57,839
Speaker 1: to get that to a theater or a screen near you.

161
00:09:58,360 --> 00:10:02,440
Speaker 1: But the technology continue to evolve. Soon people were able

162
00:10:02,480 --> 00:10:06,600
Speaker 1: to purchase televisions and instead of relying on radios, they

163
00:10:06,760 --> 00:10:10,320
Speaker 1: would be able to put an image with a sound.

164
00:10:10,760 --> 00:10:14,000
Speaker 1: This played a huge role. Like the first televised presidential debate.

165
00:10:14,280 --> 00:10:16,880
Speaker 1: You know about this one, right, Nixon and Kennedy. Oh yeah,

166
00:10:16,960 --> 00:10:20,000
Speaker 1: the just the difference in their appearances, and how much

167
00:10:20,040 --> 00:10:22,720
Speaker 1: of a difference that made outside of the words they

168
00:10:22,800 --> 00:10:25,960
Speaker 1: were saying exactly. So, people listening to the radio, regardless

169
00:10:25,960 --> 00:10:28,640
Speaker 1: of political party, thought that Nixon won the debate, and

170
00:10:28,679 --> 00:10:31,679
Speaker 1: people watching television, regardless of the political party, thought that

171
00:10:32,280 --> 00:10:35,600
Speaker 1: Kennedy won the debate. So we were able to bring

172
00:10:35,720 --> 00:10:37,960
Speaker 1: a small version in the big screen to living rooms

173
00:10:38,000 --> 00:10:41,800
Speaker 1: around the world and happily ever after, right, yes, yeah, Well,

174
00:10:42,120 --> 00:10:46,000
Speaker 1: so then we had the rise of home projectors vhs,

175
00:10:46,360 --> 00:10:48,520
Speaker 1: like you had mentioned earlier in old DVD and so on.

176
00:10:48,840 --> 00:10:51,439
Speaker 1: These allowed viewers more agency. You didn't just have to

177
00:10:51,520 --> 00:10:54,560
Speaker 1: stick to the programming dictated by your TV channels. Your

178
00:10:54,600 --> 00:10:58,080
Speaker 1: CBS is, your NBC's. It's like a rudimentary early version

179
00:10:58,160 --> 00:11:00,600
Speaker 1: of what we now think of as on demand um

180
00:11:01,080 --> 00:11:05,520
Speaker 1: consuming of of media, which has really changed the game entirely, right, right,

181
00:11:05,960 --> 00:11:09,720
Speaker 1: And we're still segments of the industry are still attempting

182
00:11:09,800 --> 00:11:13,760
Speaker 1: to keep up with that change. As film technology improved,

183
00:11:14,440 --> 00:11:19,319
Speaker 1: cost for equipment began to create her Nowadays, filmmakers don't

184
00:11:19,559 --> 00:11:21,640
Speaker 1: always have to go to a studio. You don't have

185
00:11:21,760 --> 00:11:24,920
Speaker 1: to shake the same hands at Warner or whatever, and

186
00:11:25,000 --> 00:11:29,079
Speaker 1: you could shoot a low budget movie with technology available,

187
00:11:29,400 --> 00:11:32,839
Speaker 1: and you know, make VHS copies let's say, of it,

188
00:11:32,920 --> 00:11:36,199
Speaker 1: and distribute it locally or maybe two stores in a

189
00:11:36,320 --> 00:11:38,520
Speaker 1: in a local area. I'm just saying, like, in that time,

190
00:11:38,640 --> 00:11:43,280
Speaker 1: even before you know nowadays, it was possible to escape

191
00:11:43,559 --> 00:11:48,080
Speaker 1: the studio, right, right, And this escape from the studios,

192
00:11:48,160 --> 00:11:51,719
Speaker 1: exodus from the studio, has accelerated the rise of home

193
00:11:51,800 --> 00:11:55,719
Speaker 1: and internet video commingled and it led us to the

194
00:11:55,800 --> 00:11:58,400
Speaker 1: current world, a world in which anyone with a little

195
00:11:58,400 --> 00:12:02,440
Speaker 1: bit of scratch can purchase tools to make their own video. Right.

196
00:12:02,480 --> 00:12:05,080
Speaker 1: And not only is it the democratization of creating, it's

197
00:12:05,120 --> 00:12:09,000
Speaker 1: the democratization of consuming, because you know, anyone can upload

198
00:12:09,040 --> 00:12:11,839
Speaker 1: a video to the internet for free on YouTube and

199
00:12:12,160 --> 00:12:16,480
Speaker 1: you can command your own audience depending on the quality

200
00:12:16,640 --> 00:12:19,280
Speaker 1: of your workland quality is sort of a loose term there,

201
00:12:19,360 --> 00:12:21,880
Speaker 1: I guess, but at least in terms of how uh

202
00:12:22,040 --> 00:12:25,120
Speaker 1: salacious or how kind of hooky it is and how

203
00:12:25,200 --> 00:12:28,280
Speaker 1: much it grabs people's attention, you can actually command uh

204
00:12:28,440 --> 00:12:33,440
Speaker 1: an audience of scale as a creator with very little overhead. Yeah,

205
00:12:33,520 --> 00:12:39,559
Speaker 1: with worldwide virtually instant distribution sick. Yeah, So if these

206
00:12:39,600 --> 00:12:43,479
Speaker 1: filmmakers have an Internet connection, they can bypass those antiquated

207
00:12:43,600 --> 00:12:47,160
Speaker 1: channels of distribution send their work across the planet. Anyone

208
00:12:47,240 --> 00:12:50,400
Speaker 1: else with web access can, in theory watch it to

209
00:12:50,440 --> 00:12:52,960
Speaker 1: their hearts content as much or as little as they wish.

210
00:12:53,160 --> 00:12:56,719
Speaker 1: But don't feel bad for the old guard. The democratization

211
00:12:56,800 --> 00:12:59,880
Speaker 1: of a V technology did not lead them to extinction.

212
00:13:00,360 --> 00:13:03,080
Speaker 1: They evolved as well. Yeah they had to. I mean

213
00:13:03,480 --> 00:13:06,920
Speaker 1: m a journalist in say Hong Kong now during the

214
00:13:07,200 --> 00:13:11,559
Speaker 1: protests that we're experiencing can immediately post video and help

215
00:13:11,640 --> 00:13:14,679
Speaker 1: out those news providers shedding light on events that might

216
00:13:14,800 --> 00:13:18,120
Speaker 1: have been otherwise um relegated to the shadows. Reporters in

217
00:13:18,200 --> 00:13:22,440
Speaker 1: Belgium or Bolivia can record political announcements that the capital

218
00:13:22,679 --> 00:13:26,280
Speaker 1: live and stream it to millions of viewers or followers.

219
00:13:26,520 --> 00:13:30,760
Speaker 1: The president can, you know, before having important meetings with

220
00:13:30,960 --> 00:13:34,240
Speaker 1: world leaders, decided to talk for forty five minutes because

221
00:13:34,320 --> 00:13:38,319
Speaker 1: the cameras are live, and other world leaders can be

222
00:13:38,400 --> 00:13:42,120
Speaker 1: caught doing their own rendition of mean girls to significant

223
00:13:42,160 --> 00:13:47,679
Speaker 1: Ghio political events. All at all, this is impressive stuff, right.

224
00:13:47,840 --> 00:13:53,040
Speaker 1: Live video has the potential to bring our incredibly litigious

225
00:13:53,080 --> 00:13:56,680
Speaker 1: and bellico species a little bit closer together. If everyone

226
00:13:56,880 --> 00:13:59,720
Speaker 1: can see something happening, playing his day at the same

227
00:13:59,760 --> 00:14:03,000
Speaker 1: time while it's happening, what on earth is there left

228
00:14:03,160 --> 00:14:10,480
Speaker 1: to argue about? Seeing is believing right? No? What? But

229
00:14:10,640 --> 00:14:13,000
Speaker 1: we'll talk about it right after work from our spotsor

230
00:14:16,440 --> 00:14:18,480
Speaker 1: This episode of stuff they don't want you to know

231
00:14:18,679 --> 00:14:21,920
Speaker 1: is brought to you by Express VPN. Recently, over one

232
00:14:22,560 --> 00:14:25,920
Speaker 1: million people had their personal information stolen in a major

233
00:14:26,120 --> 00:14:30,640
Speaker 1: data breach. We're talking social security numbers, contact details, credit scores,

234
00:14:30,640 --> 00:14:34,000
Speaker 1: and more, all taken from Capital One customers. Oh that's me,

235
00:14:34,240 --> 00:14:38,000
Speaker 1: and that means there's a good chance you were personally affected. Folks.

236
00:14:38,080 --> 00:14:41,480
Speaker 1: These kinds of attacks are getting more frequent and more severe.

237
00:14:41,560 --> 00:14:46,320
Speaker 1: It's not just Capital One, zac Fax, Facebook, eBay, Uber, PlayStation,

238
00:14:46,600 --> 00:14:50,200
Speaker 1: and Yahoo. They have all leaked passwords, credit card info,

239
00:14:50,520 --> 00:14:54,480
Speaker 1: and bank numbers belonging to billions of users. That's why

240
00:14:54,560 --> 00:14:58,600
Speaker 1: we use Express VPN to safeguard our personal data online.

241
00:14:58,720 --> 00:15:01,240
Speaker 1: According to recent report, it's hackers can make up to

242
00:15:01,400 --> 00:15:04,400
Speaker 1: one thousand dollars when they sell someone's personal information on

243
00:15:04,480 --> 00:15:08,880
Speaker 1: the dark web, making people like us easy lucrative targets.

244
00:15:09,040 --> 00:15:11,760
Speaker 1: Express vpn is an app, and the app connects with

245
00:15:11,880 --> 00:15:14,800
Speaker 1: just one click. It's lightning fast, and the best part

246
00:15:14,920 --> 00:15:18,200
Speaker 1: is Express vpn costs less than seven bucks a month

247
00:15:18,320 --> 00:15:21,520
Speaker 1: to use and Listen, Honestly, if a breach can happen

248
00:15:21,560 --> 00:15:24,840
Speaker 1: to Capital One, it can easily happen to anyone else.

249
00:15:25,040 --> 00:15:28,720
Speaker 1: So protect yourself with Express vpn, the number one VPN

250
00:15:28,880 --> 00:15:32,160
Speaker 1: rated by tech Radar, c Net, the Verge, and countless others.

251
00:15:32,360 --> 00:15:35,800
Speaker 1: You can use our special link Express vpn dot com

252
00:15:36,000 --> 00:15:39,600
Speaker 1: slash conspiracy right now to arm yourself with an extra

253
00:15:39,760 --> 00:15:43,360
Speaker 1: three months of Express vpn for free, Support the show

254
00:15:43,400 --> 00:15:46,640
Speaker 1: and keep your information safe. That's Express vpn dot com

255
00:15:46,760 --> 00:15:58,280
Speaker 1: slash conspiracy for an extra three months free. Here's where

256
00:15:58,360 --> 00:16:02,480
Speaker 1: it gets crazy. Hey, remember when we said that technology,

257
00:16:02,640 --> 00:16:04,840
Speaker 1: that that stuff that lets us do all the things

258
00:16:04,880 --> 00:16:08,400
Speaker 1: that we like to do. Um, it's still evolving. Remember

259
00:16:08,480 --> 00:16:12,680
Speaker 1: we talked about that. Uh, Well, we're on the precipice

260
00:16:12,760 --> 00:16:16,120
Speaker 1: of this other thing. This uh, this new shift in

261
00:16:16,320 --> 00:16:21,160
Speaker 1: video technology, and it is not um, I don't see

262
00:16:21,240 --> 00:16:25,960
Speaker 1: good things happening from it. It's gonna be like amazing

263
00:16:26,080 --> 00:16:31,200
Speaker 1: for those dank memes, but for everything else. Oh boy, Yeah,

264
00:16:31,520 --> 00:16:36,080
Speaker 1: it's an inherently conspiratorial shift. Today we are talking about

265
00:16:36,120 --> 00:16:39,520
Speaker 1: the rise of the deep fake, which sounds hyperbolic but

266
00:16:39,680 --> 00:16:42,440
Speaker 1: very much is not. What is a deep fake? Somebody

267
00:16:42,520 --> 00:16:45,040
Speaker 1: might be saying, well, we're glad you asked. Our story

268
00:16:45,120 --> 00:16:52,080
Speaker 1: starts with a fellow named Ian J. Goodfellow. I think

269
00:16:52,120 --> 00:16:55,040
Speaker 1: that's his really, But while you may not have heard

270
00:16:55,080 --> 00:16:57,880
Speaker 1: of his name before, and while you may be unfamiliar

271
00:16:57,920 --> 00:17:01,280
Speaker 1: with the concept of deep fakes, you have almost certainly

272
00:17:01,440 --> 00:17:06,320
Speaker 1: already encountered some version of Ian Goodfellow's work. He works

273
00:17:06,400 --> 00:17:10,800
Speaker 1: extensively in areas of what we call machine learning. Anybody

274
00:17:10,920 --> 00:17:16,000
Speaker 1: remembers our earlier conversation about machine consciousness with a friend

275
00:17:16,080 --> 00:17:19,159
Speaker 1: of the show, Damian Patrick Williams is probably familiar with

276
00:17:19,320 --> 00:17:23,800
Speaker 1: these uh edges of science, the bleeding edge of artificial intelligence.

277
00:17:24,000 --> 00:17:27,600
Speaker 1: Here's what he did. Essentially, he taught algorithms to play

278
00:17:27,840 --> 00:17:30,720
Speaker 1: games with each other, specifically to kind of play game theory,

279
00:17:31,160 --> 00:17:34,320
Speaker 1: which is still incredibly strange and important. And well, yeah,

280
00:17:34,400 --> 00:17:37,879
Speaker 1: let's let's talk about what deep learning is really, because

281
00:17:38,520 --> 00:17:43,080
Speaker 1: we're talking about machine learning just essentially teaching versions of

282
00:17:43,240 --> 00:17:47,720
Speaker 1: artificial intelligence how to learn. And in this case, this

283
00:17:47,960 --> 00:17:50,320
Speaker 1: is a sub field of machine learning we're gonna talk

284
00:17:50,320 --> 00:17:54,359
Speaker 1: about called deep learning. And it's fascinating stuff. It is

285
00:17:54,440 --> 00:17:57,200
Speaker 1: the it's also the stuff of nightmares. It's the stuff

286
00:17:57,240 --> 00:17:59,760
Speaker 1: of our eventual future. And there's no way around it.

287
00:18:00,320 --> 00:18:04,040
Speaker 1: But it's the concept of focusing algorithms that are or

288
00:18:04,160 --> 00:18:08,119
Speaker 1: focusing on algorithms that are inspired by the way the

289
00:18:08,400 --> 00:18:13,639
Speaker 1: human brain functions. And they're called artificial neural networks. And

290
00:18:13,720 --> 00:18:16,640
Speaker 1: if you're watching the final season of Silicon Valley, you're

291
00:18:16,680 --> 00:18:18,680
Speaker 1: getting kind of a crash course in that right now

292
00:18:18,920 --> 00:18:21,960
Speaker 1: as the well, as we're recording this, the penultimate episode

293
00:18:22,080 --> 00:18:25,960
Speaker 1: just came out. Um, But anyway, it's it's really fascinating stuff.

294
00:18:26,160 --> 00:18:29,640
Speaker 1: And good Fellow, our friend Ian J. Goodfellow, actually wrote

295
00:18:29,640 --> 00:18:33,840
Speaker 1: a book on this subject. Yeah, a book called his

296
00:18:33,920 --> 00:18:37,760
Speaker 1: book about deep learning is called deep learning. Hey, it's coining.

297
00:18:37,840 --> 00:18:42,720
Speaker 1: He's a busy guy. So he explains deep learning this way.

298
00:18:42,800 --> 00:18:45,280
Speaker 1: He thinks of it in terms of a hierarchy of concepts,

299
00:18:45,359 --> 00:18:48,480
Speaker 1: and he says having a hierarchy of concepts allows the

300
00:18:48,560 --> 00:18:52,440
Speaker 1: computer to learn these complicated concepts by building them out

301
00:18:52,520 --> 00:18:55,639
Speaker 1: of simpler ones, which is what we have spent a

302
00:18:55,680 --> 00:18:58,120
Speaker 1: lot of time doing it. How stuff works right this episode,

303
00:18:58,160 --> 00:19:01,560
Speaker 1: even which kind of yeah, that's how, because that's how

304
00:19:01,640 --> 00:19:05,159
Speaker 1: our brains often approach things. We build towards that gestalt.

305
00:19:06,080 --> 00:19:09,320
Speaker 1: So he says, if we draw a graph showing how

306
00:19:09,480 --> 00:19:12,440
Speaker 1: these concepts are built upon each other, we see that

307
00:19:12,560 --> 00:19:15,680
Speaker 1: the graph is just visually, it's deep. It has a

308
00:19:15,760 --> 00:19:17,800
Speaker 1: ton of layers. And he says, for this reason, we

309
00:19:17,880 --> 00:19:23,159
Speaker 1: call this approach to artificial intelligence deep learning. In plain English,

310
00:19:23,240 --> 00:19:27,440
Speaker 1: what that means is that we as a species have

311
00:19:27,680 --> 00:19:31,840
Speaker 1: programs that can work more and more like an organic brain,

312
00:19:31,960 --> 00:19:36,600
Speaker 1: and artificial neural network is meant to function more and

313
00:19:36,800 --> 00:19:41,480
Speaker 1: more like a brain. And he has one very well

314
00:19:41,560 --> 00:19:44,879
Speaker 1: known invention. This is the this is the engine behind

315
00:19:45,000 --> 00:19:47,320
Speaker 1: his work that you have already seen. Even if you've

316
00:19:47,359 --> 00:19:49,080
Speaker 1: never heard of a deep fake, you've never heard of

317
00:19:49,160 --> 00:19:51,760
Speaker 1: being good fellow, and you've never heard of machine or

318
00:19:51,920 --> 00:19:54,960
Speaker 1: deep learning, that's right, uh, And it's his most well

319
00:19:55,040 --> 00:20:00,560
Speaker 1: known invention innovation. It's something that's called generative ad serial

320
00:20:00,880 --> 00:20:05,080
Speaker 1: network or a general adversarial network or GAN and gans

321
00:20:05,359 --> 00:20:11,760
Speaker 1: um enable algorithms to move beyond classifying data into actually

322
00:20:12,000 --> 00:20:17,080
Speaker 1: generating or creating images. Oh yes, now, maybe you're the

323
00:20:17,400 --> 00:20:20,040
Speaker 1: gears are turning in your mind. Maybe I have seen

324
00:20:20,240 --> 00:20:22,600
Speaker 1: something like is it like the deep dream kind of

325
00:20:22,680 --> 00:20:26,359
Speaker 1: stuff Google's deep dreams, right, yeah, where it would uh

326
00:20:26,880 --> 00:20:29,760
Speaker 1: sort of take a like a face or an image

327
00:20:30,080 --> 00:20:33,600
Speaker 1: and then it would pull things from elsewhere on the

328
00:20:33,680 --> 00:20:36,439
Speaker 1: Internet that it's sort of matched up to those textures

329
00:20:36,680 --> 00:20:40,520
Speaker 1: or you know, spaces like to fill with other images

330
00:20:40,600 --> 00:20:44,240
Speaker 1: of say like dogs or slugs or what have you.

331
00:20:44,520 --> 00:20:47,000
Speaker 1: And then people started animating them and they became these

332
00:20:47,119 --> 00:20:52,480
Speaker 1: like hellscape kind of psychedelic nightmare images, very Dolly's very

333
00:20:52,760 --> 00:20:56,639
Speaker 1: essuy just like super h trippy really for lack of

334
00:20:56,680 --> 00:20:59,159
Speaker 1: a better term. Yeah, you're you're on the right track there,

335
00:20:59,240 --> 00:21:02,040
Speaker 1: because they are indeed related. In terms of the science,

336
00:21:02,400 --> 00:21:06,160
Speaker 1: deep dream is a Mikes use of rather something called

337
00:21:06,200 --> 00:21:10,960
Speaker 1: a convolutional neural network or a comv net or CNN,

338
00:21:11,560 --> 00:21:14,160
Speaker 1: which could be a little confusing. So they're they're very

339
00:21:14,240 --> 00:21:21,200
Speaker 1: similar approaches at basis. So, uh, these generational adversarial networks

340
00:21:21,440 --> 00:21:23,720
Speaker 1: are trying to trick each other. They can move beyond

341
00:21:23,880 --> 00:21:30,920
Speaker 1: classifying data into generating or creating data, generating or creating images.

342
00:21:31,400 --> 00:21:36,480
Speaker 1: So these two networks, these two generative adversarial networks, they

343
00:21:36,640 --> 00:21:39,800
Speaker 1: attempt to fool each other into thinking that a given

344
00:21:39,840 --> 00:21:43,399
Speaker 1: image is real and using as little as one image.

345
00:21:43,440 --> 00:21:47,560
Speaker 1: From that back and forth between what's called the generative

346
00:21:47,640 --> 00:21:52,760
Speaker 1: and the discriminative sides of this thing. Just using one image,

347
00:21:53,000 --> 00:21:56,320
Speaker 1: they can create a video clip of a person, so

348
00:21:56,440 --> 00:21:59,119
Speaker 1: they can animate a picture. And they also can uh,

349
00:21:59,560 --> 00:22:01,800
Speaker 1: you can all so taking a step further and have

350
00:22:02,040 --> 00:22:05,359
Speaker 1: that animation speaking and what sounds like that person or

351
00:22:05,440 --> 00:22:10,359
Speaker 1: that images voice. Samsung's AI Center released a report on

352
00:22:10,440 --> 00:22:14,680
Speaker 1: the science behind DAN and they said such an approach

353
00:22:14,800 --> 00:22:18,560
Speaker 1: is able to learn highly realistic and personalized talking head

354
00:22:18,640 --> 00:22:22,680
Speaker 1: models of new people and even portrait paintings. It's for people,

355
00:22:22,960 --> 00:22:27,480
Speaker 1: new people, created people, generated humans, and they look great.

356
00:22:27,840 --> 00:22:30,679
Speaker 1: They really do. I'll say it. Some of them are attractive.

357
00:22:32,760 --> 00:22:35,840
Speaker 1: If you didn't know, and you just saw a picture

358
00:22:36,200 --> 00:22:39,080
Speaker 1: of a of one of these generated images on your

359
00:22:39,680 --> 00:22:42,639
Speaker 1: dating app of choice, yeah, there are a couple you

360
00:22:42,680 --> 00:22:45,320
Speaker 1: would probably swipe right on. I wouldn't know which way

361
00:22:45,359 --> 00:22:47,560
Speaker 1: to swipe because I don't understand those things, but I

362
00:22:48,359 --> 00:22:51,560
Speaker 1: totally get what you're saying. So and it's startling because

363
00:22:51,600 --> 00:22:54,560
Speaker 1: now even now you can take tests where you attempt

364
00:22:54,600 --> 00:22:59,040
Speaker 1: to identify a real person from a generated talking head

365
00:22:59,119 --> 00:23:02,280
Speaker 1: or image, and it's tough. Yeah, we're getting closer and

366
00:23:02,400 --> 00:23:06,680
Speaker 1: closer to traversing the Uncanny Valley. Oh yeah, well it's

367
00:23:07,920 --> 00:23:12,200
Speaker 1: it really is frustrating that it's not easier because for

368
00:23:12,320 --> 00:23:14,960
Speaker 1: so long there and I think, you know, you referenced

369
00:23:15,000 --> 00:23:18,240
Speaker 1: Final Fantasy cut scenes or something earlier on Larry and

370
00:23:18,280 --> 00:23:21,040
Speaker 1: I remember when that Final Fantasy movie came out a

371
00:23:21,160 --> 00:23:23,680
Speaker 1: long time ago. There was all too good at that point,

372
00:23:23,720 --> 00:23:25,560
Speaker 1: well it was, yeah, it was all computer generated and

373
00:23:25,680 --> 00:23:27,879
Speaker 1: it looked fantastic. It was a feature film, and I

374
00:23:27,920 --> 00:23:31,640
Speaker 1: remember thinking how incredible it was seeing that, and then

375
00:23:32,080 --> 00:23:36,280
Speaker 1: seeing something like Avatar, where you've got these um, I

376
00:23:36,359 --> 00:23:39,960
Speaker 1: forget what they're called, the Navy, the nav that don't

377
00:23:40,080 --> 00:23:44,440
Speaker 1: look human necessarily, but they look real enough, right, And

378
00:23:44,520 --> 00:23:46,480
Speaker 1: then when you get to something like this and you're

379
00:23:46,560 --> 00:23:52,000
Speaker 1: looking at these just the portraits, even it feels it

380
00:23:52,400 --> 00:23:55,560
Speaker 1: feels pretty scary, not being able to trust your eyes

381
00:23:56,200 --> 00:23:58,440
Speaker 1: to know if something is real or not, even though

382
00:23:58,480 --> 00:24:01,440
Speaker 1: it's generated. Either way, it was an actual image of

383
00:24:01,520 --> 00:24:04,600
Speaker 1: a person and it was taken converted into data, you know,

384
00:24:04,680 --> 00:24:07,639
Speaker 1: ones and zeros, then displayed on your screen. That's not

385
00:24:07,760 --> 00:24:11,639
Speaker 1: a real person necessarily, we're looking at a representation. But

386
00:24:11,960 --> 00:24:15,359
Speaker 1: just knowing that a computer can fool you, um that

387
00:24:15,560 --> 00:24:18,600
Speaker 1: hard is is pretty creative, very easily. Yeah, that's kind

388
00:24:18,600 --> 00:24:21,720
Speaker 1: of why I wonder too, why well, you know, to

389
00:24:21,840 --> 00:24:24,119
Speaker 1: be fair, a lot of the d aging stuff in

390
00:24:24,359 --> 00:24:27,520
Speaker 1: movies is quite good, so much better than it's ever been.

391
00:24:27,720 --> 00:24:30,040
Speaker 1: But there were a couple of spots in The Irishman

392
00:24:30,119 --> 00:24:32,560
Speaker 1: where I was like, really, like, I thought the technology

393
00:24:32,680 --> 00:24:35,200
Speaker 1: was better than this, and and it is. But I

394
00:24:35,240 --> 00:24:38,280
Speaker 1: guess it's different when you're making a younger version that

395
00:24:38,400 --> 00:24:41,479
Speaker 1: has to then coincide with written lines and map up

396
00:24:41,520 --> 00:24:44,480
Speaker 1: to you know, an actor's face and you know, look

397
00:24:44,560 --> 00:24:46,399
Speaker 1: believable in terms of the way the mouth moves and

398
00:24:46,560 --> 00:24:49,200
Speaker 1: acting ticks and all of that stuff that's specific to

399
00:24:50,040 --> 00:24:53,800
Speaker 1: a scene rather than a pre existing video of some kind, right,

400
00:24:53,920 --> 00:24:57,120
Speaker 1: And it also goes down to how much existing footage

401
00:24:57,359 --> 00:25:00,679
Speaker 1: they may be able to obtain of in the person

402
00:25:00,920 --> 00:25:03,240
Speaker 1: was that age. So that's one of the reasons why

403
00:25:04,000 --> 00:25:07,600
Speaker 1: it probably works best with celebrities and political figures because

404
00:25:07,920 --> 00:25:11,080
Speaker 1: there's just you know, not everybody did taxi driver when

405
00:25:11,119 --> 00:25:14,359
Speaker 1: they were a kid, right, So so the stuff with

406
00:25:14,440 --> 00:25:17,760
Speaker 1: de Niro specifically, I would argue he just always looked

407
00:25:17,880 --> 00:25:22,480
Speaker 1: vaguely in his late thirties to early forties and somewhat perturbed,

408
00:25:22,600 --> 00:25:25,560
Speaker 1: like was it a shart or is he mad about

409
00:25:25,600 --> 00:25:28,760
Speaker 1: a relationship? You know what I mean? His face has

410
00:25:28,800 --> 00:25:31,120
Speaker 1: a story it tells, which I want to bring something

411
00:25:31,440 --> 00:25:37,000
Speaker 1: up here quickly with the the um high priced effects

412
00:25:37,080 --> 00:25:39,080
Speaker 1: that were going on in a movie like The Irishman,

413
00:25:39,160 --> 00:25:42,639
Speaker 1: with these aging effects this These are designed to be

414
00:25:42,760 --> 00:25:46,800
Speaker 1: displayed in the highest resolution possible, So you're talking four K,

415
00:25:47,240 --> 00:25:50,679
Speaker 1: ten ADP something like that. It's a designed to do that, right.

416
00:25:50,960 --> 00:25:53,399
Speaker 1: It's on a streaming service. They don't know what you're

417
00:25:53,440 --> 00:25:55,800
Speaker 1: going to play it on what screen, but it's got

418
00:25:55,920 --> 00:25:58,560
Speaker 1: to be high res, right. And what we're saying is

419
00:25:58,600 --> 00:26:01,800
Speaker 1: it's fairly easy to discern that something is going on

420
00:26:02,000 --> 00:26:04,960
Speaker 1: here at that high resolution. But what if it's a

421
00:26:05,119 --> 00:26:09,880
Speaker 1: much lower resolution, more grainy, get destroyed like a gift,

422
00:26:10,000 --> 00:26:12,920
Speaker 1: like a small YouTube video that isn't maybe ten a

423
00:26:13,000 --> 00:26:17,080
Speaker 1: DP or something, or a cell phone camera video that

424
00:26:17,200 --> 00:26:22,280
Speaker 1: then gets uploaded integrated. It changes our ability to discern

425
00:26:22,600 --> 00:26:25,359
Speaker 1: some of these things. And we'll get to this, but again,

426
00:26:25,400 --> 00:26:27,919
Speaker 1: all of this comes down to these algorithms, which, as

427
00:26:27,960 --> 00:26:31,840
Speaker 1: it turns out require uh an insane amount of computing power,

428
00:26:32,440 --> 00:26:35,200
Speaker 1: yes to to even to do in these low res

429
00:26:35,280 --> 00:26:39,399
Speaker 1: forms right for now, at least exactly. So think of

430
00:26:39,480 --> 00:26:42,280
Speaker 1: it this way now without too much of a hassle,

431
00:26:42,400 --> 00:26:46,800
Speaker 1: as you are listening to today's episode, you can get

432
00:26:46,960 --> 00:26:52,760
Speaker 1: this technology online. You can create videos that are nearly

433
00:26:52,840 --> 00:26:57,600
Speaker 1: impossible to identify as quote unquote fake. For a fun example,

434
00:26:58,040 --> 00:27:01,080
Speaker 1: we just want to keep it innocuous before we have

435
00:27:01,280 --> 00:27:03,840
Speaker 1: to strap in and go down this rabbit hole. For

436
00:27:03,960 --> 00:27:06,000
Speaker 1: fun example, let's do two fun examples. So let's say

437
00:27:06,040 --> 00:27:09,439
Speaker 1: you have a friend who knows you love Marvel movies,

438
00:27:09,800 --> 00:27:13,080
Speaker 1: and so for you know, your birthday, your King signer

439
00:27:13,240 --> 00:27:15,840
Speaker 1: or whatever, they make a deep fake video where it

440
00:27:15,920 --> 00:27:19,240
Speaker 1: looks like you're in a Marvel film. The Avengers all assemble,

441
00:27:19,320 --> 00:27:22,680
Speaker 1: and holy smokes, there's Derek. That's fun, all right? That

442
00:27:22,800 --> 00:27:26,280
Speaker 1: seems fun. What a thoughtful kind gift. Or to make

443
00:27:26,320 --> 00:27:29,840
Speaker 1: it a little more applicable to our show, one thing

444
00:27:29,920 --> 00:27:33,440
Speaker 1: that would be a fantastic deep fake present for our

445
00:27:33,520 --> 00:27:36,359
Speaker 1: very own superproducer, Paul Michig Control Decade would be to

446
00:27:36,440 --> 00:27:38,920
Speaker 1: take an Apple Bee's commercial and just put him in it.

447
00:27:39,359 --> 00:27:43,159
Speaker 1: Oh Man, and so he's the person there, you know, uh,

448
00:27:43,720 --> 00:27:48,919
Speaker 1: gesturing in amazement as the ribs in the bottomless Halopaniel

449
00:27:48,920 --> 00:27:52,720
Speaker 1: poppers or whatever come out. And we could even have

450
00:27:52,920 --> 00:27:55,480
Speaker 1: him say the tagline in his voice. That is so

451
00:27:55,640 --> 00:27:58,480
Speaker 1: much fun. But that is not the only use of

452
00:27:58,640 --> 00:28:02,080
Speaker 1: this technology. Make no mistake, the lid of Pandora's jar,

453
00:28:02,280 --> 00:28:05,560
Speaker 1: and it was a jar, is unscrewed. This technology is

454
00:28:05,640 --> 00:28:08,480
Speaker 1: no longer theoretical. It is very real, and it is

455
00:28:08,560 --> 00:28:12,800
Speaker 1: immensely dangerous. Why we'll tell you after a word from

456
00:28:12,840 --> 00:28:22,920
Speaker 1: our sponsor, psych We're not going to tell you it's

457
00:28:23,000 --> 00:28:27,639
Speaker 1: not even us what you thought this was conspiracy stuff.

458
00:28:28,560 --> 00:28:31,240
Speaker 1: We're just your car talking to you. Is my real voice,

459
00:28:31,240 --> 00:28:37,480
Speaker 1: and Robbi Deniu so so terrible denial impressions. Aside, there

460
00:28:37,520 --> 00:28:42,160
Speaker 1: are applications of deep fakes which should trouble every single

461
00:28:42,440 --> 00:28:46,800
Speaker 1: person or bought listening to this show. I did not

462
00:28:47,240 --> 00:28:50,880
Speaker 1: initially think of the first application, which was dumb and

463
00:28:51,000 --> 00:28:53,800
Speaker 1: naive of me. Apparently one of the first things people

464
00:28:53,880 --> 00:28:57,760
Speaker 1: tried to do when gan technology got out of it's

465
00:28:57,960 --> 00:29:00,360
Speaker 1: got out of its research R and D high hole

466
00:29:00,800 --> 00:29:04,560
Speaker 1: was to apply it to pornography, and pornography drives a

467
00:29:04,640 --> 00:29:07,960
Speaker 1: lot of technology. I mean, arguably, there's a very good

468
00:29:08,000 --> 00:29:10,920
Speaker 1: case to be made that the reason VHS went out

469
00:29:11,000 --> 00:29:15,240
Speaker 1: over Beta Max was because the porn studios went with VHS. Yeah.

470
00:29:15,400 --> 00:29:17,800
Speaker 1: I don't want to be crash or get too much

471
00:29:17,840 --> 00:29:22,040
Speaker 1: away here, but I do remember far before this technology

472
00:29:22,160 --> 00:29:25,480
Speaker 1: was available, when photoshopping was really the only option, there

473
00:29:25,560 --> 00:29:28,920
Speaker 1: were sites, I want to say, early on in the

474
00:29:29,040 --> 00:29:33,800
Speaker 1: Internet where it was just sites dedicated to celebrity pornography,

475
00:29:33,920 --> 00:29:38,000
Speaker 1: where it was just photoshopped images on purpose. That would

476
00:29:38,040 --> 00:29:45,080
Speaker 1: be pretty crazy applying it to video with this new technology. Yikes,

477
00:29:45,240 --> 00:29:48,480
Speaker 1: black mirror esque, right, Like, imagine you are a creep

478
00:29:48,600 --> 00:29:51,600
Speaker 1: with a crush on someone. Could be a colleague, a classmate,

479
00:29:51,640 --> 00:29:54,280
Speaker 1: a celebrity, you know what, it could be uh, an

480
00:29:54,400 --> 00:29:58,040
Speaker 1: historical figure, maybe King Touch just really does it for

481
00:29:58,160 --> 00:30:02,600
Speaker 1: you for some reason. One the actual one. We know

482
00:30:02,720 --> 00:30:05,760
Speaker 1: what the guy actually looks like now and not Steve Martin.

483
00:30:05,920 --> 00:30:09,040
Speaker 1: Not Steve Martin is one of the best banjo players

484
00:30:09,040 --> 00:30:10,440
Speaker 1: in the world, and he really does do it for

485
00:30:10,560 --> 00:30:13,520
Speaker 1: me at least. But in this case, so okay, So

486
00:30:13,720 --> 00:30:16,800
Speaker 1: let's Steve Martin example them that now there's nothing wrong

487
00:30:16,960 --> 00:30:21,240
Speaker 1: with having a crush. It becomes creepy. However, if you

488
00:30:21,480 --> 00:30:25,440
Speaker 1: use deep fake technology and put Steve Martin's face on

489
00:30:25,560 --> 00:30:29,400
Speaker 1: the face of someone in a pornographic video, especially when

490
00:30:29,480 --> 00:30:34,200
Speaker 1: the video will genuinely look real and it will sound

491
00:30:34,480 --> 00:30:38,640
Speaker 1: like them, this is not science fiction. This is the

492
00:30:38,760 --> 00:30:42,720
Speaker 1: idea behind a website with the immensely creative name deep Nudes.

493
00:30:43,680 --> 00:30:48,520
Speaker 1: Deep Nudes did exactly what we just described to her. Luckily,

494
00:30:48,960 --> 00:30:54,200
Speaker 1: the founder eventually canceled the site's launch and they had

495
00:30:54,240 --> 00:30:57,160
Speaker 1: a public statement about it. Yeah, and the founder actually

496
00:30:57,200 --> 00:31:01,600
Speaker 1: eventually canceled launching the site. Um noting that quote, the

497
00:31:01,680 --> 00:31:06,200
Speaker 1: probability that people will misuse it is too high. Oh,

498
00:31:07,080 --> 00:31:09,200
Speaker 1: I never would have never would have thought that what

499
00:31:12,000 --> 00:31:16,760
Speaker 1: somebody's gonna misuse this. There's a great Mitchell and Web

500
00:31:16,920 --> 00:31:20,560
Speaker 1: sketch about like an evil scientist where he's like, I

501
00:31:20,720 --> 00:31:23,440
Speaker 1: built the ultra violence laser to save the world, not

502
00:31:23,640 --> 00:31:27,440
Speaker 1: destroy it. I'll send it to you guys. Maybe we

503
00:31:27,480 --> 00:31:29,920
Speaker 1: could post it on here's where it gets crazy. But

504
00:31:31,040 --> 00:31:33,160
Speaker 1: that's just one use. That's the immediate one. And again,

505
00:31:33,240 --> 00:31:34,720
Speaker 1: I I don't know about you, guys, but I felt

506
00:31:34,800 --> 00:31:39,840
Speaker 1: naive for not immediately assuming that's what would happen. Yeah,

507
00:31:39,880 --> 00:31:43,200
Speaker 1: I think it's it's pretty obvious. It's just hey, cheers

508
00:31:43,240 --> 00:31:46,880
Speaker 1: guys for not always thinking about pornography. I guess. So, Yeah,

509
00:31:47,280 --> 00:31:50,280
Speaker 1: it just felt awkward when he's talking to some contacts

510
00:31:50,280 --> 00:31:53,120
Speaker 1: about this off air and they looked at me like

511
00:31:53,320 --> 00:31:57,320
Speaker 1: I was from you know, a different universe or time period.

512
00:31:57,400 --> 00:32:01,600
Speaker 1: When they said, they're like, yeah, pull arn, Ben. It's

513
00:32:02,520 --> 00:32:06,120
Speaker 1: that's why people have technology, is to get better poor,

514
00:32:07,000 --> 00:32:08,800
Speaker 1: which I don't know if, I don't know whether that's

515
00:32:08,840 --> 00:32:12,040
Speaker 1: completely true, but it was a weird night. There's another

516
00:32:12,240 --> 00:32:17,200
Speaker 1: use of deep fakes that is more apparent and has

517
00:32:17,360 --> 00:32:21,560
Speaker 1: the potential. So like this, this fake pornography or this

518
00:32:21,760 --> 00:32:25,880
Speaker 1: fake um rendition of people in these intimate times, it

519
00:32:26,000 --> 00:32:29,160
Speaker 1: can ruin an individual's life. It could be blackmail, and

520
00:32:29,240 --> 00:32:32,800
Speaker 1: it could be blackmail. But there's another version of a

521
00:32:32,880 --> 00:32:36,240
Speaker 1: deep fake, a weaponized deep fake, that could ruin the

522
00:32:36,400 --> 00:32:39,640
Speaker 1: lives of hundreds of thousands or millions of people. Yeah,

523
00:32:39,680 --> 00:32:42,880
Speaker 1: because we can't forget that a lot of countries, including US,

524
00:32:43,240 --> 00:32:48,840
Speaker 1: are very heavily entrenched um in this notion of asymmetrical

525
00:32:48,920 --> 00:32:55,360
Speaker 1: warfare called cyber war awkwardly cyber a little bit. Many

526
00:32:55,480 --> 00:32:59,760
Speaker 1: world leaders have extensive like you said, Ben, video footage

527
00:33:00,240 --> 00:33:05,160
Speaker 1: um out there of themselves at functions, giving speeches, events

528
00:33:05,240 --> 00:33:08,600
Speaker 1: and the like. There's more than enough for gan to

529
00:33:09,080 --> 00:33:14,080
Speaker 1: to work from here. Yeah, so let's say, uh, who

530
00:33:14,120 --> 00:33:17,520
Speaker 1: can we use as an example? All right? Mattuh you

531
00:33:17,600 --> 00:33:21,200
Speaker 1: did Steve barn earlier? Yeah, okay, uh okay. So let's

532
00:33:21,240 --> 00:33:25,000
Speaker 1: say let's say, Matt, you and Noel are the leaders

533
00:33:25,520 --> 00:33:28,840
Speaker 1: of these different opposing countries. It's been a lot of

534
00:33:28,960 --> 00:33:31,800
Speaker 1: tension for a while. Okay. Uh So if there was

535
00:33:31,880 --> 00:33:34,920
Speaker 1: someone either on one of your sides or a third party,

536
00:33:35,000 --> 00:33:37,400
Speaker 1: Let's say I'm a country that just messes with other

537
00:33:37,480 --> 00:33:39,800
Speaker 1: countries for fun, right, I'm Russia, I'm the U S

538
00:33:39,880 --> 00:33:42,720
Speaker 1: I'm I'm one of one of the hits, right, And

539
00:33:43,440 --> 00:33:46,240
Speaker 1: I say, you know what's gonna be great. I can't

540
00:33:46,400 --> 00:33:51,680
Speaker 1: take on uh no Landia or the Republic of Frederick

541
00:33:52,200 --> 00:33:56,440
Speaker 1: with conventional military might. So I'm going to turn them,

542
00:33:56,960 --> 00:33:59,840
Speaker 1: turn them against each other. I'm going to foment instability,

543
00:34:00,240 --> 00:34:03,640
Speaker 1: or I'm gonna mess with their election by just going

544
00:34:03,680 --> 00:34:05,400
Speaker 1: on Facebook. I don't need to launch an I C

545
00:34:05,520 --> 00:34:07,480
Speaker 1: B M. I'm just gonna go on Facebook and Twitter.

546
00:34:07,640 --> 00:34:11,120
Speaker 1: I'm gonna make fake videos of leaders of both countries

547
00:34:11,160 --> 00:34:13,160
Speaker 1: and no Landia and the Republic of Frederick, and I'm

548
00:34:13,160 --> 00:34:15,959
Speaker 1: gonna have them say things that they would never actually say.

549
00:34:16,360 --> 00:34:19,960
Speaker 1: This sounds like a Who're building Twitter comedy bit, But

550
00:34:20,080 --> 00:34:23,760
Speaker 1: there's a real fake, a real fake video for multiple

551
00:34:23,840 --> 00:34:27,600
Speaker 1: levels of Nancy Pelosi, a politician here in the US.

552
00:34:27,719 --> 00:34:31,080
Speaker 1: And have you guys seen this video? This one? So

553
00:34:31,520 --> 00:34:35,239
Speaker 1: it's uh. It came out with a nice side by

554
00:34:35,320 --> 00:34:38,560
Speaker 1: side view, but the deep fake video was also propagated

555
00:34:38,640 --> 00:34:41,480
Speaker 1: on its own. It's very interesting to watch the difference

556
00:34:41,560 --> 00:34:44,280
Speaker 1: between the two. This gives us a chance to rewrite

557
00:34:44,360 --> 00:34:48,759
Speaker 1: history in a disturbingly or welly and way. What happens if,

558
00:34:49,000 --> 00:34:53,359
Speaker 1: for example, all original let's say nol makes uh an

559
00:34:53,440 --> 00:34:57,880
Speaker 1: historic speech in no Landia that triggers a new golden

560
00:34:57,920 --> 00:35:03,080
Speaker 1: age for the country, and someone destroys the original copies

561
00:35:03,520 --> 00:35:07,839
Speaker 1: of this profound speech and replaces it with a deep fake.

562
00:35:09,320 --> 00:35:12,600
Speaker 1: And then the last living generation, the last people who

563
00:35:12,600 --> 00:35:16,560
Speaker 1: were there when the original speech was propagated, they can

564
00:35:16,640 --> 00:35:20,600
Speaker 1: keep its memory alive, but when they die, history has

565
00:35:20,640 --> 00:35:23,120
Speaker 1: in a very real way changed. Now. I don't know

566
00:35:23,160 --> 00:35:25,680
Speaker 1: if this counts or not, but uh, there was a

567
00:35:25,800 --> 00:35:27,960
Speaker 1: brief period recently where there were some what I would

568
00:35:28,000 --> 00:35:30,839
Speaker 1: consider deep fake gifts that were making the rounds. There's

569
00:35:30,840 --> 00:35:33,920
Speaker 1: like Obama on a skateboard. Uh, there's one of the

570
00:35:34,080 --> 00:35:37,279
Speaker 1: Pope doing a magic trick where he like pulls the

571
00:35:38,120 --> 00:35:41,719
Speaker 1: the tablecloth out from under some kind of votives like

572
00:35:41,880 --> 00:35:44,920
Speaker 1: at a like on an event like a live CNN stream,

573
00:35:45,120 --> 00:35:47,480
Speaker 1: which I choose to. I know it's fake, but I'm

574
00:35:47,520 --> 00:35:50,640
Speaker 1: just gonna believe in it, but they're so good. Yeah,

575
00:35:50,880 --> 00:35:53,160
Speaker 1: I I wanted to believe in it as well, especially

576
00:35:53,200 --> 00:35:55,560
Speaker 1: the Obama on a skateboard one. But does that kind

577
00:35:55,600 --> 00:35:57,560
Speaker 1: of fall does that sort of fall under these like

578
00:35:57,640 --> 00:36:01,160
Speaker 1: this category. Yeah, yeah, that's a that's a less dangerous version,

579
00:36:01,360 --> 00:36:04,600
Speaker 1: you know, because that's fun seeing seeing the Pope pull

580
00:36:04,680 --> 00:36:08,880
Speaker 1: off a dope magic trick. Is not gonna like foment

581
00:36:09,080 --> 00:36:12,920
Speaker 1: instability in South America. But going back to our example,

582
00:36:13,000 --> 00:36:16,720
Speaker 1: let's let's assume the Republic of Frederick and no Landia

583
00:36:17,120 --> 00:36:19,320
Speaker 1: are kind are beefed up to the level of like

584
00:36:19,400 --> 00:36:24,080
Speaker 1: Pakistan and India or Israel and Palestine, and all of

585
00:36:24,160 --> 00:36:28,040
Speaker 1: a sudden, in no traceable way, the public of both

586
00:36:28,120 --> 00:36:32,520
Speaker 1: countries gets hit with this barrage of videos that seemed

587
00:36:32,600 --> 00:36:36,520
Speaker 1: to say seem seemed to have you know, the benevolent

588
00:36:36,880 --> 00:36:41,759
Speaker 1: dictator h Matt Frederick and the Prime Minister of no

589
00:36:41,880 --> 00:36:44,880
Speaker 1: Landy and Noel Brown seems also a dictator by the ways,

590
00:36:45,239 --> 00:36:49,160
Speaker 1: officially not in his title. Okay, so so the uh,

591
00:36:49,440 --> 00:36:52,520
Speaker 1: let's these these guys have these videos wherein each of

592
00:36:52,640 --> 00:36:56,760
Speaker 1: them are announcing their tissue, their intention to deploy nuclear

593
00:36:56,840 --> 00:37:01,440
Speaker 1: weapons and mop the mop thecent country clean, let no

594
00:37:01,719 --> 00:37:05,200
Speaker 1: Stone stand on another How would you know if you're

595
00:37:05,239 --> 00:37:07,200
Speaker 1: the audience of this other country, how would you know

596
00:37:07,360 --> 00:37:10,320
Speaker 1: whether these were real or fake? How would you react?

597
00:37:10,520 --> 00:37:12,640
Speaker 1: How much time would you have if you think that

598
00:37:12,760 --> 00:37:16,200
Speaker 1: you literally saw the leader of the country that you

599
00:37:16,360 --> 00:37:19,880
Speaker 1: previously went to war with saying that's it, We're launching

600
00:37:19,920 --> 00:37:23,360
Speaker 1: in five Well, it's really it's really scary because the

601
00:37:23,520 --> 00:37:27,800
Speaker 1: way that would function, it would be posted somewhere and

602
00:37:27,880 --> 00:37:31,000
Speaker 1: it would become viral on social media so greatly. That's

603
00:37:31,040 --> 00:37:33,400
Speaker 1: the only way that it would propagate. But it would

604
00:37:33,480 --> 00:37:37,719
Speaker 1: propagate likely. And I'm just gonna say personally, what I

605
00:37:37,760 --> 00:37:41,160
Speaker 1: would do in that situation is go to the standby

606
00:37:41,239 --> 00:37:43,160
Speaker 1: that we discussed at the top. I would probably turn

607
00:37:43,320 --> 00:37:46,239
Speaker 1: a TV on somewhere just to check and see if

608
00:37:46,440 --> 00:37:50,160
Speaker 1: somebody is talking about it seriously, you know, on one

609
00:37:50,200 --> 00:37:53,239
Speaker 1: of the major outlets. The problem is what if you

610
00:37:53,360 --> 00:37:57,360
Speaker 1: fool them? Um, it happens all the time, even with

611
00:37:57,440 --> 00:38:00,600
Speaker 1: quote unquote fake news. Everyone so into get the scoop

612
00:38:01,000 --> 00:38:03,800
Speaker 1: with these, with the fast cycle, the turnaround of news,

613
00:38:04,239 --> 00:38:06,560
Speaker 1: that everyone wants to be the first, so they tend

614
00:38:06,640 --> 00:38:09,399
Speaker 1: to not vet things like they used to, and that's

615
00:38:09,400 --> 00:38:14,080
Speaker 1: how you often get these, uh misreporting of election results, etcetera.

616
00:38:14,360 --> 00:38:16,560
Speaker 1: You know, and this is just perfect example of someone's life.

617
00:38:16,640 --> 00:38:19,480
Speaker 1: Here's a good example. There is a deep fake Donald

618
00:38:19,520 --> 00:38:23,200
Speaker 1: Trump pe tape that's out there. Um, what if that

619
00:38:23,400 --> 00:38:27,719
Speaker 1: had been pushed out, you know by let's say, you know, uh,

620
00:38:28,200 --> 00:38:32,560
Speaker 1: some network opposing that the Trump administration that didn't happen. Um,

621
00:38:33,360 --> 00:38:35,600
Speaker 1: now that we know a little bit more about deep

622
00:38:35,640 --> 00:38:37,799
Speaker 1: fakes being a thing, maybe there's some caution. No one

623
00:38:37,840 --> 00:38:40,800
Speaker 1: wants to be the news agency that does that. But

624
00:38:41,000 --> 00:38:45,200
Speaker 1: with a inflammatory enough thing, maybe you don't have time

625
00:38:45,239 --> 00:38:47,480
Speaker 1: to think about it. Well. See, yeah, here's here's what

626
00:38:47,560 --> 00:38:51,440
Speaker 1: I wanted to bring up. In the examples we're talking

627
00:38:51,520 --> 00:38:56,040
Speaker 1: about with you know, a speech or something that exists.

628
00:38:56,080 --> 00:38:59,439
Speaker 1: So you so in theory, you would take the base

629
00:38:59,600 --> 00:39:03,480
Speaker 1: level of video of the speech, and maybe maybe that audio,

630
00:39:03,640 --> 00:39:05,960
Speaker 1: then you'd manipulate the audio and then the face right

631
00:39:06,120 --> 00:39:10,000
Speaker 1: or something to where the the words being spoken are

632
00:39:10,120 --> 00:39:14,360
Speaker 1: different on that video. I think the scariest ones, the

633
00:39:14,440 --> 00:39:18,800
Speaker 1: scariest versions of this um are where it's supposed to

634
00:39:18,840 --> 00:39:22,120
Speaker 1: be a hidden camera or something, where it's, like I

635
00:39:22,200 --> 00:39:25,520
Speaker 1: was saying before, it's so degraded it's difficult to truly

636
00:39:25,640 --> 00:39:28,120
Speaker 1: make out what's going on. But you can tell that, oh,

637
00:39:28,320 --> 00:39:31,920
Speaker 1: that's definitely Billy Eichner. And where you can you know

638
00:39:32,040 --> 00:39:33,480
Speaker 1: your brain is at least telling you that, and it

639
00:39:33,600 --> 00:39:38,920
Speaker 1: sounds like Billy Eichner's voice saying things Billy on the street.

640
00:39:40,040 --> 00:39:41,479
Speaker 1: On the street, I mean, and he was in Parks

641
00:39:41,480 --> 00:39:44,759
Speaker 1: and Rack. He's the really like wound up guy from

642
00:39:44,840 --> 00:39:48,720
Speaker 1: Eagleton that ends up working in the office when they combine.

643
00:39:49,080 --> 00:39:51,560
Speaker 1: I'm so glad you guys are here, Like I recognize

644
00:39:51,800 --> 00:39:54,759
Speaker 1: maybe three out of five. He's the guys that have

645
00:39:54,840 --> 00:39:56,840
Speaker 1: costs people on the street with a microphone and just

646
00:39:56,960 --> 00:39:59,440
Speaker 1: yell celebrity names at them and stuff. I thought that

647
00:39:59,520 --> 00:40:05,759
Speaker 1: was wilf Errol doing Harry Cary, but I'm sorry, I'm

648
00:40:05,800 --> 00:40:08,440
Speaker 1: derailate okay, But the point is there's a there's a

649
00:40:08,560 --> 00:40:10,640
Speaker 1: human being that you know that is famous for one

650
00:40:10,680 --> 00:40:14,680
Speaker 1: reason or another, possibly powerful, and their their voice is

651
00:40:14,719 --> 00:40:18,360
Speaker 1: being manipulated. But the video itself isn't something that you

652
00:40:18,480 --> 00:40:22,160
Speaker 1: can reference to like a speech or to a movie

653
00:40:22,360 --> 00:40:25,600
Speaker 1: or something that you remember you can verify with. It

654
00:40:25,719 --> 00:40:28,600
Speaker 1: looks like a brand new video, but you can still

655
00:40:28,719 --> 00:40:31,160
Speaker 1: tell that it's that person. Their mouth is moving and

656
00:40:31,239 --> 00:40:34,840
Speaker 1: the words coming out or something awful. I feel like

657
00:40:34,920 --> 00:40:38,440
Speaker 1: we're putting this in an accurate, humorous way. We do

658
00:40:38,560 --> 00:40:41,480
Speaker 1: have to emphasize this very scary thing that you've built

659
00:40:41,480 --> 00:40:45,200
Speaker 1: a beautiful example that leads us to another nefarious use

660
00:40:45,600 --> 00:40:50,440
Speaker 1: of deep fakes, which will happen. It absolutely will happen

661
00:40:51,840 --> 00:40:58,400
Speaker 1: false accusations. So that's exactly so. According to the Andrea Hickerson,

662
00:40:58,440 --> 00:41:00,480
Speaker 1: who is the director of the School of Nalism and

663
00:41:00,560 --> 00:41:04,560
Speaker 1: Mass Communications at the University of South Carolina, this is

664
00:41:04,600 --> 00:41:07,440
Speaker 1: a problem because at the most basic level, deep fakes

665
00:41:07,480 --> 00:41:11,480
Speaker 1: are lies disguise to look like truth. Uh and Hickerson says,

666
00:41:11,520 --> 00:41:14,000
Speaker 1: if we take them as truth or evidence, we can

667
00:41:14,080 --> 00:41:19,879
Speaker 1: easily make false conclusions with potentially disastrous consequences. So if

668
00:41:20,000 --> 00:41:24,680
Speaker 1: you want to ruin someone's life, you want to smear

669
00:41:24,800 --> 00:41:28,640
Speaker 1: a political opponent, or you just I don't really like

670
00:41:28,760 --> 00:41:32,120
Speaker 1: your neighbor, you know, like they're their their vibe. It

671
00:41:32,239 --> 00:41:36,080
Speaker 1: just irks you. Then you could with this, with this

672
00:41:36,200 --> 00:41:39,799
Speaker 1: capacity in the future, you would be able to make

673
00:41:39,920 --> 00:41:43,640
Speaker 1: something where it appears that they're saying something terrible, where

674
00:41:43,680 --> 00:41:47,080
Speaker 1: they're like, yeah, I don't kicking puppies, burning buildings, just

675
00:41:47,239 --> 00:41:50,040
Speaker 1: whatever to feel something, you know what I mean. Also,

676
00:41:50,400 --> 00:41:53,319
Speaker 1: I uh ili, I rent scooters and I leave them

677
00:41:53,360 --> 00:41:55,239
Speaker 1: in the middle of the sidewalk because I'm that guy.

678
00:41:55,680 --> 00:41:58,520
Speaker 1: I'm the one, and then it would look real. But

679
00:41:59,520 --> 00:42:02,000
Speaker 1: if we go back to orwell, this becomes even more

680
00:42:02,120 --> 00:42:05,760
Speaker 1: dangerous because we have to consider activism and heavy handed

681
00:42:06,280 --> 00:42:09,879
Speaker 1: state actors. It's already dangerous to be an activist. Suit

682
00:42:10,000 --> 00:42:14,200
Speaker 1: to Knowl's example about Hong Kong right people are in endangered,

683
00:42:14,239 --> 00:42:19,279
Speaker 1: people are dying, They're fighting UM an authoritarian massive state.

684
00:42:19,800 --> 00:42:24,280
Speaker 1: So currently, state corporating criminal actors seeking to silence dissidents

685
00:42:24,520 --> 00:42:28,000
Speaker 1: all use the ordinary tool kid of suppression, all the

686
00:42:28,120 --> 00:42:33,080
Speaker 1: time tested stuff, all the smooth jazz, all the hits, threats, violence, kidnapping,

687
00:42:33,320 --> 00:42:39,280
Speaker 1: smear campaigns, incarceration, disappearing, and of course assassination the breakout

688
00:42:39,520 --> 00:42:43,080
Speaker 1: single of suppression, right, you know, But soon, even as

689
00:42:43,160 --> 00:42:45,360
Speaker 1: we record this state actors will have a new and

690
00:42:45,440 --> 00:42:50,279
Speaker 1: powerful tool. If, for instance, you are protesting something and

691
00:42:50,480 --> 00:42:53,239
Speaker 1: you are the leader, the face of activism, and say

692
00:42:53,320 --> 00:42:55,600
Speaker 1: Hong Kong or one of the many other places around

693
00:42:55,600 --> 00:42:58,759
Speaker 1: the world where protests are active, the authorities or the

694
00:42:58,760 --> 00:43:01,680
Speaker 1: opposition would be able to make videos in which you

695
00:43:01,960 --> 00:43:05,640
Speaker 1: are on camera disagreeing with the status quo, saying, hey,

696
00:43:06,160 --> 00:43:08,400
Speaker 1: they were right all along. I had a change of

697
00:43:08,480 --> 00:43:11,400
Speaker 1: heart and I want to confess that my motives were

698
00:43:11,440 --> 00:43:15,040
Speaker 1: not pure. I was paid by someone else to do this,

699
00:43:15,680 --> 00:43:19,160
Speaker 1: so I apologize. I'm turning myself in. And then the

700
00:43:19,320 --> 00:43:22,880
Speaker 1: actual you finds out about this when you were confession

701
00:43:23,160 --> 00:43:27,040
Speaker 1: is aired on the news and people start contacting you. Yeah,

702
00:43:27,160 --> 00:43:33,440
Speaker 1: that is uh an intense hypothetical currently situation. It's gonna happen.

703
00:43:33,719 --> 00:43:36,400
Speaker 1: I know, I I know right now, it's just it

704
00:43:36,600 --> 00:43:39,400
Speaker 1: is possible. Well, that leads us to a lot of

705
00:43:39,480 --> 00:43:42,359
Speaker 1: things that need to happen or or potentially are going

706
00:43:42,440 --> 00:43:44,399
Speaker 1: to happen. And I want to lead with this. It's

707
00:43:44,440 --> 00:43:48,040
Speaker 1: really interesting in in the law now, um and and

708
00:43:48,239 --> 00:43:51,680
Speaker 1: lawyers out there, correct me if I'm oversimplifying this, But um,

709
00:43:52,000 --> 00:43:55,839
Speaker 1: video evidence isn't like the end all be all already, right.

710
00:43:55,960 --> 00:43:58,560
Speaker 1: It has to fit a couple of requirements in order

711
00:43:58,640 --> 00:44:01,640
Speaker 1: to be admissible. And the first place, which is relevance

712
00:44:02,080 --> 00:44:06,839
Speaker 1: and authenticity that chain, chain of custody. It's a lot

713
00:44:06,880 --> 00:44:08,040
Speaker 1: of things that go into that, and we're going to

714
00:44:08,120 --> 00:44:10,360
Speaker 1: get into that in a minute, and also potentially what

715
00:44:10,560 --> 00:44:12,879
Speaker 1: might have to happen as this technology gets better and better.

716
00:44:13,200 --> 00:44:17,799
Speaker 1: But um, video evidence can be considered hearsay if there

717
00:44:17,880 --> 00:44:21,960
Speaker 1: isn't someone to corroborate it. Um, if it's the only evidence,

718
00:44:22,280 --> 00:44:24,800
Speaker 1: that's not a great case. Uh that you know, and

719
00:44:24,840 --> 00:44:26,759
Speaker 1: eye witness, you know, I was just saying the good

720
00:44:26,800 --> 00:44:29,600
Speaker 1: crime shows and eyeball witness is really your best bet

721
00:44:29,680 --> 00:44:32,440
Speaker 1: to getting a conviction. Video and video alone if that's

722
00:44:32,440 --> 00:44:35,200
Speaker 1: all you got, especially if it's like grainy security camp footage,

723
00:44:35,360 --> 00:44:38,840
Speaker 1: a lawyer could say that, uh, is not a reasonable

724
00:44:38,920 --> 00:44:43,360
Speaker 1: representation of the subject being displayed or in question, etcetera.

725
00:44:43,680 --> 00:44:45,759
Speaker 1: But what this comes down to is life. When is

726
00:44:45,880 --> 00:44:48,440
Speaker 1: the law going to catch up to this new technology

727
00:44:48,800 --> 00:44:51,279
Speaker 1: in terms of that chain of custody, Because that's the

728
00:44:51,360 --> 00:44:54,080
Speaker 1: thing if we can't believe our eyes, like we said

729
00:44:54,080 --> 00:44:55,960
Speaker 1: at the top of the show, you said, ben seeing

730
00:44:56,080 --> 00:45:00,399
Speaker 1: isn't really believing. How do we authenticate this stuff. It's

731
00:45:00,440 --> 00:45:03,319
Speaker 1: going to have to be that chain of custody that's

732
00:45:03,360 --> 00:45:05,719
Speaker 1: kept under lock and key, so we know that the

733
00:45:05,760 --> 00:45:10,560
Speaker 1: footage we're seeing was captured and then disseminated with no

734
00:45:11,120 --> 00:45:13,920
Speaker 1: in between. Right, just video is not enough. That's why

735
00:45:14,000 --> 00:45:17,520
Speaker 1: Bigfoot will never see a day in jail ever. And

736
00:45:17,640 --> 00:45:20,759
Speaker 1: he needs to be there, he really does. But you know,

737
00:45:20,880 --> 00:45:22,839
Speaker 1: here's the other thing. We can't trust our memories either.

738
00:45:23,000 --> 00:45:26,279
Speaker 1: We're talking about in court. Video without corroboration from a

739
00:45:26,360 --> 00:45:29,560
Speaker 1: witness doesn't work, and the witness is the best way,

740
00:45:29,840 --> 00:45:34,080
Speaker 1: the eyewitness. But the eyewitness is very unreliable. It's perhaps

741
00:45:34,160 --> 00:45:37,319
Speaker 1: the most unreliable thing that exists as far as evidence goes.

742
00:45:37,760 --> 00:45:40,080
Speaker 1: So what the heck do we trust when it comes

743
00:45:40,120 --> 00:45:42,960
Speaker 1: to evidence of something that actually happened. The approach is

744
00:45:43,000 --> 00:45:46,600
Speaker 1: at uh, what of a cretion of aggregation? Right? You

745
00:45:46,680 --> 00:45:48,719
Speaker 1: have to have multiple likeness said, You've got to have

746
00:45:49,160 --> 00:45:51,960
Speaker 1: multiple things so you can say you can triangulate a

747
00:45:52,040 --> 00:45:55,800
Speaker 1: little bit, say, yeah, memories imperfect. We can't trust video alone.

748
00:45:55,840 --> 00:45:57,640
Speaker 1: But we have a video and we have a witness,

749
00:45:58,040 --> 00:46:01,520
Speaker 1: and then um, maybe you have some other forensic evidence

750
00:46:01,640 --> 00:46:06,040
Speaker 1: like a gun case, scenes of you know, spent shells

751
00:46:06,120 --> 00:46:10,200
Speaker 1: were found in they match the gun. Completely agree, and

752
00:46:10,280 --> 00:46:13,320
Speaker 1: you're right. It's just what worries me is that what

753
00:46:13,600 --> 00:46:17,040
Speaker 1: is stopping let's say, world leaders and powerful people who

754
00:46:17,200 --> 00:46:21,680
Speaker 1: do wrong things where a video is an actual video

755
00:46:21,840 --> 00:46:24,920
Speaker 1: is taken, maybe a surveillance cam video is taken of

756
00:46:25,040 --> 00:46:29,600
Speaker 1: some wrongdoing occurring, let's say as as high as an assassination,

757
00:46:29,640 --> 00:46:32,759
Speaker 1: as something as dire as that occurs on camera, but

758
00:46:32,840 --> 00:46:36,040
Speaker 1: there are no witnesses. But for sure, on camera there

759
00:46:36,200 --> 00:46:41,480
Speaker 1: is a world leader shooting somebody in the head. What

760
00:46:41,760 --> 00:46:45,120
Speaker 1: is stopping them from saying, oh, that's a deep fake video? Obviously,

761
00:46:45,360 --> 00:46:47,279
Speaker 1: I'm so glad you said that, because that's the other

762
00:46:47,360 --> 00:46:49,600
Speaker 1: side of the coin, right, If you are caught on

763
00:46:49,760 --> 00:46:53,879
Speaker 1: video doing something despicable, doing something illegal, whatever, you can

764
00:46:54,000 --> 00:46:56,919
Speaker 1: just say I have enemies, this was a deep fake.

765
00:46:57,280 --> 00:47:02,440
Speaker 1: You know, my ex hates me, or I am I

766
00:47:02,560 --> 00:47:05,560
Speaker 1: am doing work for I don't know whatever, a union

767
00:47:05,800 --> 00:47:09,440
Speaker 1: or something or an in geo. So there's not a

768
00:47:09,640 --> 00:47:12,880
Speaker 1: very good way to refute that other than having to

769
00:47:12,960 --> 00:47:16,719
Speaker 1: resort to as much other corroborating evidence as you can,

770
00:47:17,560 --> 00:47:19,640
Speaker 1: and then again you have to rely on, like to say,

771
00:47:19,680 --> 00:47:23,040
Speaker 1: those eyeball witnesses whose memories at times can be very

772
00:47:23,120 --> 00:47:28,040
Speaker 1: financially motivated. So in the end we're relying on law.

773
00:47:28,520 --> 00:47:31,000
Speaker 1: Just such an old concept, right, I think this is

774
00:47:31,239 --> 00:47:33,480
Speaker 1: I think that's where we're going, Nolan. I know this

775
00:47:33,680 --> 00:47:37,680
Speaker 1: is something we've talked about on this show so often

776
00:47:37,719 --> 00:47:42,399
Speaker 1: because it's so important. Technology will always outpace legislation. If

777
00:47:42,520 --> 00:47:45,440
Speaker 1: we finally get around to making a law about something,

778
00:47:45,880 --> 00:47:49,800
Speaker 1: the horse has already left the barn. The you know,

779
00:47:49,960 --> 00:47:53,120
Speaker 1: the detainee has already pulled the blackhood off and as

780
00:47:53,200 --> 00:47:56,600
Speaker 1: well on their way to international water. In the summer

781
00:47:56,640 --> 00:48:00,480
Speaker 1: of two thousand nineteen, the US House Representatives Intelligence Committee

782
00:48:00,840 --> 00:48:02,799
Speaker 1: sent a letter. They didn't pass anything, and they sent

783
00:48:02,840 --> 00:48:07,200
Speaker 1: a letter to the big socials Twitter, Facebook, Google, asking

784
00:48:07,280 --> 00:48:12,160
Speaker 1: them how these sites planned to fight against deep fakes,

785
00:48:12,560 --> 00:48:18,840
Speaker 1: particularly in the election. And this all came about because

786
00:48:19,040 --> 00:48:22,719
Speaker 1: remember we mentioned that deep fake video of Nancy Pelosi,

787
00:48:22,800 --> 00:48:27,000
Speaker 1: how speaker here in the US. The current president, President

788
00:48:27,120 --> 00:48:30,680
Speaker 1: Donald Trump, didn't just co sign that deep fake video.

789
00:48:30,840 --> 00:48:33,520
Speaker 1: He retweeted it. You say, he wasn't just like saying, hey,

790
00:48:33,560 --> 00:48:36,560
Speaker 1: watch out for this fake video of my friend Nancy Pelosi.

791
00:48:36,960 --> 00:48:40,960
Speaker 1: Now he put it out there like he was implying

792
00:48:41,040 --> 00:48:43,759
Speaker 1: that it was real. Yeah, he's a savage on Twitter too,

793
00:48:43,840 --> 00:48:46,680
Speaker 1: as anybody knows. On Twitter as a matter of fact,

794
00:48:46,719 --> 00:48:50,239
Speaker 1: if you look at his record, he disagrees with himself extensively. Well,

795
00:48:50,320 --> 00:48:54,840
Speaker 1: it's just it's one of those things where it really

796
00:48:54,960 --> 00:49:00,359
Speaker 1: shows how convincing these things can be. I think too.

797
00:49:00,800 --> 00:49:02,880
Speaker 1: I think we've mentioned on this show before, but the

798
00:49:03,000 --> 00:49:08,160
Speaker 1: Jordan Peel deep fake video of President Obama where it

799
00:49:08,480 --> 00:49:13,360
Speaker 1: looks like President Barack Obama sitting at a chair in

800
00:49:13,480 --> 00:49:17,120
Speaker 1: the Oval office somewhere and he's just saying weird things

801
00:49:17,680 --> 00:49:20,840
Speaker 1: and it sounds pretty close to him, but it's actually

802
00:49:20,920 --> 00:49:24,440
Speaker 1: just Jordan Peel doing a voice like a voiceover in

803
00:49:24,560 --> 00:49:28,080
Speaker 1: the character that he's played before on Key and Peel,

804
00:49:28,840 --> 00:49:32,440
Speaker 1: and it is. It's pretty disturbing, and I think it

805
00:49:32,680 --> 00:49:36,600
Speaker 1: speaks to how well Jordan's uh impression is, or how

806
00:49:36,640 --> 00:49:39,040
Speaker 1: good his impression is, But it also speaks to the

807
00:49:39,120 --> 00:49:43,680
Speaker 1: ability disability of matching that face, turning it into or

808
00:49:43,840 --> 00:49:47,040
Speaker 1: or changing the way the president's mouth is moving to

809
00:49:47,160 --> 00:49:50,880
Speaker 1: match the words, and if you imagine the technology that

810
00:49:51,120 --> 00:49:54,120
Speaker 1: is coming out right now where you can take five

811
00:49:54,360 --> 00:49:59,279
Speaker 1: seconds of any voice recording and then you can recreate

812
00:49:59,480 --> 00:50:02,759
Speaker 1: that voice saying anything you wanted to. Again, there is

813
00:50:02,960 --> 00:50:06,279
Speaker 1: a small research project coming out of a couple of universities,

814
00:50:06,640 --> 00:50:10,720
Speaker 1: but you could if you combined to that voice changing

815
00:50:10,800 --> 00:50:15,480
Speaker 1: technology with the face changing manipulating technology, you get you

816
00:50:15,600 --> 00:50:19,840
Speaker 1: get to a point where it will truly be we

817
00:50:19,960 --> 00:50:24,080
Speaker 1: will be unable to tell that's right, and we know

818
00:50:24,320 --> 00:50:27,560
Speaker 1: that therefore there's a ticking time bomb on this right.

819
00:50:28,080 --> 00:50:34,960
Speaker 1: So the presidential retweet of that deep fake video is

820
00:50:35,000 --> 00:50:38,080
Speaker 1: what inspired the House representatives to send that letter, But

821
00:50:38,320 --> 00:50:42,360
Speaker 1: this followed a request from Congress that occurred earlier this

822
00:50:42,480 --> 00:50:45,120
Speaker 1: year in January, where they asked the d and I

823
00:50:45,239 --> 00:50:49,000
Speaker 1: Director of National Intelligence to give a formal report on

824
00:50:49,239 --> 00:50:52,799
Speaker 1: deep fake technology. It was basically, explain it to us

825
00:50:53,360 --> 00:50:55,080
Speaker 1: so that we sound like we know what's going on

826
00:50:55,280 --> 00:50:58,680
Speaker 1: with our constituents. And also, of course there's more than

827
00:50:58,719 --> 00:51:01,480
Speaker 1: a little self preservation and involved, because these are members

828
00:51:01,520 --> 00:51:05,800
Speaker 1: of Congress, right, so we know that we have to

829
00:51:05,880 --> 00:51:10,040
Speaker 1: have legislation involved, but we also know there's a big

830
00:51:10,160 --> 00:51:12,480
Speaker 1: chance that it's just not going to be enough. It's

831
00:51:12,520 --> 00:51:15,719
Speaker 1: too easy to do this, it's too convenient, it's too powerful,

832
00:51:16,040 --> 00:51:22,239
Speaker 1: right really quick. In our industry is as podcasters. There's

833
00:51:22,480 --> 00:51:24,880
Speaker 1: there's an audio equivalent of this that's kind of on

834
00:51:25,040 --> 00:51:30,120
Speaker 1: the on the horizon, frankin Biting's next evolution very much so. UM.

835
00:51:30,480 --> 00:51:33,120
Speaker 1: I found out about this through a third party. I

836
00:51:33,239 --> 00:51:35,680
Speaker 1: can't name names. I don't think it's technology that's really

837
00:51:35,800 --> 00:51:39,320
Speaker 1: on the market yet, um, but literally, an algorithm that

838
00:51:39,400 --> 00:51:43,600
Speaker 1: could sweep through our catalog. Me you, Ben and Matt

839
00:51:44,160 --> 00:51:47,640
Speaker 1: Um run an algorithm on our catalog, and then you

840
00:51:47,760 --> 00:51:52,000
Speaker 1: could feed it lines uh, and it'll it'll approximate our voices.

841
00:51:52,080 --> 00:51:54,200
Speaker 1: And I've heard it in action and it does a

842
00:51:54,360 --> 00:51:56,759
Speaker 1: very convincing job. And on the one hand, we can say,

843
00:51:56,800 --> 00:51:58,920
Speaker 1: oh cool, we don't have to read ads anymore. But

844
00:51:59,040 --> 00:52:02,080
Speaker 1: on the other hand, we could say, oh no, we

845
00:52:02,120 --> 00:52:05,080
Speaker 1: don't have jobs anymore because we don't need a podcasting.

846
00:52:05,320 --> 00:52:08,840
Speaker 1: I'm just saying, like, the the uh, the implications of

847
00:52:08,920 --> 00:52:11,960
Speaker 1: stuff like this, it's always more far reaching than you

848
00:52:12,000 --> 00:52:15,520
Speaker 1: would originally think. You know, that's why Dan is ending Harmontown.

849
00:52:15,640 --> 00:52:18,160
Speaker 1: It is because he's just gonna take all the voices

850
00:52:18,239 --> 00:52:21,600
Speaker 1: and bottom together and make a whole new podcast that

851
00:52:21,760 --> 00:52:23,760
Speaker 1: he can just write. Did I tell you I watched

852
00:52:23,800 --> 00:52:27,560
Speaker 1: the last episode? I yeah, it was No, you didn't

853
00:52:27,560 --> 00:52:30,319
Speaker 1: tell me, but I'm glad that you did watch it too.

854
00:52:30,600 --> 00:52:34,400
Speaker 1: That was cool. Yeah. Uh so, I mean end of

855
00:52:34,440 --> 00:52:36,640
Speaker 1: an era for sure. But then now we have the

856
00:52:36,680 --> 00:52:39,520
Speaker 1: scary proposition, and I think I'm I think I'm familiar

857
00:52:39,560 --> 00:52:41,880
Speaker 1: with what you're talking about. No, we have the scary

858
00:52:41,920 --> 00:52:46,640
Speaker 1: proposition where if someone has enough audio footage to pull from,

859
00:52:47,239 --> 00:52:50,279
Speaker 1: then Harmontown would never have to end. It would just

860
00:52:50,360 --> 00:52:52,839
Speaker 1: get really weird because the technology you would still need

861
00:52:52,880 --> 00:52:55,160
Speaker 1: to evolve. I'm telling you that's what he's doing, guys,

862
00:52:55,280 --> 00:52:57,360
Speaker 1: That's what I'm saying. Like he's gonna make Rob Shrobs

863
00:52:57,360 --> 00:53:01,239
Speaker 1: say whatever he wants. It's gonna be amazing. Well, i'll

864
00:53:01,320 --> 00:53:03,960
Speaker 1: alterunon Altunion. If you guys are listening, let us know

865
00:53:04,520 --> 00:53:06,600
Speaker 1: we would We would love to hear it, and we

866
00:53:06,800 --> 00:53:11,160
Speaker 1: applaud you for pioneering into this brave, new strange world.

867
00:53:12,120 --> 00:53:16,400
Speaker 1: Government institutions like our favorite Matt, scientists, DARPA, and researchers

868
00:53:16,440 --> 00:53:20,040
Speaker 1: at colleges like you had mentioned Matt Carnegie Mellon University

869
00:53:20,040 --> 00:53:24,120
Speaker 1: of Washington, Stanford so on are also experimenting with deep

870
00:53:24,160 --> 00:53:28,239
Speaker 1: fake technology. They're they're experimenting in two different but very

871
00:53:28,320 --> 00:53:31,840
Speaker 1: related paths. One they want to figure out how to

872
00:53:31,960 --> 00:53:34,279
Speaker 1: use it, how to make it better, and to while

873
00:53:34,360 --> 00:53:36,040
Speaker 1: doing that, they want to figure out how to stop it.

874
00:53:36,360 --> 00:53:39,480
Speaker 1: So thank you. These goals are kind of contradictory unless

875
00:53:39,520 --> 00:53:42,600
Speaker 1: they build in some kind of equivalent of a governor switch,

876
00:53:43,080 --> 00:53:47,920
Speaker 1: you know, like for anyone unfamiliar with automobiles, uh some uh,

877
00:53:48,120 --> 00:53:51,800
Speaker 1: some engines have a specific switch in there that limits

878
00:53:51,960 --> 00:53:55,960
Speaker 1: the performance of the engine governor, like in long distance

879
00:53:56,000 --> 00:53:58,640
Speaker 1: trucks especially. There's a company. I only know this because

880
00:53:58,640 --> 00:54:00,440
Speaker 1: I knew a guy that that was a long haul

881
00:54:00,440 --> 00:54:03,960
Speaker 1: truck driver. And there's a company called Swift that everyone

882
00:54:04,239 --> 00:54:06,640
Speaker 1: jokes in the industry stands for sure, wish I had

883
00:54:06,680 --> 00:54:10,120
Speaker 1: a fast truck because they are notorious for putting governors

884
00:54:10,200 --> 00:54:12,680
Speaker 1: on there. That I mean, I guess it's a calculation

885
00:54:12,760 --> 00:54:14,919
Speaker 1: you make as a business in terms of the risk,

886
00:54:15,200 --> 00:54:17,000
Speaker 1: you know, versus like how fast can we get there

887
00:54:17,120 --> 00:54:19,520
Speaker 1: versus how likely are our drivers are gonna drive too

888
00:54:19,560 --> 00:54:22,920
Speaker 1: fast and potentially put themselves at risk and others and

889
00:54:23,000 --> 00:54:25,319
Speaker 1: you know, open up for liability and lawsuits, but yeah,

890
00:54:25,360 --> 00:54:29,399
Speaker 1: that's absolutely a thing. So a blue Bird buses would

891
00:54:29,400 --> 00:54:31,359
Speaker 1: be another great example, and they have their own very

892
00:54:31,440 --> 00:54:36,080
Speaker 1: interesting story. Bird scooters have digitally triggered governors, where like

893
00:54:36,200 --> 00:54:38,040
Speaker 1: we have an area in town called the belt line

894
00:54:38,280 --> 00:54:41,200
Speaker 1: where it GEO targets that area and when you ride

895
00:54:41,239 --> 00:54:43,120
Speaker 1: them on our belt line, which is like almost like

896
00:54:43,160 --> 00:54:45,120
Speaker 1: the high line in New York, it's like a walking trail,

897
00:54:45,160 --> 00:54:47,040
Speaker 1: but you can ride the scooters. It does not allow

898
00:54:47,120 --> 00:54:49,200
Speaker 1: you to go past a certain speed that you could

899
00:54:49,280 --> 00:54:51,920
Speaker 1: go past elsewhere in the city. Man, I remember I

900
00:54:52,000 --> 00:54:57,160
Speaker 1: had one on my Dodge caravan talked out right around.

901
00:54:59,000 --> 00:55:03,319
Speaker 1: So so we're talking about the technological analog. That's why

902
00:55:03,320 --> 00:55:06,360
Speaker 1: I'm bringing up governor switches, because really it seems like

903
00:55:06,480 --> 00:55:10,480
Speaker 1: researching the improvement of a given technology while also researching

904
00:55:10,560 --> 00:55:13,759
Speaker 1: a way to stymy that technology means that you would

905
00:55:13,800 --> 00:55:17,640
Speaker 1: ultimately build in something like that, and ideally you would

906
00:55:17,680 --> 00:55:20,560
Speaker 1: want it to be proprietary such that you would control it.

907
00:55:20,840 --> 00:55:25,040
Speaker 1: So in a way, these institutions are competing. Who will

908
00:55:25,160 --> 00:55:28,440
Speaker 1: control the nature of the truth and reality. That's a

909
00:55:28,600 --> 00:55:31,399
Speaker 1: that's a big question, but it's it's there. I think

910
00:55:31,400 --> 00:55:34,800
Speaker 1: I think we need some kind of tech Ethics board

911
00:55:35,200 --> 00:55:38,000
Speaker 1: or you know, or advisory committee like tech. It's like

912
00:55:38,360 --> 00:55:41,520
Speaker 1: like tethics, tethics, Yeah, tethics. We should do that. We

913
00:55:41,520 --> 00:55:44,080
Speaker 1: should get somebody to come through and like create a

914
00:55:44,360 --> 00:55:48,239
Speaker 1: something sign. Doesn't that have to be self imposed by

915
00:55:48,360 --> 00:55:51,040
Speaker 1: like these tech companies, you know what I mean? Like,

916
00:55:51,320 --> 00:55:53,040
Speaker 1: I don't know, if you get like a Gavin Belson

917
00:55:53,160 --> 00:55:55,399
Speaker 1: or somebody like that, you could probably get everybody else

918
00:55:55,440 --> 00:55:58,200
Speaker 1: on board, like you know, a big name. You'd have to, uh,

919
00:55:59,040 --> 00:56:02,759
Speaker 1: you have to ruin some an example, you'd have to

920
00:56:02,880 --> 00:56:05,400
Speaker 1: ruin some people in the beginning for an example probably,

921
00:56:05,840 --> 00:56:07,759
Speaker 1: But but I think you could. You could push it through.

922
00:56:07,960 --> 00:56:10,279
Speaker 1: You just you name dropped a fictional character. I just

923
00:56:10,320 --> 00:56:12,160
Speaker 1: want to point that out to wait what Yeah, just

924
00:56:12,840 --> 00:56:15,600
Speaker 1: I was putting that out there. Who any Silicon Valley

925
00:56:15,680 --> 00:56:18,200
Speaker 1: fans out trying to I'm trying to put something in

926
00:56:18,280 --> 00:56:20,359
Speaker 1: there that you literally made me get a double Wait

927
00:56:20,360 --> 00:56:22,719
Speaker 1: a minute, like is he? Who is he? How come

928
00:56:22,760 --> 00:56:26,359
Speaker 1: I haven't heard of him? It's trying to Easter Egget. Sorry, dude,

929
00:56:26,960 --> 00:56:28,840
Speaker 1: I shouldn't have said anything. He played along, but you

930
00:56:28,920 --> 00:56:31,440
Speaker 1: played along perfectly. Oh thanks man, that was beautiful. I

931
00:56:31,480 --> 00:56:35,319
Speaker 1: really screwed it all guys. Sorry, well, what what's next? Really?

932
00:56:35,440 --> 00:56:39,239
Speaker 1: I mean, I think I think right now, regardless, there

933
00:56:39,239 --> 00:56:43,160
Speaker 1: are people who are pro deep fake technology. Uh, there's

934
00:56:43,239 --> 00:56:46,040
Speaker 1: there's a very convincing argument, all right, I think a

935
00:56:46,120 --> 00:56:49,360
Speaker 1: very exciting argument that this could fundamentally change what we

936
00:56:49,480 --> 00:56:53,520
Speaker 1: think of as film as entertainment because imagine you have

937
00:56:53,680 --> 00:56:57,719
Speaker 1: the perfect role for an actor who has passed uh,

938
00:56:57,880 --> 00:57:01,360
Speaker 1: and you want them to be your film. With this

939
00:57:01,520 --> 00:57:05,960
Speaker 1: kind of technology, you can do it plausibly. And that

940
00:57:06,120 --> 00:57:10,000
Speaker 1: means that, coupled also with machine learning writing of fiction,

941
00:57:10,400 --> 00:57:13,840
Speaker 1: we could arrive at a time in our individual or

942
00:57:13,880 --> 00:57:20,280
Speaker 1: collective lives where a film is made without human involvement

943
00:57:20,400 --> 00:57:24,200
Speaker 1: on the creative end. How insane is that? Forward to

944
00:57:24,280 --> 00:57:27,040
Speaker 1: the future, I say, no, reduce. We know that other

945
00:57:27,080 --> 00:57:30,640
Speaker 1: researchers are attempting to do that, like combining the different

946
00:57:31,280 --> 00:57:34,560
Speaker 1: versions and types of neural networks, to write a screenplay,

947
00:57:35,080 --> 00:57:38,680
Speaker 1: to do some voice over at work, you know, um,

948
00:57:39,040 --> 00:57:43,480
Speaker 1: to actually shoot video and edit video. Um, It's it's

949
00:57:43,480 --> 00:57:45,640
Speaker 1: all happening. It's just a matter of time. I think

950
00:57:45,920 --> 00:57:47,720
Speaker 1: you're right on the money. They're been. So where does

951
00:57:47,760 --> 00:57:51,640
Speaker 1: this leave us for and beyond? We're at a where

952
00:57:51,720 --> 00:57:56,160
Speaker 1: to dilemma because it's really a matter of free expression

953
00:57:56,440 --> 00:58:00,600
Speaker 1: versus true deception. According to Sharon Bradford frank And as

954
00:58:00,600 --> 00:58:04,520
Speaker 1: a policy director for the Open Technology Institute out of

955
00:58:04,560 --> 00:58:08,280
Speaker 1: New America, deep fake videos threatened our civic discourse and

956
00:58:08,360 --> 00:58:12,840
Speaker 1: can cause serious reputational and psychic harmed individuals. Right. They

957
00:58:12,880 --> 00:58:15,880
Speaker 1: also make it more challenging for platforms to engage in

958
00:58:16,000 --> 00:58:19,920
Speaker 1: responsible moderation of content, which is already a huge problem

959
00:58:20,000 --> 00:58:22,160
Speaker 1: for anyone who's been paying attention to the latest news

960
00:58:22,200 --> 00:58:27,800
Speaker 1: about Facebook. While the public is understandably calling for social

961
00:58:27,880 --> 00:58:31,000
Speaker 1: media companies to develop techniques to detect and prevent the

962
00:58:31,080 --> 00:58:35,840
Speaker 1: spread of deep fakes, we must also avoid establishing legal

963
00:58:36,000 --> 00:58:38,960
Speaker 1: rules that push too far in the opposite direction and

964
00:58:39,080 --> 00:58:44,080
Speaker 1: pressure platforms to engage in censorship of free expression online. So,

965
00:58:44,360 --> 00:58:46,640
Speaker 1: to take your earlier argument, which again Matt, I love,

966
00:58:46,680 --> 00:58:49,479
Speaker 1: where someone says, that's not me, this is a deep fake.

967
00:58:50,040 --> 00:58:52,800
Speaker 1: What happens if you were posting something, let's say you're

968
00:58:52,840 --> 00:58:57,320
Speaker 1: an activist or you've you've or a whistleblower, yes even better,

969
00:58:57,640 --> 00:59:02,280
Speaker 1: and you propagate this film that is indisputable proof of

970
00:59:02,400 --> 00:59:05,200
Speaker 1: the shenanigans you said, we're carrying all along, and then

971
00:59:05,240 --> 00:59:08,120
Speaker 1: the people who have their their hands on the switches there,

972
00:59:08,320 --> 00:59:10,680
Speaker 1: you know, their fingers on the faucet, they just say, oh,

973
00:59:10,800 --> 00:59:12,640
Speaker 1: that's a deep fake and they turn it off. We

974
00:59:12,760 --> 00:59:14,880
Speaker 1: were always at war with they say, sha, you know

975
00:59:14,960 --> 00:59:20,320
Speaker 1: what I mean? Yeah, Uh, it's harrowing when you put

976
00:59:20,400 --> 00:59:23,480
Speaker 1: that at the infant. We're always in more with these, right,

977
00:59:23,920 --> 00:59:25,360
Speaker 1: I mean, but where do you where do you think

978
00:59:25,440 --> 00:59:28,720
Speaker 1: this is going? I would say one person's opinion that

979
00:59:29,080 --> 00:59:32,600
Speaker 1: it's it's going to continue, it will not disappear. I

980
00:59:32,680 --> 00:59:36,680
Speaker 1: will say, if you want to see really really um

981
00:59:37,200 --> 00:59:40,760
Speaker 1: Jarring Lee good example of what this can look like.

982
00:59:41,480 --> 00:59:47,000
Speaker 1: It's a video of um Bill Hayter on is it

983
00:59:47,160 --> 00:59:50,800
Speaker 1: the Letterman Show? No, yeah, it was an older talk

984
00:59:50,880 --> 00:59:52,600
Speaker 1: to an old clip. It was an old clip. It's

985
00:59:52,600 --> 00:59:54,600
Speaker 1: a Bill Hayter. I believe it's the Letterman Show. It

986
00:59:54,680 --> 00:59:57,800
Speaker 1: might be Conan and he's telling the story about meeting

987
00:59:57,920 --> 01:00:01,840
Speaker 1: Tom Cruise. Um a party of some kind of it

988
01:00:01,880 --> 01:00:03,840
Speaker 1: had to do with um. No, he was not what

989
01:00:03,960 --> 01:00:05,600
Speaker 1: it was. He was in that movie Tropic Thunder with

990
01:00:05,680 --> 01:00:07,640
Speaker 1: with Tom Cruise and he sees he meets Tom Cruise

991
01:00:07,920 --> 01:00:10,720
Speaker 1: at the premiere and Bill Hayter at this point isn't

992
01:00:10,760 --> 01:00:12,880
Speaker 1: like huge, he's you know, a escenal. He's known for

993
01:00:12,960 --> 01:00:15,959
Speaker 1: doing impressions, etcetera. So of course when he's telling the story,

994
01:00:16,320 --> 01:00:18,720
Speaker 1: he's doing the Tom Cruise impression. When he's doing Tom

995
01:00:18,760 --> 01:00:21,040
Speaker 1: Cruises parts of the story, he's a great impression. He's

996
01:00:21,040 --> 01:00:23,280
Speaker 1: a great impressionist. And in this deep in this video,

997
01:00:23,720 --> 01:00:26,720
Speaker 1: every time he starts to lapse into the Tom Cruise voice,

998
01:00:27,160 --> 01:00:29,720
Speaker 1: his face turns into Tom Cruise and it is it

999
01:00:29,920 --> 01:00:33,360
Speaker 1: is Jarring Lee good and it's it's Almo. It's borderline

1000
01:00:33,440 --> 01:00:36,200
Speaker 1: like it makes your brain kind of like spas them

1001
01:00:36,200 --> 01:00:38,680
Speaker 1: a little bit because it's just so good. He also

1002
01:00:38,760 --> 01:00:42,440
Speaker 1: does a great impression of Arnold Schwarzenegger and Al Pacino.

1003
01:00:42,520 --> 01:00:44,919
Speaker 1: I want to say, and you can see the same

1004
01:00:45,040 --> 01:00:47,680
Speaker 1: technology of play. The Tom Cruise when I would say,

1005
01:00:47,760 --> 01:00:50,160
Speaker 1: is the best example because their faces are already a

1006
01:00:50,240 --> 01:00:52,280
Speaker 1: little more similar this one, even though it goes a

1007
01:00:52,320 --> 01:00:54,920
Speaker 1: little further where he starts to do a Seth Rogan

1008
01:00:55,000 --> 01:00:58,120
Speaker 1: impression later and then his face becomes Seth Rogan, but

1009
01:00:58,320 --> 01:01:00,760
Speaker 1: the way it does it morphs where it just like

1010
01:01:00,920 --> 01:01:03,200
Speaker 1: for a split second it'll be Seth Rowgan and then

1011
01:01:03,240 --> 01:01:06,200
Speaker 1: it's back to Bill Hayter. But the Tom Cruise parts

1012
01:01:06,240 --> 01:01:09,680
Speaker 1: are are shockingly good. It doesn't look like mapping. It

1013
01:01:09,680 --> 01:01:12,640
Speaker 1: doesn't look like projection mapping or you know, it really

1014
01:01:12,840 --> 01:01:16,040
Speaker 1: very much is like he becomes him and everything seems

1015
01:01:16,280 --> 01:01:19,240
Speaker 1: you can tell. It's like pulling from something that Tom

1016
01:01:19,320 --> 01:01:21,800
Speaker 1: Cruise did where he was acting, you know. Like, but

1017
01:01:22,240 --> 01:01:24,760
Speaker 1: like you said, Ben, this video that we're seeing doesn't

1018
01:01:24,920 --> 01:01:27,840
Speaker 1: really exist in the wild. It's like a you know,

1019
01:01:28,200 --> 01:01:30,680
Speaker 1: um composite of like all of this stuff that's out there,

1020
01:01:30,760 --> 01:01:32,400
Speaker 1: right Are you seeing it, Matt, Well, yeah, I've seen it.

1021
01:01:32,440 --> 01:01:34,840
Speaker 1: And there's the there's one that's called the Deep Fake

1022
01:01:34,920 --> 01:01:40,040
Speaker 1: Impressionist or Deep. There's there's another impersonator. It's a Instagram

1023
01:01:40,120 --> 01:01:42,400
Speaker 1: account that I follow as well that does a lot

1024
01:01:42,480 --> 01:01:44,840
Speaker 1: of these. Okay, yeah, they're all I know is this

1025
01:01:44,880 --> 01:01:46,680
Speaker 1: is one guy that I've seen several videos of. I

1026
01:01:46,720 --> 01:01:48,600
Speaker 1: couldn't even tell you his name right now, but he

1027
01:01:48,880 --> 01:01:51,480
Speaker 1: got together with somebody. They made an entire video of

1028
01:01:51,720 --> 01:01:56,720
Speaker 1: NonStop versions of this and it really is convincing, um crazy,

1029
01:01:57,040 --> 01:02:00,160
Speaker 1: And this is just the beginning. People will look back

1030
01:02:00,320 --> 01:02:07,880
Speaker 1: on this era as the halcyon days of uh discernible fakes.

1031
01:02:08,600 --> 01:02:11,360
Speaker 1: At least if this continues where just the time of

1032
01:02:11,480 --> 01:02:14,080
Speaker 1: us knowing what was real and what was not right?

1033
01:02:14,560 --> 01:02:17,280
Speaker 1: Where where do you see this technology going? Folks who

1034
01:02:17,320 --> 01:02:19,240
Speaker 1: want to pass the torch to you, let us know,

1035
01:02:19,640 --> 01:02:22,680
Speaker 1: and on the way, send us your favorite deep fakes.

1036
01:02:22,960 --> 01:02:25,480
Speaker 1: You can post them on our community page. Here's where

1037
01:02:25,480 --> 01:02:28,280
Speaker 1: it gets crazy. On Facebook. You can find us on Instagram,

1038
01:02:28,400 --> 01:02:31,000
Speaker 1: you can find us on Twitter. You can also find

1039
01:02:31,120 --> 01:02:34,040
Speaker 1: every episode we have ever done on our website. Stuff

1040
01:02:34,040 --> 01:02:35,920
Speaker 1: they don't want you to know dot com And we

1041
01:02:36,000 --> 01:02:38,640
Speaker 1: don't put this out there enough. But as the holidays

1042
01:02:38,680 --> 01:02:41,640
Speaker 1: are coming around, remember we've got a t public store.

1043
01:02:41,840 --> 01:02:44,440
Speaker 1: So if you want to get some some awesome stuff

1044
01:02:44,480 --> 01:02:47,120
Speaker 1: they don't want you to know, iifts whatever it is

1045
01:02:47,680 --> 01:02:52,160
Speaker 1: pads do it and you know that's look candidly and upfront.

1046
01:02:52,240 --> 01:02:55,360
Speaker 1: We get a tiny, tiny, witty bitty percentage, But you

1047
01:02:55,440 --> 01:02:58,560
Speaker 1: are supporting our show by by doing this. As the

1048
01:02:58,600 --> 01:03:01,640
Speaker 1: designs are cool, they're real cool. I I genuinely wear

1049
01:03:01,920 --> 01:03:04,000
Speaker 1: my new red stuff. They don't want you to know.

1050
01:03:04,160 --> 01:03:11,640
Speaker 1: Shirt time that the text version like a Superman move right. Look,

1051
01:03:11,960 --> 01:03:14,000
Speaker 1: we feel weird wearing our own things, so a lot

1052
01:03:14,040 --> 01:03:15,880
Speaker 1: of times we do what Ben's doing. You put something

1053
01:03:15,960 --> 01:03:18,240
Speaker 1: over it. But but when you wear it, it's almost

1054
01:03:18,320 --> 01:03:21,160
Speaker 1: like for me like having the Superman's thing on. You know,

1055
01:03:21,280 --> 01:03:22,800
Speaker 1: I wear the stuff they don't want you to know,

1056
01:03:23,080 --> 01:03:26,840
Speaker 1: signature tighty whities. What the stuff they don't want you

1057
01:03:26,920 --> 01:03:30,480
Speaker 1: to know? Snuggie, Do you actually have underwear custom made? Yes?

1058
01:03:30,840 --> 01:03:33,200
Speaker 1: Oh my god. I don't think the public offers it.

1059
01:03:33,360 --> 01:03:36,400
Speaker 1: I had to outsource it. And more importantly, we have

1060
01:03:36,560 --> 01:03:39,280
Speaker 1: a good authory. I don't know if you guys get

1061
01:03:39,320 --> 01:03:41,640
Speaker 1: these texts to but every time somebody buys something from

1062
01:03:41,680 --> 01:03:44,360
Speaker 1: the store, I get a text from Connal that just

1063
01:03:44,480 --> 01:03:49,560
Speaker 1: says one more day. Yeah, I'm not on that chain,

1064
01:03:49,760 --> 01:03:52,240
Speaker 1: but I'd like to get on there. That That would

1065
01:03:52,240 --> 01:03:55,880
Speaker 1: be fun. Really, it would really just fuel my uh

1066
01:03:56,760 --> 01:04:01,959
Speaker 1: my neuroses. Right uh if what if maybe you're listening

1067
01:04:02,000 --> 01:04:04,720
Speaker 1: you're saying okay, Even if I assume that you guys

1068
01:04:04,760 --> 01:04:06,760
Speaker 1: are real and your voices are not being faked by

1069
01:04:06,800 --> 01:04:10,560
Speaker 1: that technology you alluded to earlier, I hate social media.

1070
01:04:11,000 --> 01:04:16,160
Speaker 1: I think that real conversations have to happen on the phone,

1071
01:04:16,240 --> 01:04:18,920
Speaker 1: because because that's how I am. What do you do?

1072
01:04:19,080 --> 01:04:22,800
Speaker 1: Then you could call us we are one eight three

1073
01:04:22,880 --> 01:04:27,360
Speaker 1: three st d W I T K leave a message,

1074
01:04:27,520 --> 01:04:30,280
Speaker 1: tell us what you think. Uh, tell us what you

1075
01:04:30,960 --> 01:04:35,080
Speaker 1: about a deep fake video? Maybe make a deep fake

1076
01:04:35,240 --> 01:04:37,720
Speaker 1: video and send it our way. Whatever, it doesn't matter.

1077
01:04:37,920 --> 01:04:40,160
Speaker 1: Just call us, tell us you're what you think about

1078
01:04:40,160 --> 01:04:43,920
Speaker 1: this episode and give us suggestions for other episodes. Remember

1079
01:04:43,960 --> 01:04:48,600
Speaker 1: when photoshop was like the original deep fake? Yes, you know, definitely.

1080
01:04:48,640 --> 01:04:52,560
Speaker 1: I used to photoshop images of Josh Clark on The Hulk.

1081
01:04:52,720 --> 01:04:54,240
Speaker 1: I used to do that a lot. I used to

1082
01:04:54,320 --> 01:04:58,120
Speaker 1: photoshop images of you guys onto things. That was one

1083
01:04:58,160 --> 01:04:59,880
Speaker 1: of my favorite things to do. Put it in videos.

1084
01:05:00,120 --> 01:05:02,600
Speaker 1: I remember we were mentioning that earlier in the episode

1085
01:05:03,240 --> 01:05:05,680
Speaker 1: with photoshop, and I was I was thinking the exact

1086
01:05:05,800 --> 01:05:07,960
Speaker 1: same thing. I didn't. They don't want to play your

1087
01:05:08,040 --> 01:05:10,840
Speaker 1: hand for you. But for anyone who doesn't know, Matt

1088
01:05:11,000 --> 01:05:13,880
Speaker 1: is actually quite talented at photoshop. I think a lot

1089
01:05:13,960 --> 01:05:15,919
Speaker 1: of people in this office are you got You would

1090
01:05:15,920 --> 01:05:17,840
Speaker 1: be selling yourself short as if you said you didn't

1091
01:05:17,880 --> 01:05:22,760
Speaker 1: have them, Mr Boland or Brown. Uh So, anyway, call us.

1092
01:05:22,920 --> 01:05:24,480
Speaker 1: That was the end of that. That's just you know,

1093
01:05:24,520 --> 01:05:26,120
Speaker 1: you can call us. If you don't want to do that,

1094
01:05:26,240 --> 01:05:30,040
Speaker 1: you can always reach us via our email. That's right,

1095
01:05:30,120 --> 01:05:33,520
Speaker 1: we have an email. It is conspiracy at iHeart radio

1096
01:05:33,680 --> 01:05:54,520
Speaker 1: dot com stuff they don't want you to know. As

1097
01:05:54,560 --> 01:05:57,280
Speaker 1: a production of iHeart Radio's How Stuff Works. For more

1098
01:05:57,360 --> 01:06:00,000
Speaker 1: podcasts from my heart Radio, visit the iHeart radio app,

1099
01:06:00,160 --> 01:06:03,040
Speaker 1: Apple Podcasts, or wherever you listen to your favorite shows.