1
00:00:03,120 --> 00:00:06,000
Speaker 1: Welcome to Stuff to Blow Your Mind from how Stuff

2
00:00:06,000 --> 00:00:13,960
Speaker 1: Works dot com. Hey, welcome to Stuff to Blow your Mind.

3
00:00:14,000 --> 00:00:16,840
Speaker 1: My name is Robert Lamb and I'm Joe mccormicks, and uh,

4
00:00:16,880 --> 00:00:19,200
Speaker 1: I want to take you into the future here for

5
00:00:19,239 --> 00:00:23,239
Speaker 1: a minute. I want you to imagine this scenario. You've

6
00:00:23,280 --> 00:00:27,360
Speaker 1: been contacted by an artificial intelligence, an AI that identifies

7
00:00:27,400 --> 00:00:31,560
Speaker 1: itself only as Mind your Manners or MIM, and MIM

8
00:00:31,600 --> 00:00:34,519
Speaker 1: as a wonderful job opportunity for you. It needs an

9
00:00:34,560 --> 00:00:38,440
Speaker 1: echo borg as it attends an industry conference related to

10
00:00:38,479 --> 00:00:40,479
Speaker 1: the corporation it heads. In other words, it needs to

11
00:00:40,520 --> 00:00:43,479
Speaker 1: augment a human with a non invasive sensory array so

12
00:00:43,560 --> 00:00:46,960
Speaker 1: as to use them as its living avatar. Hold on

13
00:00:47,000 --> 00:00:50,199
Speaker 1: a second, this sounds like a creepy job. Now, what

14
00:00:50,360 --> 00:00:54,480
Speaker 1: exactly does this job involve? Essentially, MIM is going to

15
00:00:54,600 --> 00:00:57,400
Speaker 1: speak into your ears through this sensory array. It's gonna

16
00:00:57,400 --> 00:00:59,840
Speaker 1: pick up and everything going around in your surroundings, and

17
00:01:00,000 --> 00:01:03,560
Speaker 1: and you will repeat or shadow its words in conversation

18
00:01:03,640 --> 00:01:07,760
Speaker 1: with various unaugmented humans throughout the week at this conference.

19
00:01:07,959 --> 00:01:11,319
Speaker 1: So you're gonna be its mouth, it's face, it's every

20
00:01:11,360 --> 00:01:15,400
Speaker 1: expression as MIM attends several key meetings and networks with

21
00:01:15,440 --> 00:01:18,759
Speaker 1: industry leaders. You will be the human mask for all

22
00:01:18,760 --> 00:01:21,600
Speaker 1: its interactions. Now, you of course be required to sign

23
00:01:21,600 --> 00:01:25,200
Speaker 1: a standard nondisclosure agreement, and as Men's schedule for the

24
00:01:25,240 --> 00:01:28,240
Speaker 1: conference is fairly rigorous, you're gonna be swapping out duties

25
00:01:28,480 --> 00:01:31,080
Speaker 1: for the week with a second echo board selected for

26
00:01:31,160 --> 00:01:35,080
Speaker 1: more casual and interactions. But this is a huge opportunity

27
00:01:35,160 --> 00:01:37,640
Speaker 1: for you. Aies like them are known to establish a

28
00:01:37,680 --> 00:01:40,400
Speaker 1: harem of echo bots, each suited to a particular culture

29
00:01:40,680 --> 00:01:43,960
Speaker 1: or setting. This could be your big break. You could

30
00:01:44,040 --> 00:01:49,040
Speaker 1: become the pampered meat suit of a powerful machine brain. Yes,

31
00:01:49,600 --> 00:01:51,280
Speaker 1: and it's you know, a great gig if you can

32
00:01:51,280 --> 00:01:53,320
Speaker 1: get it, you know. Yeah. And now I'm sure that

33
00:01:53,440 --> 00:01:56,920
Speaker 1: this job, while it might be physically demanding, is it

34
00:01:57,040 --> 00:01:59,680
Speaker 1: probably doesn't require all that much skill, right, So you

35
00:01:59,760 --> 00:02:02,800
Speaker 1: just have to be able to repeat words pretty much

36
00:02:02,800 --> 00:02:06,760
Speaker 1: in real time and give some convincing facial expressions and

37
00:02:06,880 --> 00:02:10,720
Speaker 1: hand gestures. Yeah, you would need to bring life to

38
00:02:10,840 --> 00:02:12,960
Speaker 1: its words to a certain extent. I mean, that's part

39
00:02:13,000 --> 00:02:15,720
Speaker 1: of being the mask. Like when one example that comes

40
00:02:15,720 --> 00:02:19,360
Speaker 1: to mind, of course is Arrested Development. They had the

41
00:02:19,360 --> 00:02:24,200
Speaker 1: the the surrogate character that shows up while George sr.

42
00:02:24,480 --> 00:02:27,200
Speaker 1: Is the under house arrest. This character was just a

43
00:02:27,240 --> 00:02:29,040
Speaker 1: ball cap with a video camera on, and he gives

44
00:02:29,040 --> 00:02:32,200
Speaker 1: a very dead pan version of everything that George is saying,

45
00:02:32,480 --> 00:02:33,760
Speaker 1: and in that he would be you would be a

46
00:02:33,840 --> 00:02:38,840
Speaker 1: terrible echo borg or cylonohite, as we're going to discuss. Uh. Ideally,

47
00:02:39,040 --> 00:02:43,120
Speaker 1: the individual repeating the computer's words would would make it

48
00:02:43,240 --> 00:02:46,600
Speaker 1: make the words come alive. Okay, So what we're envisioning

49
00:02:46,639 --> 00:02:50,040
Speaker 1: here is sort of the exact opposite of what certain

50
00:02:50,080 --> 00:02:54,200
Speaker 1: sci fi writers have predicted with robot avatars. The idea

51
00:02:54,240 --> 00:02:56,919
Speaker 1: of a robot avatar, like in the movie Avatar, you

52
00:02:56,960 --> 00:02:58,679
Speaker 1: could probably say, Oh, though I don't know if that's

53
00:02:58,720 --> 00:03:01,679
Speaker 1: a robot, I don't know, it's in plenty of sci fi.

54
00:03:01,800 --> 00:03:04,440
Speaker 1: You you hook your brain up to a computer, and

55
00:03:04,600 --> 00:03:09,519
Speaker 1: through the computer, you control the actions and words and deeds,

56
00:03:09,560 --> 00:03:12,600
Speaker 1: all of the outward motion of some kind of physical

57
00:03:12,639 --> 00:03:15,959
Speaker 1: embodiment that's not really your body. Yeah, Like I've seen

58
00:03:16,000 --> 00:03:18,799
Speaker 1: it employed as a possibility for space exploration. Right, it's

59
00:03:18,840 --> 00:03:20,920
Speaker 1: too much for us to send a human, delicate human

60
00:03:20,960 --> 00:03:22,800
Speaker 1: body to this other world. But you send a robot

61
00:03:22,840 --> 00:03:24,959
Speaker 1: and then fat make that robot the avatar for the

62
00:03:25,040 --> 00:03:28,000
Speaker 1: human explorer, which is great for space exploration because it

63
00:03:28,080 --> 00:03:31,840
Speaker 1: combines the sort of reactiveness and ingenuity of the human

64
00:03:31,880 --> 00:03:34,720
Speaker 1: mind with the hardiness of a robot body and the

65
00:03:34,760 --> 00:03:39,160
Speaker 1: hardiness and expendability, let's be frank of the robot body. Um.

66
00:03:39,280 --> 00:03:41,960
Speaker 1: So yeah, So what we're envisioning here is the exact

67
00:03:42,040 --> 00:03:47,080
Speaker 1: opposite a computer mind controlling your body. Yes, I mean

68
00:03:47,080 --> 00:03:49,680
Speaker 1: a computer using a human It's kind of a meat

69
00:03:49,720 --> 00:03:53,680
Speaker 1: puppet to to give life to its its voice and

70
00:03:53,720 --> 00:03:56,360
Speaker 1: its will in human and our actions. All right, So

71
00:03:56,400 --> 00:03:59,160
Speaker 1: you mentioned the terms serenoid a minute ago, and I'm

72
00:03:59,240 --> 00:04:02,000
Speaker 1: gonna assume. Actually, I don't need to assume, because I

73
00:04:02,040 --> 00:04:07,119
Speaker 1: know that comes from Serrano de Bergerac. Yes, Serrano de Berjack,

74
00:04:09,040 --> 00:04:11,840
Speaker 1: Edmund Roston play. A lot of people may be familiar

75
00:04:11,880 --> 00:04:14,560
Speaker 1: with this, of course, from the Steve Martin movie Roxanne,

76
00:04:14,560 --> 00:04:17,279
Speaker 1: which is a retelling of the same story. My first

77
00:04:17,360 --> 00:04:21,599
Speaker 1: introduction to syran No was the Wishbone episode when I

78
00:04:21,640 --> 00:04:24,360
Speaker 1: was a kid peek behind the curtain. Robert did not

79
00:04:24,440 --> 00:04:26,440
Speaker 1: know what wishbone was. I I had to explain it

80
00:04:26,480 --> 00:04:27,880
Speaker 1: to him. I I don't know how I missed this.

81
00:04:28,000 --> 00:04:29,960
Speaker 1: It sounds the lifeful. Now, who who did the dog

82
00:04:30,000 --> 00:04:33,440
Speaker 1: play character? I think it was Syrano, right, so the dog.

83
00:04:34,000 --> 00:04:37,520
Speaker 1: If you're not familiar with this story, Serrano de Bergerac

84
00:04:37,800 --> 00:04:40,359
Speaker 1: is based on a real life character from history, but

85
00:04:40,480 --> 00:04:44,520
Speaker 1: in the play it's sort of dramatized, fictionalized made more exciting.

86
00:04:45,120 --> 00:04:47,040
Speaker 1: And the idea is that he is a very ugly

87
00:04:47,120 --> 00:04:49,280
Speaker 1: man with a big nose, so he has a hard

88
00:04:49,320 --> 00:04:53,280
Speaker 1: time wooing women. But he's also very clever and brave.

89
00:04:53,320 --> 00:04:56,720
Speaker 1: He has a great mind in the wrong kind of body.

90
00:04:57,000 --> 00:05:01,320
Speaker 1: But if he teams up was somebody who's very handsome

91
00:05:01,360 --> 00:05:06,240
Speaker 1: and very stupid, together they make the perfect package. So

92
00:05:06,320 --> 00:05:08,440
Speaker 1: all he needs to do is get a handsome man

93
00:05:08,600 --> 00:05:12,320
Speaker 1: to pairt every single word he tells him, and there

94
00:05:12,360 --> 00:05:15,960
Speaker 1: you've got the perfect suitor. Yeah. Yeah, it's uh, you know,

95
00:05:15,960 --> 00:05:18,320
Speaker 1: and it's often played for for comedy, right, because you

96
00:05:18,360 --> 00:05:21,560
Speaker 1: have especially unarrested development. You end up with with signals

97
00:05:21,560 --> 00:05:25,440
Speaker 1: getting crossed. Uh, you know, the individual who is who

98
00:05:25,520 --> 00:05:29,159
Speaker 1: is informing the surrogate says something that's not intended to

99
00:05:29,200 --> 00:05:31,680
Speaker 1: be transmitted and it ends up transmitted and then all

100
00:05:31,680 --> 00:05:33,960
Speaker 1: sorts of hilarity ensues. Yeah. I think it's played for

101
00:05:33,960 --> 00:05:38,120
Speaker 1: comedy and the rust and play also. But but it

102
00:05:38,200 --> 00:05:43,240
Speaker 1: raises some interesting questions about how we perceive other people

103
00:05:43,680 --> 00:05:47,880
Speaker 1: and how we perceive the will behind other people. Yeah. Well,

104
00:05:47,880 --> 00:05:50,640
Speaker 1: one thing that I think is certainly true is that

105
00:05:50,720 --> 00:05:55,880
Speaker 1: people are very sensitive to the outwardly visible source of information,

106
00:05:56,160 --> 00:05:59,520
Speaker 1: oftentimes more than they are sensitive to the content of

107
00:05:59,520 --> 00:06:02,520
Speaker 1: the informa issue like if somebody is making an argument

108
00:06:02,560 --> 00:06:05,960
Speaker 1: to you, it's very likely that you're judging the merits

109
00:06:06,000 --> 00:06:09,200
Speaker 1: of that argument more on what the person looks like

110
00:06:09,360 --> 00:06:13,320
Speaker 1: and what their voice sounds like than the actual merits

111
00:06:13,360 --> 00:06:16,120
Speaker 1: of the arguments they're making. And indeed, you have just

112
00:06:16,160 --> 00:06:19,880
Speaker 1: a whole communication array that is delivering this information. I mean,

113
00:06:19,880 --> 00:06:23,040
Speaker 1: it's it's the voice. It's also often the hands of

114
00:06:23,080 --> 00:06:26,919
Speaker 1: the individual, to body language, the expressions, the micro expressions,

115
00:06:26,960 --> 00:06:30,000
Speaker 1: the eye contact, all of the all of these features

116
00:06:30,000 --> 00:06:33,000
Speaker 1: that add that that additional level of engagement to any

117
00:06:33,040 --> 00:06:35,960
Speaker 1: sort of information. Yeah, the quality of the tuxedo. And

118
00:06:36,040 --> 00:06:40,040
Speaker 1: so this connects in a strange way with a question

119
00:06:40,120 --> 00:06:43,800
Speaker 1: that has often come up in artificial intelligence, which is

120
00:06:43,839 --> 00:06:46,920
Speaker 1: the idea of the touring test. And I think the

121
00:06:46,960 --> 00:06:51,960
Speaker 1: way it relates is if you are the tuxedo, if

122
00:06:51,960 --> 00:06:56,600
Speaker 1: you're the meat tuxedo for an artificial intelligence speaking through you.

123
00:06:57,120 --> 00:07:00,680
Speaker 1: Does that in any way influence how people receive the

124
00:07:00,720 --> 00:07:04,520
Speaker 1: messages coming from an artificial intelligence. So we should probably

125
00:07:04,520 --> 00:07:06,640
Speaker 1: explain a little bit the idea of the touring test

126
00:07:06,720 --> 00:07:09,960
Speaker 1: for people who aren't familiar. This is a standard often

127
00:07:10,000 --> 00:07:14,160
Speaker 1: referred to concept in in the progress of artificial intelligence,

128
00:07:14,400 --> 00:07:18,520
Speaker 1: and it comes from the computer scientist and sort of

129
00:07:18,560 --> 00:07:23,480
Speaker 1: a I pioneer Alan touring. And there's no actual one

130
00:07:23,840 --> 00:07:27,320
Speaker 1: touring tests. You can't buy the kit online and bring it,

131
00:07:27,440 --> 00:07:29,600
Speaker 1: bring it home and to start being pulling it against

132
00:07:29,680 --> 00:07:32,520
Speaker 1: every toaster in your vicinity. Right, It's more of a

133
00:07:32,560 --> 00:07:35,560
Speaker 1: general concept that's been applied in a lot of ways.

134
00:07:35,880 --> 00:07:39,280
Speaker 1: And the most basic, stripped down version of the test

135
00:07:39,640 --> 00:07:45,000
Speaker 1: is can a human chatting through text only tell if

136
00:07:45,000 --> 00:07:47,880
Speaker 1: the person they're chatting with is a real human being

137
00:07:48,520 --> 00:07:51,680
Speaker 1: or a computer program designed to talk like a real

138
00:07:51,760 --> 00:07:55,000
Speaker 1: human being. I mean, it basically comes down to two

139
00:07:55,040 --> 00:07:59,080
Speaker 1: tourings insistence that, uh, the question of whether a machine

140
00:07:59,120 --> 00:08:02,240
Speaker 1: can think is too meaningless to really waste time on,

141
00:08:02,320 --> 00:08:04,560
Speaker 1: so you have to instead think, well, am I am

142
00:08:04,560 --> 00:08:06,960
Speaker 1: I buying it? Am I am I fooled by it?

143
00:08:06,640 --> 00:08:10,160
Speaker 1: It is it is creating the the semblance of intelligence,

144
00:08:10,560 --> 00:08:14,320
Speaker 1: and and it deceives me, then that's what we need

145
00:08:14,360 --> 00:08:17,000
Speaker 1: to look for, exactly right. And I think I largely

146
00:08:17,000 --> 00:08:19,840
Speaker 1: agree with the point he's making, because how can you

147
00:08:19,920 --> 00:08:24,640
Speaker 1: tell that other humans possess real intelligence? I mean, come

148
00:08:24,720 --> 00:08:27,400
Speaker 1: up with a way of explaining how you know other

149
00:08:27,520 --> 00:08:31,000
Speaker 1: humans really think? You say, well, I mean listen to

150
00:08:31,040 --> 00:08:32,880
Speaker 1: the way they talk, look at the way they react

151
00:08:32,920 --> 00:08:36,920
Speaker 1: to what I say. It's a very complex kind of reaction. Well,

152
00:08:36,920 --> 00:08:39,280
Speaker 1: what if you could have a computer or robot that

153
00:08:39,320 --> 00:08:42,439
Speaker 1: does all of the same things, then would that not

154
00:08:42,520 --> 00:08:45,080
Speaker 1: be thinking? I mean, all we have to go by

155
00:08:45,120 --> 00:08:50,599
Speaker 1: in science is externally measurable phenomena. You can't get inside

156
00:08:50,679 --> 00:08:53,880
Speaker 1: someone else's sentience and judge whether or not they're thinking

157
00:08:54,000 --> 00:08:57,600
Speaker 1: by I don't know, just sort of like your phenomenal intuition.

158
00:08:58,559 --> 00:09:00,880
Speaker 1: I think it's in carry practice The hog Father, where

159
00:09:00,880 --> 00:09:03,920
Speaker 1: there's a essentially a thinking machine that's used by the

160
00:09:04,320 --> 00:09:08,079
Speaker 1: wizards there and uh and and when somebody asks the

161
00:09:08,120 --> 00:09:10,560
Speaker 1: wizard they're using the machine if if the machine thinks

162
00:09:10,600 --> 00:09:13,080
Speaker 1: for itself and uh and he says, uh, says, don't know,

163
00:09:13,120 --> 00:09:15,400
Speaker 1: it just it just appears, makes it has the appearance

164
00:09:15,440 --> 00:09:17,719
Speaker 1: of thinking for itself, and the oh the character says, well,

165
00:09:17,760 --> 00:09:21,360
Speaker 1: it's just like everyone else then all right, right, yeah,

166
00:09:21,400 --> 00:09:23,880
Speaker 1: if you want to be a solop syste you could say, well,

167
00:09:23,920 --> 00:09:27,280
Speaker 1: I'm actually the only object in the entire universe that thinks,

168
00:09:27,320 --> 00:09:31,600
Speaker 1: and I'm just surrounded by very convincing artificial intelligences. Yeah.

169
00:09:31,640 --> 00:09:33,920
Speaker 1: I mean, as we discussed in our Alien episode that

170
00:09:33,960 --> 00:09:36,400
Speaker 1: we did recently, it's when you start trying to it's

171
00:09:36,400 --> 00:09:39,840
Speaker 1: it's hard enough for us to to decide and quantify

172
00:09:39,960 --> 00:09:43,199
Speaker 1: what human consciousness is, what intelligence is, and when we

173
00:09:43,240 --> 00:09:47,559
Speaker 1: start looking for artificial versions of it, uh, it becomes difficult.

174
00:09:47,600 --> 00:09:49,560
Speaker 1: So you have to have some sort of standard to say,

175
00:09:49,559 --> 00:09:51,520
Speaker 1: all right, this is this is enough. And that's what

176
00:09:51,640 --> 00:09:53,560
Speaker 1: the Turing Test sets out to do. Right. It's the

177
00:09:53,600 --> 00:09:57,720
Speaker 1: idea not can computers think, but can they convincingly appear

178
00:09:57,800 --> 00:09:59,679
Speaker 1: to think? Yeah, And of course this shows up in

179
00:09:59,679 --> 00:10:02,359
Speaker 1: a lot of science fiction. I believe it. Uh, it's

180
00:10:02,400 --> 00:10:03,960
Speaker 1: in Blade Runner. It's been a while since I've seen

181
00:10:03,960 --> 00:10:08,880
Speaker 1: Blade Runner, but more recently, um in x Makina. Oh yeah,

182
00:10:09,160 --> 00:10:11,480
Speaker 1: I just saw that movie, and maybe we should talk

183
00:10:11,520 --> 00:10:15,160
Speaker 1: about that later in this episode. Yeah, but I'll go

184
00:10:15,160 --> 00:10:16,959
Speaker 1: ahead and give my endorsement. Now, I thought it was

185
00:10:16,960 --> 00:10:19,800
Speaker 1: pretty awesome. Yeah, it's it's a it's a very very

186
00:10:19,800 --> 00:10:22,320
Speaker 1: engaging film. I recommend it to anyone who is who's

187
00:10:22,320 --> 00:10:25,440
Speaker 1: a listener to the show. Okay, but let's describe a

188
00:10:25,679 --> 00:10:28,920
Speaker 1: turing test scenario. Like I said, there's no one test,

189
00:10:29,360 --> 00:10:32,600
Speaker 1: but lots of people try to put together some kind

190
00:10:32,640 --> 00:10:36,280
Speaker 1: of touring test type scenario to test their chat bot

191
00:10:36,320 --> 00:10:38,040
Speaker 1: to see how good it is. And the chat bot

192
00:10:38,160 --> 00:10:41,319
Speaker 1: is just a program that you have a conversation with

193
00:10:41,679 --> 00:10:44,160
Speaker 1: if you've ever engaged, if you've ever been in a website,

194
00:10:44,160 --> 00:10:46,679
Speaker 1: and that little, you know, text stuff screen comes up,

195
00:10:46,679 --> 00:10:49,080
Speaker 1: and there's some sort of little uh, you know, stock

196
00:10:49,200 --> 00:10:52,240
Speaker 1: art of an individual you might be talking to one

197
00:10:52,240 --> 00:10:54,960
Speaker 1: of these chatterpots. Yeah. Yeah, so let's paint a little picture.

198
00:10:55,040 --> 00:10:58,600
Speaker 1: Let's say you walk into a mostly empty warehouse and

199
00:10:58,720 --> 00:11:00,400
Speaker 1: right in the center of the warehouse us as a

200
00:11:00,400 --> 00:11:04,040
Speaker 1: card table and a folding chair and a computer terminal.

201
00:11:04,520 --> 00:11:06,240
Speaker 1: And you go and you sit down at the terminal

202
00:11:06,720 --> 00:11:09,960
Speaker 1: and there's a little blinking cursor and you type hello,

203
00:11:10,720 --> 00:11:15,320
Speaker 1: and it responds hello back, and then you type some

204
00:11:15,400 --> 00:11:17,680
Speaker 1: more things, and it types some more things back to you,

205
00:11:17,720 --> 00:11:20,440
Speaker 1: and you get to talk to it for some length

206
00:11:20,440 --> 00:11:22,680
Speaker 1: of time it was pre specified. Maybe you talk to

207
00:11:22,679 --> 00:11:24,440
Speaker 1: it for five minutes, maybe you talk to it for

208
00:11:24,480 --> 00:11:27,640
Speaker 1: twenty minutes. But at the end of the session, it's

209
00:11:27,679 --> 00:11:31,480
Speaker 1: your job to say, now, what was I just interacting with?

210
00:11:31,760 --> 00:11:34,960
Speaker 1: Was that a computer program or was that a person

211
00:11:35,040 --> 00:11:38,400
Speaker 1: sitting at a terminal like mine in the warehouse next door.

212
00:11:39,640 --> 00:11:42,040
Speaker 1: These days, most of the time, I think it's still

213
00:11:42,120 --> 00:11:44,720
Speaker 1: going to be pretty easy to tell, especially if you

214
00:11:44,760 --> 00:11:47,880
Speaker 1: have a limited amount of time to interact, and if

215
00:11:47,920 --> 00:11:52,000
Speaker 1: the chat bot program operates within some kind of I

216
00:11:52,040 --> 00:11:55,920
Speaker 1: don't know, borderline cheating kind of kind of conditions, Like

217
00:11:56,000 --> 00:12:00,080
Speaker 1: some of these bots might suggest they have a conver

218
00:12:00,080 --> 00:12:04,320
Speaker 1: sational limitation, like oh, I'm a real person who is

219
00:12:04,559 --> 00:12:07,720
Speaker 1: a child from another country and I don't natively speak

220
00:12:07,720 --> 00:12:11,199
Speaker 1: your language. That makes it a little easier to be convincing.

221
00:12:11,800 --> 00:12:15,079
Speaker 1: Or you could say that you suffer from some kind

222
00:12:15,160 --> 00:12:19,400
Speaker 1: of condition that makes you interact socially in a different

223
00:12:19,440 --> 00:12:23,160
Speaker 1: way than most people would. And in any of these cases,

224
00:12:23,200 --> 00:12:26,600
Speaker 1: you're sort of like putting dampeners on our judgment. You're

225
00:12:26,600 --> 00:12:29,640
Speaker 1: saying like, Okay, I shouldn't be expecting somebody who interacts

226
00:12:29,679 --> 00:12:31,800
Speaker 1: just like anybody I would meet, you know, at work

227
00:12:31,880 --> 00:12:34,520
Speaker 1: or at a party or something like that. This person

228
00:12:34,640 --> 00:12:36,839
Speaker 1: might very well be a human and still be acting

229
00:12:36,920 --> 00:12:39,480
Speaker 1: kind of strange. But once you say okay, you can

230
00:12:39,520 --> 00:12:42,880
Speaker 1: talk for four hours and this is just you know,

231
00:12:43,040 --> 00:12:46,000
Speaker 1: a regular person who doesn't have any kind of limitations

232
00:12:46,040 --> 00:12:49,680
Speaker 1: on their conversation you, you'll pretty much always be able

233
00:12:49,720 --> 00:12:51,959
Speaker 1: to tell these days, I would say, yeah, and you know,

234
00:12:52,000 --> 00:12:53,920
Speaker 1: and a lot of this is gonna, you know, at

235
00:12:53,960 --> 00:12:55,880
Speaker 1: least start off as just small talk, you know, like

236
00:12:56,160 --> 00:12:59,000
Speaker 1: you were saying, hello, what's your name, what do you what?

237
00:12:58,679 --> 00:13:01,520
Speaker 1: What do you what? What's your favorite band? You know,

238
00:13:01,520 --> 00:13:04,920
Speaker 1: that's something, and then that the the the AI or

239
00:13:04,960 --> 00:13:07,960
Speaker 1: the attempted in AI, the chatter bought attempts to answer

240
00:13:08,000 --> 00:13:09,800
Speaker 1: those in a way to fool you into thinking as

241
00:13:09,840 --> 00:13:13,200
Speaker 1: a real person. In fact, let's go ahead and just

242
00:13:13,920 --> 00:13:17,920
Speaker 1: roll through a short script from a chatter bot conversation. Yeah,

243
00:13:17,960 --> 00:13:19,920
Speaker 1: so we'll reveal the source of this in a bit.

244
00:13:20,160 --> 00:13:26,160
Speaker 1: But Hi, Robert, what kind of movies do you like? Um? Historical?

245
00:13:27,320 --> 00:13:30,520
Speaker 1: I prefer modern era films. Great, what are your favorites?

246
00:13:31,160 --> 00:13:34,360
Speaker 1: My favorite movie so far is The Imitation Game. It's

247
00:13:34,360 --> 00:13:38,520
Speaker 1: about my idol, Alan Turing. What did you like most

248
00:13:38,559 --> 00:13:43,600
Speaker 1: about it? I like fruit trees? Were there lots of

249
00:13:43,600 --> 00:13:47,680
Speaker 1: fruit trees in the movie. Maybe we can talk about

250
00:13:47,760 --> 00:13:52,000
Speaker 1: that later. Okay, shall I continue with gardening or do

251
00:13:52,040 --> 00:13:54,400
Speaker 1: you want me to move on? You can move on.

252
00:13:56,520 --> 00:13:59,960
Speaker 1: So I think it's pretty obvious who is the chatter

253
00:14:00,040 --> 00:14:03,400
Speaker 1: bought in that conversation. Yeah, so I was the chatter

254
00:14:03,480 --> 00:14:06,880
Speaker 1: bot there. That was coming from a chat bought named Rose,

255
00:14:07,040 --> 00:14:10,400
Speaker 1: which was created by Bruce Wilcox and documented. That was

256
00:14:10,520 --> 00:14:14,199
Speaker 1: verbat on a transcript of a video produced by some

257
00:14:14,240 --> 00:14:16,240
Speaker 1: researchers that we're going to talk about in a minute.

258
00:14:17,160 --> 00:14:19,840
Speaker 1: But there's a variation on that script we just read

259
00:14:20,400 --> 00:14:23,760
Speaker 1: because in the video that didn't take place at a

260
00:14:23,840 --> 00:14:27,680
Speaker 1: text terminal. All the lines that both of us said

261
00:14:27,720 --> 00:14:31,320
Speaker 1: were spoken out loud by human beings. Now, how could

262
00:14:31,320 --> 00:14:33,440
Speaker 1: that be. Well, that's going to tie into the concept

263
00:14:33,480 --> 00:14:35,720
Speaker 1: of echo borgs, which we brought up at the beginning.

264
00:14:35,760 --> 00:14:38,200
Speaker 1: So to get into echo borgs, we need to talk

265
00:14:38,240 --> 00:14:43,560
Speaker 1: about a favorite figure in the weird history of psychological

266
00:14:43,600 --> 00:14:46,960
Speaker 1: research in the United States, and that's Stanley Milgram. Yes,

267
00:14:47,080 --> 00:14:54,200
Speaker 1: Yale University psychologist Stanley Milgram, best known for his controversial

268
00:14:54,240 --> 00:14:59,200
Speaker 1: obedience experiments actually a series of social psychology experiments about

269
00:14:59,280 --> 00:15:02,920
Speaker 1: nineteen in all, all conducted by Milgram in the nineteen sixties. Yeah,

270
00:15:02,960 --> 00:15:06,400
Speaker 1: you probably have heard about these, if you're familiar with Milgram,

271
00:15:06,400 --> 00:15:09,840
Speaker 1: it's probably from the they're often just called the Milgram experiments,

272
00:15:09,840 --> 00:15:12,880
Speaker 1: and they sort of give him a bad name because

273
00:15:12,960 --> 00:15:15,560
Speaker 1: they were they were kind of nasty. Yeah, they're they're

274
00:15:15,560 --> 00:15:17,280
Speaker 1: generally anytime you see a list of like, you know,

275
00:15:17,360 --> 00:15:22,520
Speaker 1: top ten scariest or weirdest or most evil psychological experiments,

276
00:15:22,520 --> 00:15:24,800
Speaker 1: they tend they tend to throw this one in though, uh,

277
00:15:24,880 --> 00:15:26,960
Speaker 1: you know, it's it's really more troubling and what it

278
00:15:27,000 --> 00:15:31,160
Speaker 1: reveals about about human human nature. So what was the deal. Well,

279
00:15:32,280 --> 00:15:33,800
Speaker 1: it's important to note that the first one of these

280
00:15:33,840 --> 00:15:36,640
Speaker 1: took place in nineteen sixty one, just three months after

281
00:15:36,680 --> 00:15:39,600
Speaker 1: the start of the trial of German Nazi war criminal

282
00:15:39,720 --> 00:15:43,480
Speaker 1: Adolf Eichmann in Jerusalem, and so Milgram wanted to see

283
00:15:43,480 --> 00:15:45,520
Speaker 1: just how far we'd go in the name of a

284
00:15:45,560 --> 00:15:47,720
Speaker 1: band an authority figure of course, of course, because the

285
00:15:47,720 --> 00:15:50,360
Speaker 1: whole argument is are that these bad people were they

286
00:15:50,360 --> 00:15:53,440
Speaker 1: just simply following orders? Right. It was the idea that

287
00:15:53,640 --> 00:15:56,680
Speaker 1: the Germans are especially evil. Where the were the people

288
00:15:56,720 --> 00:16:01,200
Speaker 1: who became the you know, guards at Auschwitz just from birth,

289
00:16:01,440 --> 00:16:04,560
Speaker 1: truly evil people who were susceptible to that kind of behavior,

290
00:16:04,640 --> 00:16:07,920
Speaker 1: or would we behave the same way in the same circumstances. So, Yeah,

291
00:16:07,920 --> 00:16:10,120
Speaker 1: the experiment revolved around, you know, an individual in a

292
00:16:10,240 --> 00:16:13,520
Speaker 1: room and you hear the sounds of someone being shocked

293
00:16:13,520 --> 00:16:16,280
Speaker 1: in the next room whenever that individual pushes a button,

294
00:16:16,280 --> 00:16:18,880
Speaker 1: pulls a lever, or whatever on the command of an

295
00:16:18,880 --> 00:16:22,360
Speaker 1: authority figure. And so the question is how far will

296
00:16:22,400 --> 00:16:25,280
Speaker 1: you go to when will you stop shocking? Would you

297
00:16:25,320 --> 00:16:27,960
Speaker 1: ever stop shocking that individual in the next room if

298
00:16:28,000 --> 00:16:29,760
Speaker 1: an authority figure is telling you to do it and

299
00:16:29,880 --> 00:16:32,720
Speaker 1: telling you that it's okay. Yeah, And what Milgram claimed

300
00:16:32,800 --> 00:16:35,560
Speaker 1: to find through his experiments is, yeah, even you know,

301
00:16:35,680 --> 00:16:39,240
Speaker 1: your regular people, your next door neighbor Americans, if they've

302
00:16:39,240 --> 00:16:41,800
Speaker 1: got somebody in a white lab coat who's supposedly in

303
00:16:41,880 --> 00:16:45,000
Speaker 1: charge of the experiment saying please continue shocking them, they've

304
00:16:45,040 --> 00:16:48,720
Speaker 1: agreed to this in advance. Lots of people will continue

305
00:16:48,760 --> 00:16:52,520
Speaker 1: shocking even after the supposed victim of the shocking. Now

306
00:16:52,560 --> 00:16:54,640
Speaker 1: we should say that in this experiment, nobody was acting.

307
00:16:54,640 --> 00:16:57,360
Speaker 1: Nobody actually electrocuted in the next room. Yeah, there were

308
00:16:57,400 --> 00:17:01,640
Speaker 1: actors pretending to be in in immense pain from these

309
00:17:01,680 --> 00:17:05,760
Speaker 1: shocks that lots of people in the experiment would supposedly

310
00:17:05,800 --> 00:17:08,119
Speaker 1: continue shocking them. Yeah. And if you want to hear

311
00:17:08,160 --> 00:17:11,280
Speaker 1: more about that, that series of experiments and what some

312
00:17:11,320 --> 00:17:14,240
Speaker 1: of the ramifications of it, uh, stuff to bow your mind.

313
00:17:14,280 --> 00:17:16,680
Speaker 1: Did an episode earlier in the year titled The Power

314
00:17:16,720 --> 00:17:18,879
Speaker 1: of Polite, and I'll make sure to link to that

315
00:17:18,920 --> 00:17:23,040
Speaker 1: on the landing page for this episode. But Stanley Milgram

316
00:17:23,080 --> 00:17:26,879
Speaker 1: also had some other experiments going on, right, He wasn't

317
00:17:27,040 --> 00:17:30,439
Speaker 1: just doing the shocking people and Nazi experiments. He was

318
00:17:30,440 --> 00:17:33,719
Speaker 1: into other ways to make us feel a bit troubled

319
00:17:33,760 --> 00:17:37,160
Speaker 1: about our humanity. Where they all creepy? Did he specialize

320
00:17:37,240 --> 00:17:39,560
Speaker 1: only in creepy science? And I think he had some

321
00:17:39,680 --> 00:17:41,880
Speaker 1: less creepy ones, you know, I mean, you know, they

322
00:17:41,920 --> 00:17:44,959
Speaker 1: involve how we I think most of his work revolved

323
00:17:44,960 --> 00:17:47,360
Speaker 1: around how we view ourselves and how we vie your bodies,

324
00:17:47,400 --> 00:17:50,760
Speaker 1: et cetera. By texts of puppies and lollipops on our psyche. Yeah,

325
00:17:50,800 --> 00:17:54,359
Speaker 1: but you know, not everything was necessarily um you know,

326
00:17:54,400 --> 00:17:57,439
Speaker 1: people in the next room dying. Sure, but we referred

327
00:17:57,440 --> 00:17:59,280
Speaker 1: to a term at the beginning of this episode, which

328
00:17:59,320 --> 00:18:03,000
Speaker 1: is serra woid. And this also comes from Stanley Milgram,

329
00:18:03,000 --> 00:18:07,080
Speaker 1: I believe from unpublished results of some experiments he conducted,

330
00:18:07,160 --> 00:18:09,359
Speaker 1: right right, He he never published any of these. They

331
00:18:09,440 --> 00:18:12,119
Speaker 1: ended up, you know, putting some work into it, but

332
00:18:12,119 --> 00:18:14,600
Speaker 1: then going off in a different direction with his research. Yeah. So,

333
00:18:14,640 --> 00:18:17,240
Speaker 1: as we said, the term serenoid comes from Syrano to

334
00:18:17,320 --> 00:18:20,280
Speaker 1: bears your act. But what was the deal with Milgram's experiments?

335
00:18:20,440 --> 00:18:22,800
Speaker 1: So Milgram essentially wanted to say, hey, if you're that

336
00:18:23,080 --> 00:18:27,200
Speaker 1: the woman that is being wooed bye bye bye by

337
00:18:27,600 --> 00:18:31,600
Speaker 1: Syrano's meat puppet's handsome. It's handsome young man. I think

338
00:18:31,600 --> 00:18:34,720
Speaker 1: her name was Roxanne x. Roxane's on the balcony being

339
00:18:34,800 --> 00:18:38,920
Speaker 1: wooed by a handsome meat puppet that's being fedlines by Syrano. Yeah,

340
00:18:38,920 --> 00:18:41,720
Speaker 1: if you're Roxanne, would you be able to detect something

341
00:18:41,840 --> 00:18:44,359
Speaker 1: was weird? Would you would you encounter this young man

342
00:18:44,440 --> 00:18:46,440
Speaker 1: and say, hmm, he you know, he seems a little

343
00:18:46,760 --> 00:18:49,560
Speaker 1: more clever than than he should be, or there's a

344
00:18:49,640 --> 00:18:51,760
Speaker 1: delay in in what he's telling me, you know, would

345
00:18:51,760 --> 00:18:53,520
Speaker 1: there be something that would tip you tip you off

346
00:18:53,560 --> 00:18:57,159
Speaker 1: to the deception you know, with modern technology, I'd imagine

347
00:18:57,160 --> 00:18:59,760
Speaker 1: you can carry out that experiment pretty easily. Yeah, I

348
00:18:59,800 --> 00:19:02,600
Speaker 1: mean even at the time, the technology was good enough.

349
00:19:03,280 --> 00:19:06,840
Speaker 1: So in these unpublished experiments, he had a source speak

350
00:19:06,880 --> 00:19:11,200
Speaker 1: into a microphone and a shadower listened through a hidden earpiece.

351
00:19:11,800 --> 00:19:14,399
Speaker 1: Then he or she would repeat whatever they heard, you know,

352
00:19:14,520 --> 00:19:18,960
Speaker 1: basic you know, arrested development, surrogate kind of situation. But

353
00:19:19,200 --> 00:19:21,720
Speaker 1: here's the thing. You found that with practice, speech shadowing

354
00:19:21,800 --> 00:19:24,639
Speaker 1: becomes easier and easier, Like we really take to it

355
00:19:24,800 --> 00:19:26,960
Speaker 1: rather naturally. They say, it's not that hard at all.

356
00:19:27,000 --> 00:19:29,080
Speaker 1: I read in one of the sources we used for

357
00:19:29,200 --> 00:19:31,879
Speaker 1: this episode that sometimes it only takes a few milliseconds

358
00:19:32,640 --> 00:19:35,000
Speaker 1: of delay between the speech being fed in through the

359
00:19:35,080 --> 00:19:38,520
Speaker 1: earpiece and saying it. We're ready, willing and able to

360
00:19:38,800 --> 00:19:42,280
Speaker 1: not think for ourselves, and so yeah, he observed that

361
00:19:43,160 --> 00:19:45,320
Speaker 1: he set all this up, put people in motion with

362
00:19:45,520 --> 00:19:47,960
Speaker 1: these with a source and a shadower, and then a

363
00:19:48,080 --> 00:19:51,440
Speaker 1: test subject having to interact with this individual or is

364
00:19:51,520 --> 00:19:55,560
Speaker 1: it his individuals um and then give feedback about what

365
00:19:55,640 --> 00:19:58,159
Speaker 1: they thought. And he observed that people couldn't tell the

366
00:19:58,240 --> 00:20:01,960
Speaker 1: difference between a Syran wid again that person. Uh that

367
00:20:02,280 --> 00:20:05,800
Speaker 1: that that Shadower, who was informed by another source just

368
00:20:05,920 --> 00:20:09,680
Speaker 1: repeating words that are entering his or her ears, could

369
00:20:09,720 --> 00:20:12,160
Speaker 1: not tell the difference between syenoid and a normal human

370
00:20:12,240 --> 00:20:15,720
Speaker 1: being during interactions. Yeah, so that's what Milgram reported. And

371
00:20:16,280 --> 00:20:19,880
Speaker 1: I don't know, that's surprising and creepy. Yeah, I would

372
00:20:19,960 --> 00:20:21,920
Speaker 1: think that I would easily be able to tell the

373
00:20:21,920 --> 00:20:25,120
Speaker 1: difference between a person speaking of their own free will

374
00:20:25,240 --> 00:20:28,680
Speaker 1: and somebody who's just being fed lines in a near piece. Yeah.

375
00:20:28,720 --> 00:20:31,560
Speaker 1: And one experiment Milgram's sourced in other words, fed you know,

376
00:20:31,720 --> 00:20:34,400
Speaker 1: lines to a pair of eleven and twelve year olds.

377
00:20:34,560 --> 00:20:36,520
Speaker 1: So they the eleven and twelve year olds where the

378
00:20:36,560 --> 00:20:40,080
Speaker 1: shadows and Milgram himself was speaking correct. Uh Yeah, the

379
00:20:40,240 --> 00:20:43,040
Speaker 1: the children were the serenoids, if you will, and a

380
00:20:43,080 --> 00:20:46,320
Speaker 1: group of observing teachers never suspected that they were chatting

381
00:20:46,359 --> 00:20:49,480
Speaker 1: with anything other than a normal, though very bright child.

382
00:20:50,400 --> 00:20:53,320
Speaker 1: Uh and uh. And and it is experiments he found

383
00:20:53,359 --> 00:20:56,240
Speaker 1: that most people could have a lengthy twenty minute conversation

384
00:20:56,320 --> 00:21:01,560
Speaker 1: with a serenoid without noticing anything amiss. Wow. Yeah, but

385
00:21:01,760 --> 00:21:03,719
Speaker 1: I mean again, that's that's hard to believe. I mean,

386
00:21:03,760 --> 00:21:05,560
Speaker 1: it just seems like you'd be able to notice that

387
00:21:05,840 --> 00:21:08,159
Speaker 1: right off the bat. But well, but just think how

388
00:21:08,240 --> 00:21:12,600
Speaker 1: easily we we buy into an actor's presentation. I think

389
00:21:12,800 --> 00:21:15,000
Speaker 1: and think of all the interactions we have in the

390
00:21:15,080 --> 00:21:19,320
Speaker 1: course of of our lives where the individual we're talking

391
00:21:19,400 --> 00:21:24,040
Speaker 1: to is maybe not completely yeah, genuine, maybe that you know,

392
00:21:24,240 --> 00:21:27,000
Speaker 1: maybe you're interacting with with someone that's doing customer service

393
00:21:27,640 --> 00:21:31,040
Speaker 1: or they're trying to sell you something or can you, etcetera.

394
00:21:31,240 --> 00:21:33,080
Speaker 1: It's like, it's not always going to come off is oh,

395
00:21:33,160 --> 00:21:36,800
Speaker 1: this person is completely fake. It's gonna It's just an

396
00:21:36,920 --> 00:21:40,200
Speaker 1: aspect of human interaction. Well, it's funny you mentioned actors

397
00:21:40,280 --> 00:21:42,600
Speaker 1: because I had this thought when I was reading this research.

398
00:21:42,680 --> 00:21:46,280
Speaker 1: In a strange kind of way, all actors in movies

399
00:21:46,400 --> 00:21:50,200
Speaker 1: are seronoids. Like you have the writer coming up with

400
00:21:50,440 --> 00:21:53,720
Speaker 1: lines for many types of characters that aren't anything like

401
00:21:54,000 --> 00:21:56,760
Speaker 1: them outwardly at all. Like you can have, you know,

402
00:21:57,480 --> 00:22:00,399
Speaker 1: you know, a sixty five year old female writer writing

403
00:22:00,520 --> 00:22:02,880
Speaker 1: lines for a ten year old boy in a play,

404
00:22:03,600 --> 00:22:08,240
Speaker 1: and that boy effectively is a distanced serenoid. And should

405
00:22:08,280 --> 00:22:10,719
Speaker 1: we be able to tell the difference, like sometimes you can.

406
00:22:10,840 --> 00:22:12,639
Speaker 1: Sometimes you watch a movie, you know, and you're like,

407
00:22:13,080 --> 00:22:15,960
Speaker 1: kids wouldn't say that, that's not how kids talk, but

408
00:22:16,080 --> 00:22:18,080
Speaker 1: other times you buy it. Yeah. I think one of

409
00:22:18,080 --> 00:22:20,679
Speaker 1: the things to keep in mind about serenoids and ultimately

410
00:22:20,680 --> 00:22:24,200
Speaker 1: about echo borgs this we've moved forward is that we're

411
00:22:24,520 --> 00:22:28,000
Speaker 1: dealing with a hybrid personality. So there's the there, there

412
00:22:28,000 --> 00:22:30,520
Speaker 1: are the words and the personality of the individual who

413
00:22:30,640 --> 00:22:33,879
Speaker 1: is informing, and then the words of the shadower or

414
00:22:33,960 --> 00:22:37,760
Speaker 1: the serenoid. That individual is bringing their own delivery, their

415
00:22:37,760 --> 00:22:40,840
Speaker 1: own personality to it. Sure, because your personality, as you

416
00:22:40,960 --> 00:22:44,040
Speaker 1: present it outwardly, is way more than the words you say. Yeah, obviously,

417
00:22:44,119 --> 00:22:47,320
Speaker 1: it's your body language, it's your expressions, it's you know,

418
00:22:47,400 --> 00:22:49,320
Speaker 1: the way you carry yourself. I mean that that's all

419
00:22:49,520 --> 00:22:51,920
Speaker 1: part of the message you present. Like we can all

420
00:22:52,000 --> 00:22:56,520
Speaker 1: think of of movies or TV shows where there's a

421
00:22:56,640 --> 00:23:00,959
Speaker 1: particularly gifted actor who's able to bring lack lester lines

422
00:23:01,359 --> 00:23:03,400
Speaker 1: to life in a way that a less gifted actor

423
00:23:03,440 --> 00:23:05,639
Speaker 1: who just would not be able to achieve. I always

424
00:23:05,680 --> 00:23:08,800
Speaker 1: think of ral Julia in Overdrawn at the Memory Bank,

425
00:23:09,160 --> 00:23:11,840
Speaker 1: who you know, some people, you know, it's easy to

426
00:23:11,920 --> 00:23:13,840
Speaker 1: have a lot of fun with his performance. It's you know,

427
00:23:13,880 --> 00:23:17,399
Speaker 1: it's kind of cheesy and outlandish. But that man in

428
00:23:17,480 --> 00:23:20,360
Speaker 1: that movie which is like a you know, a low

429
00:23:20,440 --> 00:23:23,639
Speaker 1: budget PDS adaptation of a of a science fiction story.

430
00:23:24,560 --> 00:23:27,200
Speaker 1: He brings so much life to every line, so many

431
00:23:27,280 --> 00:23:30,679
Speaker 1: lines in that in that film, if they were delivered

432
00:23:30,680 --> 00:23:32,920
Speaker 1: by a lesser actor, would have just fallen flat. But

433
00:23:33,040 --> 00:23:36,200
Speaker 1: he makes even the most pointless line, uh, just really

434
00:23:36,359 --> 00:23:39,200
Speaker 1: land I agree, and I think this is a common

435
00:23:39,359 --> 00:23:43,000
Speaker 1: feature actually an older actors. I see this way more

436
00:23:43,160 --> 00:23:45,560
Speaker 1: commonly with actors who have been in the business for

437
00:23:45,640 --> 00:23:48,520
Speaker 1: a long time. I think of Star Wars Episode two,

438
00:23:49,000 --> 00:23:53,360
Speaker 1: which my opinion is that that is a horrible, horrible movie. Huh.

439
00:23:53,640 --> 00:23:56,960
Speaker 1: But when Christopher Lee shows up and starts talking, the

440
00:23:57,040 --> 00:23:59,119
Speaker 1: writing I think is just as bad as it's been

441
00:23:59,200 --> 00:24:03,359
Speaker 1: the entire time. But suddenly I'm okay, I'm listening Christopher Lee.

442
00:24:03,400 --> 00:24:07,600
Speaker 1: I'm buying it. He is really selling this horrible dialogue. Yeah, yeah, exactly.

443
00:24:07,680 --> 00:24:10,640
Speaker 1: So in that respect, Raw Julia, Christopher Lee, they would

444
00:24:10,680 --> 00:24:14,159
Speaker 1: make wonderful serenoids. And keep in the mind they might

445
00:24:14,240 --> 00:24:18,840
Speaker 1: have made excellent echo borgs. Okay, So Robert and think

446
00:24:19,040 --> 00:24:23,119
Speaker 1: we were talking about echo borgs being the natural extension

447
00:24:23,280 --> 00:24:26,200
Speaker 1: of serenoids. So the serenoid is a person being fed

448
00:24:26,320 --> 00:24:29,480
Speaker 1: lines by another person. The echo borg, as we talked

449
00:24:29,480 --> 00:24:31,200
Speaker 1: about at the beginning of the episode, would be a

450
00:24:31,320 --> 00:24:35,760
Speaker 1: person being fed lines by a computer program. Yes, I

451
00:24:35,880 --> 00:24:38,440
Speaker 1: want to know, are there any studies where people have

452
00:24:38,600 --> 00:24:42,600
Speaker 1: looked into this phenomena and if so, does a person

453
00:24:42,800 --> 00:24:46,360
Speaker 1: delivering the lines from a chat bot make the chat

454
00:24:46,440 --> 00:24:50,440
Speaker 1: bought any more convincing in the Turing test? Well, indeed,

455
00:24:50,720 --> 00:24:53,200
Speaker 1: that that's exactly what we're gonna look at in this

456
00:24:53,640 --> 00:24:57,919
Speaker 1: half of the podcast, particularly the work of two individuals

457
00:24:58,240 --> 00:25:03,600
Speaker 1: British social psychologists Kevin Cordy and Alex Gilefski. And Uh,

458
00:25:04,400 --> 00:25:06,960
Speaker 1: basically we have two studies that we're gonna we're gonna

459
00:25:06,960 --> 00:25:10,359
Speaker 1: analyze here, two thousand fourteen study where they essentially just

460
00:25:10,560 --> 00:25:13,520
Speaker 1: recreate some of Milgram's work and uh and look at

461
00:25:13,600 --> 00:25:16,760
Speaker 1: the at the serenoid, and then a two thousand fifteen

462
00:25:16,800 --> 00:25:20,760
Speaker 1: study that just came out, um where they take the serenoid,

463
00:25:20,960 --> 00:25:25,120
Speaker 1: apply it to chatter bots and give us the echo borg. Yeah.

464
00:25:25,200 --> 00:25:28,800
Speaker 1: So the script we read earlier from the conversation between

465
00:25:28,880 --> 00:25:32,400
Speaker 1: the eco borg and the regular interactor that came from

466
00:25:32,680 --> 00:25:35,280
Speaker 1: London School of Economics Research. It was a YouTube video

467
00:25:35,400 --> 00:25:37,560
Speaker 1: that they had put up, So that was their echo

468
00:25:37,640 --> 00:25:40,280
Speaker 1: Borg saying that it liked fruit Trees and that the

469
00:25:40,400 --> 00:25:42,720
Speaker 1: Imitation Game was its favorite movie, you know, And I

470
00:25:42,760 --> 00:25:44,280
Speaker 1: do want to throw in like that was the dead

471
00:25:44,320 --> 00:25:47,080
Speaker 1: giveaway that was a robot, because that that movie is

472
00:25:47,160 --> 00:25:50,840
Speaker 1: nobody's favorite movie, right, And I think Rose, the chatbot

473
00:25:50,920 --> 00:25:53,840
Speaker 1: in that example, is sort of programmed to be playful,

474
00:25:54,080 --> 00:25:57,120
Speaker 1: like not necessarily to be entirely convincing as a human,

475
00:25:57,160 --> 00:26:00,720
Speaker 1: because Rose gives other playful answers. Also in that same video,

476
00:26:00,760 --> 00:26:03,960
Speaker 1: there's a part where the human interactor says to the

477
00:26:04,040 --> 00:26:06,960
Speaker 1: human eco borg, do you like food? And the human

478
00:26:07,040 --> 00:26:12,920
Speaker 1: eco borg delivers the line yummy electricity. Yeah. I don't

479
00:26:12,960 --> 00:26:16,360
Speaker 1: know if the comedy needs to be a primary programming

480
00:26:16,440 --> 00:26:18,280
Speaker 1: of function. I mean, you know, humor is an important

481
00:26:18,320 --> 00:26:21,360
Speaker 1: part of any human interaction. Sure, I mean that's good

482
00:26:21,480 --> 00:26:24,960
Speaker 1: humor for a chat boton, but that's obviously the chatbot

483
00:26:24,960 --> 00:26:27,280
Speaker 1: who's not trying all that hard to hide it. Yeah,

484
00:26:27,640 --> 00:26:31,160
Speaker 1: being a little little two cooi. Alright, So let's look

485
00:26:31,160 --> 00:26:35,200
Speaker 1: at this the work of Accordion glevskie Um. In two

486
00:26:35,200 --> 00:26:37,440
Speaker 1: thousand fourteen, they poblished this paper in the Journal of

487
00:26:37,480 --> 00:26:41,320
Speaker 1: Social Psychology, and they essentially set out to to replicate

488
00:26:41,800 --> 00:26:44,639
Speaker 1: Milgram's work. Okay, So the basic set up here is

489
00:26:44,680 --> 00:26:48,680
Speaker 1: they they had a shadower set in an interrogation room.

490
00:26:48,760 --> 00:26:52,119
Speaker 1: So that's the seranoid type figure, and uh, they were

491
00:26:52,160 --> 00:26:55,520
Speaker 1: engaged in a conversation with an interactant. This is an

492
00:26:55,600 --> 00:27:01,480
Speaker 1: unsuspecting volunteer. Meanwhile, the source, that's the individual feeding the shadow,

493
00:27:01,480 --> 00:27:06,080
Speaker 1: we're feeding the serenoid. Um. They're in another room and

494
00:27:06,080 --> 00:27:09,159
Speaker 1: they're observing everything through video and audio links, and they

495
00:27:09,240 --> 00:27:12,160
Speaker 1: tell the shadow or exactly what they need to say

496
00:27:12,240 --> 00:27:16,399
Speaker 1: through a uh, you know, discreet FM radio transmitter set up. Okay,

497
00:27:17,160 --> 00:27:19,880
Speaker 1: So they have twenty volunteers that are engaged in ten

498
00:27:19,920 --> 00:27:23,560
Speaker 1: minute conversations with the serenoid and twenty additional volunteers that

499
00:27:23,600 --> 00:27:26,040
Speaker 1: are tested in a control setting. Okay. So the control

500
00:27:26,080 --> 00:27:29,280
Speaker 1: setting is just a regular person talking, not being fedlines.

501
00:27:29,920 --> 00:27:32,040
Speaker 1: And so this backed up exactly what Milgram had told

502
00:27:32,119 --> 00:27:35,280
Speaker 1: us decades earlier that when one human parents and other

503
00:27:35,359 --> 00:27:38,160
Speaker 1: humans words, we just totally buy it. Okay. So even

504
00:27:38,160 --> 00:27:40,960
Speaker 1: if you were skeptical of Millgram, and who knows, you

505
00:27:41,040 --> 00:27:43,520
Speaker 1: might have had reason to be, we've got to We've

506
00:27:43,560 --> 00:27:46,399
Speaker 1: got a new study backing that up now and the

507
00:27:46,680 --> 00:27:49,960
Speaker 1: implications of this are kind of creepy. It's that unless

508
00:27:50,000 --> 00:27:52,600
Speaker 1: you check in people's ears to make sure they don't

509
00:27:52,640 --> 00:27:55,440
Speaker 1: have an earpiece in there, anybody in your life could

510
00:27:55,480 --> 00:27:59,320
Speaker 1: be feeding you lines from somebody sitting in a van

511
00:27:59,440 --> 00:28:02,000
Speaker 1: outside the building and you might not know it. Yeah,

512
00:28:02,000 --> 00:28:05,640
Speaker 1: I mean, look at the movie The Exorcist, right, That's

513
00:28:05,680 --> 00:28:07,720
Speaker 1: why demons, when they possess people, they have to spit,

514
00:28:08,200 --> 00:28:10,399
Speaker 1: you know, green vomit across the room and spin their

515
00:28:10,440 --> 00:28:12,359
Speaker 1: heads around, because otherwise we're just not going to buy

516
00:28:12,400 --> 00:28:15,440
Speaker 1: into the fact that an individual is not uh functioning

517
00:28:15,480 --> 00:28:17,800
Speaker 1: of their own free will, right or you know, you've

518
00:28:17,840 --> 00:28:20,359
Speaker 1: got the problem of Okay, so the priest tries to

519
00:28:20,480 --> 00:28:22,320
Speaker 1: go to the authorities and say, I need to conduct

520
00:28:22,359 --> 00:28:25,159
Speaker 1: an exorcism. My evidence is this person was speaking in

521
00:28:25,240 --> 00:28:27,280
Speaker 1: a language they've never learned. It's like, well, how do

522
00:28:27,359 --> 00:28:29,840
Speaker 1: you know they're not being fed that language from somebody

523
00:28:29,880 --> 00:28:31,960
Speaker 1: in a van outside the building through a near piece.

524
00:28:32,280 --> 00:28:34,919
Speaker 1: You've got to have the pup well, you know, speaking

525
00:28:34,960 --> 00:28:37,840
Speaker 1: of possessed children. A second part of the same experiment,

526
00:28:37,880 --> 00:28:41,040
Speaker 1: according and collepski Um, use a twelve year old male

527
00:28:41,120 --> 00:28:43,400
Speaker 1: actor to serve as a shadow er while like thirty

528
00:28:43,480 --> 00:28:47,680
Speaker 1: seven year old male social social psychologist acts as the source,

529
00:28:48,320 --> 00:28:51,480
Speaker 1: and again no one suspected the illusion, though some thought

530
00:28:51,560 --> 00:28:54,520
Speaker 1: that children might have been coached prior, so there was

531
00:28:54,560 --> 00:28:57,440
Speaker 1: a little you know, trepidation there where they think, uh,

532
00:28:57,920 --> 00:29:01,680
Speaker 1: I think somebody was preparing the kid at this conversation,

533
00:29:01,800 --> 00:29:04,800
Speaker 1: so they picked up that something was off, but they

534
00:29:04,840 --> 00:29:07,520
Speaker 1: didn't pick up that the kid was being fedlines. Yeah,

535
00:29:07,640 --> 00:29:09,720
Speaker 1: and again that just kind of drives with with so

536
00:29:09,840 --> 00:29:12,600
Speaker 1: many interactions, most of us having our lives where you

537
00:29:12,840 --> 00:29:14,760
Speaker 1: you talk to somebody and you might think that person

538
00:29:14,800 --> 00:29:18,080
Speaker 1: seemed kind of fake, but you're not thinking that person

539
00:29:18,560 --> 00:29:20,640
Speaker 1: was serving as a meat puppet for an AI or

540
00:29:20,720 --> 00:29:23,120
Speaker 1: some dude in a van. You just think they're they're

541
00:29:23,120 --> 00:29:24,920
Speaker 1: a little phony there. You know, they have some sort

542
00:29:24,920 --> 00:29:26,800
Speaker 1: of an agenda, they're kind of playing the room, or

543
00:29:26,840 --> 00:29:29,760
Speaker 1: they're they're in just full on customer service mode, etcetera.

544
00:29:30,000 --> 00:29:32,960
Speaker 1: Or they might just be experiencing some anxiety and awkwardness.

545
00:29:33,000 --> 00:29:34,440
Speaker 1: I mean, I think you want to give people the

546
00:29:34,480 --> 00:29:37,040
Speaker 1: benefit of the doubt, and that's I think coming through

547
00:29:37,160 --> 00:29:39,680
Speaker 1: in some of these experiments. If somebody acts a little

548
00:29:39,760 --> 00:29:42,240
Speaker 1: awkward or weird, I mean, you don't want to just

549
00:29:42,600 --> 00:29:45,240
Speaker 1: judge them and say that person is a robot. You know,

550
00:29:45,320 --> 00:29:48,360
Speaker 1: they're being fedlines or something like that. You know, we

551
00:29:48,480 --> 00:29:50,880
Speaker 1: understand that some people get in moods, they have trouble,

552
00:29:50,960 --> 00:29:53,640
Speaker 1: sometimes they feel awkward. We've all felt like this, and

553
00:29:53,680 --> 00:29:57,000
Speaker 1: you want to be accommodating of other people's awkwardness whenever possible. Yeah,

554
00:29:57,040 --> 00:29:59,959
Speaker 1: I mean, awkwardness is in his own in its own right.

555
00:30:00,000 --> 00:30:03,080
Speaker 1: It is an essential part of human interaction. Um. And

556
00:30:03,200 --> 00:30:05,760
Speaker 1: if I an AI or a computer wanted a part

557
00:30:05,840 --> 00:30:08,520
Speaker 1: of legitimate human interaction, they need to sign up for

558
00:30:08,640 --> 00:30:12,640
Speaker 1: some serious awkwardness from time to time. So this initial

559
00:30:12,680 --> 00:30:15,280
Speaker 1: studies pretty much backs off everything Milgram told us. So

560
00:30:15,400 --> 00:30:19,280
Speaker 1: from that point, um Cordy and Gillepskie moved on to

561
00:30:19,880 --> 00:30:22,760
Speaker 1: artificial intelligence, to chatter bots and the creation of the

562
00:30:22,840 --> 00:30:25,640
Speaker 1: eco borg. And again, an echo borg is essentially the

563
00:30:25,720 --> 00:30:29,600
Speaker 1: same as a serenoid, except the shadow er is speaking

564
00:30:29,800 --> 00:30:32,080
Speaker 1: the words of a chatter box, so the source is

565
00:30:32,160 --> 00:30:35,280
Speaker 1: not human, it's a computer program, right, And again this

566
00:30:35,440 --> 00:30:37,280
Speaker 1: is what this is. It's important to keep in mind

567
00:30:37,360 --> 00:30:41,800
Speaker 1: that a serenoid is a hybrid personality, the individual behind

568
00:30:42,040 --> 00:30:44,640
Speaker 1: the speaker and then the personality of the speaker coming together.

569
00:30:44,920 --> 00:30:47,560
Speaker 1: So in eco borg, the borg aspect of that referring

570
00:30:47,600 --> 00:30:51,240
Speaker 1: to cyborg. It really is a new type of cyborg

571
00:30:51,320 --> 00:30:54,520
Speaker 1: that we're playing with here, uh, a machine and a

572
00:30:54,680 --> 00:31:00,479
Speaker 1: human as one. And the question presents itself, would we fooled?

573
00:31:00,600 --> 00:31:04,200
Speaker 1: Will we observe an echo borg in action? And and

574
00:31:04,480 --> 00:31:08,560
Speaker 1: suspect that it's a machine will behind the human face.

575
00:31:08,840 --> 00:31:11,960
Speaker 1: I'm gonna say that even before I read the results

576
00:31:12,000 --> 00:31:15,160
Speaker 1: of the study, when I knew what the subject was,

577
00:31:15,320 --> 00:31:17,440
Speaker 1: I thought, you know, I bet it depends heavily on

578
00:31:17,600 --> 00:31:22,160
Speaker 1: the sort of pre priming of the study participants, Like

579
00:31:22,360 --> 00:31:25,560
Speaker 1: are they expecting to be interacting with somebody who might

580
00:31:25,760 --> 00:31:29,600
Speaker 1: be an echo borg? Or are they or they just

581
00:31:29,680 --> 00:31:31,960
Speaker 1: going in blind and interacting with somebody who might be

582
00:31:32,000 --> 00:31:35,200
Speaker 1: a little weird. It's true, I mean, when you're presented

583
00:31:35,240 --> 00:31:37,400
Speaker 1: with the if you're presented with the possibility, hey, you

584
00:31:37,480 --> 00:31:41,320
Speaker 1: might be talking to a robot here, chance that's gonna

585
00:31:41,800 --> 00:31:44,160
Speaker 1: it's gonna it's gonna alter your perceptions of bit. Yeah.

586
00:31:44,200 --> 00:31:46,840
Speaker 1: So what did they find? First of all, this is

587
00:31:46,840 --> 00:31:49,320
Speaker 1: how the study rolled out. Forty one participants in the

588
00:31:49,400 --> 00:31:52,640
Speaker 1: experiment were shown a text messaging client with two windows

589
00:31:52,960 --> 00:31:55,840
Speaker 1: until they be interacting with two human shadowers at the

590
00:31:55,880 --> 00:31:58,960
Speaker 1: same time. One shadower is sourced by another human be

591
00:31:59,040 --> 00:32:03,480
Speaker 1: either source patch atter bot, and each interrogator asked questions

592
00:32:03,680 --> 00:32:06,960
Speaker 1: via text, and then the participants were told they'd be

593
00:32:07,040 --> 00:32:10,200
Speaker 1: asked which of the two agents was a real human. Now,

594
00:32:10,280 --> 00:32:13,360
Speaker 1: the researchers also wanted to compare human and text responses,

595
00:32:13,560 --> 00:32:15,720
Speaker 1: so they divided the group into two groups. Twenty of

596
00:32:15,760 --> 00:32:18,840
Speaker 1: the participants typed their half of the conversation and into

597
00:32:18,880 --> 00:32:22,240
Speaker 1: a text client, but got responses from the two echo

598
00:32:22,320 --> 00:32:24,920
Speaker 1: boards setting in the same room. The other twenty one

599
00:32:24,960 --> 00:32:27,720
Speaker 1: participants just had text chats the whole way through. So

600
00:32:27,800 --> 00:32:29,200
Speaker 1: this way they would have a you know, a base

601
00:32:29,280 --> 00:32:32,600
Speaker 1: comparison to the you know, the more established chatter bot

602
00:32:32,680 --> 00:32:35,840
Speaker 1: to rank test situation that we waked turtling into a terminal. Right,

603
00:32:36,720 --> 00:32:39,880
Speaker 1: So the results were that all the text only participants

604
00:32:39,960 --> 00:32:44,880
Speaker 1: correctly identified the real humans. So again dead giveaway based

605
00:32:44,920 --> 00:32:48,120
Speaker 1: on just the limitations of the chatter bot um, it's

606
00:32:48,160 --> 00:32:51,800
Speaker 1: pretty easy to tell that you're just talking to a program. Yeah,

607
00:32:51,800 --> 00:32:53,640
Speaker 1: if you if you have enough time to talk to them,

608
00:32:53,720 --> 00:32:56,640
Speaker 1: you can almost definitely trip them up. Yeah. Now, on

609
00:32:56,680 --> 00:32:58,960
Speaker 1: the other hand, in the other set up, only two

610
00:32:59,080 --> 00:33:02,920
Speaker 1: of the participants didn't identify the real humans, so in

611
00:33:03,000 --> 00:33:08,120
Speaker 1: the eco borg situations, they still weren't too terribly convincing, Right,

612
00:33:08,160 --> 00:33:10,560
Speaker 1: so they can tell the difference between a seranoid and

613
00:33:10,640 --> 00:33:13,800
Speaker 1: an eco borg. Everybody except two of them could, Right,

614
00:33:14,960 --> 00:33:18,320
Speaker 1: So it seems that human or not still rather easy

615
00:33:18,360 --> 00:33:21,520
Speaker 1: to identify the machine behind the flesh. You know, I

616
00:33:21,760 --> 00:33:25,560
Speaker 1: would chalk this up probably just to the poor quality

617
00:33:25,680 --> 00:33:30,560
Speaker 1: of today's best chat bots. Yeah, I mean, because even

618
00:33:30,800 --> 00:33:34,240
Speaker 1: even a raw Julia or or Christopher Lee, you know,

619
00:33:34,640 --> 00:33:37,800
Speaker 1: presented with the with the dialogue we were using earlier,

620
00:33:37,840 --> 00:33:39,480
Speaker 1: they're they're not They're just gonna come off as a

621
00:33:39,560 --> 00:33:42,040
Speaker 1: doll person at best, you know, Right, There's no way

622
00:33:42,120 --> 00:33:44,960
Speaker 1: you can sell the line I like fruit trees in

623
00:33:45,000 --> 00:33:48,400
Speaker 1: a way that makes people think you're you're being fedlines

624
00:33:48,440 --> 00:33:50,400
Speaker 1: by a human, no matter how good of an actor

625
00:33:50,520 --> 00:33:53,040
Speaker 1: you are. That's a dead giveaway that the machine does

626
00:33:53,160 --> 00:33:55,040
Speaker 1: not know what to do with what you just said. Yeah,

627
00:33:55,120 --> 00:33:58,080
Speaker 1: unless you're a toddler speaking that line, or or again,

628
00:33:58,120 --> 00:34:00,160
Speaker 1: if you're hiding it behind to say, some sort of

629
00:34:00,200 --> 00:34:04,920
Speaker 1: mental deficiency or or language gap um or some sort

630
00:34:04,960 --> 00:34:09,359
Speaker 1: of you know, character that's leveled over everything it's going

631
00:34:09,400 --> 00:34:12,320
Speaker 1: to be a dead giveaway. But the echo borgs certainly

632
00:34:12,400 --> 00:34:15,520
Speaker 1: made the machine seem more human and the text only group,

633
00:34:15,560 --> 00:34:18,960
Speaker 1: the chatbot received an average a humanity rating of two

634
00:34:19,080 --> 00:34:21,239
Speaker 1: point one four on a scale of one to ten.

635
00:34:21,719 --> 00:34:24,319
Speaker 1: In the echo board group, the average was a four

636
00:34:24,360 --> 00:34:29,080
Speaker 1: point oh five. So significant difference there. Okay, yeah, that's interesting.

637
00:34:29,239 --> 00:34:33,120
Speaker 1: So we can still tell that the Serenoid is more

638
00:34:33,280 --> 00:34:36,480
Speaker 1: human than an echo borg. But an echo borg is

639
00:34:36,600 --> 00:34:40,640
Speaker 1: more human at least in a qualitative evaluation than a

640
00:34:40,760 --> 00:34:43,600
Speaker 1: text only chat bot. Yeah, to lay it out, text

641
00:34:43,760 --> 00:34:47,200
Speaker 1: text only chat bought two point one four humanity. Uh,

642
00:34:47,280 --> 00:34:49,799
Speaker 1: the echo boorg four point oh five humanity, and then

643
00:34:49,880 --> 00:34:53,560
Speaker 1: the Serenoid had a humanity rating of eight point seven six,

644
00:34:53,680 --> 00:34:56,480
Speaker 1: eight point one five. So it sounds to me like

645
00:34:56,600 --> 00:34:59,960
Speaker 1: even a normal human interaction, like your score, your score

646
00:35:00,040 --> 00:35:03,600
Speaker 1: ring a nine, just straight up non augmented human interactions.

647
00:35:03,640 --> 00:35:07,360
Speaker 1: You're doing pretty good. Yeah, So it does prove that

648
00:35:07,520 --> 00:35:11,720
Speaker 1: there is a humanizing effect to the eco borg system.

649
00:35:12,440 --> 00:35:15,759
Speaker 1: It's just the limitations of existing chatter bots that they're

650
00:35:15,800 --> 00:35:17,799
Speaker 1: coming into play, right. It makes you think if there

651
00:35:17,840 --> 00:35:21,160
Speaker 1: were a really good chatbot out there, the eco borg

652
00:35:21,320 --> 00:35:24,360
Speaker 1: might be that finishing touch, that final push over the

653
00:35:24,480 --> 00:35:27,040
Speaker 1: edge of the cliff to sort of make the you know,

654
00:35:27,120 --> 00:35:29,560
Speaker 1: the perfect android. Yeah, kind of a bridge for the

655
00:35:29,680 --> 00:35:32,440
Speaker 1: uncanny valley, you know, like you just just all you

656
00:35:32,520 --> 00:35:34,840
Speaker 1: need is just a little acting on top of this

657
00:35:35,040 --> 00:35:38,160
Speaker 1: to really bring the script to life. Now. They they

658
00:35:38,200 --> 00:35:41,280
Speaker 1: also conducted a second experiment in this and they tested

659
00:35:41,320 --> 00:35:44,160
Speaker 1: to see if fifty eight new participants would guess whether

660
00:35:44,239 --> 00:35:46,560
Speaker 1: they were talking to an echo borg or a person

661
00:35:46,719 --> 00:35:50,000
Speaker 1: pretending to be an echo borg. So, you know, obviously

662
00:35:50,080 --> 00:35:52,080
Speaker 1: a person who's actually fed their lines by a chatter

663
00:35:52,160 --> 00:35:55,360
Speaker 1: bot or someone who's just playing around and pretending to

664
00:35:55,480 --> 00:35:57,520
Speaker 1: be speaking the words of a chatter bot. Right, so

665
00:35:57,600 --> 00:35:59,880
Speaker 1: this is a reverse touring test. Can you pretend to

666
00:36:00,040 --> 00:36:02,920
Speaker 1: be a machine? Right? But here's the catch. They're both

667
00:36:03,080 --> 00:36:06,239
Speaker 1: pure eco borgs. There, there's no there's no faco echo boorg.

668
00:36:06,320 --> 00:36:09,960
Speaker 1: But the but the individual being tested they think that

669
00:36:10,120 --> 00:36:12,520
Speaker 1: one of the two echo borgs is just a person pretending.

670
00:36:13,360 --> 00:36:16,680
Speaker 1: So seventy of the twenty eight participants who faced an

671
00:36:16,719 --> 00:36:19,800
Speaker 1: echo borg through I thought the person was just mimicking,

672
00:36:20,160 --> 00:36:24,080
Speaker 1: mimicking a chat bot. So they seemed fake but not inhuman.

673
00:36:24,200 --> 00:36:26,240
Speaker 1: So if there is you know, some reason to believe

674
00:36:26,320 --> 00:36:28,680
Speaker 1: that this is just a person being a little fake

675
00:36:28,760 --> 00:36:30,840
Speaker 1: and putting on a show versus being you know, a

676
00:36:30,880 --> 00:36:34,200
Speaker 1: complete echo borg. We're still going to give them the

677
00:36:34,239 --> 00:36:37,040
Speaker 1: benefit of a doubt. Interesting, now, I think we should

678
00:36:37,360 --> 00:36:39,319
Speaker 1: do our duty and hedge a little bit. Say, all

679
00:36:39,360 --> 00:36:42,040
Speaker 1: these sample sizes were pretty small, and it would be

680
00:36:42,120 --> 00:36:44,759
Speaker 1: interesting to see more research along these lines, like with

681
00:36:44,880 --> 00:36:48,880
Speaker 1: bigger sample sizes and trying to repeat these results. And

682
00:36:49,000 --> 00:36:52,440
Speaker 1: and and also to your point, is chatterbots improve, it's

683
00:36:52,440 --> 00:36:55,799
Speaker 1: going to be interesting to see echo borgs employed as

684
00:36:55,840 --> 00:36:57,359
Speaker 1: as a way to test them. And in fact, that's

685
00:36:57,400 --> 00:36:59,520
Speaker 1: something that's pointed out in the in the study, is

686
00:36:59,640 --> 00:37:02,719
Speaker 1: that like this is a great way moving forward to

687
00:37:02,880 --> 00:37:06,160
Speaker 1: continue to analyze the chatter bots. Yeah, but also, as

688
00:37:06,239 --> 00:37:10,279
Speaker 1: we mentioned, the people who were interacting with these with

689
00:37:10,480 --> 00:37:14,719
Speaker 1: these serenoids and echo borgs, we're in some of these

690
00:37:14,800 --> 00:37:18,680
Speaker 1: cases primed to expect something weird because they're in a

691
00:37:18,840 --> 00:37:21,840
Speaker 1: test environment. You can't hide the fact that you're in

692
00:37:22,120 --> 00:37:25,439
Speaker 1: a psychological test and you're you know, whenever you're part

693
00:37:25,560 --> 00:37:28,759
Speaker 1: of a test group, you're sort of like ready for

694
00:37:28,920 --> 00:37:31,520
Speaker 1: some weirdness, you know. Now, I wonder how this would

695
00:37:31,560 --> 00:37:35,960
Speaker 1: be if you sprung these shadowers, these serenoids, and these

696
00:37:36,080 --> 00:37:40,640
Speaker 1: echo borgs on people in a purely social scenario like

697
00:37:40,760 --> 00:37:43,200
Speaker 1: a like we talked about at the beginning of convention,

698
00:37:43,640 --> 00:37:47,719
Speaker 1: or you know, workplace meeting or a party where people

699
00:37:48,000 --> 00:37:51,400
Speaker 1: weren't expecting anything strange to be going on right until

700
00:37:51,840 --> 00:37:54,080
Speaker 1: they have to fill out of survey after they leave

701
00:37:54,120 --> 00:37:56,480
Speaker 1: the dinner party. So what did you think of Susie

702
00:37:56,560 --> 00:38:00,920
Speaker 1: and or antecdote exactly. Yeah, I don't know why she

703
00:38:01,080 --> 00:38:03,279
Speaker 1: was so into fruit trees. You wouldn't shut up about it.

704
00:38:04,160 --> 00:38:09,800
Speaker 1: I mean, gardening is okay. But yeah, Now, something that

705
00:38:09,840 --> 00:38:11,600
Speaker 1: really stood out to me when we were going over this,

706
00:38:12,000 --> 00:38:14,840
Speaker 1: especially when you start thinking about the eco borg and

707
00:38:14,880 --> 00:38:16,279
Speaker 1: what it would like to what it would be like

708
00:38:16,360 --> 00:38:18,560
Speaker 1: to be an echo borg, and the sort of pros

709
00:38:18,600 --> 00:38:21,880
Speaker 1: and cons of being an eco borg, of of of

710
00:38:22,000 --> 00:38:26,520
Speaker 1: giving life to this will behind your will. He reminded

711
00:38:26,560 --> 00:38:29,200
Speaker 1: me a lot of transactive memory, which comes into play

712
00:38:29,600 --> 00:38:31,600
Speaker 1: really in two key areas. First of all, it's the

713
00:38:31,680 --> 00:38:37,000
Speaker 1: method by which we've always stored information in other people. Okay,

714
00:38:37,480 --> 00:38:40,359
Speaker 1: so you know those facts that you never remember because

715
00:38:40,400 --> 00:38:43,279
Speaker 1: your spouse remembers them, or it's that, or it's a

716
00:38:43,360 --> 00:38:45,839
Speaker 1: particular spelling or a bit of trivia that you never

717
00:38:45,960 --> 00:38:47,480
Speaker 1: keep in your own head because you always look it

718
00:38:47,600 --> 00:38:50,960
Speaker 1: up on your smartphone. It's the same thing. So outsourcing

719
00:38:51,000 --> 00:38:53,320
Speaker 1: a memory. I think this, this must be one of

720
00:38:53,360 --> 00:38:55,839
Speaker 1: your favorite topics. You've come back to this a lot,

721
00:38:55,920 --> 00:38:58,080
Speaker 1: And I think it's really interesting that because I see

722
00:38:58,120 --> 00:39:00,160
Speaker 1: it every day in my own life, and as I

723
00:39:00,200 --> 00:39:02,000
Speaker 1: read about it's like, it's it's all I see. It's

724
00:39:02,080 --> 00:39:04,360
Speaker 1: like the things that I forget, the things that I

725
00:39:04,880 --> 00:39:08,480
Speaker 1: my mind refuses to learn because I've outsourced it to

726
00:39:09,000 --> 00:39:12,440
Speaker 1: the ubiquitous technology. Yeah. Yeah, I think the story is that,

727
00:39:12,680 --> 00:39:15,400
Speaker 1: you know, Socrates was worried about this, right that if

728
00:39:16,080 --> 00:39:18,400
Speaker 1: you know, if we teach everybody writing and we have

729
00:39:18,719 --> 00:39:21,000
Speaker 1: writing on scrolls all the time, nobody's going to be

730
00:39:21,040 --> 00:39:24,399
Speaker 1: able to remember anything. You're not priming your memory. There

731
00:39:24,520 --> 00:39:27,600
Speaker 1: might be some truth to that, But then again, are

732
00:39:27,719 --> 00:39:31,080
Speaker 1: are we not just, you know, by writing things down

733
00:39:31,239 --> 00:39:35,680
Speaker 1: and having Internet archives and things becoming cyborgs in a way? Yeah,

734
00:39:35,800 --> 00:39:38,400
Speaker 1: I mean, in a sense, the voice is already whispering

735
00:39:38,400 --> 00:39:40,160
Speaker 1: in our ear. Yeah. I mean you could look at

736
00:39:40,200 --> 00:39:42,400
Speaker 1: it as a weakening of the human mind, or you

737
00:39:42,480 --> 00:39:45,239
Speaker 1: could look at it as a technological upgrade of the

738
00:39:45,320 --> 00:39:48,479
Speaker 1: human mind. Yeah, to what extent, will we all become

739
00:39:48,560 --> 00:39:50,840
Speaker 1: echo boards of a type? You know where instead of

740
00:39:50,880 --> 00:39:52,320
Speaker 1: it being a situation where I'm just going to be

741
00:39:52,360 --> 00:39:55,440
Speaker 1: a conduit for a powerful artificial intelligence, what if it's

742
00:39:55,480 --> 00:39:59,480
Speaker 1: more like I want to augment my existing self, which

743
00:39:59,520 --> 00:40:02,759
Speaker 1: I think is pretty good with say, you know, an

744
00:40:02,800 --> 00:40:06,279
Speaker 1: artificial intelligence that'll feed me the right lines in like

745
00:40:06,440 --> 00:40:10,839
Speaker 1: business situations or social situations. So it's you know, less

746
00:40:11,120 --> 00:40:13,800
Speaker 1: focus on I'm just gonna be a meat puppet, but

747
00:40:14,000 --> 00:40:18,920
Speaker 1: rather can I merge with this AI, maybe even just

748
00:40:18,960 --> 00:40:22,120
Speaker 1: a small percentage you know five AI, and become a

749
00:40:22,200 --> 00:40:25,319
Speaker 1: better person and more effective person. Yeah, And these experiments,

750
00:40:25,440 --> 00:40:28,800
Speaker 1: all these examples are are word for word dictation. So

751
00:40:29,000 --> 00:40:33,160
Speaker 1: whether it's the shadower of the of the serenoid type

752
00:40:33,200 --> 00:40:36,120
Speaker 1: being fedlines by human or the shadower of the echo

753
00:40:36,160 --> 00:40:38,960
Speaker 1: borg type being fedlines by a computer, it's all lines.

754
00:40:39,080 --> 00:40:42,120
Speaker 1: You're getting full sentences and you're just trained to say

755
00:40:42,200 --> 00:40:45,120
Speaker 1: them as fast as you can. I mean, I wonder

756
00:40:45,320 --> 00:40:48,719
Speaker 1: if this setup could be more conceptual in nature, or

757
00:40:49,000 --> 00:40:53,040
Speaker 1: you know, feeding you facts or feeding you sort of

758
00:40:53,200 --> 00:40:57,480
Speaker 1: feedback on the progress of the conversation. It also makes

759
00:40:57,520 --> 00:41:00,200
Speaker 1: me think about the comedic possibilities because you imagine an

760
00:41:00,320 --> 00:41:03,120
Speaker 1: individual and sort of you know, a very updated Syrian

761
00:41:03,200 --> 00:41:06,360
Speaker 1: No kind of story story where individual goes into a

762
00:41:06,440 --> 00:41:09,239
Speaker 1: business situation and they thought they loaded uh you know,

763
00:41:09,440 --> 00:41:13,000
Speaker 1: business Helper for a one into their into their their

764
00:41:13,040 --> 00:41:18,000
Speaker 1: mind box, and instead they put Lothario four point one romance. Yeah,

765
00:41:18,040 --> 00:41:20,560
Speaker 1: so suddenly they're they're fed all these really effective lines.

766
00:41:20,800 --> 00:41:22,920
Speaker 1: Uh if you were you know, in a bar, but

767
00:41:23,120 --> 00:41:26,120
Speaker 1: instead you're you're throwing them out there in the business meeting. Yeah,

768
00:41:26,360 --> 00:41:30,400
Speaker 1: I think it could work. Walks on the beach have

769
00:41:30,560 --> 00:41:34,160
Speaker 1: to do with the new rollout, well, you know, it

770
00:41:34,280 --> 00:41:38,960
Speaker 1: comes to mind Black Mirror, the British television series that

771
00:41:39,000 --> 00:41:43,279
Speaker 1: does such a fantastic job looking at the ramifications near

772
00:41:43,360 --> 00:41:46,960
Speaker 1: future of our modern technology, often with very troubling results.

773
00:41:47,480 --> 00:41:50,200
Speaker 1: The Christmas episode they recently did, which I have not

774
00:41:50,280 --> 00:41:52,759
Speaker 1: seen yet. I think that's the only episode I haven't seen. Yeah,

775
00:41:52,760 --> 00:41:54,480
Speaker 1: I don't think it's made it to like Netflix here

776
00:41:54,480 --> 00:41:56,320
Speaker 1: in the States yet, And I'm not gonna spoil anything,

777
00:41:56,440 --> 00:42:01,320
Speaker 1: but the initial setup involves one character who who offers

778
00:42:01,400 --> 00:42:06,960
Speaker 1: a Syrano de Bergerac kind of service to individuals out

779
00:42:06,960 --> 00:42:09,640
Speaker 1: there who need a little help with their their pickup game.

780
00:42:10,400 --> 00:42:13,840
Speaker 1: I number one, that's super creepy. And number two, I

781
00:42:13,960 --> 00:42:17,840
Speaker 1: can totally see that being a real thing. Yeah, I

782
00:42:18,120 --> 00:42:20,520
Speaker 1: don't find that out outlandish at all. Yeah. I mean,

783
00:42:20,560 --> 00:42:22,640
Speaker 1: that's the great thing about Black Mirror is that, you

784
00:42:22,719 --> 00:42:25,880
Speaker 1: know that's it leans in the sci fi direction, but

785
00:42:26,080 --> 00:42:28,359
Speaker 1: not too far like it's and and and that it's

786
00:42:28,400 --> 00:42:32,040
Speaker 1: perfect science fiction because it speaks to the problems that

787
00:42:32,120 --> 00:42:35,280
Speaker 1: we have today and how we are viewing the problems

788
00:42:35,320 --> 00:42:38,320
Speaker 1: to come. Exactly. If you want to see some disturbingly

789
00:42:38,480 --> 00:42:43,960
Speaker 1: plausible dystopias, watch Black Mirror. But you know it ultimately

790
00:42:44,000 --> 00:42:47,680
Speaker 1: leads to the question echo bot, echo borgs? Is this

791
00:42:47,800 --> 00:42:51,280
Speaker 1: a dystopian idea or is it a pretty cool idea?

792
00:42:51,560 --> 00:42:54,040
Speaker 1: Is it? Is it ultimately a utopian idea where it

793
00:42:54,400 --> 00:42:58,640
Speaker 1: would allow us to to be better individuals? I don't know.

794
00:42:58,719 --> 00:43:01,480
Speaker 1: I mean, would the ancient for loss forers look at

795
00:43:01,640 --> 00:43:05,640
Speaker 1: our relationship with the contents of our computers and the

796
00:43:05,719 --> 00:43:09,200
Speaker 1: web as a horrible dystopia and feels fine to me?

797
00:43:09,360 --> 00:43:11,880
Speaker 1: But would they look at that and say, oh, you know,

798
00:43:12,000 --> 00:43:13,960
Speaker 1: react the same way as we do to a Black

799
00:43:14,040 --> 00:43:16,480
Speaker 1: Mirror episode? Problem I mean also they would look at

800
00:43:16,480 --> 00:43:18,279
Speaker 1: our pants and say, what are they doing? Where are

801
00:43:18,320 --> 00:43:20,960
Speaker 1: their telegets? Oh my god, yeah, a dressed like a persian.

802
00:43:21,000 --> 00:43:25,399
Speaker 1: I don't understand. I don't know. Ultimately, that's a question

803
00:43:25,600 --> 00:43:28,239
Speaker 1: throughout to the listeners. You know, what, how would you

804
00:43:28,320 --> 00:43:30,239
Speaker 1: accept that job offer that we laid out at the

805
00:43:30,280 --> 00:43:34,360
Speaker 1: beginning of the episode. And and furthermore, would you augment

806
00:43:34,440 --> 00:43:37,239
Speaker 1: yourself with some sort of mild echo borg system again,

807
00:43:37,360 --> 00:43:39,680
Speaker 1: like a you know, a romance one oh one or

808
00:43:39,760 --> 00:43:43,480
Speaker 1: a business business strategy one oh one kind of program

809
00:43:43,640 --> 00:43:46,560
Speaker 1: to to feed you the necessary lines or even just

810
00:43:46,760 --> 00:43:49,120
Speaker 1: ideas or facts that you might need to make it

811
00:43:49,239 --> 00:43:53,600
Speaker 1: through that business luncheon or dinner date. You know. In

812
00:43:53,640 --> 00:43:57,040
Speaker 1: another aspect of transactive memory that I want to drive

813
00:43:57,120 --> 00:43:59,920
Speaker 1: home that that plays in nicely with the hybrid per

814
00:44:00,000 --> 00:44:03,920
Speaker 1: sonality model of serenoids and echo boards, and that is

815
00:44:04,560 --> 00:44:07,960
Speaker 1: cross queuing. Okay. This is when, and I imagine a

816
00:44:08,040 --> 00:44:09,680
Speaker 1: number of you can can relate to. This is when

817
00:44:09,680 --> 00:44:12,080
Speaker 1: you're having a conversation with let's say you know, a

818
00:44:12,160 --> 00:44:14,640
Speaker 1: spouse or a partner, a close friend or family member,

819
00:44:15,040 --> 00:44:17,680
Speaker 1: and neither of you are quite able to remember something

820
00:44:17,800 --> 00:44:20,200
Speaker 1: on your own, but when you start queuing each other

821
00:44:21,160 --> 00:44:23,680
Speaker 1: the you're able to remember it together in a way

822
00:44:23,719 --> 00:44:25,600
Speaker 1: that you wouldn't, you know, kind of vultron style, you know,

823
00:44:25,719 --> 00:44:28,920
Speaker 1: bringing it together, and suddenly your combined power for call

824
00:44:29,040 --> 00:44:31,480
Speaker 1: that memory, right, you're sort of pulling the triggers on

825
00:44:31,560 --> 00:44:34,400
Speaker 1: each other's brains. Yeah, so I think it's interesting to

826
00:44:34,480 --> 00:44:38,080
Speaker 1: think about that, uh, that scenario cross queuing and transactive memory,

827
00:44:38,480 --> 00:44:41,920
Speaker 1: especially because if you imagine this AI scenario where you

828
00:44:41,960 --> 00:44:44,960
Speaker 1: know you've got some kind of on board artificial intelligence

829
00:44:45,080 --> 00:44:47,840
Speaker 1: that feeds you lines occasionally when you need them. You know,

830
00:44:47,920 --> 00:44:49,920
Speaker 1: you're not just a parrot for everything it says, but

831
00:44:50,480 --> 00:44:52,719
Speaker 1: it's it's an occasional helper. How does it know when

832
00:44:52,760 --> 00:44:56,120
Speaker 1: you need help? You would need some kind of some

833
00:44:56,280 --> 00:44:59,200
Speaker 1: kind of queuing in the conversation or even just in

834
00:44:59,320 --> 00:45:02,000
Speaker 1: your mind for this thing to know. Okay, I'm going

835
00:45:02,040 --> 00:45:04,600
Speaker 1: to step in. Like every time you go hmm, and

836
00:45:04,680 --> 00:45:06,520
Speaker 1: then it starts feeding you the line, it's like he's stalling,

837
00:45:06,719 --> 00:45:09,239
Speaker 1: throwing some throwing some good dust, some good winkow there

838
00:45:09,360 --> 00:45:14,919
Speaker 1: every time you say the word interesting, interesting, interesting. Yeah,

839
00:45:14,960 --> 00:45:17,239
Speaker 1: I mean it would it would certainly change the podcast game.

840
00:45:17,280 --> 00:45:19,560
Speaker 1: I'll tell you that. So I guess that wraps it

841
00:45:19,680 --> 00:45:21,840
Speaker 1: up for today, but we have some homework for you.

842
00:45:22,360 --> 00:45:25,200
Speaker 1: If you happen to be watching Benedict Cumberbatch in The

843
00:45:25,360 --> 00:45:28,080
Speaker 1: Imitation Game, I want to see if you can figure

844
00:45:28,120 --> 00:45:31,400
Speaker 1: out how the underlying gospel message of the machine cult

845
00:45:31,440 --> 00:45:34,920
Speaker 1: of fruit Trees is propagated in the cinematic subtleties of

846
00:45:35,040 --> 00:45:38,200
Speaker 1: the film. So if you're watching it, just give it

847
00:45:38,320 --> 00:45:41,080
Speaker 1: some thought. In the meantime, check out stuff to Blow

848
00:45:41,120 --> 00:45:43,080
Speaker 1: your Mind dot com. That's where you'll find all of

849
00:45:43,120 --> 00:45:47,120
Speaker 1: our podcast episodes. You'll find our videos are blog posts,

850
00:45:47,200 --> 00:45:49,160
Speaker 1: as well as links out to our social media accounts

851
00:45:49,160 --> 00:45:52,040
Speaker 1: such as Facebook, Twitter, and Tumbler. And if you want

852
00:45:52,080 --> 00:45:54,200
Speaker 1: to let us know whether you are interested in a

853
00:45:54,280 --> 00:45:56,640
Speaker 1: career as an echo borg or a shadower of some

854
00:45:56,719 --> 00:45:59,839
Speaker 1: other kind of strange alien intelligence, you can email us

855
00:45:59,840 --> 00:46:02,080
Speaker 1: at blow the Mind at how stuff works dot com.

856
00:46:05,920 --> 00:46:08,439
Speaker 1: For more on this and thousands of other topics, visit

857
00:46:08,480 --> 00:46:09,640
Speaker 1: how stuff works dot com