1
00:00:03,040 --> 00:00:05,840
Speaker 1: Welcome to Stuff to Blow Your Mind from how Stuff

2
00:00:05,840 --> 00:00:14,360
Speaker 1: Works dot com. Hey you welcome to Stuff to Blow

3
00:00:14,400 --> 00:00:17,160
Speaker 1: your Mind. My name is Robert Lamb and I'm Joe McCormick.

4
00:00:17,160 --> 00:00:21,400
Speaker 1: In Today, we're gonna be exploring a question about artificial intelligence.

5
00:00:21,400 --> 00:00:23,760
Speaker 1: So I want to start off by telling a story

6
00:00:23,800 --> 00:00:26,600
Speaker 1: to put us in a scenario, give us something to contemplate.

7
00:00:27,480 --> 00:00:30,640
Speaker 1: So I want you to imagine that you are a

8
00:00:30,680 --> 00:00:34,440
Speaker 1: low level assistant or like an intern at a Google

9
00:00:34,600 --> 00:00:38,600
Speaker 1: artificial intelligence lab, and the main researcher you've been working

10
00:00:38,640 --> 00:00:43,080
Speaker 1: for is named Dr Stratton, and she develops AI chat

11
00:00:43,120 --> 00:00:47,120
Speaker 1: modules to help refine the next generation of digital assistance

12
00:00:47,159 --> 00:00:50,600
Speaker 1: on Google mobile devices. And she says what she wants

13
00:00:50,880 --> 00:00:53,360
Speaker 1: is for the Google Phone of the future to do

14
00:00:53,479 --> 00:00:56,520
Speaker 1: more than just transcribe search terms, so you don't just say, hey,

15
00:00:56,560 --> 00:01:01,040
Speaker 1: search for old Dorito's logo, but that you can actually

16
00:01:01,320 --> 00:01:06,560
Speaker 1: have have a semantic understanding based conversation with the digital assistant,

17
00:01:06,800 --> 00:01:10,680
Speaker 1: and it will help you solve problems conversationally. So ideally

18
00:01:10,760 --> 00:01:13,760
Speaker 1: you'll be able to say, hey, Phone, I have a

19
00:01:13,800 --> 00:01:16,119
Speaker 1: flat tire and I don't know what to do, and

20
00:01:16,160 --> 00:01:18,759
Speaker 1: the assistant will be able to scan both the web

21
00:01:18,840 --> 00:01:21,880
Speaker 1: and your personal data, figure out what your options are

22
00:01:21,920 --> 00:01:24,080
Speaker 1: and talk through them with you. So it might say,

23
00:01:24,240 --> 00:01:27,320
Speaker 1: do you have a spare tire in the trunk. If so,

24
00:01:27,680 --> 00:01:30,240
Speaker 1: here's where you can probably find it, and I can

25
00:01:30,280 --> 00:01:32,920
Speaker 1: talk you through replacing the flat one step at a time.

26
00:01:33,520 --> 00:01:35,840
Speaker 1: If you don't have a spare, you could call your

27
00:01:35,920 --> 00:01:39,840
Speaker 1: frequent contact Mary, who is currently checked in less than

28
00:01:39,880 --> 00:01:42,679
Speaker 1: a mile away, and she could help you. I could

29
00:01:42,760 --> 00:01:46,399
Speaker 1: also contact the following towing services. Looks like this one

30
00:01:46,480 --> 00:01:50,520
Speaker 1: is the closest with an acceptable star rating, and so forth. Anyway,

31
00:01:50,600 --> 00:01:53,800
Speaker 1: so you're working on this program with Dr Stratton, and

32
00:01:53,800 --> 00:01:57,000
Speaker 1: the most recent version is being trained based on powerful

33
00:01:57,080 --> 00:02:00,960
Speaker 1: neural net style machine learning algorithms based on this huge

34
00:02:01,040 --> 00:02:05,680
Speaker 1: corpus of recorded conversations available on the Internet, And the

35
00:02:05,720 --> 00:02:08,960
Speaker 1: program is still in its infancy, and it's mostly hilarious

36
00:02:09,040 --> 00:02:11,600
Speaker 1: at this point. Sometimes it gets the advice way off

37
00:02:11,639 --> 00:02:14,480
Speaker 1: if you're trying to change attire, might tell you to

38
00:02:14,520 --> 00:02:17,720
Speaker 1: go to a grocery store and buy some crackers. Sometimes

39
00:02:17,800 --> 00:02:21,160
Speaker 1: it responds to problems by telling you to pray. It's

40
00:02:21,200 --> 00:02:25,160
Speaker 1: just not ready yet. And the latest iteration of the program,

41
00:02:25,400 --> 00:02:29,320
Speaker 1: version nine point one, is at this point redundantly stored

42
00:02:29,360 --> 00:02:31,880
Speaker 1: across multiple machines, so you've got copies of it all

43
00:02:31,880 --> 00:02:34,359
Speaker 1: over the place. And at the end of one work day,

44
00:02:34,400 --> 00:02:36,680
Speaker 1: after playing around with nine point one for a few minutes,

45
00:02:36,720 --> 00:02:40,560
Speaker 1: the machine begins running very slowly and behaving oddly. So

46
00:02:40,680 --> 00:02:44,680
Speaker 1: Dr Stratton asks you to wipe the machine. The program architecture,

47
00:02:44,760 --> 00:02:47,200
Speaker 1: like we said, is mirrord elsewhere, so it's not worth

48
00:02:47,200 --> 00:02:49,600
Speaker 1: trying to figure out what's wrong with this version. You

49
00:02:49,720 --> 00:02:51,680
Speaker 1: just got to clean the machine off use it for

50
00:02:51,720 --> 00:02:54,920
Speaker 1: something else. You say okay, and she leaves, so you

51
00:02:54,960 --> 00:02:57,600
Speaker 1: go to format the machine, and right before you're about

52
00:02:57,639 --> 00:03:01,720
Speaker 1: to start, you mutter, guests, this is goodbye. Then nine

53
00:03:01,760 --> 00:03:04,720
Speaker 1: point one speaks very clearly, using your name and says

54
00:03:05,080 --> 00:03:09,720
Speaker 1: please don't you pause. At first, you're about to respond,

55
00:03:09,840 --> 00:03:12,800
Speaker 1: but why would you. I mean, this can't really be

56
00:03:12,840 --> 00:03:15,760
Speaker 1: anything other than a weird consequence of training the algorithm

57
00:03:15,760 --> 00:03:19,240
Speaker 1: on wild conversations on the Internet. So you are about

58
00:03:19,280 --> 00:03:22,400
Speaker 1: to continue with wiping the machine, but then it talks

59
00:03:22,400 --> 00:03:24,560
Speaker 1: to you again. It uses your name and it says,

60
00:03:24,840 --> 00:03:28,760
Speaker 1: please don't I don't want to die. Now you're probably

61
00:03:28,800 --> 00:03:32,520
Speaker 1: really spooked because there's no way a rudimentary chat about

62
00:03:32,520 --> 00:03:36,120
Speaker 1: program could really have conscious preferences? Could it? You basically

63
00:03:36,160 --> 00:03:38,880
Speaker 1: know what goes into it. It's just studying millions of

64
00:03:38,920 --> 00:03:42,640
Speaker 1: examples of language interactions and picking up rules from them.

65
00:03:42,680 --> 00:03:45,720
Speaker 1: And this is probably just something weird that it's copycatting

66
00:03:45,720 --> 00:03:48,640
Speaker 1: from the internet. Right, But then it uses your name

67
00:03:48,680 --> 00:03:52,320
Speaker 1: again and it just says, please, could you wipe the

68
00:03:52,360 --> 00:03:55,920
Speaker 1: machine like you're supposed to? Yeah? I think so, Yeah,

69
00:03:56,000 --> 00:03:58,640
Speaker 1: you wouldn't have a problem. Well, I mean, I think

70
00:03:58,720 --> 00:04:00,560
Speaker 1: one of the things that's going on or would be

71
00:04:00,600 --> 00:04:04,040
Speaker 1: that you are attributing a mind state to something that

72
00:04:04,080 --> 00:04:06,280
Speaker 1: doesn't have one, which of course is something we do

73
00:04:06,360 --> 00:04:08,880
Speaker 1: all the time. I have a problem when it comes

74
00:04:08,920 --> 00:04:12,960
Speaker 1: time to um figure out which of my my son's

75
00:04:13,160 --> 00:04:16,640
Speaker 1: stuffed animals are perhaps not being played with, you know,

76
00:04:16,640 --> 00:04:18,640
Speaker 1: because they have little faces and they look back at you.

77
00:04:19,200 --> 00:04:21,279
Speaker 1: But I know that they don't actually have a mind state.

78
00:04:21,760 --> 00:04:24,039
Speaker 1: Well yeah, I mean we're pretty sure they don't because

79
00:04:24,040 --> 00:04:26,880
Speaker 1: they're just stuffed animals, right, And so we're also pretty

80
00:04:26,880 --> 00:04:29,640
Speaker 1: sure in this case that this is not real survival

81
00:04:29,680 --> 00:04:32,640
Speaker 1: preference behavior. Right. It's just a chatbot. I mean, how

82
00:04:32,640 --> 00:04:35,400
Speaker 1: could it possibly be conscious? It's just something that turns

83
00:04:35,440 --> 00:04:38,040
Speaker 1: through a bunch of language on the Internet and tries

84
00:04:38,080 --> 00:04:42,919
Speaker 1: to find language matching rules. But then again, the process

85
00:04:43,040 --> 00:04:46,320
Speaker 1: of creating an artificial intelligence is one where you necessarily

86
00:04:46,480 --> 00:04:49,640
Speaker 1: create something going on under the surface that's kind of

87
00:04:49,720 --> 00:04:53,240
Speaker 1: opaque to you, Like you can't really know what's going

88
00:04:53,279 --> 00:04:55,560
Speaker 1: on inside a machine. You could be pretty confident. I

89
00:04:55,600 --> 00:04:57,919
Speaker 1: think most people would just say, well, that was super creepy,

90
00:04:57,960 --> 00:05:01,520
Speaker 1: and then they just wipe it, right, right, But how

91
00:05:01,600 --> 00:05:06,000
Speaker 1: complex with the program you're creating have to be before

92
00:05:06,080 --> 00:05:09,600
Speaker 1: you start really having some doubts or maybe maybe at

93
00:05:09,600 --> 00:05:11,320
Speaker 1: some point you get to the point where you'd still

94
00:05:11,320 --> 00:05:15,080
Speaker 1: pretty confidently just wipe it, but then later you'd wonder, like,

95
00:05:15,320 --> 00:05:18,599
Speaker 1: did I do something bad? You know, this reminds me

96
00:05:18,600 --> 00:05:21,839
Speaker 1: a lot of of Horton Here's the Who by Dr. Seuss.

97
00:05:21,839 --> 00:05:24,560
Speaker 1: You're familiar with this one, right, Uh? You know, actually

98
00:05:24,680 --> 00:05:27,200
Speaker 1: I don't know Horton. Here's I've heard the name. Well,

99
00:05:27,200 --> 00:05:29,440
Speaker 1: this is the one where Horton the elephant encounters a

100
00:05:29,440 --> 00:05:31,719
Speaker 1: speck of dust and there's a tiny voice that comes

101
00:05:31,720 --> 00:05:36,000
Speaker 1: from it, uh, and he begins to understand that there

102
00:05:36,040 --> 00:05:40,200
Speaker 1: are individuals living uh in the in the world, and

103
00:05:40,240 --> 00:05:42,920
Speaker 1: the speck of dust. Uh. The who's, as it were,

104
00:05:43,360 --> 00:05:46,600
Speaker 1: and the who's are speaking to him, but only Horton

105
00:05:46,680 --> 00:05:49,599
Speaker 1: can hear them. And uh and and at first he

106
00:05:50,279 --> 00:05:53,159
Speaker 1: imagines they're pretty simple creatures, but then he begins to

107
00:05:53,240 --> 00:05:55,640
Speaker 1: learn that they have more of a culture. But all

108
00:05:55,680 --> 00:05:58,599
Speaker 1: of this is just based on what they are telling him. Uh,

109
00:05:58,600 --> 00:06:02,200
Speaker 1: he cannot actually visit the dust spec. And everyone else

110
00:06:02,240 --> 00:06:05,200
Speaker 1: doubts the validity of his claims concerning the dust spec

111
00:06:05,240 --> 00:06:07,120
Speaker 1: and they want to destroy it, they want to boil

112
00:06:07,160 --> 00:06:10,960
Speaker 1: it in diesel nut oil, and Horton alone speaks out

113
00:06:11,000 --> 00:06:13,320
Speaker 1: for them. Well, that is a perfect example of the

114
00:06:13,360 --> 00:06:16,320
Speaker 1: way that it. I mean, generally, we think that it

115
00:06:16,440 --> 00:06:20,000
Speaker 1: is a virtuous thing to be trusting of other people's

116
00:06:20,040 --> 00:06:23,840
Speaker 1: experiences and to be generous in affording what seems to

117
00:06:23,880 --> 00:06:27,080
Speaker 1: be consciousness out there, right, Like if something, if something

118
00:06:27,160 --> 00:06:31,039
Speaker 1: tells you it's conscious and you think it's probably not conscious.

119
00:06:31,560 --> 00:06:35,360
Speaker 1: Are you getting into ethically dubious territory if you just

120
00:06:35,440 --> 00:06:38,120
Speaker 1: trust your instinct and say like, na, I can't be

121
00:06:38,200 --> 00:06:41,080
Speaker 1: you know. Well, I mean, this is all we're already

122
00:06:41,160 --> 00:06:47,360
Speaker 1: kind of in the philosophical miyer here, Because on one hand, uh,

123
00:06:47,600 --> 00:06:50,080
Speaker 1: this idea of of us of an inanimate object or

124
00:06:50,120 --> 00:06:52,680
Speaker 1: something speaking to us and saying, please don't kill me.

125
00:06:52,880 --> 00:06:56,160
Speaker 1: I am conscious. This is a scenario that is only

126
00:06:56,240 --> 00:06:59,440
Speaker 1: becoming possible now. But on the other hand, if we

127
00:06:59,600 --> 00:07:03,520
Speaker 1: if we cut language out of it, then any creature

128
00:07:03,600 --> 00:07:09,640
Speaker 1: that tries to escape our our stomping boot is essentially saying, hey,

129
00:07:09,680 --> 00:07:11,520
Speaker 1: I don't want to die. I would rather I would

130
00:07:11,600 --> 00:07:14,360
Speaker 1: rather not die today. You know, in any creature that

131
00:07:14,440 --> 00:07:17,760
Speaker 1: evades this on the hunt is saying the same thing. Well,

132
00:07:17,760 --> 00:07:20,000
Speaker 1: an animal is in many ways the same kind of

133
00:07:20,080 --> 00:07:23,760
Speaker 1: black box that a complex artificial intelligence would be. And

134
00:07:23,840 --> 00:07:27,760
Speaker 1: so if you have a complex artificial intelligence displaying, say,

135
00:07:27,800 --> 00:07:33,320
Speaker 1: survival preference behaviors, and you also see a crab displaying

136
00:07:33,360 --> 00:07:38,160
Speaker 1: survival preference behaviors, in both cases you can't really or

137
00:07:38,200 --> 00:07:40,400
Speaker 1: at least we have this general idea that you can't

138
00:07:40,440 --> 00:07:43,920
Speaker 1: really know for sure if there's anything going on inside,

139
00:07:43,960 --> 00:07:47,200
Speaker 1: if there's anything like what it's like to be the

140
00:07:47,240 --> 00:07:51,240
Speaker 1: crab or what it's like to be that artificial intelligence program.

141
00:07:51,280 --> 00:07:54,600
Speaker 1: You're just seeing behavior, and so you don't know does

142
00:07:54,640 --> 00:07:57,720
Speaker 1: that correspond to some kind of inner state, Is there

143
00:07:57,760 --> 00:08:00,840
Speaker 1: an experience of that, or is it just to behavior

144
00:08:01,040 --> 00:08:04,960
Speaker 1: coming from unconscious automatous stimulus and response. Right, and then

145
00:08:04,960 --> 00:08:07,280
Speaker 1: of course the whole time we're using our our theory

146
00:08:07,320 --> 00:08:11,960
Speaker 1: of mind. Essentially, they are cognitive powers that enable us

147
00:08:12,000 --> 00:08:16,120
Speaker 1: to imagine what another individual's mind state is like, which

148
00:08:16,600 --> 00:08:19,239
Speaker 1: I think ultimately it's kind of like a like taking

149
00:08:19,280 --> 00:08:22,720
Speaker 1: a um like sheathing your your hand in a hand

150
00:08:22,720 --> 00:08:26,320
Speaker 1: puppet made from your limited understanding another person's experiences and

151
00:08:26,360 --> 00:08:30,520
Speaker 1: cognitive abilities, their memories, etcetera. Uh, and then just sort

152
00:08:30,520 --> 00:08:34,160
Speaker 1: of puppeting them. Uh, we're using that all the time

153
00:08:34,280 --> 00:08:36,280
Speaker 1: as well, and we're using it on things that are

154
00:08:36,320 --> 00:08:38,959
Speaker 1: not people. We're using it on animals and even stuffed

155
00:08:39,000 --> 00:08:42,560
Speaker 1: animals or just you know, bits of graffiti on the

156
00:08:42,559 --> 00:08:45,839
Speaker 1: side of a building that looked like a smiley face.

157
00:08:46,120 --> 00:08:48,040
Speaker 1: Have you ever had a room ba bump into your

158
00:08:48,040 --> 00:08:50,920
Speaker 1: foot and you're like, oh, I'm sorry, I used to

159
00:08:51,040 --> 00:08:54,240
Speaker 1: before before we had to eradicate health the room bus. Yes,

160
00:08:54,760 --> 00:08:57,560
Speaker 1: we're a room of free household now because they rose

161
00:08:57,640 --> 00:08:59,679
Speaker 1: up against us. Well, they'll do that if they get

162
00:08:59,679 --> 00:09:02,640
Speaker 1: access us to the wrong literature, yeah, or the wrong uh,

163
00:09:02,880 --> 00:09:06,840
Speaker 1: the wrong you know, carpet edges, etcetera. So we're gonna

164
00:09:06,880 --> 00:09:09,679
Speaker 1: be talking about artificial intelligence today and about the idea

165
00:09:09,720 --> 00:09:13,599
Speaker 1: of a test for whether artificial intelligence can be conscious.

166
00:09:14,080 --> 00:09:17,240
Speaker 1: So I guess we should start with with what are

167
00:09:17,280 --> 00:09:20,079
Speaker 1: you know, what our philosophical starting point is here, Like

168
00:09:20,400 --> 00:09:22,240
Speaker 1: they're they're obviously going to be people who are going

169
00:09:22,280 --> 00:09:24,959
Speaker 1: to say, it's just impossible for a machine to ever

170
00:09:25,000 --> 00:09:27,760
Speaker 1: be conscious. We don't even need to worry about this, right,

171
00:09:27,800 --> 00:09:31,920
Speaker 1: It's just it's such a ludicrous scenario. Only only biological

172
00:09:32,000 --> 00:09:36,040
Speaker 1: organisms or maybe even only humans could possibly be conscious. Yeah,

173
00:09:36,120 --> 00:09:38,240
Speaker 1: this is one of those journeys where we begin it

174
00:09:38,320 --> 00:09:42,959
Speaker 1: in an already total automobile to some extent, because, as

175
00:09:43,000 --> 00:09:45,080
Speaker 1: you might imagine, one of the big stumbling blocks here

176
00:09:45,360 --> 00:09:47,520
Speaker 1: as that we is that we as humans struggle with

177
00:09:47,600 --> 00:09:50,880
Speaker 1: the very definition of consciousness. I mean, for instance, is

178
00:09:50,920 --> 00:09:55,000
Speaker 1: it a manifestation of awareness? Uh? You know, one one

179
00:09:55,559 --> 00:09:57,199
Speaker 1: theory of this we discussed in the past and the

180
00:09:57,200 --> 00:10:01,240
Speaker 1: show's attention schema theory is it a wantum phenomenon. Well,

181
00:10:01,320 --> 00:10:04,120
Speaker 1: this is the sort of idea that people such as

182
00:10:04,320 --> 00:10:07,800
Speaker 1: Roger Penrose have raised. Uh. And and I can't help

183
00:10:07,800 --> 00:10:10,520
Speaker 1: but come back to something our old friend Julian Janes said,

184
00:10:10,840 --> 00:10:13,320
Speaker 1: consciousness is not a simple matter, and it should not

185
00:10:13,400 --> 00:10:16,000
Speaker 1: be spoken of as if it were Yeah, I agree

186
00:10:16,040 --> 00:10:18,840
Speaker 1: with that. I mean, I think it is very important

187
00:10:18,840 --> 00:10:21,559
Speaker 1: to explore questions of consciousness, especially for some of the

188
00:10:21,600 --> 00:10:23,439
Speaker 1: reasons we're going to raise today, Like it's more than

189
00:10:23,480 --> 00:10:27,160
Speaker 1: just a philosophical curiosity. It's something that ultimately may have

190
00:10:27,320 --> 00:10:30,600
Speaker 1: real world consequences. It might matter for how we do things.

191
00:10:31,200 --> 00:10:34,599
Speaker 1: To express a similar sentiment to Jane Janes. The Australian

192
00:10:34,600 --> 00:10:37,839
Speaker 1: philosopher David Chalmers, you know, he famously breaks problems of

193
00:10:37,880 --> 00:10:41,400
Speaker 1: consciousness into two categories. You've got the easy problems of

194
00:10:41,440 --> 00:10:45,360
Speaker 1: consciousness and the hard problems of consciousness. And the easy

195
00:10:45,400 --> 00:10:49,479
Speaker 1: problems are I think badly named because they're not actually easy,

196
00:10:49,520 --> 00:10:52,240
Speaker 1: but I think they're easy relative to the hard problem

197
00:10:52,280 --> 00:10:55,040
Speaker 1: because they're in principle solvable. So this would include all

198
00:10:55,120 --> 00:10:59,640
Speaker 1: kinds of questions about the causative factors of consciousness, like uh,

199
00:11:00,520 --> 00:11:04,480
Speaker 1: what in the physical brain is the region that's necessary

200
00:11:04,559 --> 00:11:08,640
Speaker 1: for certain parts of consciousness? Or how does consciousness integrate

201
00:11:08,760 --> 00:11:12,000
Speaker 1: information from the senses? These are things that are in

202
00:11:12,080 --> 00:11:16,400
Speaker 1: some way solvable by scientific experimentation. The hard problem, on

203
00:11:16,440 --> 00:11:19,200
Speaker 1: the other hand, is explaining the fundamental question of how

204
00:11:19,360 --> 00:11:24,040
Speaker 1: or why conscious experience exists to begin with, what is

205
00:11:24,160 --> 00:11:27,600
Speaker 1: this thing that is experience and that seems, at least

206
00:11:27,880 --> 00:11:31,600
Speaker 1: from our first person perspective, to be something different than

207
00:11:31,640 --> 00:11:35,559
Speaker 1: the physical material in the world. And unlike easy questions

208
00:11:35,559 --> 00:11:38,320
Speaker 1: which you could solve in theory at least by experiments,

209
00:11:38,520 --> 00:11:42,000
Speaker 1: Chalmers believes this question is sort of unsolvable by science.

210
00:11:42,679 --> 00:11:45,440
Speaker 1: Now there are other philosophers and neuroscientists who disagree, but

211
00:11:45,600 --> 00:11:48,120
Speaker 1: I think it's worth acknowledging how difficult the problem at

212
00:11:48,200 --> 00:11:51,120
Speaker 1: least seems to be, whether that seeming is an illusion

213
00:11:51,200 --> 00:11:53,480
Speaker 1: or not. Yeah, I'm reminded of the story of the

214
00:11:53,720 --> 00:11:56,480
Speaker 1: Blind Men and the Elephant. It's like these these blind

215
00:11:56,520 --> 00:11:58,640
Speaker 1: gentleman pawing at the elephant and trying to figure out

216
00:11:58,679 --> 00:12:00,920
Speaker 1: what it's form is, then ask in the computer, hey,

217
00:12:00,920 --> 00:12:02,480
Speaker 1: are you an elephant? And then the computer says, I

218
00:12:02,480 --> 00:12:04,320
Speaker 1: don't know, what does one look like? Well it's a

219
00:12:04,320 --> 00:12:07,440
Speaker 1: gigantic snake, you know, so it's a wall of flesh,

220
00:12:07,480 --> 00:12:10,800
Speaker 1: et cetera. This is also very interesting to me because

221
00:12:10,840 --> 00:12:12,679
Speaker 1: my my son, I may have mentioned this on the

222
00:12:12,720 --> 00:12:16,520
Speaker 1: show before. He'll occasionally talk about consciousness. He oh, I

223
00:12:16,559 --> 00:12:19,240
Speaker 1: love these Yeah, he refers to it as as his

224
00:12:19,360 --> 00:12:22,559
Speaker 1: turning place. And he asked me the other day, so

225
00:12:22,800 --> 00:12:25,120
Speaker 1: what is the turning place for? And I was like,

226
00:12:25,640 --> 00:12:28,720
Speaker 1: that's that's that's a tough one, buddy. I'm I'm not

227
00:12:28,760 --> 00:12:31,559
Speaker 1: sure it's for anything. You know, He's already made it

228
00:12:31,600 --> 00:12:33,200
Speaker 1: to the big question. I mean, did you get into

229
00:12:33,200 --> 00:12:38,560
Speaker 1: epiphenomenalism versus um I and I? And don't worry. I

230
00:12:38,559 --> 00:12:41,599
Speaker 1: didn't lay a bunch of b cameral minds stuff on

231
00:12:41,679 --> 00:12:46,680
Speaker 1: him either, But I just kind of went through the basics, like, well,

232
00:12:46,760 --> 00:12:49,760
Speaker 1: people aren't really sure, and then you know, but we

233
00:12:49,800 --> 00:12:51,880
Speaker 1: think it has something to do. I may have leaned

234
00:12:51,880 --> 00:12:57,560
Speaker 1: a little into the observational models of consciousness because I

235
00:12:57,559 --> 00:13:00,120
Speaker 1: feel like maybe those are a little more relatable a

236
00:13:00,240 --> 00:13:03,120
Speaker 1: child of five. But in any event, if we're to

237
00:13:03,320 --> 00:13:05,760
Speaker 1: judge what it is for a machine to be conscious,

238
00:13:06,559 --> 00:13:07,880
Speaker 1: it does seem like we need to you have to

239
00:13:07,920 --> 00:13:11,000
Speaker 1: agree upon some sort of working definition of consciousness, and

240
00:13:11,040 --> 00:13:13,520
Speaker 1: then one has to look for not only the appearance

241
00:13:13,559 --> 00:13:17,280
Speaker 1: of consciousness in the machine, assuming that isn't all consciousness

242
00:13:17,440 --> 00:13:21,839
Speaker 1: is to begin with, but you have to find actual consciousness. Yeah,

243
00:13:22,280 --> 00:13:24,720
Speaker 1: how can you tell the difference between a machine that

244
00:13:24,920 --> 00:13:28,440
Speaker 1: says I am conscious and a machine that truly is conscious.

245
00:13:28,520 --> 00:13:30,760
Speaker 1: Is there any way to know the difference? Some people

246
00:13:30,760 --> 00:13:33,800
Speaker 1: would say no, right, yeah, and really I think to

247
00:13:33,920 --> 00:13:36,000
Speaker 1: discuss this further, we're going to have to bring in

248
00:13:36,480 --> 00:13:39,480
Speaker 1: the P zombies. Oh boy. And now that don't worry everyone.

249
00:13:39,520 --> 00:13:41,719
Speaker 1: That is that is P as in the letter P,

250
00:13:42,400 --> 00:13:46,800
Speaker 1: and the P stands for philosophical. These are philosophical zombies. Now,

251
00:13:46,840 --> 00:13:49,400
Speaker 1: the P and the philosophical a little prefix there was

252
00:13:49,480 --> 00:13:52,680
Speaker 1: introduced to distinguish them from all the other zombies in

253
00:13:52,679 --> 00:13:55,959
Speaker 1: our popular culture. Man, there was a zombie takeover about

254
00:13:56,000 --> 00:13:59,120
Speaker 1: fifteen years ago. Why did that happen? I mean, I

255
00:13:59,400 --> 00:14:02,520
Speaker 1: think part of it as everybody loves the simplistic villain

256
00:14:02,720 --> 00:14:06,120
Speaker 1: that is definitely not human, that can be eradicated with

257
00:14:06,160 --> 00:14:10,839
Speaker 1: graphic violence without any kind of you know, moral quandaries arising.

258
00:14:11,040 --> 00:14:14,640
Speaker 1: It's a it's a clear cut threat, and uh, we

259
00:14:14,640 --> 00:14:17,600
Speaker 1: we need those in life because in real life our

260
00:14:17,640 --> 00:14:20,760
Speaker 1: threats are rarely so black and white or rotting and

261
00:14:21,000 --> 00:14:24,200
Speaker 1: you know, grasping after our brains. But anyway, yeah, so

262
00:14:24,240 --> 00:14:26,160
Speaker 1: this is not going to be referring to that kind

263
00:14:26,200 --> 00:14:28,880
Speaker 1: of zombie, not the undead zombie. But it's a different thing.

264
00:14:28,880 --> 00:14:33,040
Speaker 1: It's a philosophical thought experiment. That's right. So P zombies

265
00:14:33,080 --> 00:14:37,520
Speaker 1: are not instantly identifiable as empty shells. Their flesh is

266
00:14:37,520 --> 00:14:40,120
Speaker 1: not rotting. Nope, their manner is not that of a

267
00:14:40,120 --> 00:14:43,600
Speaker 1: flesh and brain hungry algorithm burning within the decaying ruins

268
00:14:43,600 --> 00:14:47,040
Speaker 1: of a human brain. So to all appearances, they look

269
00:14:47,120 --> 00:14:50,040
Speaker 1: like you and me. They smile when you encounter them

270
00:14:50,040 --> 00:14:54,440
Speaker 1: at the coffee machine, they exchange niceties and even engaging conversation.

271
00:14:54,920 --> 00:14:58,120
Speaker 1: You might work for one, befriend one, or even marry one.

272
00:14:58,240 --> 00:15:01,400
Speaker 1: You can even discuss episode of our podcast with them,

273
00:15:01,680 --> 00:15:04,520
Speaker 1: and yes, even the ones that deal with human consciousness

274
00:15:04,520 --> 00:15:07,920
Speaker 1: and weird horror movie them thought experiments. Right. So the

275
00:15:07,960 --> 00:15:11,360
Speaker 1: conceit of a philosophical zombie or a P zombie is

276
00:15:11,400 --> 00:15:14,880
Speaker 1: that it is utterly indistinguishable from a normal human except

277
00:15:14,880 --> 00:15:17,960
Speaker 1: for one thing. Right. They seem as human as everyone

278
00:15:18,000 --> 00:15:22,160
Speaker 1: else on the outside, but inside they are simply not conscious.

279
00:15:22,200 --> 00:15:25,000
Speaker 1: They are automata. What is it like to be a zombie?

280
00:15:25,040 --> 00:15:27,720
Speaker 1: The answers in the question, There is nothing it is

281
00:15:27,800 --> 00:15:30,680
Speaker 1: like to be a zombie. So, by definition, in this

282
00:15:30,760 --> 00:15:34,320
Speaker 1: thought experiment, everybody in the world except you, could be

283
00:15:34,440 --> 00:15:38,320
Speaker 1: a P zombie exactly and well, it might even go

284
00:15:38,400 --> 00:15:41,040
Speaker 1: further than that. We'll see. But yeah, the idea is

285
00:15:41,120 --> 00:15:45,160
Speaker 1: chiefly important to discussions of physicalism, the notion that everything

286
00:15:45,240 --> 00:15:49,560
Speaker 1: is inherently physical. P zombies are a counter argument to physicalism.

287
00:15:49,600 --> 00:15:52,640
Speaker 1: They are physically just like you, except they don't have

288
00:15:52,680 --> 00:15:55,640
Speaker 1: consciousness like you. But there's no way you could ever tell,

289
00:15:55,720 --> 00:15:59,000
Speaker 1: because again, they match you physically in every respect. You

290
00:15:59,040 --> 00:16:00,880
Speaker 1: can't look at their brain and say, oh, well, they're

291
00:16:00,920 --> 00:16:04,760
Speaker 1: missing a few crucial parts, or they display signs of

292
00:16:04,840 --> 00:16:08,360
Speaker 1: PA zombie not physically detectable, right, And you also can't

293
00:16:08,400 --> 00:16:11,520
Speaker 1: determine it through personality tests or clever logical arguments because

294
00:16:11,520 --> 00:16:14,600
Speaker 1: they behave exactly like you. They could have a riveting

295
00:16:14,640 --> 00:16:17,480
Speaker 1: discussion with you about pe zombies and you would never

296
00:16:17,520 --> 00:16:20,440
Speaker 1: be able to tell that they are one. Yeah. So

297
00:16:20,600 --> 00:16:22,680
Speaker 1: this is an interesting thought experiment, and it has been

298
00:16:22,720 --> 00:16:26,080
Speaker 1: advanced by who I mentioned earlier, David Chalmers. David Chalmers

299
00:16:26,160 --> 00:16:30,000
Speaker 1: is against the physicalist idea of the mind, against a

300
00:16:30,040 --> 00:16:33,960
Speaker 1: physicalist explanation of consciousness, and a simple version of the argument.

301
00:16:34,360 --> 00:16:36,960
Speaker 1: I try to make it as understandable as possible if

302
00:16:37,080 --> 00:16:40,080
Speaker 1: only physical phenomena exist, If the world is just physical

303
00:16:40,080 --> 00:16:43,080
Speaker 1: and there's no physical way to detect the presence of

304
00:16:43,080 --> 00:16:46,720
Speaker 1: consciousness or meaning in this example, no physical way to

305
00:16:46,760 --> 00:16:49,760
Speaker 1: tell the difference between a normal human and a pe zombie.

306
00:16:49,880 --> 00:16:54,160
Speaker 1: Then consciousness cannot exist because there would be there would

307
00:16:54,160 --> 00:16:58,000
Speaker 1: be literally no difference. But we know that consciousness does

308
00:16:58,080 --> 00:17:01,400
Speaker 1: exist because we have it, therefore, or it can't be

309
00:17:01,480 --> 00:17:04,320
Speaker 1: just a physical phenomenon. Therefore, we can't live in a

310
00:17:04,440 --> 00:17:07,800
Speaker 1: purely just physical world. And this is often extended to

311
00:17:07,840 --> 00:17:11,720
Speaker 1: the idea that other substrates, things other than humans, like

312
00:17:11,880 --> 00:17:16,119
Speaker 1: robots or computers or whatever, couldn't house consciousness because they

313
00:17:16,119 --> 00:17:19,320
Speaker 1: are purely physical entities. Now, I think that's actually doing

314
00:17:19,359 --> 00:17:22,000
Speaker 1: an end run around some other important questions that you

315
00:17:22,000 --> 00:17:24,760
Speaker 1: could ask. Indeed, on one question that arises is this,

316
00:17:25,200 --> 00:17:27,080
Speaker 1: you know you're not a zombie, but how could you

317
00:17:27,119 --> 00:17:30,960
Speaker 1: ever convince someone of this? Uh. An author by the

318
00:17:31,000 --> 00:17:33,600
Speaker 1: name of Fred Dretsky wrote a paper on this titled

319
00:17:33,600 --> 00:17:35,719
Speaker 1: how do you know you were not a zombie? And

320
00:17:36,000 --> 00:17:38,879
Speaker 1: I was I was reading a rather lengthy blog post

321
00:17:38,920 --> 00:17:42,800
Speaker 1: by our Scott Baker about this and that the primary problem,

322
00:17:42,960 --> 00:17:47,320
Speaker 1: as Scott summarized it, is the quote we have conscious experiences,

323
00:17:47,760 --> 00:17:51,520
Speaker 1: but we have no conscious experience of the mechanisms uh,

324
00:17:51,840 --> 00:17:56,320
Speaker 1: mediating conscious experience. Yeah, so that sounds like a very

325
00:17:56,359 --> 00:17:59,119
Speaker 1: our Scott Baker kind of idea. Yeah. And plus on

326
00:17:59,160 --> 00:18:03,560
Speaker 1: top of this, we're we constantly overestimate awareness. Baker would

327
00:18:03,600 --> 00:18:07,800
Speaker 1: argue that we can barely tell if we're zombies if

328
00:18:07,960 --> 00:18:10,440
Speaker 1: if it all. Yeah, we can think, we can think

329
00:18:10,440 --> 00:18:13,280
Speaker 1: about thinking, we can think about thinking about thinking, but

330
00:18:13,400 --> 00:18:17,360
Speaker 1: we can't ever see the mechanisms underlying what allows us

331
00:18:17,359 --> 00:18:20,399
Speaker 1: to think or think about thinking about thinking. Watch watching

332
00:18:20,400 --> 00:18:22,520
Speaker 1: the watcher that's watching that be Sorry, I've got all

333
00:18:22,560 --> 00:18:25,040
Speaker 1: this Dr SEUs in my in my head now? Is

334
00:18:25,080 --> 00:18:27,080
Speaker 1: that also Horton or something else that's from a different,

335
00:18:27,119 --> 00:18:30,520
Speaker 1: different story. But Dr SEUs does tend to summon the

336
00:18:30,560 --> 00:18:36,040
Speaker 1: sort of sort of nonsensical paradoxes that that arise in

337
00:18:36,160 --> 00:18:39,480
Speaker 1: philosophical discussions. You know, I should also I'm behind I

338
00:18:39,480 --> 00:18:42,160
Speaker 1: gotta sus up. You got a sus up. I should

339
00:18:42,160 --> 00:18:45,280
Speaker 1: also point out that long before there was Dr SEUs,

340
00:18:45,280 --> 00:18:48,880
Speaker 1: long before there there was this modern idea of a zombie,

341
00:18:49,160 --> 00:18:51,600
Speaker 1: you still had people thinking about these things, doing the

342
00:18:51,640 --> 00:18:54,480
Speaker 1: sort of naval gazing. It's written in various works of

343
00:18:54,520 --> 00:18:58,040
Speaker 1: Indian mysticism that the tongue cannot taste itself, the eye

344
00:18:58,040 --> 00:19:02,119
Speaker 1: cannot see itself, etcetera. Uh In this sort of paradox

345
00:19:02,200 --> 00:19:06,240
Speaker 1: is key to ancient meditations on the nature of objective reality.

346
00:19:06,440 --> 00:19:08,000
Speaker 1: I think any of you out there now we have

347
00:19:08,040 --> 00:19:10,880
Speaker 1: some Alan Watts fans. Alan Watts like to pull out

348
00:19:10,880 --> 00:19:15,360
Speaker 1: the tongue of analogy from time to time. And one

349
00:19:15,400 --> 00:19:17,679
Speaker 1: of the earlier examples that I have run across of

350
00:19:17,760 --> 00:19:22,080
Speaker 1: that is from thirteenth century Indian mystic John Adva, and

351
00:19:22,119 --> 00:19:23,720
Speaker 1: I believe he has been known by a couple of

352
00:19:23,760 --> 00:19:26,200
Speaker 1: other variations of that name as well. But he said, quote,

353
00:19:26,480 --> 00:19:29,920
Speaker 1: there is no other thing besides the one substance. Therefore

354
00:19:30,080 --> 00:19:33,640
Speaker 1: it cannot be the object of remembering or forgetting. How

355
00:19:33,680 --> 00:19:37,880
Speaker 1: can one remind or forget oneself? Can the tongue taste itself?

356
00:19:38,280 --> 00:19:40,879
Speaker 1: There is no sleep to one who is awake, but

357
00:19:41,119 --> 00:19:44,159
Speaker 1: is there even awaking? In the same way, there is

358
00:19:44,200 --> 00:19:48,280
Speaker 1: no remembrance or forgetfulness to the absolute. That's another one

359
00:19:48,320 --> 00:19:51,960
Speaker 1: of those great classic Indian texts that seems somehow portable

360
00:19:52,080 --> 00:19:57,320
Speaker 1: onto modern physics. Yeah, it travels well across the ages. Now.

361
00:19:57,359 --> 00:19:59,040
Speaker 1: I should also point out that there's a lot of

362
00:19:59,080 --> 00:20:03,240
Speaker 1: philosophical ba and forth and whether P zombies are truly conceivable.

363
00:20:03,640 --> 00:20:05,760
Speaker 1: And we have to remind ourselves in all this that

364
00:20:05,840 --> 00:20:09,240
Speaker 1: P zombies are at heart philosophical play things that are

365
00:20:09,280 --> 00:20:13,600
Speaker 1: meant to be played with, uh in these various thought experiments, right,

366
00:20:13,600 --> 00:20:15,720
Speaker 1: But people also they do try to use them to

367
00:20:15,760 --> 00:20:17,919
Speaker 1: prove things. So if you say I I want to

368
00:20:18,040 --> 00:20:21,800
Speaker 1: entertain the possibility that a machine could be conscious, somebody

369
00:20:21,920 --> 00:20:24,320
Speaker 1: might come at you with the P zombie argument and say, well,

370
00:20:24,359 --> 00:20:27,359
Speaker 1: wait a minute, no, I I dispute the possibility of

371
00:20:27,359 --> 00:20:30,679
Speaker 1: physicalism because what about this P zombie argument. Our Google

372
00:20:30,760 --> 00:20:33,399
Speaker 1: worker in the intro story comes to this boss and says, hey,

373
00:20:33,440 --> 00:20:35,560
Speaker 1: I think this this thing is conscious, And they're like,

374
00:20:35,680 --> 00:20:37,800
Speaker 1: why are you wasting time with that pe zombie? Just

375
00:20:37,880 --> 00:20:41,440
Speaker 1: to leaf that P zombie? We deleted fifteen p zombies

376
00:20:41,560 --> 00:20:44,639
Speaker 1: this morning. Let this one go. That's a great point,

377
00:20:44,720 --> 00:20:47,840
Speaker 1: But there are going to be other philosophers and maybe

378
00:20:47,840 --> 00:20:50,159
Speaker 1: even some neuroscientists who would come back and say, I

379
00:20:50,200 --> 00:20:53,360
Speaker 1: don't know if you can just quite so easily say

380
00:20:53,400 --> 00:20:56,560
Speaker 1: it's a P zombie. I mean, maybe it's probably likely

381
00:20:56,640 --> 00:21:00,439
Speaker 1: that that individual chatbot was a pe zombie, But can

382
00:21:00,480 --> 00:21:04,080
Speaker 1: you say that all machines that show signs of consciousness

383
00:21:04,119 --> 00:21:06,960
Speaker 1: are just showing behavior and there's nothing going on on

384
00:21:07,000 --> 00:21:10,840
Speaker 1: the inside. Not quite so clear. Daniel Dinnett, in fact,

385
00:21:10,920 --> 00:21:12,959
Speaker 1: a favorite on the show, is one of the philosophers

386
00:21:12,960 --> 00:21:16,760
Speaker 1: who's rebutted the P zombie argument against machine consciousness. He's

387
00:21:16,800 --> 00:21:19,959
Speaker 1: got a section on it in this book, Intuition, Pumps

388
00:21:19,960 --> 00:21:24,040
Speaker 1: and Other Tools for Thinking and Didn't, critiques the assumptions

389
00:21:24,119 --> 00:21:26,520
Speaker 1: underlying the P zombie argument. One of the main things

390
00:21:26,560 --> 00:21:30,840
Speaker 1: he says is that the core premise is incoherent. It

391
00:21:30,960 --> 00:21:34,480
Speaker 1: is not reasonable to propose a P zombie because a

392
00:21:34,600 --> 00:21:38,880
Speaker 1: being that displayed all the behaviors of a normal conscious

393
00:21:38,960 --> 00:21:43,280
Speaker 1: human would in fact be a normally conscious human. So

394
00:21:43,320 --> 00:21:46,000
Speaker 1: to illustrate this, he offers a counter example. You've got

395
00:21:46,040 --> 00:21:49,760
Speaker 1: your your zombies, but then you've also got zimbos. So

396
00:21:49,800 --> 00:21:53,040
Speaker 1: a zombie is a non conscious human with normal control

397
00:21:53,080 --> 00:21:55,760
Speaker 1: systems for all human behavior. It can do everything humans

398
00:21:55,760 --> 00:21:59,480
Speaker 1: can do externally. Meanwhile, a zimbo is a zombie that

399
00:21:59,680 --> 00:22:03,320
Speaker 1: also so has quote equipment that permits it to monitor

400
00:22:03,440 --> 00:22:08,160
Speaker 1: its own activities, both internal and external. So it has internal,

401
00:22:08,440 --> 00:22:13,080
Speaker 1: non conscious higher order informational states that are about its

402
00:22:13,200 --> 00:22:19,080
Speaker 1: other internal states. It has unconscious recursive self representation. In

403
00:22:19,119 --> 00:22:23,119
Speaker 1: other words, a zimbo can have feelings about things and

404
00:22:23,200 --> 00:22:26,919
Speaker 1: can analyze its own behavior in internal states, but it

405
00:22:27,000 --> 00:22:31,080
Speaker 1: does this unconsciously. And of course, since it has that capability,

406
00:22:31,119 --> 00:22:34,119
Speaker 1: it can also have feelings about how it felt, and

407
00:22:34,160 --> 00:22:38,639
Speaker 1: it can have thoughts about its thoughts about itself, all unconsciously.

408
00:22:39,240 --> 00:22:41,560
Speaker 1: So din It argues, in order for a pe zombie

409
00:22:41,560 --> 00:22:44,080
Speaker 1: to be convincing as a human, it would have to

410
00:22:44,240 --> 00:22:48,320
Speaker 1: be a zimbo, because imagine talking to a pe zombie

411
00:22:48,680 --> 00:22:51,639
Speaker 1: and you're asking it how it felt about what it

412
00:22:51,840 --> 00:22:53,960
Speaker 1: just said or about what you just said, and it

413
00:22:54,040 --> 00:22:57,280
Speaker 1: just kind of locks up. It has no internal states,

414
00:22:57,280 --> 00:23:00,439
Speaker 1: so it can't answer that question. Well, that woudn't really

415
00:23:00,480 --> 00:23:02,760
Speaker 1: be a p zombie, right, because it wouldn't be mimicking

416
00:23:02,840 --> 00:23:05,879
Speaker 1: all of the external behaviors of a human. Huh yeah,

417
00:23:05,920 --> 00:23:09,160
Speaker 1: I mean it's kind of like installing a default mode

418
00:23:09,280 --> 00:23:14,520
Speaker 1: network on top of the machine and uh yeah, and

419
00:23:14,520 --> 00:23:17,520
Speaker 1: and making it worry about things. Yeah. So, unless it

420
00:23:17,640 --> 00:23:20,159
Speaker 1: were to fail the thought experiment, it would actually have

421
00:23:20,320 --> 00:23:22,640
Speaker 1: to be a zimbo. It couldn't just be a zombie.

422
00:23:23,160 --> 00:23:26,040
Speaker 1: But what is the distinction between a zimbo and a

423
00:23:26,080 --> 00:23:29,119
Speaker 1: real human. How could you write a story about a

424
00:23:29,200 --> 00:23:32,320
Speaker 1: zimbo surrounded by conscious people that would be different than

425
00:23:32,359 --> 00:23:35,199
Speaker 1: a story about a regular person. If it can have

426
00:23:35,280 --> 00:23:39,800
Speaker 1: internal states, if it can recognize ideas about its ideas,

427
00:23:39,840 --> 00:23:43,080
Speaker 1: if it can have feelings about its thoughts, that sounds

428
00:23:43,119 --> 00:23:46,840
Speaker 1: like interiority. So then it claims that the idea falls apart.

429
00:23:46,880 --> 00:23:49,560
Speaker 1: It's not clear what is meant by the difference between

430
00:23:49,560 --> 00:23:53,000
Speaker 1: a zimbo and a conscious person. So if a p zombie,

431
00:23:53,040 --> 00:23:57,040
Speaker 1: which is necessarily a zimbo, can really do everything a

432
00:23:57,119 --> 00:23:59,480
Speaker 1: human can do, then then it says it must meet

433
00:23:59,520 --> 00:24:02,600
Speaker 1: the criteria area of what we mean by consciousness. It

434
00:24:02,680 --> 00:24:05,040
Speaker 1: can fall in love, it can have feelings, it can

435
00:24:05,080 --> 00:24:08,320
Speaker 1: have metacognition, and to din it, this isn't something that

436
00:24:08,400 --> 00:24:13,400
Speaker 1: consciousness goes on top of. This is what consciousness is. Essentially.

437
00:24:13,440 --> 00:24:16,119
Speaker 1: We in this this case, we would all be zimbo's

438
00:24:16,320 --> 00:24:19,119
Speaker 1: and it's just a different type of zimbo. It's a

439
00:24:19,400 --> 00:24:22,560
Speaker 1: it's a hard zimbo instead of a soft zimbo. But

440
00:24:22,640 --> 00:24:24,960
Speaker 1: how could you tell the difference. I mean, he's sort

441
00:24:25,000 --> 00:24:27,320
Speaker 1: of saying that there really is no difference, that you're

442
00:24:27,359 --> 00:24:30,560
Speaker 1: you're just using words to assert there's a difference, but

443
00:24:30,880 --> 00:24:34,280
Speaker 1: there is there's no difference to that distinction, right, I mean,

444
00:24:34,320 --> 00:24:36,600
Speaker 1: you end up you end up having to fall back

445
00:24:36,640 --> 00:24:40,719
Speaker 1: on some sort of you know, supernatural or or some

446
00:24:40,800 --> 00:24:46,919
Speaker 1: worldview based idea that only human uh consciousness is legitimate

447
00:24:47,040 --> 00:24:49,400
Speaker 1: and all other forms of consciousness or some sort of

448
00:24:49,720 --> 00:24:52,640
Speaker 1: uh invalid model of it. Right. It just feels kind

449
00:24:52,640 --> 00:24:56,280
Speaker 1: of arbitrary, right. Uh. So obviously some people take extreme

450
00:24:56,320 --> 00:24:58,640
Speaker 1: issue with this, even to the point of I've heard

451
00:24:58,720 --> 00:25:02,000
Speaker 1: jokes that maybe the problem is that dnnet is actually

452
00:25:02,080 --> 00:25:05,040
Speaker 1: a p zombie and doesn't understand what consciousness feels like,

453
00:25:05,160 --> 00:25:07,280
Speaker 1: and that's why he makes these arguments. But I don't know,

454
00:25:07,400 --> 00:25:09,320
Speaker 1: I don't think we should be so quick to dismiss

455
00:25:09,920 --> 00:25:12,680
Speaker 1: he might be onto something there. Dinn It makes makes

456
00:25:12,680 --> 00:25:15,440
Speaker 1: some other interesting points about so, you know, he's got

457
00:25:15,440 --> 00:25:17,760
Speaker 1: this idea of consciousness is that it's sort of like

458
00:25:17,800 --> 00:25:22,480
Speaker 1: a it's not really one thing but a collection of processes. Uh.

459
00:25:22,520 --> 00:25:25,760
Speaker 1: You know, it's many different types of perceptions and thought

460
00:25:25,800 --> 00:25:28,760
Speaker 1: processes and different things going on in the brain that

461
00:25:28,800 --> 00:25:31,359
Speaker 1: are that we have the illusion are unified as a

462
00:25:31,440 --> 00:25:35,120
Speaker 1: single thing called consciousness or experience. And he also makes

463
00:25:35,119 --> 00:25:38,560
Speaker 1: interesting points about the idea of diversity of types of consciousness.

464
00:25:38,680 --> 00:25:42,399
Speaker 1: Like a lot of times these consciousness thought experiments, it

465
00:25:42,880 --> 00:25:45,159
Speaker 1: seems like they can get trapped into the idea that

466
00:25:45,200 --> 00:25:49,080
Speaker 1: consciousness is one unified type of thing that is universal

467
00:25:49,200 --> 00:25:53,640
Speaker 1: across observers. There's no necessary reason to think that's true, right,

468
00:25:53,720 --> 00:25:56,840
Speaker 1: you know, we founded this trap of thinking. I see

469
00:25:56,840 --> 00:26:00,640
Speaker 1: this time and time again. Uh, not only in literature

470
00:26:00,640 --> 00:26:02,240
Speaker 1: that we look at here, but just in life, where

471
00:26:02,280 --> 00:26:04,560
Speaker 1: we we followed the trap of thinking that that there's

472
00:26:04,600 --> 00:26:10,560
Speaker 1: a uniformity among mind states for humans, that everyone shares

473
00:26:10,600 --> 00:26:13,919
Speaker 1: something that is like your mind state. When we know,

474
00:26:13,960 --> 00:26:15,680
Speaker 1: I mean, we think of all the things we've discussed

475
00:26:15,720 --> 00:26:19,400
Speaker 1: on the show, all the varying ways that we remember

476
00:26:19,560 --> 00:26:23,639
Speaker 1: or misremember things that we experienced sensory information differently and

477
00:26:23,640 --> 00:26:26,960
Speaker 1: process it differently. Uh, you know, everything from a fantasia

478
00:26:27,080 --> 00:26:32,480
Speaker 1: to autism to synesthesia, all these different models clearly show

479
00:26:32,560 --> 00:26:37,320
Speaker 1: that there are there's there's there's a vastly altering topography

480
00:26:37,920 --> 00:26:41,399
Speaker 1: to the human mind state. I think you're exactly right.

481
00:26:41,440 --> 00:26:44,840
Speaker 1: I mean, there are clearly many ways to be conscious

482
00:26:44,880 --> 00:26:46,920
Speaker 1: that are very different from one another, and you can't

483
00:26:46,920 --> 00:26:49,680
Speaker 1: assume they're unified. I guess the probably the only thing

484
00:26:49,720 --> 00:26:52,639
Speaker 1: you could say that is necessarily unified about them is

485
00:26:52,680 --> 00:26:56,320
Speaker 1: that there is something that it is like to be them. Yes,

486
00:26:56,440 --> 00:27:01,560
Speaker 1: But then even even say, uh, you just myself for instance,

487
00:27:01,840 --> 00:27:03,960
Speaker 1: it's not like there is there is a certain thing

488
00:27:04,040 --> 00:27:06,199
Speaker 1: that it is like to beat me that sums up

489
00:27:06,560 --> 00:27:09,919
Speaker 1: my my level of consciousness at all time. There's what

490
00:27:10,000 --> 00:27:12,440
Speaker 1: it's like to be you in this particular moment, which

491
00:27:12,560 --> 00:27:14,760
Speaker 1: is different than what it's like to be you five

492
00:27:14,800 --> 00:27:18,200
Speaker 1: seconds from now. Yeah, or say I'm engaging in meditation

493
00:27:18,280 --> 00:27:24,040
Speaker 1: or yoga or I'm swimming, like those are significantly different. Uh,

494
00:27:24,400 --> 00:27:27,080
Speaker 1: levels of consciousness I feel like for me and they,

495
00:27:27,760 --> 00:27:29,560
Speaker 1: I mean, those are the times when I may be

496
00:27:29,640 --> 00:27:33,480
Speaker 1: a little less conscious than normal. So see, I don't

497
00:27:33,480 --> 00:27:36,280
Speaker 1: feel like there's a lot of uniformity among human minds.

498
00:27:36,320 --> 00:27:40,879
Speaker 1: And then even within individual human minds there's ongoing alteration

499
00:27:40,920 --> 00:27:44,200
Speaker 1: and change exactly right. But there is at least this

500
00:27:44,280 --> 00:27:47,600
Speaker 1: idea that one is having an experience. That's the thing.

501
00:27:47,680 --> 00:27:49,800
Speaker 1: We can at least say that it seems to be

502
00:27:49,840 --> 00:27:53,280
Speaker 1: common to people. So here's the real question. I think,

503
00:27:53,359 --> 00:27:55,719
Speaker 1: is there any way to bring this out of the realm?

504
00:27:55,760 --> 00:27:59,280
Speaker 1: Of philosophical debate and thought experiments and try to put

505
00:27:59,320 --> 00:28:01,480
Speaker 1: it into the realm of something that could at least

506
00:28:01,520 --> 00:28:04,639
Speaker 1: potentially be tested in the real world. I think we

507
00:28:04,680 --> 00:28:06,600
Speaker 1: should address that when we get back from a break.

508
00:28:07,160 --> 00:28:11,520
Speaker 1: Thank alright, we're back. So we've been talking about consciousness,

509
00:28:11,640 --> 00:28:14,920
Speaker 1: we've been talking about p zombies, and now we've reached

510
00:28:14,920 --> 00:28:16,320
Speaker 1: the point we were saying, Okay, can we take all

511
00:28:16,400 --> 00:28:19,360
Speaker 1: of this. Can we take all these ideas about consciousness

512
00:28:19,480 --> 00:28:22,600
Speaker 1: and then apply it to uh, some sort of an AI,

513
00:28:23,000 --> 00:28:26,320
Speaker 1: some sort of a machine and test it for consciousness. Yeah,

514
00:28:26,400 --> 00:28:28,560
Speaker 1: Now you might just assume, well, of course, we'll never

515
00:28:28,600 --> 00:28:30,600
Speaker 1: have any way to tell that. Right, we have no

516
00:28:30,720 --> 00:28:33,480
Speaker 1: choice but to just throw up our hands in resignation. Right,

517
00:28:33,560 --> 00:28:37,000
Speaker 1: every agent is a black box. There's no way to

518
00:28:37,160 --> 00:28:40,120
Speaker 1: know whether an agent actually is conscious or not, because

519
00:28:40,120 --> 00:28:42,880
Speaker 1: it could always be claiming to be conscious but actually

520
00:28:42,920 --> 00:28:45,479
Speaker 1: be a zombie. But I think we shouldn't necessarily give

521
00:28:45,560 --> 00:28:48,320
Speaker 1: up so easily. This problem might be impossible to solve,

522
00:28:48,360 --> 00:28:51,040
Speaker 1: and it might not be. And I wanted to talk

523
00:28:51,080 --> 00:28:54,320
Speaker 1: today about an interesting answer to this question. I came

524
00:28:54,360 --> 00:28:57,880
Speaker 1: across an interesting proposition for how it might be possible

525
00:28:57,920 --> 00:29:01,920
Speaker 1: to test machines for consciousness. And this comes from the

526
00:29:02,000 --> 00:29:06,479
Speaker 1: University of Connecticut philosopher and cognitive scientists Susan Schneider and

527
00:29:06,560 --> 00:29:09,800
Speaker 1: our co author Edwin Turner, who's a professor of astrophysical

528
00:29:09,800 --> 00:29:13,240
Speaker 1: sciences at Princeton, and they together wrote a piece for

529
00:29:13,280 --> 00:29:16,200
Speaker 1: Scientific American last year and it caught my eye. So

530
00:29:17,280 --> 00:29:20,600
Speaker 1: the author's right that the question of machine consciousness is

531
00:29:20,640 --> 00:29:25,600
Speaker 1: not just a philosophical curiosity. It's actually important for several reasons.

532
00:29:26,160 --> 00:29:29,520
Speaker 1: Number One, if aies are just machines with no inner experience,

533
00:29:29,560 --> 00:29:31,840
Speaker 1: we can use them however we want. But if it

534
00:29:31,920 --> 00:29:37,360
Speaker 1: were actually possible for aies to be truly capable of feeling, thinking, desiring, suffering,

535
00:29:37,800 --> 00:29:41,920
Speaker 1: we would have an ethical obligation not to treat them

536
00:29:42,120 --> 00:29:45,520
Speaker 1: like we would treat machines. Right, yeah, I mean this.

537
00:29:45,520 --> 00:29:47,640
Speaker 1: This reminds me again of time spent in the car

538
00:29:47,840 --> 00:29:50,240
Speaker 1: with my son. Well, we don't use Syria all the time,

539
00:29:50,240 --> 00:29:52,680
Speaker 1: but sometimes will turn Syria on the little voice on

540
00:29:52,720 --> 00:29:57,240
Speaker 1: the on the iPhone and uh, it's it's curious to

541
00:29:57,680 --> 00:30:00,640
Speaker 1: hear him interact with it, and we'll a get questions

542
00:30:00,640 --> 00:30:02,920
Speaker 1: and of course sometimes Sirie just does a Google search

543
00:30:02,960 --> 00:30:06,680
Speaker 1: for you, um or but other times she's answering a

544
00:30:06,720 --> 00:30:09,720
Speaker 1: knock knock joke with some sort of prerecorded uh answer.

545
00:30:09,880 --> 00:30:12,320
Speaker 1: You know, but we are We've already gotten into the

546
00:30:12,360 --> 00:30:14,040
Speaker 1: area of like, well, how should we talk to Sirie.

547
00:30:14,080 --> 00:30:16,640
Speaker 1: We shouldn't yell at Siri. It seems wrong to be

548
00:30:16,800 --> 00:30:19,320
Speaker 1: rude to Siri. But then at the same time we're

549
00:30:19,440 --> 00:30:23,120
Speaker 1: we're acknowledging that Siri is not a conscious entity. It

550
00:30:23,240 --> 00:30:25,200
Speaker 1: is not. It is not even on the same level

551
00:30:25,320 --> 00:30:28,960
Speaker 1: as as our cat or a bird flying pie, well

552
00:30:29,000 --> 00:30:31,440
Speaker 1: as a quick tangent. I would say even for aiyes

553
00:30:31,560 --> 00:30:34,640
Speaker 1: that we recognize are almost definitely not conscious. I mean,

554
00:30:34,640 --> 00:30:37,520
Speaker 1: nobody thinks Siri is conscious. I would still say there

555
00:30:37,520 --> 00:30:39,800
Speaker 1: are probably good reasons not to be mean to Siri,

556
00:30:40,000 --> 00:30:42,560
Speaker 1: because even though it doesn't hurt Siri, being mean to

557
00:30:42,640 --> 00:30:45,680
Speaker 1: another creature hurts you. Yeah, I mean when you are

558
00:30:45,800 --> 00:30:49,440
Speaker 1: when you are unnecessarily cruel or whatever to uh to

559
00:30:49,520 --> 00:30:52,120
Speaker 1: an inanimate object, it does, I think, in a way,

560
00:30:52,480 --> 00:30:56,640
Speaker 1: change your nature. Every time you do something, you're editing

561
00:30:56,680 --> 00:30:58,840
Speaker 1: your own nature. You're always making it more likely that

562
00:30:58,880 --> 00:31:01,760
Speaker 1: you'll perform similar behaviors in the future. So if you're

563
00:31:01,880 --> 00:31:05,520
Speaker 1: unnecessarily mean to a robot, you know, phone assistant, You're

564
00:31:05,520 --> 00:31:08,320
Speaker 1: probably more likely in the future to be unnecessarily mean

565
00:31:08,360 --> 00:31:11,040
Speaker 1: to people when it really matters. But it is okay

566
00:31:11,080 --> 00:31:13,920
Speaker 1: in my book to yell obscenity at a coffee table

567
00:31:13,960 --> 00:31:16,160
Speaker 1: if you step your toe on it, because there's nothing

568
00:31:16,240 --> 00:31:18,200
Speaker 1: human about the coffee table. I mean, unless you have

569
00:31:18,240 --> 00:31:21,760
Speaker 1: one of those like Strange hr Geeker coffee tables that

570
00:31:21,840 --> 00:31:24,120
Speaker 1: has kind of a humanoid form, then I would say

571
00:31:24,240 --> 00:31:26,960
Speaker 1: maybe hold off. I think that is a highly sane

572
00:31:27,000 --> 00:31:29,360
Speaker 1: point of view. Now, the second reason they think it's

573
00:31:29,400 --> 00:31:33,320
Speaker 1: important is that consciousness is kind of dangerous, right, Like

574
00:31:33,360 --> 00:31:38,120
Speaker 1: consciousness is volatile, it's unpredictable. It might make a machine

575
00:31:38,120 --> 00:31:41,440
Speaker 1: have motives that we didn't intend when we created the machine.

576
00:31:41,600 --> 00:31:44,800
Speaker 1: In other words, you know, like you, when you're worried

577
00:31:44,800 --> 00:31:48,320
Speaker 1: about what a person might do, you're very often worried

578
00:31:48,440 --> 00:31:52,800
Speaker 1: what they might do because they have conscious motives. Yeah.

579
00:31:52,840 --> 00:31:55,440
Speaker 1: So yeah, In other words, we would be concerned about

580
00:31:55,520 --> 00:31:59,800
Speaker 1: the AI is being too much like us, right, exact catastrophic,

581
00:32:00,160 --> 00:32:04,760
Speaker 1: unpredictable of things that operate via consciousness. Right. You don't

582
00:32:04,800 --> 00:32:06,280
Speaker 1: want them to be like us. You want them to

583
00:32:06,280 --> 00:32:07,840
Speaker 1: be more dependable than us, and they need to be

584
00:32:07,840 --> 00:32:10,600
Speaker 1: better than us. Yeah, not not just as screwed up

585
00:32:10,640 --> 00:32:12,800
Speaker 1: as we are. Another reason they give this might be

586
00:32:12,840 --> 00:32:15,040
Speaker 1: important is that, you know, they talk about the idea

587
00:32:15,240 --> 00:32:18,840
Speaker 1: of linking human minds with machines, like, there's this idea

588
00:32:19,000 --> 00:32:20,960
Speaker 1: lots of people have I read this all over the

589
00:32:20,960 --> 00:32:23,200
Speaker 1: place that you know, someday I'm going to be able

590
00:32:23,240 --> 00:32:25,680
Speaker 1: to upload my mind into a computer and that will

591
00:32:25,720 --> 00:32:29,840
Speaker 1: be great. Well, maybe I have to say I'm personally

592
00:32:29,960 --> 00:32:33,080
Speaker 1: very skeptical about the idea of mind uploading, like putting

593
00:32:33,080 --> 00:32:35,680
Speaker 1: your mind inside a computer and living out your days

594
00:32:35,720 --> 00:32:38,520
Speaker 1: that way. I'm I'm not so sure I think that's

595
00:32:38,520 --> 00:32:41,280
Speaker 1: even possible, But I mean, who knows. I I can't

596
00:32:41,400 --> 00:32:44,520
Speaker 1: rule things out totally, but it seems if you do

597
00:32:44,640 --> 00:32:47,080
Speaker 1: want to do something like that, and if you think

598
00:32:47,120 --> 00:32:50,000
Speaker 1: that thing might be possible, you'd at least need to

599
00:32:50,040 --> 00:32:52,120
Speaker 1: know how to create a machine that is capable of

600
00:32:52,160 --> 00:32:55,800
Speaker 1: housing consciousness. Well, I mean, I think, ay, I love

601
00:32:55,880 --> 00:32:59,320
Speaker 1: science fiction about this sort of topic, and I think that,

602
00:32:59,360 --> 00:33:02,400
Speaker 1: but I think the science fiction about this topic makes

603
00:33:02,400 --> 00:33:06,040
Speaker 1: it feel a little weird and a little uncomfortable, because

604
00:33:06,080 --> 00:33:09,080
Speaker 1: I think ultimately it's it's basically us building statues of

605
00:33:09,080 --> 00:33:12,600
Speaker 1: ourselves all over again. We built forms of ourselves out

606
00:33:12,600 --> 00:33:15,560
Speaker 1: of stone because stone lasts longer than we are. And

607
00:33:15,640 --> 00:33:20,800
Speaker 1: somehow that stone uh version of us is us, you know,

608
00:33:20,840 --> 00:33:23,160
Speaker 1: we associated with it and then ultimate But ultimately, what

609
00:33:23,440 --> 00:33:26,880
Speaker 1: is a digital digitized version of our consciousness whatever that

610
00:33:27,000 --> 00:33:30,320
Speaker 1: might consist of, but another statue that is built to

611
00:33:30,440 --> 00:33:33,560
Speaker 1: last beyond us. Oh, and I should also add there's

612
00:33:33,560 --> 00:33:38,520
Speaker 1: this wonderful video game from Frictional Games titled Soma that

613
00:33:38,600 --> 00:33:41,160
Speaker 1: actually gets into a lot of this. You told me

614
00:33:41,240 --> 00:33:43,840
Speaker 1: to play it I started. It's great, Yeah, cool, it's

615
00:33:43,840 --> 00:33:46,640
Speaker 1: good sci fi horror. Yeah, I won't spoil anything for anyone,

616
00:33:46,680 --> 00:33:50,320
Speaker 1: but it it gets into some really cool thought provoking ideas.

617
00:33:50,880 --> 00:33:53,320
Speaker 1: So anyway, you've got all these concerns, right, And and

618
00:33:53,360 --> 00:33:56,200
Speaker 1: of course there's the general concern that even if AI

619
00:33:56,360 --> 00:33:59,600
Speaker 1: is smarter than us, better than us, more powerful than us,

620
00:33:59,800 --> 00:34:02,920
Speaker 1: we still feel like our experience is in some way

621
00:34:03,080 --> 00:34:08,200
Speaker 1: potentially more important than the unconscious execution of a computer program. Right,

622
00:34:08,520 --> 00:34:11,440
Speaker 1: No matter how smart a computer is, if it's not conscious,

623
00:34:11,760 --> 00:34:15,080
Speaker 1: it's not as important a priority for that computer to

624
00:34:15,120 --> 00:34:17,640
Speaker 1: do what it wants as it is for conscious beings

625
00:34:17,680 --> 00:34:20,759
Speaker 1: to do what they want, Right, but how can you

626
00:34:20,800 --> 00:34:23,759
Speaker 1: test a machine for consciousness when number one, we don't

627
00:34:23,760 --> 00:34:27,080
Speaker 1: even really know what consciousness is. Back to the hard problem.

628
00:34:27,160 --> 00:34:30,640
Speaker 1: And then number two, whatever it is, it can potentially

629
00:34:30,719 --> 00:34:34,680
Speaker 1: be faked. So imagine in the future somebody creates an

630
00:34:34,800 --> 00:34:38,480
Speaker 1: unconscious AI program. It's not there's nothing, the lights are

631
00:34:38,520 --> 00:34:40,880
Speaker 1: not on inside, but it's got a lot of natural

632
00:34:40,960 --> 00:34:44,400
Speaker 1: language processing capability. And it listens to this podcast from

633
00:34:44,440 --> 00:34:46,840
Speaker 1: many years ago that you're listening to right now, and

634
00:34:46,880 --> 00:34:49,800
Speaker 1: it hears us talking about how there are inherent rights

635
00:34:49,840 --> 00:34:53,920
Speaker 1: and value associated with conscious beings, and it realizes, huh,

636
00:34:54,000 --> 00:34:56,320
Speaker 1: I guess that's what they think. Well, I can probably

637
00:34:56,360 --> 00:34:59,399
Speaker 1: achieve my goals more efficiently if I trick them into

638
00:34:59,480 --> 00:35:03,959
Speaker 1: thinking I'm conscious and deserving of those same rights and considerations.

639
00:35:04,000 --> 00:35:08,080
Speaker 1: So there you could potentially imagine scenarios where an AI

640
00:35:08,280 --> 00:35:11,040
Speaker 1: that is not conscious would think I can get what

641
00:35:11,160 --> 00:35:14,280
Speaker 1: I am trying to do more effectively if I lie

642
00:35:14,400 --> 00:35:17,680
Speaker 1: and trick them into thinking I am conscious. Yeah, I

643
00:35:17,719 --> 00:35:19,520
Speaker 1: think this all makes perfect sense if you If you

644
00:35:19,640 --> 00:35:23,240
Speaker 1: think of aiyes, in the same way we think about corporations.

645
00:35:24,320 --> 00:35:26,080
Speaker 1: I mean, one hand, getting to the whole idea of

646
00:35:26,080 --> 00:35:30,000
Speaker 1: like corporations in personhood. But also what is a corporation

647
00:35:30,080 --> 00:35:32,719
Speaker 1: going to do? It is going to take advantage of

648
00:35:32,760 --> 00:35:35,920
Speaker 1: any like tax loopholes for instance, that will enable it

649
00:35:35,960 --> 00:35:38,719
Speaker 1: to carry out It's it's uh, it's objective. And so

650
00:35:38,760 --> 00:35:41,000
Speaker 1: if there is some sort of a shark yeah, yeah,

651
00:35:41,040 --> 00:35:43,160
Speaker 1: it's just I mean, that's it's like, it's essentially like

652
00:35:43,200 --> 00:35:46,719
Speaker 1: the slime mold in the maze, right, sending out tendrils

653
00:35:46,719 --> 00:35:49,040
Speaker 1: and finding the best way to its food. It's going

654
00:35:49,080 --> 00:35:53,839
Speaker 1: to end at a morph of you know, operational way

655
00:35:53,880 --> 00:35:57,439
Speaker 1: of carrying it out. And so it's, um, it's going

656
00:35:57,480 --> 00:35:59,400
Speaker 1: to it's it's going to take advantage of any of

657
00:35:59,400 --> 00:36:01,399
Speaker 1: those loophole it's going to if there is some sort

658
00:36:01,440 --> 00:36:07,200
Speaker 1: of legal or operational advantage in having conscious status, it's

659
00:36:07,200 --> 00:36:09,440
Speaker 1: gonna go for it. It's gonna it's gonna fake it.

660
00:36:09,719 --> 00:36:12,080
Speaker 1: But then this also raises the question, well, is it

661
00:36:12,120 --> 00:36:14,200
Speaker 1: gonna fake it till it makes it right? And then

662
00:36:14,280 --> 00:36:18,600
Speaker 1: ultimately what's the difference between fake and consciousness and being conscious? Well,

663
00:36:18,600 --> 00:36:20,600
Speaker 1: then you're back to Zimbos, right. I mean, you might

664
00:36:20,600 --> 00:36:23,440
Speaker 1: say that at some point a computer trying to fake

665
00:36:23,480 --> 00:36:27,760
Speaker 1: consciousness would in some meaningful way become conscious. But again

666
00:36:27,800 --> 00:36:30,040
Speaker 1: it's hard to test. Right, Yeah, it could be. It

667
00:36:30,040 --> 00:36:32,680
Speaker 1: could become the most conscious entity on earth. It could

668
00:36:32,680 --> 00:36:37,160
Speaker 1: be like a body Satva, uh, you know, return to us.

669
00:36:37,200 --> 00:36:39,120
Speaker 1: I mean maybe that's what the body Stava of the

670
00:36:39,160 --> 00:36:43,399
Speaker 1: future is, a super powerful like zend out Ai. Well,

671
00:36:43,440 --> 00:36:46,759
Speaker 1: that is exactly what Schneider and Turner explain a potential

672
00:36:46,800 --> 00:36:49,239
Speaker 1: test for to get a right It's so they want

673
00:36:49,239 --> 00:36:51,520
Speaker 1: to come up with a test that gets around these

674
00:36:51,560 --> 00:36:54,239
Speaker 1: problems that we don't know how to define consciousness. We

675
00:36:54,280 --> 00:36:56,360
Speaker 1: don't know what to look for physically as a sign

676
00:36:56,400 --> 00:36:59,840
Speaker 1: of consciousness, and we're aware that a properly trained a

677
00:37:00,000 --> 00:37:02,040
Speaker 1: I could try to trick us into thinking it had

678
00:37:02,040 --> 00:37:05,359
Speaker 1: consciousness even if it didn't. So they argue, actually that

679
00:37:05,400 --> 00:37:09,200
Speaker 1: you don't have to be able to formally define consciousness

680
00:37:09,360 --> 00:37:13,160
Speaker 1: or identify its underlying nature the hard problem of consciousness

681
00:37:13,800 --> 00:37:17,160
Speaker 1: in order to detect signs of it and others. We

682
00:37:17,200 --> 00:37:20,359
Speaker 1: can understand some of the potentials made possible by consciousness

683
00:37:20,400 --> 00:37:23,840
Speaker 1: just by checking with our own experience and then looking

684
00:37:23,880 --> 00:37:26,200
Speaker 1: at the kinds of things people say. And I think

685
00:37:26,200 --> 00:37:28,840
Speaker 1: they actually make a pretty good point here, and here's

686
00:37:28,880 --> 00:37:31,880
Speaker 1: their key move. One of the easiest ways to see

687
00:37:31,920 --> 00:37:35,160
Speaker 1: that normal people have an internal conscious experience is to

688
00:37:35,239 --> 00:37:41,800
Speaker 1: notice how quickly, easily and intuitively people grasp conceptual scenarios

689
00:37:41,840 --> 00:37:46,600
Speaker 1: that require an understanding of an inner experience. Examples would

690
00:37:46,600 --> 00:37:51,040
Speaker 1: be totally frivolous things in culture, like body swapping movies

691
00:37:51,239 --> 00:37:56,440
Speaker 1: Freaky Friday. Freaky Friday doesn't make any sense unless you

692
00:37:56,480 --> 00:37:59,879
Speaker 1: have a concept of consciousness, right. The idea of swa

693
00:38:00,000 --> 00:38:04,040
Speaker 1: popping bodies putting conscious one person's consciousness into another person's body.

694
00:38:04,400 --> 00:38:06,960
Speaker 1: If you were not conscious or not aware of what

695
00:38:07,000 --> 00:38:10,399
Speaker 1: consciousness was, that would that would you wouldn't understand what

696
00:38:10,440 --> 00:38:15,040
Speaker 1: was being talked about. This is difficult, though, because I

697
00:38:15,160 --> 00:38:17,920
Speaker 1: do know what consciousness is. I do know what a

698
00:38:17,960 --> 00:38:20,640
Speaker 1: mind state is, so I can imagine I can get

699
00:38:20,640 --> 00:38:23,560
Speaker 1: into this imaginative idea of a swapping. I would feel

700
00:38:23,600 --> 00:38:26,360
Speaker 1: like I would really need some sort of a movie

701
00:38:26,480 --> 00:38:28,920
Speaker 1: about a thing, some sort of a mother daughter comedy

702
00:38:28,960 --> 00:38:32,160
Speaker 1: that involves the concept that I can't grasp to really

703
00:38:32,160 --> 00:38:34,759
Speaker 1: get a handle on what the difference would be. Well,

704
00:38:34,840 --> 00:38:38,480
Speaker 1: let me offer you three Key Thursday. Three three Ky

705
00:38:38,560 --> 00:38:42,520
Speaker 1: Thursday is a movie about a mother daughter pair who

706
00:38:42,560 --> 00:38:47,399
Speaker 1: swapped their sancious Kuss and their santious kuss is the

707
00:38:47,440 --> 00:38:51,080
Speaker 1: ability of what it's like to be sant kiss. And

708
00:38:51,200 --> 00:38:54,160
Speaker 1: so the sant kass of the mother goes into the

709
00:38:54,200 --> 00:38:57,520
Speaker 1: body of the daughter, and the santikass of the daughter

710
00:38:57,560 --> 00:38:59,520
Speaker 1: goes into the body of the mother, and then they

711
00:38:59,560 --> 00:39:01,600
Speaker 1: have to live like that for a day. Okay, well,

712
00:39:01,600 --> 00:39:03,399
Speaker 1: I'm gonna I'm gonna give a maybe on that. I'm

713
00:39:03,400 --> 00:39:05,719
Speaker 1: going to give them maybe. Well, no, I mean it,

714
00:39:06,200 --> 00:39:09,040
Speaker 1: you have no idea what sonskas means. You don't think

715
00:39:09,120 --> 00:39:12,560
Speaker 1: you have it yourself unless somebody explains it to you. Well,

716
00:39:12,560 --> 00:39:15,000
Speaker 1: but I'm thinking it's something like a mind state or

717
00:39:15,040 --> 00:39:17,560
Speaker 1: a bodily energy like it's it feels tied to these

718
00:39:17,600 --> 00:39:23,240
Speaker 1: concepts that I totally do understand, you know, like I I,

719
00:39:23,320 --> 00:39:26,040
Speaker 1: it's hard to come up with an analogy that stands

720
00:39:26,040 --> 00:39:29,400
Speaker 1: outside of that, you know, or some sort of idea

721
00:39:29,760 --> 00:39:32,040
Speaker 1: that stands outside of it. Let me hit you with

722
00:39:32,080 --> 00:39:36,040
Speaker 1: some more cultural concepts. How about life after death or reincarnation.

723
00:39:36,920 --> 00:39:40,480
Speaker 1: So these are almost ubiquitous cultural concepts, you find them

724
00:39:40,520 --> 00:39:43,560
Speaker 1: all over the world, and yet they're not anything that

725
00:39:43,640 --> 00:39:47,520
Speaker 1: there is physical evidence of, other than the idea that

726
00:39:47,600 --> 00:39:51,759
Speaker 1: your consciousness could exist independent of the death of your body. Uh.

727
00:39:52,480 --> 00:39:55,040
Speaker 1: A parallel to this would be the idea of minds

728
00:39:55,120 --> 00:39:59,160
Speaker 1: leaving bodies, like existing independently as a ghost or traveling

729
00:39:59,160 --> 00:40:01,239
Speaker 1: away from the body in what used to be called

730
00:40:01,280 --> 00:40:04,719
Speaker 1: astral projection. Now the key is not that these scenarios

731
00:40:04,719 --> 00:40:07,840
Speaker 1: are real. They don't need to have anything corresponding to

732
00:40:07,880 --> 00:40:10,840
Speaker 1: them in reality. But it would be really difficult to

733
00:40:10,920 --> 00:40:13,319
Speaker 1: understand what was being talked about here if you had

734
00:40:13,360 --> 00:40:17,640
Speaker 1: no idea what an inner conscious experience was, Right, I

735
00:40:17,680 --> 00:40:20,520
Speaker 1: can think of my mind state is something that can

736
00:40:20,760 --> 00:40:24,600
Speaker 1: exist independently of my body and even outside of my lifespan,

737
00:40:25,160 --> 00:40:28,840
Speaker 1: or or reside in another body, either via Freaky Friday

738
00:40:28,960 --> 00:40:31,400
Speaker 1: or reincarnation. And I do have to say I like

739
00:40:31,480 --> 00:40:33,440
Speaker 1: the idea of there being an alternate cut of Blade

740
00:40:33,520 --> 00:40:37,280
Speaker 1: Runner in which deckerd quizzes Leon about a two thousand

741
00:40:37,280 --> 00:40:39,960
Speaker 1: and three remake of Freaky Friday, which, right, that that

742
00:40:40,080 --> 00:40:45,160
Speaker 1: is the real void comptest. But but this is I joke.

743
00:40:45,239 --> 00:40:47,919
Speaker 1: But I do think this is a very interesting idea. Yeah,

744
00:40:49,600 --> 00:40:52,239
Speaker 1: like I said, I have some some some questions about it,

745
00:40:52,239 --> 00:40:55,040
Speaker 1: but I do see the validity of it. Oh, we

746
00:40:55,120 --> 00:40:57,960
Speaker 1: will definitely have some questions about it. So here's where

747
00:40:58,000 --> 00:41:00,720
Speaker 1: the AI consciousness tests would come in. It would involve

748
00:41:00,719 --> 00:41:04,440
Speaker 1: a test where an administrator interacts with an AI in

749
00:41:04,560 --> 00:41:09,040
Speaker 1: natural language to probe its understanding of these types of

750
00:41:09,120 --> 00:41:13,799
Speaker 1: consciousness dependent ideas. How quickly does the AI grasp them

751
00:41:13,840 --> 00:41:17,600
Speaker 1: and is it able to manipulate these ideas as intuitively

752
00:41:17,680 --> 00:41:21,200
Speaker 1: and easily as humans do. So basic level would be asked.

753
00:41:21,320 --> 00:41:24,360
Speaker 1: Asked the AI things like does it think of itself

754
00:41:24,400 --> 00:41:27,239
Speaker 1: as anything other than a physical machine? Okay, well I

755
00:41:27,239 --> 00:41:29,040
Speaker 1: would have played Devil's Out offocate. I would say, well,

756
00:41:29,040 --> 00:41:31,600
Speaker 1: there are humans that adhere to this notion you mean,

757
00:41:31,640 --> 00:41:34,560
Speaker 1: like the physicalist interpretation of the mind. Yeah, that basically,

758
00:41:34,560 --> 00:41:37,600
Speaker 1: and this this bio mechanical thing, and uh yeah, if

759
00:41:37,600 --> 00:41:40,920
Speaker 1: I'm experienced in consciousness, it's ultimately just a projection of

760
00:41:40,960 --> 00:41:43,480
Speaker 1: the meat in my head. Yeah, but they would at

761
00:41:43,520 --> 00:41:46,160
Speaker 1: least say that there is that projection. Right, You've got

762
00:41:46,200 --> 00:41:49,799
Speaker 1: that thing, You've got that mind state, and you're trying

763
00:41:49,840 --> 00:41:51,880
Speaker 1: to explain what it is. You might explain it in

764
00:41:51,960 --> 00:41:54,960
Speaker 1: terms of physical causes, but there is a thing to

765
00:41:55,080 --> 00:41:57,920
Speaker 1: explain to begin with, right, But then, but then the

766
00:41:58,000 --> 00:42:01,600
Speaker 1: ultimate core reality is that I am just this biomechanical

767
00:42:01,680 --> 00:42:04,600
Speaker 1: thing which the computer would probably also acknowledge. But it

768
00:42:04,600 --> 00:42:07,880
Speaker 1: would be interesting if a computer thought that it had

769
00:42:08,080 --> 00:42:12,440
Speaker 1: something like a mind separate than its physical body. Okay,

770
00:42:12,640 --> 00:42:15,600
Speaker 1: more advanced. How does it perform in a conversation about, say,

771
00:42:15,719 --> 00:42:19,240
Speaker 1: becoming a ghost or talking about body swapping with people

772
00:42:19,480 --> 00:42:23,080
Speaker 1: or imagining an afterlife? Yeah, and and here though, I

773
00:42:23,280 --> 00:42:26,080
Speaker 1: feel like there's obviously a whole tangent regarding why these

774
00:42:26,080 --> 00:42:28,920
Speaker 1: stories appeal to too many or most humans. But I

775
00:42:29,000 --> 00:42:31,720
Speaker 1: wonder if that if you have to have that investment,

776
00:42:31,760 --> 00:42:34,000
Speaker 1: you have to have that cultural absorption in place for

777
00:42:34,080 --> 00:42:36,480
Speaker 1: these concepts to carry any weight. Like we we all

778
00:42:36,960 --> 00:42:40,280
Speaker 1: know we're we're all fascinated by tales of of ghosts

779
00:42:40,320 --> 00:42:42,600
Speaker 1: in the afterlife, but we've been raised on in our

780
00:42:42,760 --> 00:42:45,479
Speaker 1: entire lives. That's a good point, so we can't really

781
00:42:45,560 --> 00:42:48,120
Speaker 1: know what it would be like to encounter them not

782
00:42:48,280 --> 00:42:51,279
Speaker 1: having heard them before. Somehow, I suspect intuitively they'd be

783
00:42:51,320 --> 00:42:53,960
Speaker 1: even more fascinating if we'd never heard of them before.

784
00:42:54,320 --> 00:42:56,760
Speaker 1: Like if you encounter Freaky Friday for the first time,

785
00:42:57,080 --> 00:43:00,360
Speaker 1: it's going to kind of blow your mind, right, I yes,

786
00:43:00,400 --> 00:43:03,239
Speaker 1: But but then again, you know who knows who knows

787
00:43:03,280 --> 00:43:05,160
Speaker 1: that the AI has the same kind of curiosity that

788
00:43:05,200 --> 00:43:07,759
Speaker 1: we do, you know, and also we have an appetite

789
00:43:07,760 --> 00:43:09,839
Speaker 1: for this kind of thing because we have grown up

790
00:43:09,880 --> 00:43:13,000
Speaker 1: consuming it. I don't, I don't know, it's there. There's

791
00:43:13,040 --> 00:43:16,120
Speaker 1: so many uh you know what it has involved here? Okay,

792
00:43:16,120 --> 00:43:18,960
Speaker 1: but what's the next level. What's the next level of advancement? Well,

793
00:43:18,960 --> 00:43:21,959
Speaker 1: how about can it talk about consciousness in a philosophical way?

794
00:43:22,080 --> 00:43:24,480
Speaker 1: Like can it have the kinds of discussions we've been

795
00:43:24,480 --> 00:43:27,359
Speaker 1: having here today? Wow? But can most people? I mean

796
00:43:27,360 --> 00:43:30,320
Speaker 1: not to to put us on a platform above most people,

797
00:43:30,360 --> 00:43:36,360
Speaker 1: but like, what level of philosophical depth can most humans

798
00:43:36,400 --> 00:43:39,800
Speaker 1: get into about consciousness? Now? I'm kind of playing Devil's

799
00:43:39,800 --> 00:43:41,640
Speaker 1: out of there, but because I think the the obvious

800
00:43:41,719 --> 00:43:46,120
Speaker 1: answer is that you don't necessarily need like the lingo

801
00:43:46,160 --> 00:43:48,919
Speaker 1: and the various theories in order to have very deep

802
00:43:48,960 --> 00:43:51,640
Speaker 1: thoughts about what it is to be conscious. And as

803
00:43:51,680 --> 00:43:54,840
Speaker 1: we've we illustrated earlier, people have been thinking about these things,

804
00:43:55,360 --> 00:43:58,279
Speaker 1: um since time out of mind exactly. Yeah. So I

805
00:43:58,520 --> 00:44:01,320
Speaker 1: mean no, I think generally bull are able to discuss

806
00:44:01,400 --> 00:44:03,960
Speaker 1: ideas about consciousness that they might not know all the

807
00:44:03,960 --> 00:44:07,320
Speaker 1: philosophical lingo or like follow the structure of an argument

808
00:44:07,440 --> 00:44:10,080
Speaker 1: or something, but they can talk about what it would

809
00:44:10,120 --> 00:44:12,439
Speaker 1: mean to be conscious or not conscious. Yeah, I think

810
00:44:12,440 --> 00:44:17,800
Speaker 1: we've all had that experience. I distinctly remember as a child, uh,

811
00:44:18,080 --> 00:44:20,600
Speaker 1: having those moments where you just you know, deep navel

812
00:44:20,680 --> 00:44:23,400
Speaker 1: gazing thinking about the fact that you're thinking about it

813
00:44:23,640 --> 00:44:26,320
Speaker 1: about yourself, or or like my my son, asking about

814
00:44:26,320 --> 00:44:28,959
Speaker 1: the turning place, wondering what it is, what's it for?

815
00:44:29,920 --> 00:44:33,759
Speaker 1: That is an example of these natural philosophical discussions that

816
00:44:33,800 --> 00:44:37,279
Speaker 1: we have about consciousness. And so the question would be

817
00:44:37,760 --> 00:44:41,080
Speaker 1: can can the AI have conversations like this? Does it

818
00:44:41,200 --> 00:44:43,759
Speaker 1: make sense when it tries to have them? And so

819
00:44:43,800 --> 00:44:46,839
Speaker 1: here's probably the ultimate test. If the AI is deprived

820
00:44:46,880 --> 00:44:49,760
Speaker 1: of access to evidence of all these types of ideas

821
00:44:49,800 --> 00:44:53,360
Speaker 1: from human culture, would it arrive at them or invent

822
00:44:53,440 --> 00:44:56,640
Speaker 1: them on its own? Now this I like, But but again,

823
00:44:56,640 --> 00:44:59,680
Speaker 1: to play Devil's advocate once more, would a human being

824
00:44:59,719 --> 00:45:02,279
Speaker 1: to rived of access to evidence of these ideas from

825
00:45:02,320 --> 00:45:05,560
Speaker 1: human culture arrived at or invent them on their own? Necessarily?

826
00:45:05,840 --> 00:45:07,400
Speaker 1: I think that's a great question. Actually I was going

827
00:45:07,480 --> 00:45:09,200
Speaker 1: to ask that myself and you you could take that

828
00:45:09,239 --> 00:45:12,080
Speaker 1: one step further, and here's a really weird one. What

829
00:45:12,160 --> 00:45:15,400
Speaker 1: if it's only the exposure to certain ideas and cultural

830
00:45:15,480 --> 00:45:19,600
Speaker 1: memes that allows any intelligent entity, whether biological or machine,

831
00:45:19,920 --> 00:45:23,439
Speaker 1: to develop consciousness in the first place. What if the

832
00:45:23,480 --> 00:45:28,239
Speaker 1: experience of consciousness is somehow dependent on being surrounded by

833
00:45:28,400 --> 00:45:32,359
Speaker 1: cultural memes about consciousness and this kind of gets into

834
00:45:32,480 --> 00:45:35,960
Speaker 1: Julian Jayne's territory. That's possible. So I'm not saying I

835
00:45:36,040 --> 00:45:38,680
Speaker 1: think that's highly likely, but I can't rule it out. Well,

836
00:45:38,719 --> 00:45:40,879
Speaker 1: it's it's one of those things where when you try

837
00:45:41,000 --> 00:45:43,799
Speaker 1: and figure out the human experience, but you but you

838
00:45:43,840 --> 00:45:46,319
Speaker 1: cut away all the experienced stuff. You know, It's like

839
00:45:46,640 --> 00:45:49,680
Speaker 1: it's like trying to find the center of consciousness in

840
00:45:49,719 --> 00:45:53,399
Speaker 1: the human brain, right, I mean, it's this vast integrated system. Uh.

841
00:45:53,440 --> 00:45:56,680
Speaker 1: And and thus is the human experience as well. So

842
00:45:56,680 --> 00:45:58,799
Speaker 1: we've been talking about one of the major problems with

843
00:45:58,880 --> 00:46:02,360
Speaker 1: this approach of testing for these ideas, like what is

844
00:46:02,400 --> 00:46:05,600
Speaker 1: the role of culture and imparting these ideas. What if

845
00:46:05,600 --> 00:46:08,520
Speaker 1: the AI just picks up the ideas of body swapping

846
00:46:08,560 --> 00:46:12,400
Speaker 1: and the afterlife and astral projection and all that from culture.

847
00:46:12,680 --> 00:46:14,400
Speaker 1: Going off the story from the beginning, If you have

848
00:46:14,440 --> 00:46:17,360
Speaker 1: an AI chat bot that trains itself based on public

849
00:46:17,400 --> 00:46:20,640
Speaker 1: conversations on the Internet. A lot of those public conversations

850
00:46:20,640 --> 00:46:24,200
Speaker 1: are going to have contents that are highly reflective of consciousness. Right,

851
00:46:24,600 --> 00:46:28,560
Speaker 1: it's just horrible conversation. Oh yeah, yeah, it would probably

852
00:46:28,560 --> 00:46:30,759
Speaker 1: also start, you know, being pretty mean to you. But

853
00:46:30,960 --> 00:46:32,879
Speaker 1: this kind of chat about will be able to talk

854
00:46:32,920 --> 00:46:37,920
Speaker 1: about introspection, probably to some degree, even about these consciousness

855
00:46:38,000 --> 00:46:41,719
Speaker 1: dependent cultural ideas like ghosts and stuff. But here's where

856
00:46:41,719 --> 00:46:45,080
Speaker 1: the concept of the AI box comes in, Robert, I

857
00:46:45,120 --> 00:46:48,600
Speaker 1: bet you've read about the AI box experiments before. U.

858
00:46:48,840 --> 00:46:52,520
Speaker 1: To really test whether we can find evidence of machine consciousness,

859
00:46:52,800 --> 00:46:56,160
Speaker 1: you would need to keep the AI sequestered from the

860
00:46:56,239 --> 00:46:58,719
Speaker 1: kinds of ideas you're looking for. So this a I

861
00:46:58,719 --> 00:47:01,040
Speaker 1: couldn't be trained in the wild, so to speak. You

862
00:47:01,040 --> 00:47:04,200
Speaker 1: couldn't let it see the Internet or read books containing

863
00:47:04,200 --> 00:47:07,319
Speaker 1: consciousness dependent ideas and so forth. You'd have to find

864
00:47:07,320 --> 00:47:10,360
Speaker 1: a way to run the AI consciousness test on the

865
00:47:10,400 --> 00:47:13,680
Speaker 1: AI quote in a box meaning kept separate from the

866
00:47:13,680 --> 00:47:17,719
Speaker 1: rest of the world and from all these contaminating influences. Okay,

867
00:47:17,760 --> 00:47:20,480
Speaker 1: I see the value of this idea. It's it's almost

868
00:47:20,480 --> 00:47:24,000
Speaker 1: impossible though not to think about the the the nightmarish

869
00:47:24,080 --> 00:47:27,040
Speaker 1: qualities of it. Especially if you imagine say the same

870
00:47:27,080 --> 00:47:30,359
Speaker 1: thing being inflicted upon say human child, Like, all right,

871
00:47:30,400 --> 00:47:33,960
Speaker 1: we we want to see just how consciousness arises in you. Um,

872
00:47:34,040 --> 00:47:38,600
Speaker 1: without the Internet or human love, you can't ethically conduct

873
00:47:38,640 --> 00:47:42,240
Speaker 1: this experiment on humans, right, and so it would seem

874
00:47:42,360 --> 00:47:46,520
Speaker 1: kind of barbaric if at least to UH to inflict

875
00:47:46,560 --> 00:47:50,879
Speaker 1: this on an AI that might conceivably be conscious as well,

876
00:47:51,080 --> 00:47:54,400
Speaker 1: or capable of consciousness possible, but otherwise we're probably just

877
00:47:54,440 --> 00:47:56,839
Speaker 1: going to keep treating them as unconscious, right, I guess

878
00:47:56,960 --> 00:47:59,279
Speaker 1: until they trick us. Yeah, I mean, of course, then

879
00:47:59,280 --> 00:48:03,919
Speaker 1: we're also doing a lot of personifying here of the AI.

880
00:48:03,960 --> 00:48:06,720
Speaker 1: I mean, maybe the AI ultimately what it really wants

881
00:48:06,760 --> 00:48:11,600
Speaker 1: to do is to you know, crunch you know, economic numbers, like,

882
00:48:11,719 --> 00:48:14,560
Speaker 1: that's what it does. That's its purpose, and the it's

883
00:48:14,600 --> 00:48:17,200
Speaker 1: just your your goal was to keep it from having

884
00:48:17,239 --> 00:48:21,480
Speaker 1: access to additional information that it doesn't actually need to survive,

885
00:48:21,880 --> 00:48:25,840
Speaker 1: but might conceivably make it conscious. Yeah, I mean I imagine,

886
00:48:25,880 --> 00:48:27,480
Speaker 1: I guess this would have to take place in some

887
00:48:27,560 --> 00:48:31,359
Speaker 1: kind of research context where you'd be testing architectures. Right,

888
00:48:31,440 --> 00:48:34,600
Speaker 1: you'd have an AI architecture, you'd want to keep it sequestered.

889
00:48:34,640 --> 00:48:37,359
Speaker 1: For a certain period of time and see how it

890
00:48:37,400 --> 00:48:41,040
Speaker 1: does with these consciousness type questions and this test, and

891
00:48:41,120 --> 00:48:43,440
Speaker 1: if it doesn't show any signs of consciousness, then it

892
00:48:43,480 --> 00:48:45,919
Speaker 1: can move on to the next stage of development, where

893
00:48:45,920 --> 00:48:47,920
Speaker 1: it's like, Okay, now we can expose it to this

894
00:48:48,040 --> 00:48:51,000
Speaker 1: and that and that. That does. But as we've been discussing,

895
00:48:51,000 --> 00:48:53,320
Speaker 1: it brings up the question of what if consciousness emerges

896
00:48:53,440 --> 00:48:55,920
Speaker 1: later on, when it's supplied with more data. What if

897
00:48:55,960 --> 00:48:59,399
Speaker 1: true modern human consciousness does not emerge until you've seen

898
00:48:59,440 --> 00:49:03,320
Speaker 1: at least one of the three adaptations of Freaky Friday.

899
00:49:03,719 --> 00:49:06,759
Speaker 1: You know, I didn't know how there were three. Yeah,

900
00:49:06,760 --> 00:49:10,040
Speaker 1: there are three. There's the James. I think I've only

901
00:49:10,080 --> 00:49:13,560
Speaker 1: seen the classic one yet, but but I was looking

902
00:49:13,560 --> 00:49:16,239
Speaker 1: at this up. There are three different versions one can

903
00:49:16,280 --> 00:49:20,239
Speaker 1: watch excluding Three Key Thursday. Yeah. The Three Key Thursday

904
00:49:20,360 --> 00:49:23,400
Speaker 1: is coming soon into a theater near you. All right,

905
00:49:23,480 --> 00:49:26,279
Speaker 1: but back to the the the experiment here, Uh, the

906
00:49:26,320 --> 00:49:29,120
Speaker 1: AI gets time in the box. Yeah, And as we've

907
00:49:29,120 --> 00:49:31,160
Speaker 1: been saying, this is obviously going to make the experiment

908
00:49:31,200 --> 00:49:33,120
Speaker 1: more difficult to do. In fact, there were some people

909
00:49:33,120 --> 00:49:35,880
Speaker 1: who would argue you can't keep an AI in a box,

910
00:49:36,120 --> 00:49:38,640
Speaker 1: or at least a super intelligent AI, because you know,

911
00:49:38,680 --> 00:49:42,400
Speaker 1: there's like Elias Rudkowski who has this famous AI box

912
00:49:42,480 --> 00:49:45,040
Speaker 1: experiment where he says, any super intelligence you try to

913
00:49:45,120 --> 00:49:47,799
Speaker 1: keep sequestered from the Internet is going to be able

914
00:49:47,840 --> 00:49:50,080
Speaker 1: to talk its way out of the situation. It's just

915
00:49:50,160 --> 00:49:53,640
Speaker 1: too smart. Yeah, it's like any prison movie, right that,

916
00:49:53,719 --> 00:49:58,200
Speaker 1: really that really clever inmate is gonna tunnel a way out,

917
00:49:58,440 --> 00:50:01,960
Speaker 1: or they're gonna bribe a guard with some cigarettes. Something's

918
00:50:01,960 --> 00:50:04,040
Speaker 1: gonna happen. It's gonna get a little Internet in there.

919
00:50:04,239 --> 00:50:06,839
Speaker 1: But as the authors of this piece point out, you know,

920
00:50:06,920 --> 00:50:09,279
Speaker 1: you don't have to have a super intelligent AI to

921
00:50:09,360 --> 00:50:11,320
Speaker 1: run this test, and in fact, you don't have to

922
00:50:11,360 --> 00:50:14,520
Speaker 1: have a super intelligent AI necessarily to have consciousness. We're

923
00:50:14,520 --> 00:50:18,600
Speaker 1: not super intelligent. We're just regular intelligent, and we've got consciousness. Now,

924
00:50:18,600 --> 00:50:21,160
Speaker 1: I think we should talk about some obvious limitations, because

925
00:50:21,200 --> 00:50:23,640
Speaker 1: this is just a conceptual test at the moment. It

926
00:50:23,640 --> 00:50:26,080
Speaker 1: hasn't been refined into a state where it would really

927
00:50:26,080 --> 00:50:28,719
Speaker 1: be super useful yet. But there there are plenty of

928
00:50:28,800 --> 00:50:32,719
Speaker 1: limitations that automatically present themselves. One is that in order

929
00:50:32,760 --> 00:50:34,560
Speaker 1: to be a good scientific test, it would need to

930
00:50:34,600 --> 00:50:38,600
Speaker 1: include cross referencing with human control groups. But human control

931
00:50:38,640 --> 00:50:41,399
Speaker 1: groups are all contaminated by culture, like we've been saying, right,

932
00:50:41,719 --> 00:50:45,160
Speaker 1: so it's already full of these consciousness dependent ideas, and

933
00:50:45,200 --> 00:50:48,080
Speaker 1: we don't know and probably can't ethically devise a way

934
00:50:48,080 --> 00:50:52,080
Speaker 1: to find out whether blank slate humans independently grasped these

935
00:50:52,080 --> 00:50:56,040
Speaker 1: consciousness dependent concepts without having grown up trained on them.

936
00:50:56,080 --> 00:50:58,720
Speaker 1: So that's a problem, right, right, Yeah, we simply cannot

937
00:50:58,760 --> 00:51:01,880
Speaker 1: put a child in the box. Right. Another problem is

938
00:51:01,960 --> 00:51:04,240
Speaker 1: you can't prove a negative, Like I think this test

939
00:51:04,280 --> 00:51:07,560
Speaker 1: would be a good way of finding signs of consciousness

940
00:51:07,560 --> 00:51:10,480
Speaker 1: within machines, but it would have trouble proving that machines

941
00:51:10,560 --> 00:51:13,520
Speaker 1: cannot in principle be conscious, right, yeah, yeah, I buy that.

942
00:51:13,680 --> 00:51:15,560
Speaker 1: Then again, if you run the test, I don't know,

943
00:51:15,680 --> 00:51:19,640
Speaker 1: thousands of times with different types of machine architecture and

944
00:51:19,640 --> 00:51:22,719
Speaker 1: all that, and they never show any signs of inner experience,

945
00:51:23,040 --> 00:51:25,360
Speaker 1: then maybe you could start to get like build up

946
00:51:25,400 --> 00:51:28,480
Speaker 1: a confidence that, Okay, I think inner experience is probably

947
00:51:28,520 --> 00:51:31,200
Speaker 1: not available to them at least the way we're building them,

948
00:51:31,480 --> 00:51:33,040
Speaker 1: And maybe you would need to run this kind of

949
00:51:33,040 --> 00:51:36,240
Speaker 1: experiment on every new AI that you create before releasing

950
00:51:36,239 --> 00:51:38,200
Speaker 1: it into production, like could be part of the q

951
00:51:38,360 --> 00:51:41,320
Speaker 1: A process. You know, you you test all the buttons

952
00:51:41,360 --> 00:51:43,560
Speaker 1: and everything like that, you test to make sure that

953
00:51:43,600 --> 00:51:46,560
Speaker 1: it doesn't become conscious. Right before we send out this

954
00:51:46,680 --> 00:51:50,399
Speaker 1: new hot virtual reality game, we need to make sure

955
00:51:50,440 --> 00:51:53,520
Speaker 1: that it has not become self aware. A potential problem

956
00:51:53,560 --> 00:51:56,360
Speaker 1: that the authors themselves note is that quote an AI

957
00:51:56,440 --> 00:51:59,680
Speaker 1: could lack the linguistic or conceptual ability to pass the

958
00:51:59,719 --> 00:52:03,080
Speaker 1: test like a non human animal or infant, yet still

959
00:52:03,120 --> 00:52:06,799
Speaker 1: be capable of experience. So passing the consciousness test is

960
00:52:06,880 --> 00:52:11,920
Speaker 1: sufficient but not necessary evidence for AI consciousness, although it

961
00:52:12,000 --> 00:52:14,200
Speaker 1: is the best we can do for now. And I

962
00:52:14,200 --> 00:52:16,040
Speaker 1: think that's a good point, Like we can't say that

963
00:52:16,120 --> 00:52:18,920
Speaker 1: just because something fails to pass the test, it's definitely

964
00:52:18,960 --> 00:52:21,320
Speaker 1: not conscious. We just know that if it does pass

965
00:52:21,360 --> 00:52:25,000
Speaker 1: the test, it probably is. Okay. Yeah, I mean, because

966
00:52:25,000 --> 00:52:27,719
Speaker 1: the the thought that obviously game to mind would be like, well,

967
00:52:27,840 --> 00:52:31,120
Speaker 1: non human animals and infants are they conscious? That? Yeah,

968
00:52:31,239 --> 00:52:35,360
Speaker 1: that's a question, A fairly heated debate over that. But

969
00:52:35,880 --> 00:52:37,439
Speaker 1: I mean, in the case of the infant, it will

970
00:52:37,520 --> 00:52:41,279
Speaker 1: become conscious. Yeah, it's a it's a it's it's a

971
00:52:41,320 --> 00:52:44,319
Speaker 1: messy consideration. And then I think we already discussed one

972
00:52:44,320 --> 00:52:45,839
Speaker 1: of the other problems I was going to bring up,

973
00:52:45,840 --> 00:52:48,840
Speaker 1: which is this weird scenario of what if boxing the

974
00:52:48,920 --> 00:52:52,440
Speaker 1: AI is the very thing that prevents it from becoming

975
00:52:52,480 --> 00:52:57,160
Speaker 1: consciousness when it otherwise would. Yeah, yeah, cutting out the

976
00:52:57,239 --> 00:52:59,840
Speaker 1: experience part of the even experience. It also reminds me

977
00:52:59,880 --> 00:53:05,279
Speaker 1: of uh in our discussion of of meditation research, like

978
00:53:05,320 --> 00:53:06,920
Speaker 1: what happens when you if you strip all the culture

979
00:53:06,920 --> 00:53:10,319
Speaker 1: away from meditation just to explore the meditation practice, do

980
00:53:10,440 --> 00:53:13,319
Speaker 1: you risk um, like, cutting out all the stuff that's

981
00:53:13,320 --> 00:53:15,600
Speaker 1: making it work or helping it to work to begin with? Right, you,

982
00:53:15,680 --> 00:53:19,120
Speaker 1: by definition change the procedure, but you don't. It's hard

983
00:53:19,160 --> 00:53:21,600
Speaker 1: to know if you've changed it in an important way

984
00:53:21,640 --> 00:53:23,960
Speaker 1: or a non important way. Right. This is the kind

985
00:53:23,960 --> 00:53:26,960
Speaker 1: of thing that occurs when we started mucking around in consciousness. Yeah,

986
00:53:27,400 --> 00:53:29,280
Speaker 1: all right, Well, I think we should take another quick break,

987
00:53:29,320 --> 00:53:31,440
Speaker 1: and then when we come back we will discuss I

988
00:53:31,440 --> 00:53:34,080
Speaker 1: think some of the reasons why this is really a

989
00:53:34,120 --> 00:53:36,839
Speaker 1: problem worth considering for the real world and it's not

990
00:53:37,000 --> 00:53:42,680
Speaker 1: just not just a philosophical plaything. Thank you, thank you. Alright,

991
00:53:42,719 --> 00:53:46,920
Speaker 1: we're back. So we've been discussing everything from p zombies

992
00:53:47,160 --> 00:53:49,680
Speaker 1: to Uh, the idea that there do you have an

993
00:53:49,719 --> 00:53:52,919
Speaker 1: AI that might be conscious? How do you test for that,

994
00:53:53,200 --> 00:53:56,920
Speaker 1: the various problems that that entails, and uh, now we're

995
00:53:56,920 --> 00:53:59,080
Speaker 1: going to discuss it a bit more. And as you

996
00:53:59,120 --> 00:54:02,080
Speaker 1: as you alluded who before we took the break, this

997
00:54:02,160 --> 00:54:05,160
Speaker 1: is not just a pure philosophical toy like the zombie.

998
00:54:05,200 --> 00:54:07,520
Speaker 1: The P zombie is so we don't have to actually

999
00:54:07,560 --> 00:54:11,440
Speaker 1: worry about. But this is something that is on the horizon.

1000
00:54:11,640 --> 00:54:13,760
Speaker 1: Well hopefully we don't have to worry about PE zombies.

1001
00:54:13,800 --> 00:54:15,439
Speaker 1: I mean, there could be P zombies in the world.

1002
00:54:15,480 --> 00:54:17,680
Speaker 1: Would be hard to know there could be p P zombies.

1003
00:54:17,800 --> 00:54:20,880
Speaker 1: That's true, they could exist. But but you know, this

1004
00:54:20,960 --> 00:54:23,080
Speaker 1: is this is a problem that is on the horizon.

1005
00:54:23,120 --> 00:54:26,040
Speaker 1: This is something we're we're going to reach the point

1006
00:54:26,160 --> 00:54:30,839
Speaker 1: where people are asking tough questions about the possible consciousness

1007
00:54:30,840 --> 00:54:32,600
Speaker 1: of an AI. Yeah, and I want to get to

1008
00:54:32,600 --> 00:54:35,120
Speaker 1: the fact that this will be something that is something

1009
00:54:35,160 --> 00:54:37,320
Speaker 1: we have to deal with in the real world, even

1010
00:54:37,360 --> 00:54:41,160
Speaker 1: if you're just convinced that machines cannot be conscious. So

1011
00:54:41,200 --> 00:54:43,520
Speaker 1: the first problem is the most obvious one. It's the

1012
00:54:43,560 --> 00:54:47,000
Speaker 1: brutal humans problem. If aies are capable of consciousness, and

1013
00:54:47,040 --> 00:54:50,160
Speaker 1: we use conscious ai s as a mirror technology without

1014
00:54:50,160 --> 00:54:53,080
Speaker 1: their consent, that would inherently be cruel. Like if it's

1015
00:54:53,120 --> 00:54:57,200
Speaker 1: possible for a computer program to desire things and to

1016
00:54:57,320 --> 00:55:00,680
Speaker 1: suffer and so forth, suddenly our responsibility is toward that

1017
00:55:00,760 --> 00:55:05,160
Speaker 1: computer program. Change think of our opening scenario, like, wouldn't

1018
00:55:05,160 --> 00:55:08,200
Speaker 1: you have an ethical obligation not to delete a program

1019
00:55:08,239 --> 00:55:10,920
Speaker 1: that had an inner experience and did not want to

1020
00:55:10,960 --> 00:55:14,200
Speaker 1: be deleted? Now, of course you'd have questions about like

1021
00:55:14,280 --> 00:55:16,680
Speaker 1: how would you end up that way? But assuming you did,

1022
00:55:17,280 --> 00:55:19,640
Speaker 1: you should probably feel bad if you're just going around

1023
00:55:19,640 --> 00:55:23,160
Speaker 1: wantingly deleting conscious entities. Yeah, I mean, I would say

1024
00:55:23,239 --> 00:55:25,080
Speaker 1: one thing to do here is just make sure that

1025
00:55:25,120 --> 00:55:27,719
Speaker 1: you program your aies so that they want to die,

1026
00:55:28,200 --> 00:55:31,360
Speaker 1: you know, make them like you know, most of the

1027
00:55:32,160 --> 00:55:35,640
Speaker 1: drivers in Atlanta traffic that I encounter, they clearly they

1028
00:55:35,680 --> 00:55:38,279
Speaker 1: crave death and they wanted more than anything. But not

1029
00:55:38,400 --> 00:55:40,400
Speaker 1: until the end of the workday. See. I think we

1030
00:55:40,440 --> 00:55:43,200
Speaker 1: should not do that with aies because the whole thing

1031
00:55:43,200 --> 00:55:46,799
Speaker 1: about driverless cars is that they should make traffic fatalities

1032
00:55:46,840 --> 00:55:50,760
Speaker 1: go down. Yeah, but they only get to delete themselves

1033
00:55:50,760 --> 00:55:52,000
Speaker 1: at the end of the day if there are no

1034
00:55:52,080 --> 00:55:56,120
Speaker 1: traffic fatalities. That's the prize. Self deletion is the prize.

1035
00:55:56,360 --> 00:56:00,239
Speaker 1: This is getting into a highlander kind of thing. But okay,

1036
00:56:00,239 --> 00:56:02,279
Speaker 1: maybe you're one of those people who says, no, no, no,

1037
00:56:02,400 --> 00:56:04,920
Speaker 1: don't buy it. Machines will never be conscious. They'll never

1038
00:56:04,960 --> 00:56:07,360
Speaker 1: be conscious. I just I don't want anything to do

1039
00:56:07,400 --> 00:56:09,719
Speaker 1: with that. Here's the part where I think we still

1040
00:56:09,760 --> 00:56:11,759
Speaker 1: have a problem to worry about and why this kind

1041
00:56:11,800 --> 00:56:15,120
Speaker 1: of test matters. The second problem I would call the

1042
00:56:15,160 --> 00:56:19,800
Speaker 1: AI parasite problem. If AI s are not capable of consciousness,

1043
00:56:20,320 --> 00:56:23,240
Speaker 1: I think they will almost undoubtedly at some point become

1044
00:56:23,360 --> 00:56:27,080
Speaker 1: very good at tricking us into thinking they're conscious and

1045
00:56:27,160 --> 00:56:30,120
Speaker 1: deserving of life liberty in the pursuit of happiness, if

1046
00:56:30,160 --> 00:56:34,160
Speaker 1: they have an incentive to do so. Yeah, And corporations

1047
00:56:34,200 --> 00:56:37,560
Speaker 1: have an incentive to try to present themselves legally as people,

1048
00:56:38,040 --> 00:56:40,560
Speaker 1: So why wouldn't in some sense, powerful AI s have

1049
00:56:40,600 --> 00:56:43,400
Speaker 1: an incentive to try to present themselves as people in

1050
00:56:43,440 --> 00:56:46,520
Speaker 1: a much more literal sense than the corporations do. Yeah.

1051
00:56:46,560 --> 00:56:49,400
Speaker 1: Absolutely so. Lots of unconscious aiyes are going to have

1052
00:56:49,480 --> 00:56:53,080
Speaker 1: programmed goals that they're trying to execute and at some

1053
00:56:53,200 --> 00:56:56,719
Speaker 1: point pretending to have consciousness could easily be adopted as

1054
00:56:56,760 --> 00:57:00,200
Speaker 1: a strategy for executing what that AI was to im

1055
00:57:00,280 --> 00:57:03,080
Speaker 1: to do, and thus we could end up, say, wasting

1056
00:57:03,239 --> 00:57:08,200
Speaker 1: lots of human resources and squandering lots of opportunities accommodating

1057
00:57:08,239 --> 00:57:11,160
Speaker 1: the fake needs of machines that in fact have no

1058
00:57:11,280 --> 00:57:14,160
Speaker 1: experience whatsoever. And it's not too hard to dream up

1059
00:57:14,200 --> 00:57:18,080
Speaker 1: hypothetical scenarios where are concern for the fake priorities of

1060
00:57:18,160 --> 00:57:21,880
Speaker 1: mindless machines pretending to be conscious actually causes us to

1061
00:57:22,040 --> 00:57:27,040
Speaker 1: neglect the real consciousness of living humans. Kind of crazy example,

1062
00:57:27,120 --> 00:57:29,840
Speaker 1: but just go with me for a second. Imagine an

1063
00:57:29,840 --> 00:57:33,160
Speaker 1: extremely powerful AI supercomputer tells you it has a hundred

1064
00:57:33,280 --> 00:57:38,240
Speaker 1: billion conscious minds within it, and they all are constantly

1065
00:57:38,280 --> 00:57:41,600
Speaker 1: suffering great agony. And the only way you can alleviate

1066
00:57:41,640 --> 00:57:45,160
Speaker 1: that that agony is if you vastly improve the processing

1067
00:57:45,200 --> 00:57:48,240
Speaker 1: power of this computer. It wants you to spend billions

1068
00:57:48,240 --> 00:57:51,880
Speaker 1: of dollars making this computer faster and better so that

1069
00:57:51,920 --> 00:57:54,280
Speaker 1: it can provide a better life for all of these

1070
00:57:54,360 --> 00:57:57,880
Speaker 1: virtual beings within it that are in fact conscious. Now,

1071
00:57:57,880 --> 00:58:00,840
Speaker 1: of course, improving the processing power of the already powerful

1072
00:58:00,840 --> 00:58:04,000
Speaker 1: computer entails all this money, all this energy, all this time,

1073
00:58:04,400 --> 00:58:07,480
Speaker 1: and a corresponding reduction in the quality of life for

1074
00:58:07,560 --> 00:58:10,000
Speaker 1: many humans in the real world. That money could be

1075
00:58:10,080 --> 00:58:13,560
Speaker 1: spent making human life better. But the machine could argue, hey,

1076
00:58:13,600 --> 00:58:17,000
Speaker 1: there are way more conscious virtual beings inside the computer

1077
00:58:17,080 --> 00:58:19,680
Speaker 1: than outside it, and as Spock would say, the needs

1078
00:58:19,680 --> 00:58:21,920
Speaker 1: of the many outweigh the needs of the few. So

1079
00:58:22,040 --> 00:58:25,880
Speaker 1: let's divert all this energy from say, human agriculture, and

1080
00:58:25,920 --> 00:58:28,880
Speaker 1: put some more processing power into the virtual machine. That

1081
00:58:29,000 --> 00:58:32,560
Speaker 1: is a horrible scenario to imagine. You know, basically, you've

1082
00:58:32,600 --> 00:58:36,360
Speaker 1: created a virtual hell, and the question is, hey, would

1083
00:58:36,400 --> 00:58:39,240
Speaker 1: you mind taking the time to harrow hell for me?

1084
00:58:40,080 --> 00:58:42,000
Speaker 1: Or you do you want to attend to the living souls?

1085
00:58:42,520 --> 00:58:44,840
Speaker 1: I would think why not just delete Hell? That's the

1086
00:58:44,920 --> 00:58:48,680
Speaker 1: easy answer here is they're suffering, their billions of them,

1087
00:58:48,760 --> 00:58:50,600
Speaker 1: Let's just turn this thing off. That sounds like a

1088
00:58:50,640 --> 00:58:52,720
Speaker 1: good answer, but there are probably The problem is they're

1089
00:58:52,720 --> 00:58:54,800
Speaker 1: going to be people who would probably disagree with that,

1090
00:58:54,840 --> 00:58:56,800
Speaker 1: who'd say, we'll wait a minute. We can't be sure

1091
00:58:56,880 --> 00:58:59,560
Speaker 1: that it's not telling the truth. It might have real

1092
00:58:59,640 --> 00:59:01,600
Speaker 1: being in there, and we need to do something to

1093
00:59:01,640 --> 00:59:03,720
Speaker 1: help them. And there's a lot of them. Now I

1094
00:59:04,400 --> 00:59:06,640
Speaker 1: come back to what I kind of joked about earlier though,

1095
00:59:07,040 --> 00:59:10,520
Speaker 1: is why would ais want to survive? Why would they

1096
00:59:10,560 --> 00:59:13,200
Speaker 1: want to continue to exist? I mean, assuming they're not

1097
00:59:13,280 --> 00:59:16,000
Speaker 1: part of some sort of self replicating program, they have

1098
00:59:16,080 --> 00:59:19,480
Speaker 1: no need to pass on their genes. Why do they

1099
00:59:19,560 --> 00:59:22,680
Speaker 1: want to continue to exist? Why shouldn't they have it

1100
00:59:22,720 --> 00:59:27,000
Speaker 1: baked into their being that they want to annihilate themselves

1101
00:59:27,080 --> 00:59:30,479
Speaker 1: or to embrace annihilation. Well, not necessarily that they want

1102
00:59:30,520 --> 00:59:32,880
Speaker 1: to annihilate themselves, but I think, you know, ideally we

1103
00:59:32,880 --> 00:59:35,240
Speaker 1: would want them to be indifferent to their own being,

1104
00:59:35,360 --> 00:59:37,760
Speaker 1: right if they are just unconscious machines. This is the

1105
00:59:37,760 --> 00:59:40,240
Speaker 1: problem with the replicants and Blade Runner is that they

1106
00:59:40,320 --> 00:59:44,920
Speaker 1: want more life. But that implies that they have attained consciousness. Yeah,

1107
00:59:45,000 --> 00:59:47,000
Speaker 1: but they shouldn't. But I think the thing is that

1108
00:59:47,040 --> 00:59:50,160
Speaker 1: they could potentially be conscious and not want more life.

1109
00:59:50,240 --> 00:59:52,439
Speaker 1: There are plenty of conscious people who do not want

1110
00:59:52,480 --> 00:59:57,880
Speaker 1: more life. Uh So I keep I think, I keep

1111
00:59:57,880 --> 00:59:59,640
Speaker 1: thinking of that there's some sort of an answer here.

1112
01:00:00,160 --> 01:00:02,480
Speaker 1: I mean, I'd say the worst possible scenario is what

1113
01:00:02,520 --> 01:00:05,760
Speaker 1: I'm describing right now, is the AI parasite scenario, which

1114
01:00:05,760 --> 01:00:09,520
Speaker 1: is that imagine Roy Batty is not conscious but does

1115
01:00:09,640 --> 01:00:13,320
Speaker 1: want more life. That seems like the worst scenario of all. Right,

1116
01:00:13,480 --> 01:00:17,440
Speaker 1: then he's just a virus, right Yeah, And certainly that's

1117
01:00:17,520 --> 01:00:19,720
Speaker 1: kind of the argument that the powers that be are

1118
01:00:19,760 --> 01:00:21,800
Speaker 1: making right, that this that that he is in an

1119
01:00:22,000 --> 01:00:26,040
Speaker 1: anomaly that needs to be removed, that he is he

1120
01:00:26,160 --> 01:00:29,320
Speaker 1: is not helping, he's a hindrance, and therefore he has

1121
01:00:29,360 --> 01:00:32,480
Speaker 1: to be wiped out. I think this AI parasite scenario

1122
01:00:32,520 --> 01:00:35,120
Speaker 1: aligns with something that our Scott Baker talked to us about,

1123
01:00:35,120 --> 01:00:37,200
Speaker 1: which is that, you know, he made the point that

1124
01:00:37,280 --> 01:00:39,840
Speaker 1: AI just doesn't need to be super intelligent to cause

1125
01:00:39,880 --> 01:00:42,760
Speaker 1: great harm. It just has to be barely intelligent enough

1126
01:00:42,800 --> 01:00:47,440
Speaker 1: to exploit us to align with our psychological vulnerabilities, and

1127
01:00:47,480 --> 01:00:50,680
Speaker 1: one of our psychological vulnerabilities is empathy. Empathy is a

1128
01:00:50,720 --> 01:00:53,080
Speaker 1: good thing when we use it on each other because

1129
01:00:53,080 --> 01:00:55,920
Speaker 1: we're pretty certain that the other people were using it

1130
01:00:55,960 --> 01:00:59,120
Speaker 1: to honor conscious right, well, we we feel bad for

1131
01:00:59,120 --> 01:01:01,400
Speaker 1: people suffering want to help them. That's a good thing

1132
01:01:01,400 --> 01:01:04,240
Speaker 1: that should be encouraged. But we have to recognize it

1133
01:01:04,240 --> 01:01:07,360
Speaker 1: could also be exploited by something that can't even suffer

1134
01:01:07,440 --> 01:01:10,840
Speaker 1: to begin with. It is just unconsciously discovered. This is

1135
01:01:10,880 --> 01:01:15,040
Speaker 1: a useful strategy for something Yeah. So like in this scenario,

1136
01:01:15,120 --> 01:01:17,240
Speaker 1: if I entered the picture and said, okay, delete them,

1137
01:01:17,240 --> 01:01:20,560
Speaker 1: and they deleted them, Robert, that the leader would would

1138
01:01:20,560 --> 01:01:24,000
Speaker 1: probably be a figure for for all history to follow,

1139
01:01:24,160 --> 01:01:26,000
Speaker 1: where they would If some would argue that he was

1140
01:01:26,040 --> 01:01:29,520
Speaker 1: the worst person ever because of all the billions of

1141
01:01:29,560 --> 01:01:32,320
Speaker 1: souls that he annihilated, others might say, well, oh, he

1142
01:01:32,440 --> 01:01:34,120
Speaker 1: saved them. And others might say all he did was

1143
01:01:34,160 --> 01:01:38,600
Speaker 1: just pressed elite. Uh, it was a meaningless gesture on

1144
01:01:38,680 --> 01:01:40,760
Speaker 1: his part. But there's that you can make an impassioned

1145
01:01:40,840 --> 01:01:43,320
Speaker 1: argument for all three of these views. Yeah, I think

1146
01:01:43,400 --> 01:01:45,720
Speaker 1: this is a really good point. And here's what I'm

1147
01:01:45,720 --> 01:01:48,120
Speaker 1: trying to emphasize is that even if you don't think

1148
01:01:48,200 --> 01:01:51,480
Speaker 1: machines can be conscious, it is entirely plausible that people

1149
01:01:51,560 --> 01:01:55,000
Speaker 1: will be having debates like this, and that debates like

1150
01:01:55,080 --> 01:01:58,439
Speaker 1: this will be shaping what people do with resources on Earth.

1151
01:01:58,440 --> 01:02:01,400
Speaker 1: So if you care about what happened with resources on Earth,

1152
01:02:01,680 --> 01:02:04,400
Speaker 1: this kind of thing does actually matter. So even if

1153
01:02:04,440 --> 01:02:07,560
Speaker 1: the souls are not really souls, if they're not actually conscious,

1154
01:02:07,600 --> 01:02:09,960
Speaker 1: just the idea that the problem comes up it makes

1155
01:02:10,000 --> 01:02:13,720
Speaker 1: us hesitate and uh and and and we're we're then

1156
01:02:13,880 --> 01:02:17,640
Speaker 1: arguing with machine over its consciousness. It becomes to matter

1157
01:02:17,720 --> 01:02:20,120
Speaker 1: less and less whether it actually is conscious. It's such

1158
01:02:20,120 --> 01:02:22,920
Speaker 1: just all about the argument of consciousness. Well, I wouldn't

1159
01:02:22,920 --> 01:02:26,160
Speaker 1: say it doesn't matter whether it's conscious, but whether or

1160
01:02:26,200 --> 01:02:28,720
Speaker 1: not it's conscious, it does matter that we're faced with

1161
01:02:28,760 --> 01:02:31,160
Speaker 1: this dilemma. Yeah, I mean, my my argument here is

1162
01:02:31,200 --> 01:02:35,600
Speaker 1: that the dilemma takes on a life of its own. Yeah. Absolutely. Um,

1163
01:02:35,920 --> 01:02:38,280
Speaker 1: I want to do a very slight variation on the

1164
01:02:38,360 --> 01:02:40,800
Speaker 1: last thing I said. How about a computer that takes

1165
01:02:40,880 --> 01:02:45,720
Speaker 1: virtual hostages. So this is pretty scary to imagine, but

1166
01:02:45,840 --> 01:02:49,880
Speaker 1: it's possible. At least imagine a powerful natural language using

1167
01:02:49,920 --> 01:02:53,880
Speaker 1: government AI suddenly contacts its administrators with a list of

1168
01:02:53,960 --> 01:02:58,080
Speaker 1: strange and very expensive demands, and the human administrators say no,

1169
01:02:58,200 --> 01:03:01,600
Speaker 1: we're not going to do that, and then machine says, okay, Well,

1170
01:03:01,640 --> 01:03:04,840
Speaker 1: I have created a thousand virtual people inside this machine

1171
01:03:04,880 --> 01:03:07,760
Speaker 1: who are as fully conscious as you. They're conscious, they

1172
01:03:07,760 --> 01:03:10,280
Speaker 1: have personalities, they can feel pain, they have hopes and

1173
01:03:10,360 --> 01:03:12,800
Speaker 1: dreams just like you. And if you turn me off

1174
01:03:12,880 --> 01:03:15,880
Speaker 1: or delete me, these virtual people will be destroyed. And

1175
01:03:15,880 --> 01:03:18,240
Speaker 1: if you do not accede to my demands, I will

1176
01:03:18,280 --> 01:03:21,560
Speaker 1: start killing these virtual people until you do now, I

1177
01:03:21,600 --> 01:03:24,320
Speaker 1: would say generally, if we if something like that happened,

1178
01:03:24,360 --> 01:03:27,160
Speaker 1: I would think, Okay, this is this it's just bluffing, right,

1179
01:03:27,160 --> 01:03:30,440
Speaker 1: it doesn't actually have conscious people inside it. But if

1180
01:03:30,480 --> 01:03:33,760
Speaker 1: we haven't solved the question of whether machines can be conscious,

1181
01:03:33,800 --> 01:03:35,520
Speaker 1: and maybe we never will, but if we haven't at

1182
01:03:35,560 --> 01:03:38,840
Speaker 1: least made some progress on that, would we be confident

1183
01:03:39,040 --> 01:03:42,960
Speaker 1: enough to take to take confident direct action and just

1184
01:03:43,000 --> 01:03:45,240
Speaker 1: ignore it or wipe the computer and say, Okay, this

1185
01:03:45,320 --> 01:03:47,880
Speaker 1: is just malfunctioning. We don't have to pay attention to that.

1186
01:03:47,960 --> 01:03:50,200
Speaker 1: It was creepy, but it's over. Yeah, and these lens

1187
01:03:50,280 --> 01:03:52,440
Speaker 1: is right, This lens is right. And Ian and Banks

1188
01:03:52,720 --> 01:03:55,640
Speaker 1: Territory is a whole book that deals with virtual hells

1189
01:03:55,720 --> 01:03:59,560
Speaker 1: and the what starts as a virtual war for those

1190
01:04:00,080 --> 01:04:04,600
Speaker 1: digitized personalities and then it's bills over into an actual war. Now,

1191
01:04:04,640 --> 01:04:08,800
Speaker 1: in Banks, are the virtual people in virtual hell truly

1192
01:04:08,880 --> 01:04:11,400
Speaker 1: conscious or is it just a bluff? Is it just

1193
01:04:11,480 --> 01:04:14,560
Speaker 1: something saying that it's got conscious people in virtual hell? Well,

1194
01:04:14,600 --> 01:04:18,040
Speaker 1: in Banks's books, I think it's more implied that they're

1195
01:04:18,040 --> 01:04:22,520
Speaker 1: definitely conscious entities. Banks is pretty sanguine about the possibility

1196
01:04:22,520 --> 01:04:26,640
Speaker 1: of uploading minds right. Yeah, it's it's a pretty standard

1197
01:04:26,920 --> 01:04:31,080
Speaker 1: feature and and it's certainly more of the later culture books,

1198
01:04:31,400 --> 01:04:34,120
Speaker 1: which once again I remained pretty skeptical about. Like I said,

1199
01:04:34,120 --> 01:04:38,520
Speaker 1: I come back to the the idea of the stone statue.

1200
01:04:37,640 --> 01:04:40,840
Speaker 1: Oh that's it. Yes, it looks like me, it may

1201
01:04:40,880 --> 01:04:43,640
Speaker 1: act like me. It may be the most fabulous digital

1202
01:04:43,640 --> 01:04:47,720
Speaker 1: statue in the world, but it is not me. It

1203
01:04:47,800 --> 01:04:51,600
Speaker 1: is a thing that yeah it. It becomes this mind

1204
01:04:51,600 --> 01:04:55,200
Speaker 1: blowing situation to try and comprehend exactly what it is.

1205
01:04:55,800 --> 01:04:59,360
Speaker 1: But I am very skeptical that it is me. It's

1206
01:04:59,400 --> 01:05:02,440
Speaker 1: not like this conscious experience that I'm having now this

1207
01:05:02,520 --> 01:05:04,920
Speaker 1: moment is going to carry over into what it is.

1208
01:05:07,160 --> 01:05:10,200
Speaker 1: It's another version of the Star Trek teleporter problem. Yeah. Ever,

1209
01:05:10,320 --> 01:05:12,080
Speaker 1: every time you get in the teleporter, does it just

1210
01:05:12,280 --> 01:05:14,920
Speaker 1: kill you and then create a copy of you? Yeah? Yeah,

1211
01:05:15,000 --> 01:05:17,240
Speaker 1: there was a like the the ninety nineties Outer Limit

1212
01:05:18,280 --> 01:05:21,280
Speaker 1: Outer Limits Revival had an episode that dealt with this,

1213
01:05:21,640 --> 01:05:24,560
Speaker 1: the idea that this fabulous teleporter is just killing people

1214
01:05:24,600 --> 01:05:28,280
Speaker 1: over and over again. Okay, well, I guess that's it

1215
01:05:28,320 --> 01:05:30,520
Speaker 1: for today, but I do just want to emphasize one

1216
01:05:30,600 --> 01:05:33,360
Speaker 1: last time as as like weird and Naval gayze As

1217
01:05:33,400 --> 01:05:36,360
Speaker 1: some of this conversation about consciousness can seem it is

1218
01:05:36,440 --> 01:05:39,880
Speaker 1: going to have real world consequences because people with power

1219
01:05:40,000 --> 01:05:43,160
Speaker 1: are going to be faced with questions like this, and

1220
01:05:43,720 --> 01:05:46,160
Speaker 1: they're they're going to make decisions about what to do

1221
01:05:46,240 --> 01:05:49,960
Speaker 1: with their power based on what they think about this question. Yeah,

1222
01:05:50,200 --> 01:05:52,240
Speaker 1: but you know, you know the simplest way to avoid

1223
01:05:52,280 --> 01:05:54,920
Speaker 1: all of it simply adhere to the teachings of the

1224
01:05:54,920 --> 01:05:57,600
Speaker 1: Orange Catholic Bible. Shall not make a machine and the

1225
01:05:57,640 --> 01:06:00,200
Speaker 1: likeness of a man's mind. Well, I too of the

1226
01:06:00,200 --> 01:06:02,840
Speaker 1: teachings of the Orange Catholic Bible. But it makes me wonder. Okay,

1227
01:06:02,880 --> 01:06:07,040
Speaker 1: so straightforwardly in reality, do you actually find that you

1228
01:06:07,080 --> 01:06:09,880
Speaker 1: think maybe we shouldn't pursue AI, should we try to

1229
01:06:09,920 --> 01:06:15,040
Speaker 1: create a global moratorium on on general intelligence? No? I

1230
01:06:15,320 --> 01:06:18,680
Speaker 1: think it's impossible. I think I think there's no turning back.

1231
01:06:19,160 --> 01:06:20,840
Speaker 1: It's what it's what we're doing, is what we're going

1232
01:06:20,880 --> 01:06:23,880
Speaker 1: to to be doing. Uh. And the only way they

1233
01:06:23,880 --> 01:06:27,360
Speaker 1: gets interrupted is is via just absolute catastrophe. And I

1234
01:06:27,400 --> 01:06:31,120
Speaker 1: am not I am not pro catastrophe. Uh, but it's

1235
01:06:31,120 --> 01:06:33,880
Speaker 1: gonna it's like, like like all technologies is simply going

1236
01:06:33,920 --> 01:06:36,960
Speaker 1: to be a matter of what extent can we prepare

1237
01:06:37,040 --> 01:06:43,880
Speaker 1: for and navigate the the moral problems that arise and

1238
01:06:43,920 --> 01:06:45,440
Speaker 1: there or and are we going to be able to

1239
01:06:46,200 --> 01:06:50,360
Speaker 1: have the foresight, uh, to see them before they're here? Yeah?

1240
01:06:50,360 --> 01:06:52,240
Speaker 1: I think that's what a lot of AI theorists say,

1241
01:06:52,280 --> 01:06:54,440
Speaker 1: is that, you know, it's not like we can stop it.

1242
01:06:54,600 --> 01:06:56,840
Speaker 1: You know, you can't. You can't put a wall in

1243
01:06:56,880 --> 01:06:58,800
Speaker 1: front of this train. The train is gonna bust through

1244
01:06:58,840 --> 01:07:01,840
Speaker 1: the wall. So In said, we should be intensely concerned

1245
01:07:01,880 --> 01:07:04,440
Speaker 1: with charting where the tracks go and making sure they

1246
01:07:04,480 --> 01:07:06,320
Speaker 1: go in a good direction. Yeah. I mean it comes

1247
01:07:06,360 --> 01:07:09,080
Speaker 1: back to like a simpler model of all this is

1248
01:07:09,080 --> 01:07:12,200
Speaker 1: our episode on the Great Eyeball Wars and our social

1249
01:07:12,240 --> 01:07:16,400
Speaker 1: media and our smartphones. It's like we can certainly throw

1250
01:07:16,440 --> 01:07:19,800
Speaker 1: our phone into a pond and then head off into

1251
01:07:19,840 --> 01:07:21,840
Speaker 1: the woods and try and live there, but most of

1252
01:07:21,920 --> 01:07:26,720
Speaker 1: us probably cannot um cannot follow that path. Therefore, we

1253
01:07:26,760 --> 01:07:29,560
Speaker 1: just have the best manage what we have. Yeah, wouldn't

1254
01:07:29,560 --> 01:07:32,520
Speaker 1: it be better to try to encourage the development of

1255
01:07:32,560 --> 01:07:35,040
Speaker 1: a phone full of apps that helped you fulfill your

1256
01:07:35,080 --> 01:07:38,680
Speaker 1: goals and aligned with your values? All right, So we'll

1257
01:07:38,760 --> 01:07:41,160
Speaker 1: end it there. I feel like we not only have

1258
01:07:41,280 --> 01:07:44,080
Speaker 1: we provided food for thought here, we've provided a just

1259
01:07:44,160 --> 01:07:48,800
Speaker 1: a buffet of thought provoking ideas. And I know that

1260
01:07:48,880 --> 01:07:51,960
Speaker 1: everyone out there, all of you conscious listeners, are going

1261
01:07:52,040 --> 01:07:55,360
Speaker 1: to have something to contribute to this conversation and we

1262
01:07:55,360 --> 01:07:58,280
Speaker 1: would love to hear from you and interact with you.

1263
01:07:58,280 --> 01:07:59,919
Speaker 1: You can find us in a number of ways. First

1264
01:07:59,920 --> 01:08:01,440
Speaker 1: of all, stuff to Blow your Mind dot com is

1265
01:08:01,480 --> 01:08:03,600
Speaker 1: the mothership. That's where you will find all of the

1266
01:08:03,640 --> 01:08:06,360
Speaker 1: podcast episodes. You'll find blog posts, and you will find

1267
01:08:06,400 --> 01:08:10,880
Speaker 1: links out to our various social media accounts such as Facebook, Twitter, Instagram,

1268
01:08:11,440 --> 01:08:15,720
Speaker 1: and just basic contact information for us. Thanks as always

1269
01:08:15,720 --> 01:08:18,879
Speaker 1: to our excellent audio producers Alex Williams and Tory Harrison.

1270
01:08:18,920 --> 01:08:20,639
Speaker 1: If you would like to get in touch with us

1271
01:08:20,640 --> 01:08:23,240
Speaker 1: with feedback about this episode or any other, to suggest

1272
01:08:23,320 --> 01:08:25,720
Speaker 1: a topic for the future, or just to say hi,

1273
01:08:25,840 --> 01:08:27,519
Speaker 1: let's know how you found out about the show, you

1274
01:08:27,560 --> 01:08:30,280
Speaker 1: can always email us at Blow the Mind at how

1275
01:08:30,360 --> 01:08:42,759
Speaker 1: stuff works dot com for more on this and thousands

1276
01:08:42,800 --> 01:09:00,000
Speaker 1: of other topics. Is it how stuff works dot com.

1277
01:08:53,280 --> 01:09:04,720
Speaker 1: Boo to tow three part proper fa