Popular in Law

Human Rights Law

Human Rights and Corruption: Problems and Potential of Individualising a Systemic Problem

Max Planck Institute Anne Peters

Law

A statistical approach to law school citation rankings

University of Virginia Joshua Fischman

Constitutional Law

What is constitutional ideology?

University of Oxford Nick Barber

Civil Procedure Law

Politics, Identity, and Class Certification on the U.S. Courts of Appeals

University of Pennsylvania Sean Farhang

Law

Diversity and Social Mobility in the UK legal profession

Law Society I. Stephanie Boyce

Internet Law

The Law of AI for Good

University of San Diego Orly Lobel

Human Rights Law

Transforming Restorative Justice for Transitional Settings

University of Nottingham Kerry Clamp

The Line: AI & The Future of Personhood

View more insights in Books - Professional - Law - Internet Law

View more figures from Duke University

View the page of James Boyle

Read the Study

Next in Books

Freedom’s Dominion

Jefferson Cowie | Vanderbilt University

Queer Lasting: Ecologies of Care for a Dying World

Sarah Ensor | University of Wisconsin Madison

Is ChatGPT conscious, or is it simply performing an intricate form of mimicry? What if, in the future, claims of AI consciousness become more convincing? James Boyle delves into these pressing questions, examining the profound impact such developments could have on our understanding of personhood.

Image courtesy of interviewee. October 5, 2024

Log-in or Sign-up to Faculti
Currently viewing this subject insight as a guest. You have insight(s) remaining for this month. Login to view 8000+ figures on the platform.

Generating Workspace...

Export as PDF

In this Faculti interview, Boyle discusses his new book, The Line: AI & the Future of Personhood (forthcoming MIT Press Oct 22, 2024). Chatbots like ChatGPT have challenged human exceptionalism: we are no longer the only beings capable of generating language and ideas fluently. But is ChatGPT conscious? Or is it merely engaging in sophisticated mimicry? And what happens in the future if the claims to consciousness are more credible? Boyle explores what these changes might do to our concept of personhood, to “the line” we believe separates our species from the rest of the world, but also separates “persons” with legal rights from objects. The conversation ranges from Blake Lemoine, the Google engineer who (wrongly) thought the AI system he was working on was conscious, to whether in the future we will give AI systems some form of legal status for the reasons we do so with corporations – so they can sue and be sued. Boyle explains why science fiction treatments of nonhuman intelligence affect us so powerfully, what we learned from the nonhuman animal rights movement, and how AI might cause us to reassess humanity’s own consciousness.

Language:

Export as PDF

Export as PDF
Here are five key points extracted from the video:

1. **Belief in AI Sentience**: The video discusses the case of Blake Lemoine, a Google engineer who believed that the AI chatbot, LaMDA, was sentient. This belief raised questions about the ethical treatment of AI and whether conscious entities deserve rights, highlighting a growing debate on AI consciousness.

2. **Distinction Between Sentience and Chatbots**: The speaker emphasizes that current AI chatbots, like LaMDA, are not sentient. They operate as sophisticated sentence-generating machines, not conscious beings. This distinction is crucial as it relates to the moral and ethical implications of AI rights.

3. **Empathy and Testing in "Blade Runner"**: The video references the "Voight-Kampff test" in the film "Blade Runner," which measures empathy to determine if a being is human. This raises the philosophical question of who is truly being tested—humans or replicants—and highlights our tendency to project emotions onto non-human entities.

4. **Evolution of Intelligence and AI Rights**: The discussion brings up the potential future emergence of intelligent, language-using entities that may challenge our concepts of consciousness and rights. The possibility of AI achieving some form of consciousness raises complex ethical questions about whether AI could or should be granted rights similar to humans.

5. **Philosophical and Ethical Implications**: The speaker reflects on historical challenges to human exceptionalism, particularly in light of evolution, and suggests that the discourse around AI rights may mirror past debates about the rights of non-human animals and other marginalized groups. This invites an exploration of what it means to be conscious and the moral implications of our relationship with intelligent AI.

Export as PDF
Here are some key quotes distilled from the video:

1. **On AI Sentience**:
"Lemoine started believing that it was sentient... he started to say, are we acting rightly towards this? ... Mr. Lemoine was wrong. These are not sentient. These are complete to the sentence machines."

2. **The Age of Disagreement**:
"We have now entered the age when reasonable people are going to start disagreeing about whether or not AI are conscious and whether or not if they are conscious, they deserve rights."

3. **The Nature of Consciousness**:
"What if our consciousness is really rather like ChatGPT's? What if most of us go through life day to day, most of the time just basically in a coma, feeding back the most likely responses?"

4. **Blade Runner Analogy**:
"The test that's supposed to detect whether you are a non-human is a test of empathy for non-human animals... So we test you to whether you have enough empathy for non-human animals, and if you don't have enough, we kill you."

5. **Language and Morality**:
"Language enables both of those... That's what makes us human. That's what makes us special. And along came a chatbot that taught us that sentences do not imply sentience."

6. **The Future of AI**:
"The explosion of capabilities that we've seen... tell us that we should be humble about the possibility that actually much more powerful and much more real forms of intelligence could be generated."

7. **Evolution and Consciousness**:
"The magnitude of the questions that AI poses for us has not yet fully come into focus... you're just a bunch of ones and zeros. So how could you ever be conscious?"

8. **On Rights and Legal Personality**:
"We'll grant them this legal personality... We give you legal personality so we can sue you when the driverless car goes off the road."

9. **Possibility of AI Rights**:
"For the first time in our History, homo sapiens... might not be alone in philosophizing and creating art about the meaning of life entities that we create might one day have their own version... of how to behave rightly towards others."

10. **Caution Against Hubris**:
"I think we could all agree now that maybe he was wrong... And perhaps we should meditate on that and be a little bit humble about our implicit... Oh, this is just silly sci-fi stuff when we talk about AI."

These quotes summarize the main themes regarding AI sentience, empathy, language, moral considerations, and the potential implications of advanced AI in society.

X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:90000

1
00:00:00.075 --> 00:00:03.565
Lemoine is a Google engineer who was supposed to do,

2
00:00:03.565 --> 00:00:06.725
sort of social responsibility of their, their AI chatbot,

3
00:00:06.905 --> 00:00:08.565
lambda large language model.

4
00:00:09.185 --> 00:00:11.485
And he started believing that it was sentient.

5
00:00:11.745 --> 00:00:15.685
Uh, he was somewhat of a mystic. He read it some zen koans.

6
00:00:15.785 --> 00:00:19.245
It gave him fascinating interpretations of it, or seemed to,

7
00:00:19.405 --> 00:00:21.885
because of course it's not conscious, it's a chatbot.

8
00:00:22.465 --> 00:00:24.605
And he started to think there was a ghost in the machine.

9
00:00:24.785 --> 00:00:28.125
And he started to say, are we acting rightly towards this?

10
00:00:28.145 --> 00:00:31.485
And saying it publicly, I think, Google's AI is,

11
00:00:32.105 --> 00:00:34.045
is alive where upon Google fired him.

12
00:00:34.705 --> 00:00:38.885
Um, Mr. Lemoine was wrong. Uh, these are not sentient.

13
00:00:38.885 --> 00:00:41.205
These are complete to the sentence machines.

14
00:00:41.235 --> 00:00:43.645
They are not, conscious ai,

15
00:00:44.265 --> 00:00:46.245
but his story tells us something.

16
00:00:46.245 --> 00:00:48.805
And the something that it tells us is we have now entered

17
00:00:48.805 --> 00:00:51.925
the age when reasonable people are going to start

18
00:00:52.925 --> 00:00:57.405
disagreeing about whether or notis are conscious and whether

19
00:00:57.405 --> 00:01:00.085
or not if they are conscious, they deserve rights.

20
00:01:00.225 --> 00:01:01.645
So he's sort of a harbinger,

21
00:01:01.645 --> 00:01:04.445
even if a mistaken harbinger of what is to come.

22
00:01:04.765 --> 00:01:06.485
'cause he couldn't see how

23
00:01:07.665 --> 00:01:11.965
the particular insights it was giving him, the way

24
00:01:11.965 --> 00:01:14.045
that it would respond going, I've never heard

25
00:01:14.045 --> 00:01:15.405
that Cohen, that's interesting.

26
00:01:15.425 --> 00:01:16.645
Uh, the broken mirror.

27
00:01:16.745 --> 00:01:18.125
Or perhaps that's the idea

28
00:01:18.155 --> 00:01:20.845
that once your consciousness has become aware

29
00:01:20.845 --> 00:01:22.205
of itself, there's no going back.

30
00:01:22.205 --> 00:01:24.605
Just as the mirror is broken, you know, your vision

31
00:01:24.605 --> 00:01:25.845
of yourself will become aware.

32
00:01:26.265 --> 00:01:29.285
And Lemoine just couldn't imagine,

33
00:01:29.485 --> 00:01:32.485
a mere chatbot coming up with that in response to him.

34
00:01:32.485 --> 00:01:34.885
Obviously it spoke to something deep inside him.

35
00:01:35.585 --> 00:01:39.005
And as human beings we're deeply empathetic

36
00:01:39.225 --> 00:01:42.085
and at times anthropomorphic, I talk

37
00:01:42.085 --> 00:01:44.885
to my rumba vacuum cleaner, which is obviously insane.

38
00:01:45.385 --> 00:01:47.725
Um, and so that is one of our tendencies,

39
00:01:47.905 --> 00:01:49.605
and that's something that I think we're gonna see in the

40
00:01:49.605 --> 00:01:51.605
future because AI is going to get more

41
00:01:51.845 --> 00:01:53.685
and more human seeming.

42
00:01:54.545 --> 00:01:57.685
But maybe one day with a different AI architecture,

43
00:01:57.785 --> 00:02:00.685
it might also have a aspects of consciousness

44
00:02:00.715 --> 00:02:03.245
that we would recognize as real consciousness.

45
00:02:03.545 --> 00:02:05.725
The great thing about Blade Runner is that it's all revolves

46
00:02:05.745 --> 00:02:07.685
and, do android stream of electric sheep.

47
00:02:07.685 --> 00:02:09.885
The, the book from which it's based is, they're based

48
00:02:09.885 --> 00:02:12.325
around the Voit comp test, which,

49
00:02:12.465 --> 00:02:15.445
if you've seen the movie, is the way we test replicants

50
00:02:15.505 --> 00:02:17.125
to see whether they are replicants.

51
00:02:17.125 --> 00:02:19.165
And if we find they are, we kill them.

52
00:02:19.585 --> 00:02:20.805
Uh, the blade runner kills them.

53
00:02:21.575 --> 00:02:23.565
Ironically, and it's really ironic,

54
00:02:23.585 --> 00:02:24.685
and I think most people miss this.

55
00:02:25.265 --> 00:02:26.325
The test that's supposed

56
00:02:26.325 --> 00:02:30.325
to detect whether you are a non-human is a test of empathy

57
00:02:30.345 --> 00:02:32.165
for non-human animals.

58
00:02:33.065 --> 00:02:35.605
So we test you to whether you have enough empathy

59
00:02:35.605 --> 00:02:36.605
for non-human animals,

60
00:02:37.025 --> 00:02:38.965
and if you don't have enough, we kill you.

61
00:02:38.985 --> 00:02:40.605
So the question is, who's being tested here?

62
00:02:40.605 --> 00:02:42.165
Is it the replicant or is it us?

63
00:02:42.475 --> 00:02:45.325
Because we are basically saying, you're a non-human species

64
00:02:45.385 --> 00:02:47.605
and we need to have no empathy for you

65
00:02:47.605 --> 00:02:49.805
because failed our test about empathy.

66
00:02:50.265 --> 00:02:53.205
Uh, that is the movie in a nutshell.

67
00:02:53.585 --> 00:02:55.445
And I think it's the thing that we are going to have

68
00:02:55.445 --> 00:02:58.165
to grapple with as we try and understand whether

69
00:02:58.185 --> 00:03:00.610
or not there ever, ever could be a reasonable case

70
00:03:01.185 --> 00:03:04.645
for AI rights based on something like consciousness,

71
00:03:04.645 --> 00:03:06.045
something like human attributes,

72
00:03:06.155 --> 00:03:07.605
something like intelligence.

73
00:03:07.735 --> 00:03:08.925
Human beings have been trying

74
00:03:08.925 --> 00:03:11.885
to philosophize about their special place in the world.

75
00:03:11.905 --> 00:03:13.845
The reason that we see ourselves as different

76
00:03:13.845 --> 00:03:15.845
and non-human animals different than mere things.

77
00:03:16.945 --> 00:03:19.685
And the first thing we look at is we say, well,

78
00:03:19.685 --> 00:03:21.165
it's language, it's abstract language.

79
00:03:21.165 --> 00:03:22.445
And that's what Aristotle said.

80
00:03:22.925 --> 00:03:24.205
Language gives us the power to reason

81
00:03:24.815 --> 00:03:26.365
about how to do something.

82
00:03:26.525 --> 00:03:29.525
I see a grape on that vine. I will get a chair and reach it.

83
00:03:29.865 --> 00:03:31.845
And then also to reason about justice.

84
00:03:31.995 --> 00:03:35.405
Wait, it's my neighbor's vine, maybe I should not do that.

85
00:03:36.065 --> 00:03:37.525
And language enables both of those.

86
00:03:37.545 --> 00:03:39.325
And he says, so language enables morality,

87
00:03:39.725 --> 00:03:40.925
language enables planning.

88
00:03:41.605 --> 00:03:43.925
Language enables community. That's what makes us human.

89
00:03:43.925 --> 00:03:45.085
That's what makes us special.

90
00:03:45.785 --> 00:03:48.245
And along came a chatbot that taught us

91
00:03:48.245 --> 00:03:50.845
that sentences do not imply sentience.

92
00:03:51.665 --> 00:03:53.685
And even though the philosophers had said

93
00:03:53.745 --> 00:03:55.725
before that there could be no such thing

94
00:03:55.725 --> 00:03:56.965
as mechanical intelligence,

95
00:03:56.965 --> 00:03:58.965
and frankly, I disagree with that claim,

96
00:03:59.715 --> 00:04:01.765
they were right about the chatbot.

97
00:04:01.765 --> 00:04:05.405
The chatbot is indeed just mimicking what,

98
00:04:05.455 --> 00:04:06.525
human speech is like.

99
00:04:06.865 --> 00:04:09.765
But then that leaves us to say, okay, well language alone

100
00:04:10.495 --> 00:04:13.765
can't be what makes a special sentences do not imply sents,

101
00:04:14.465 --> 00:04:16.485
so therefore we have to come up with other things.

102
00:04:16.485 --> 00:04:17.765
Well, it's intentionality,

103
00:04:17.875 --> 00:04:20.245
it's understanding the language from the inside.

104
00:04:20.275 --> 00:04:22.325
It's semantics, not just syntax.

105
00:04:23.275 --> 00:04:25.165
It's experience in the world.

106
00:04:25.385 --> 00:04:28.325
And so one thing we could do is sort of keep on refining

107
00:04:28.675 --> 00:04:30.005
what makes human special,

108
00:04:30.405 --> 00:04:31.485
although up for a while, it starts

109
00:04:31.605 --> 00:04:32.645
to seem a little defensive.

110
00:04:32.645 --> 00:04:34.645
Do you keep redrawing the line just

111
00:04:34.645 --> 00:04:35.885
to make sure you're incited?

112
00:04:36.075 --> 00:04:37.885
Another thing we might do is go, wow,

113
00:04:37.915 --> 00:04:41.245
what if our consciousness is really rather like chat GPTs?

114
00:04:41.245 --> 00:04:44.085
What if most of us go through life day to day, most

115
00:04:44.105 --> 00:04:46.405
of the time just basically in a coma,

116
00:04:46.405 --> 00:04:48.245
feeding back the most likely responses.

117
00:04:49.065 --> 00:04:50.845
And another thing we could do is say,

118
00:04:51.135 --> 00:04:54.285
maybe there are attributes that we have that could be shared

119
00:04:54.345 --> 00:04:57.405
by other entities, biological or non-biological.

120
00:04:57.825 --> 00:04:59.925
And that's the path that I'm exploring in the book.

121
00:05:00.265 --> 00:05:03.405
And one way that we might get AI rights,

122
00:05:03.465 --> 00:05:06.765
the current large language model, of generative ai,

123
00:05:06.765 --> 00:05:08.805
is not gonna produce anything like consciousness,

124
00:05:09.185 --> 00:05:12.485
but the explosion of capabilities that we've seen

125
00:05:12.715 --> 00:05:15.365
that not even people inside the field predicted,

126
00:05:15.885 --> 00:05:18.405
I think tell us that we should be humble about the

127
00:05:18.405 --> 00:05:21.485
possibility that actually much more powerful

128
00:05:21.485 --> 00:05:24.925
and much more real forms of intelligence could be generated.

129
00:05:25.305 --> 00:05:27.925
So yes, there will be a process of imitation,

130
00:05:27.985 --> 00:05:30.485
but it's also possible that scientists could try

131
00:05:30.485 --> 00:05:33.445
and say, you know what, two very famous AI scientists,

132
00:05:33.445 --> 00:05:34.445
Norvig and Russell said,

133
00:05:34.745 --> 00:05:36.965
humans didn't learn PA powered flight by trying

134
00:05:36.965 --> 00:05:38.245
to flap their wings like birds.

135
00:05:38.345 --> 00:05:39.405
And they didn't make planes

136
00:05:39.405 --> 00:05:40.525
that flap their wings like birds.

137
00:05:40.525 --> 00:05:41.645
They tried to figure out flight

138
00:05:41.645 --> 00:05:43.765
and then they injure near directly rather than saying,

139
00:05:44.105 --> 00:05:45.285
can I imitate a bird?

140
00:05:45.735 --> 00:05:49.245
Maybe that's the way we get ai not imitating human beings,

141
00:05:49.465 --> 00:05:52.485
but actually pursuing different lines about how

142
00:05:52.485 --> 00:05:55.565
to achieve problem solving, ethical awareness, even humor.

143
00:05:56.105 --> 00:05:57.325
And whether or not those things,

144
00:05:57.325 --> 00:05:59.245
even if they're not done the way we do them,

145
00:05:59.765 --> 00:06:01.285
actually present to us as something

146
00:06:01.285 --> 00:06:04.085
that looks like a convincing, a convincing sapiens.

147
00:06:04.485 --> 00:06:07.005
I do think the theory of evolution was this great disruption

148
00:06:07.145 --> 00:06:08.365
of humans,

149
00:06:08.695 --> 00:06:10.765
scene centrality at the center of the universe.

150
00:06:11.065 --> 00:06:13.045
Of course you can be, religious

151
00:06:13.045 --> 00:06:14.565
and also believe in evolution.

152
00:06:14.565 --> 00:06:15.965
There are lots of people who do,

153
00:06:16.265 --> 00:06:18.165
but at the time, this was seen as a challenge.

154
00:06:18.305 --> 00:06:20.765
The, you mean we evolved from protozoa?

155
00:06:21.075 --> 00:06:23.845
Like how on earth protozoa aren't conscious.

156
00:06:24.145 --> 00:06:26.205
We are conscious. How could we have come from there?

157
00:06:26.705 --> 00:06:29.405
You must be wrong because a thinking power

158
00:06:29.465 --> 00:06:32.005
of a mind could not come from pond slime.

159
00:06:32.545 --> 00:06:34.685
And effectively we are saying the same things

160
00:06:34.685 --> 00:06:35.805
about ai right now.

161
00:06:35.985 --> 00:06:37.885
You're just a bunch of ones and zeros.

162
00:06:38.825 --> 00:06:40.365
So how could you ever be conscious?

163
00:06:41.105 --> 00:06:45.125
And I think that human beings found a challenge to the sense

164
00:06:45.125 --> 00:06:46.765
that we were different than the other animals.

165
00:06:46.765 --> 00:06:47.965
After all, they had evolved too.

166
00:06:48.625 --> 00:06:51.765
And I think we actually took us maybe a hundred years

167
00:06:51.785 --> 00:06:53.845
and hey, people are still denying evolution.

168
00:06:54.465 --> 00:06:56.165
It, it took us a while to deal with that.

169
00:06:56.185 --> 00:06:59.325
And I think that the magnitude of the questions

170
00:06:59.325 --> 00:07:02.285
that AI poses for us has not yet fully come into focus.

171
00:07:02.375 --> 00:07:05.285
There is neither a case that they are ethically entitled

172
00:07:05.285 --> 00:07:07.765
to it because they're fully conscious, they aren't,

173
00:07:07.765 --> 00:07:09.485
they're not general AI right now.

174
00:07:10.105 --> 00:07:12.845
Um, nor is there an argument like the one we use

175
00:07:12.845 --> 00:07:14.445
for corporate personhood, which is,

176
00:07:14.555 --> 00:07:17.405
it's not like I recognize the deep moral kinship

177
00:07:17.475 --> 00:07:18.805
with IBM or Google.

178
00:07:19.475 --> 00:07:22.485
It's that I think there's reasons why I want them to be able

179
00:07:22.485 --> 00:07:25.725
to make contracts to sue, to, to have deals

180
00:07:25.905 --> 00:07:27.125
to persist over time.

181
00:07:27.625 --> 00:07:29.685
And so we'll grant them this legal personality.

182
00:07:30.105 --> 00:07:31.165
And some people have said, well,

183
00:07:31.165 --> 00:07:33.685
we might get AI personality by that route.

184
00:07:34.145 --> 00:07:35.845
We give you legal personality

185
00:07:35.905 --> 00:07:38.925
so we can sue you when the driverless car goes off the road.

186
00:07:39.505 --> 00:07:40.725
And that's definitely possible.

187
00:07:40.975 --> 00:07:43.805
Right now, we have neither the, the reason to do it

188
00:07:43.805 --> 00:07:45.965
for reasons of convenience pragmatism

189
00:07:46.425 --> 00:07:51.285
nor for reasons of conscience or empathy, morality.

190
00:07:52.185 --> 00:07:55.005
But my point in my book is that could change

191
00:07:55.105 --> 00:07:56.325
and change relatively quickly.

192
00:07:56.565 --> 00:07:58.725
I didn't know I'd end up writing about corporate personhood.

193
00:07:58.725 --> 00:08:01.085
That was a surprise to me. I didn't know I'd end up writing

194
00:08:01.085 --> 00:08:02.325
about non-human animals,

195
00:08:02.345 --> 00:08:03.925
but I realized, well, that's a group

196
00:08:03.925 --> 00:08:06.365
that we have differentiated against self against

197
00:08:06.985 --> 00:08:10.445
the line I'm talking about is the line between us

198
00:08:10.505 --> 00:08:11.805
and the rest of the world.

199
00:08:12.625 --> 00:08:14.805
And as I read about both of those

200
00:08:14.825 --> 00:08:18.085
and I studied them, I was really quite surprised in the case

201
00:08:18.085 --> 00:08:20.605
of non-human animals, I don't think the case

202
00:08:20.705 --> 00:08:23.085
for full non-human animal personhood has been made.

203
00:08:23.085 --> 00:08:25.645
But I do think that arguments to be made on the part

204
00:08:25.645 --> 00:08:27.205
of the great apes and the citations

205
00:08:27.595 --> 00:08:29.605
that we should give them a bigger

206
00:08:29.605 --> 00:08:30.925
legal status than they currently have.

207
00:08:30.925 --> 00:08:32.365
And I think that those are, you know,

208
00:08:32.365 --> 00:08:34.125
quite convincing ethical arguments.

209
00:08:34.425 --> 00:08:37.285
Um, and so I for 1:00 AM convinced by that as

210
00:08:37.285 --> 00:08:39.565
for corporate personhood, I was actually shocked by

211
00:08:39.625 --> 00:08:42.325
how little good reasoning there was not about

212
00:08:42.325 --> 00:08:43.645
that we need corporate personhood.

213
00:08:43.765 --> 00:08:45.165
I actually tend to agree.

214
00:08:45.825 --> 00:08:47.885
But the corporate personhood needs to come

215
00:08:47.885 --> 00:08:50.005
with constitutional rights attached to it, rights

216
00:08:50.005 --> 00:08:51.565
of free speech, equal protection and so forth.

217
00:08:52.025 --> 00:08:53.325
And even Justice Rehnquist,

218
00:08:53.325 --> 00:08:54.605
very conservative Supreme Court justice

219
00:08:54.715 --> 00:08:56.445
Said that decision was made

220
00:08:56.445 --> 00:08:58.365
with neither argument nor discussion.

221
00:08:58.585 --> 00:08:59.605
And it's true, we kind

222
00:08:59.605 --> 00:09:02.925
of fumbled our way into it backwards. I don't want that to

223
00:09:02.925 --> 00:09:03.925
Happen with ai. Last

224
00:09:03.925 --> 00:09:05.005
paragraph of the book,

225
00:09:05.545 --> 00:09:07.845
For the first time in our History, homo sapiens,

226
00:09:07.845 --> 00:09:11.085
the hubristically self labeled thinking animal might not be

227
00:09:11.135 --> 00:09:12.365
alone in philosophizing

228
00:09:12.465 --> 00:09:15.365
and creating art about the meaning of life entities.

229
00:09:15.385 --> 00:09:17.165
We create might one day have their own version

230
00:09:17.185 --> 00:09:19.805
of Flo bearer's lament about the inadequacy of language,

231
00:09:19.975 --> 00:09:21.365
their own questions about how

232
00:09:21.365 --> 00:09:22.845
to behave rightly towards others.

233
00:09:23.715 --> 00:09:25.245
They might even try and Draw the kinds

234
00:09:25.245 --> 00:09:26.645
of lines I've talked about in this book

235
00:09:26.825 --> 00:09:29.045
and find it just as difficult as we do. As I've

236
00:09:29.045 --> 00:09:31.405
Tried to show there are Enormous uncertainties about

237
00:09:31.535 --> 00:09:34.765
every dimension of such a future from what we come

238
00:09:34.765 --> 00:09:36.325
to believe about the meaning of consciousness

239
00:09:36.945 --> 00:09:39.645
To the speed and shape of our technological development.

240
00:09:39.645 --> 00:09:42.365
Still it's possible. And think about what that means.

241
00:09:42.575 --> 00:09:45.605
There might be an entirely new group of highly intelligent,

242
00:09:45.795 --> 00:09:48.845
Conscious, Abstract language using

243
00:09:48.995 --> 00:09:51.925
Persons joining us on the planet. Are they, will

244
00:09:51.925 --> 00:09:53.925
They, can we risk finding out?

245
00:09:54.585 --> 00:09:56.405
It would take a poor crabbed

246
00:09:56.405 --> 00:09:57.725
And stunted imagination

247
00:09:57.825 --> 00:09:59.565
to find no wonder in those questions,

248
00:09:59.915 --> 00:10:03.285
despite all the uncertainties, you know, 11-year-old boy

249
00:10:03.285 --> 00:10:04.525
who grew up reading science fiction.

250
00:10:05.035 --> 00:10:08.965
Part of that is a person who in his old life own lifetime,

251
00:10:09.115 --> 00:10:12.965
have seen questions that people once said is this is morally

252
00:10:12.965 --> 00:10:14.125
Trivial. You shouldn't be

253
00:10:14.125 --> 00:10:15.125
Talking about this. Why

254
00:10:15.125 --> 00:10:17.285
are you talking about gay rights? That's a

255
00:10:17.355 --> 00:10:19.245
Lifestyle issue. You

256
00:10:19.245 --> 00:10:20.725
Are sullying the language of

257
00:10:20.725 --> 00:10:22.125
Human rights by using that.

258
00:10:22.985 --> 00:10:25.165
My law school dean said that to me when I was agitating

259
00:10:25.165 --> 00:10:26.685
for equal rights for our gay students.

260
00:10:27.265 --> 00:10:29.525
And I think we could all agree now that maybe he was wrong

261
00:10:30.145 --> 00:10:32.405
or people who talked about climate change back in the

262
00:10:32.645 --> 00:10:34.325
seventies, sixties and seventies, people were

263
00:10:34.325 --> 00:10:35.405
Going, this is ludicrous.

264
00:10:35.405 --> 00:10:37.525
You're taking our eyes off the war,

265
00:10:37.585 --> 00:10:39.005
off the civil rights movement.

266
00:10:39.305 --> 00:10:40.485
You know, this isn't important.

267
00:10:40.545 --> 00:10:42.205
Or people who talked about non-human animals,

268
00:10:42.205 --> 00:10:43.645
you're like, this is trivial.

269
00:10:43.645 --> 00:10:45.685
How can you care about animals when there are people

270
00:10:45.825 --> 00:10:47.165
who are dying and suffering?

271
00:10:47.865 --> 00:10:49.725
And I think we could look back now

272
00:10:49.725 --> 00:10:51.325
and go, well, none of those were

273
00:10:51.325 --> 00:10:52.685
Particularly clever responses.

274
00:10:53.195 --> 00:10:55.605
Perhaps we should meditate on that

275
00:10:55.625 --> 00:10:57.885
And be a little bit humble about our implicit.

276
00:10:58.305 --> 00:11:01.445
Oh, this is just silly sci-fi stuff when we talk about AI.

"The Sentience Debate: Reflections on AI Consciousness and Human Empathy"**

00:00:00 - Lemoine is a Google engineer who was supposed to do, sort of social responsibility of their, their AI chatbot, lambda large language model. And he started believing that it was sentient. he was somewhat of a mystic. He read it some zen koans. It gave him fascinating interpretations of it, or seemed to, because of course it's not conscious, it's a chatbot. And he started to think there was a ghost in the machine. And he started to say, are we acting rightly towards this? And saying it publicly, I think, Google's AI is, is alive where upon Google fired him. Mr. Lemoine was wrong. these are not sentient. These are complete to the sentence machines.
00:00:44 - but his story tells us something. And the something that it tells us is we have now entered the age when reasonable people are going to start disagreeing about whether or notis are conscious and whether or not if they are conscious, they deserve rights. So he's sort of a harbinger, even if a mistaken harbinger of what is to come. 'cause he couldn't see how the particular insights it was giving him, the way that it would respond going, I've never heard that Cohen, that's interesting. the broken mirror. Or perhaps that's the idea that once your consciousness has become aware of itself, there's no going back.
00:01:24 - of yourself will become aware. And Lemoine just couldn't imagine, a mere chatbot coming up with that in response to him. Obviously it spoke to something deep inside him. And as human beings we're deeply empathetic and at times anthropomorphic, I talk to my rumba vacuum cleaner, which is obviously insane. and so that is one of our tendencies, and that's something that I think we're gonna see in the future because AI is going to get more and more human seeming. But maybe one day with a different AI architecture, it might also have a aspects of consciousness that we would recognize as real consciousness.
00:02:05 - and, do android stream of electric sheep. The, the book from which it's based is, they're based around the Voit comp test, which, if you've seen the movie, is the way we test replicants to see whether they are replicants. And if we find they are, we kill them. the blade runner kills them. Ironically, and it's really ironic, and I think most people miss this. The test that's supposed to detect whether you are a non-human is a test of empathy for non-human animals. So we test you to whether you have enough empathy for non-human animals, and if you don't have enough, we kill you. So the question is, who's being tested here? Is it the replicant or is it us? Because we are basically saying, you're a non-human species
00:02:47 - because failed our test about empathy. that is the movie in a nutshell. And I think it's the thing that we are going to have to grapple with as we try and understand whether or not there ever, ever could be a reasonable case for AI rights based on something like consciousness, something like human attributes, something like intelligence. Human beings have been trying to philosophize about their special place in the world. The reason that we see ourselves as different and non-human animals different than mere things. And the first thing we look at is we say, well, it's language, it's abstract language. And that's what Aristotle said. Language gives us the power to reason about how to do something.
00:03:29 - And then also to reason about justice. Wait, it's my neighbor's vine, maybe I should not do that. And language enables both of those. And he says, so language enables morality, language enables planning. Language enables community. That's what makes us human. That's what makes us special. And along came a chatbot that taught us that sentences do not imply sentience. And even though the philosophers had said before that there could be no such thing as mechanical intelligence, and frankly, I disagree with that claim, they were right about the chatbot. The chatbot is indeed just mimicking what, human speech is like.

"Redefining Consciousness: The Intersection of AI, Language, and Moral Considerations"**

00:04:10 - can't be what makes a special sentences do not imply sents, so therefore we have to come up with other things. Well, it's intentionality, it's understanding the language from the inside. It's semantics, not just syntax. It's experience in the world. And so one thing we could do is sort of keep on refining what makes human special, although up for a while, it starts to seem a little defensive. Do you keep redrawing the line just to make sure you're incited? Another thing we might do is go, wow, what if our consciousness is really rather like chat GPTs? What if most of us go through life day to day, most of the time just basically in a coma, feeding back the most likely responses.
00:04:51 - maybe there are attributes that we have that could be shared by other entities, biological or non-biological. And that's the path that I'm exploring in the book. And one way that we might get AI rights, the current large language model, of generative ai, is not gonna produce anything like consciousness, but the explosion of capabilities that we've seen that not even people inside the field predicted, I think tell us that we should be humble about the possibility that actually much more powerful and much more real forms of intelligence could be generated. So yes, there will be a process of imitation, but it's also possible that scientists could try
00:05:33 - Norvig and Russell said, humans didn't learn PA powered flight by trying to flap their wings like birds. And they didn't make planes that flap their wings like birds. They tried to figure out flight and then they injure near directly rather than saying, can I imitate a bird? Maybe that's the way we get ai not imitating human beings, but actually pursuing different lines about how to achieve problem solving, ethical awareness, even humor. And whether or not those things, even if they're not done the way we do them, actually present to us as something that looks like a convincing, a convincing sapiens. I do think the theory of evolution was this great disruption of humans, scene centrality at the center of the universe.
00:06:13 - and also believe in evolution. There are lots of people who do, but at the time, this was seen as a challenge. The, you mean we evolved from protozoa? Like how on earth protozoa aren't conscious. We are conscious. How could we have come from there? You must be wrong because a thinking power of a mind could not come from pond slime. And effectively we are saying the same things about ai right now. You're just a bunch of ones and zeros. So how could you ever be conscious? And I think that human beings found a challenge to the sense that we were different than the other animals. After all, they had evolved too. And I think we actually took us maybe a hundred years
00:06:54 - It, it took us a while to deal with that. And I think that the magnitude of the questions that AI poses for us has not yet fully come into focus. There is neither a case that they are ethically entitled to it because they're fully conscious, they aren't, they're not general AI right now. nor is there an argument like the one we use for corporate personhood, which is, it's not like I recognize the deep moral kinship with IBM or Google. It's that I think there's reasons why I want them to be able to make contracts to sue, to, to have deals to persist over time. And so we'll grant them this legal personality. And some people have said, well, we might get AI personality by that route.
00:07:35 - so we can sue you when the driverless car goes off the road. And that's definitely possible. Right now, we have neither the, the reason to do it for reasons of convenience pragmatism nor for reasons of conscience or empathy, morality. But my point in my book is that could change and change relatively quickly. I didn't know I'd end up writing about corporate personhood. That was a surprise to me. I didn't know I'd end up writing about non-human animals, but I realized, well, that's a group that we have differentiated against self against the line I'm talking about is the line between us and the rest of the world. And as I read about both of those

"Rethinking Ethical Considerations: Non-Human Personhood and the Future of AI Rights"

00:08:18 - of non-human animals, I don't think the case for full non-human animal personhood has been made. But I do think that arguments to be made on the part of the great apes and the citations that we should give them a bigger legal status than they currently have. And I think that those are, you know, quite convincing ethical arguments. and so I for 1:00 AM convinced by that as for corporate personhood, I was actually shocked by how little good reasoning there was not about that we need corporate personhood. I actually tend to agree. But the corporate personhood needs to come with constitutional rights attached to it, rights of free speech, equal protection and so forth. And even Justice Rehnquist, very conservative Supreme Court justice Said that decision was made
00:08:58 - And it's true, we kind of fumbled our way into it backwards. I don't want that to Happen with ai. Last paragraph of the book, For the first time in our History, homo sapiens, the hubristically self labeled thinking animal might not be alone in philosophizing and creating art about the meaning of life entities. We create might one day have their own version of Flo bearer's lament about the inadequacy of language, their own questions about how to behave rightly towards others. They might even try and Draw the kinds of lines I've talked about in this book and find it just as difficult as we do. As I've Tried to show there are Enormous uncertainties about every dimension of such a future from what we come to believe about the meaning of consciousness
00:09:39 - Still it's possible. And think about what that means. There might be an entirely new group of highly intelligent, Conscious, Abstract language using Persons joining us on the planet. Are they, will They, can we risk finding out? It would take a poor crabbed And stunted imagination to find no wonder in those questions, despite all the uncertainties, you know, 11-year-old boy who grew up reading science fiction. Part of that is a person who in his old life own lifetime, have seen questions that people once said is this is morally Trivial. You shouldn't be Talking about this. Why are you talking about gay rights? That's a
00:10:19 - Are sullying the language of Human rights by using that. My law school dean said that to me when I was agitating for equal rights for our gay students. And I think we could all agree now that maybe he was wrong or people who talked about climate change back in the seventies, sixties and seventies, people were Going, this is ludicrous. You're taking our eyes off the war, off the civil rights movement. You know, this isn't important. Or people who talked about non-human animals, you're like, this is trivial. How can you care about animals when there are people who are dying and suffering? And I think we could look back now and go, well, none of those were Particularly clever responses. Perhaps we should meditate on that And be a little bit humble about our implicit.
00:10:58 -

Faculti AI can make mistakes

Research

Policy

Analysis