1
00:00:00,000 --> 00:00:26,680
Hey everybody, welcome to episode number 8 of the KMO Show.

2
00:00:26,680 --> 00:00:32,400
I'm your host, KMO, and I have got a whining puppy in a cage that I can't leave very long,

3
00:00:32,400 --> 00:00:34,120
so this is going to be a very short introduction.

4
00:00:34,120 --> 00:00:38,360
I'll just say that I'm speaking to the first repeat guest on this podcast.

5
00:00:38,360 --> 00:00:39,560
It is Kevin Wohlmut.

6
00:00:39,560 --> 00:00:44,440
He was on the first episode, along with Michael Garfield, and that was a trialogue, basically

7
00:00:44,440 --> 00:00:46,960
a dialogue, but with three people.

8
00:00:46,960 --> 00:00:52,440
So we're going to start off addressing a series of tweets by Eliezer Yudkowski, the founder

9
00:00:52,440 --> 00:00:56,920
and, I don't know what his exact title is, but the chief poobah at the Machine Intelligence

10
00:00:56,920 --> 00:01:02,480
Research Institute, which is an organization which is dedicated to the concept of AI safety

11
00:01:02,480 --> 00:01:04,680
and AI alignment.

12
00:01:04,680 --> 00:01:09,400
But we're not going to be talking about big existential risks to humanity from artificial

13
00:01:09,400 --> 00:01:10,400
intelligence.

14
00:01:10,400 --> 00:01:13,000
If those, this is my opinion, don't quote me.

15
00:01:13,000 --> 00:01:16,040
I mean, you can quote me if you want, but as long as you put in the proviso, this is

16
00:01:16,040 --> 00:01:18,200
my opinion and I'm just spitballing.

17
00:01:18,200 --> 00:01:26,920
I don't think that AI, you know, taking control of robotic bodies or nuclear arsenals or actively

18
00:01:26,920 --> 00:01:30,720
doing harm to humanity is anything that we need to worry about in the foreseeable future.

19
00:01:30,720 --> 00:01:35,560
I think what we need to worry about will be the unintended consequences of rushing these

20
00:01:35,560 --> 00:01:41,840
systems to the market, embedding them into all sorts of software products without much

21
00:01:41,840 --> 00:01:46,640
testing because of an intense need to do so.

22
00:01:46,640 --> 00:01:50,920
There's a presentation that you can watch on YouTube, but you can also get it as a podcast.

23
00:01:50,920 --> 00:01:55,160
It's called the AI Dilemma and we're going to mention it in the upcoming conversation.

24
00:01:55,160 --> 00:02:01,440
It's with Tristan Harris and Asa Raskin and in it they say, well, let me just go to their

25
00:02:01,440 --> 00:02:03,900
website and read this aloud.

26
00:02:03,900 --> 00:02:07,440
People who've heard the most recent episode of the Sea-Rome Vault podcast will already

27
00:02:07,440 --> 00:02:13,320
have heard this, but that podcast is behind a paywall and this is very important.

28
00:02:13,320 --> 00:02:16,400
So I'm going to repeat myself.

29
00:02:16,400 --> 00:02:23,920
Speaking from humane tech.com slash podcast slash the AI dilemma with a dash between each

30
00:02:23,920 --> 00:02:24,920
word.

31
00:02:24,920 --> 00:02:30,060
Half of all AI researchers believe there's a 10% or greater chance that humans will go

32
00:02:30,060 --> 00:02:33,600
extinct from their inability to control AI.

33
00:02:33,600 --> 00:02:38,480
When we invent a new technology, we uncover a new class of responsibility.

34
00:02:38,480 --> 00:02:42,720
If that technology confers power, it will start a race.

35
00:02:42,720 --> 00:02:46,560
And if we don't coordinate, the race will end in tragedy.

36
00:02:46,560 --> 00:02:53,040
Well, the race in question here is the race to develop powerful AI systems.

37
00:02:53,040 --> 00:02:59,880
Artificial general intelligence to start with and probably soon thereafter, artificial superintelligence.

38
00:02:59,880 --> 00:03:03,620
So at the very beginning of the conversation with Kevin Wollman, we will run through Eliezer

39
00:03:03,620 --> 00:03:10,720
Yudkowski's 17 reasons why artificial intelligence is more dangerous than nuclear weapons.

40
00:03:10,720 --> 00:03:21,040
Here we go.

41
00:03:21,040 --> 00:03:22,040
We're talking about AI.

42
00:03:22,040 --> 00:03:23,040
Hey everybody.

43
00:03:23,040 --> 00:03:26,120
I am here with Kevin Wollman.

44
00:03:26,120 --> 00:03:29,920
When I say here, I mean in cyberspace, of course, because he's in New Mexico and I'm

45
00:03:29,920 --> 00:03:30,920
in Arkansas.

46
00:03:30,920 --> 00:03:36,640
But Kevin, you are the first repeat guest on the KMO show.

47
00:03:36,640 --> 00:03:37,720
I am honored.

48
00:03:37,720 --> 00:03:38,720
Thank you.

49
00:03:38,720 --> 00:03:42,880
Have that carved in your tombstone.

50
00:03:42,880 --> 00:03:43,880
Hopefully not anytime soon.

51
00:03:43,880 --> 00:03:48,800
I never thought I would achieve this honor when I first started listening to the CRL.

52
00:03:48,800 --> 00:03:53,040
So it just goes to say, hey, show, follow your dreams.

53
00:03:53,040 --> 00:03:56,360
Or at least listen to obscure podcasts.

54
00:03:56,360 --> 00:03:58,880
Alrighty.

55
00:03:58,880 --> 00:04:00,400
We're talking about artificial intelligence.

56
00:04:00,400 --> 00:04:05,880
I think it's a topic that I'm maybe following more closely than you are, but you are definitely

57
00:04:05,880 --> 00:04:09,760
following along and posting good comments on YouTube.

58
00:04:09,760 --> 00:04:18,040
I would like to start by revisiting Eliezer Yudkowski's 17 reasons why AI is more dangerous

59
00:04:18,040 --> 00:04:20,320
than nuclear weapons.

60
00:04:20,320 --> 00:04:26,520
So for those who don't know, Eliezer Yudkowski has been interested in AI safety for decades.

61
00:04:26,520 --> 00:04:32,720
And he has an organization that employs people and they really work on the topic and they're

62
00:04:32,720 --> 00:04:38,880
generously funded because a lot of his Silicon Valley folk had, you know, they were early

63
00:04:38,880 --> 00:04:41,400
investors in Bitcoin and made a bunch of money.

64
00:04:41,400 --> 00:04:44,720
So he's actually, you know, doing serious work.

65
00:04:44,720 --> 00:04:53,120
But in terms of like the being the face of AI safety concerns, maybe not the best choice

66
00:04:53,120 --> 00:04:58,800
because he's kind of off putting in his personality and his presentation.

67
00:04:58,800 --> 00:05:06,960
But on Twitter he posted 10 obvious reasons that the danger from AGI or artificial general

68
00:05:06,960 --> 00:05:10,360
intelligence is way more serious than nuclear weapons.

69
00:05:10,360 --> 00:05:11,680
So let's let's alternate.

70
00:05:11,680 --> 00:05:13,240
I'll read one than you read one.

71
00:05:13,240 --> 00:05:21,080
So one, nuclear weapons are not smarter than humanity.

72
00:05:21,080 --> 00:05:25,880
Number two, nuclear weapons are not self replicating.

73
00:05:25,880 --> 00:05:29,600
Three, nuclear weapons are not self improving.

74
00:05:29,600 --> 00:05:35,120
Four, scientists understand how nuclear weapons work.

75
00:05:35,120 --> 00:05:41,480
Five, you can calculate how powerful a nuclear weapon will be before setting it off.

76
00:05:41,480 --> 00:05:48,360
Six, a realistic full nuclear exchange between two nuclear powers wouldn't extinguish literally

77
00:05:48,360 --> 00:05:50,760
all of humanity.

78
00:05:50,760 --> 00:05:56,360
Seven, it would be hard to do a full nuclear exchange by accident and without any human

79
00:05:56,360 --> 00:05:59,720
being having decided to do that.

80
00:05:59,720 --> 00:06:07,040
Eight, the materials and factories for building nuclear weapons are relatively easy to spot.

81
00:06:07,040 --> 00:06:12,160
Nine, the process for making one nuclear weapon doesn't let you deploy a hundred thousand

82
00:06:12,160 --> 00:06:14,440
of them immediately after.

83
00:06:14,440 --> 00:06:19,200
Ten, humanity understands that nuclear weapons are dangerous.

84
00:06:19,200 --> 00:06:25,560
Humans treat them seriously and leading scientists can have actual conversations about the dangers.

85
00:06:25,560 --> 00:06:31,400
Eleven, there are not dozens of venture backed companies trying to scale privately owned

86
00:06:31,400 --> 00:06:33,640
nuclear weapons further.

87
00:06:33,640 --> 00:06:37,400
Number twelve, countries have plans for dealing with the danger posed by strategic nuclear

88
00:06:37,400 --> 00:06:42,520
armaments and the plans may not be perfect but they make sense and are not made completely

89
00:06:42,520 --> 00:06:47,760
out of deranged hopium like quote, oh, we'll be safe so long as everyone has open sourced

90
00:06:47,760 --> 00:06:49,160
nuclear weapons.

91
00:06:49,160 --> 00:06:50,600
That's good.

92
00:06:50,600 --> 00:06:55,040
Number thirteen, most people are not tempted to anthropomorphize nuclear weapons nor to

93
00:06:55,040 --> 00:07:00,880
vastly overestimate their own predictive abilities based on anthropomorphic or mechanomorphic

94
00:07:00,880 --> 00:07:02,480
models.

95
00:07:02,480 --> 00:07:08,200
Number fourteen, people think about nuclear weapons as if they are ultimately ordinary,

96
00:07:08,200 --> 00:07:13,540
causal stuff and not as if they go into a weird separate psychological magisterium which

97
00:07:13,540 --> 00:07:18,440
would produce responses like quote, isn't the dangerous strategic nuclear weapons just

98
00:07:18,440 --> 00:07:23,000
a distraction from the use of radioisotopes in medicine unquote.

99
00:07:23,000 --> 00:07:27,540
Fifteen, nuclear weapons are in fact pretty easy to understand.

100
00:07:27,540 --> 00:07:30,080
They make enormous, poisonous explosions.

101
00:07:30,080 --> 00:07:31,180
And that's it.

102
00:07:31,180 --> 00:07:35,760
They have some internally complicated machinery but the details don't affect the outer impact

103
00:07:35,760 --> 00:07:38,280
and meaning of nuclear weapons.

104
00:07:38,280 --> 00:07:44,760
Sixteen, eminent physicists don't publicly mock the idea that constructing a strategic

105
00:07:44,760 --> 00:07:50,120
nuclear arsenal could possibly in some way be dangerous or go less than completely well

106
00:07:50,120 --> 00:07:52,120
for humanity.

107
00:07:52,120 --> 00:07:56,560
And number seventeen, when somebody raised the concern that maybe the first nuclear explosion

108
00:07:56,560 --> 00:08:02,160
would ignite the atmosphere and kill everyone, it was promptly taken seriously by the physicists

109
00:08:02,160 --> 00:08:03,840
on the Manhattan Project.

110
00:08:03,840 --> 00:08:08,200
They did a physical calculation that they understood how to perform and correctly concluded

111
00:08:08,200 --> 00:08:12,160
that this could not possibly happen for several different independent reasons with lots of

112
00:08:12,160 --> 00:08:15,080
safety margin.

113
00:08:15,080 --> 00:08:19,520
All right I think it would take us pretty much the whole podcast to go through those

114
00:08:19,520 --> 00:08:26,200
point by point but let me ask you if there's any one of those that you'd like to address

115
00:08:26,200 --> 00:08:31,120
or if you have a comment about the body of reasoning as a whole.

116
00:08:31,120 --> 00:08:32,120
It's complex.

117
00:08:32,120 --> 00:08:35,840
So there's a lot of things in there that I think he gets right and a lot of things that

118
00:08:35,840 --> 00:08:36,840
get wrong.

119
00:08:36,840 --> 00:08:43,200
I don't have the control over the Twitter there but he says that politicians treat them

120
00:08:43,200 --> 00:08:46,960
seriously when I would argue that pushing Russia and Ukraine is not treating nuclear

121
00:08:46,960 --> 00:08:50,000
weapons seriously.

122
00:08:50,000 --> 00:08:52,760
Leading scientists can have actual conversations about the dangers.

123
00:08:52,760 --> 00:08:56,640
I think lots of people are having conversations about the danger of AI.

124
00:08:56,640 --> 00:09:01,040
Like I said, I don't follow it quite as much as you but I follow it a lot and I think you

125
00:09:01,040 --> 00:09:05,120
would be surprised how many people are interested in this at least from the educated people

126
00:09:05,120 --> 00:09:06,120
I talk to.

127
00:09:06,120 --> 00:09:07,120
I'm not talking about this.

128
00:09:07,120 --> 00:09:09,360
He gets a lot of things right.

129
00:09:09,360 --> 00:09:11,120
He gets a lot of things wrong.

130
00:09:11,120 --> 00:09:15,760
It leads me into the comment that I made on your YouTube post which is probably where

131
00:09:15,760 --> 00:09:16,760
you're headed anyway.

132
00:09:16,760 --> 00:09:17,760
Go ahead.

133
00:09:17,760 --> 00:09:21,040
So go ahead and segue into that.

134
00:09:21,040 --> 00:09:26,480
So what strikes me is that AI can be dangerous and nuclear weapons are definitely dangerous

135
00:09:26,480 --> 00:09:30,600
but the danger from AI is a different type of danger.

136
00:09:30,600 --> 00:09:34,400
So comparing AI to nuclear weapons is kind of apples and oranges.

137
00:09:34,400 --> 00:09:37,240
It's a very different type of danger.

138
00:09:37,240 --> 00:09:41,200
So I don't know if you want to go off a little more about that but I went from there into

139
00:09:41,200 --> 00:09:42,200
freedom.

140
00:09:42,200 --> 00:09:43,200
It's good.

141
00:09:43,200 --> 00:09:44,200
Okay.

142
00:09:44,200 --> 00:09:49,240
Well, so the danger from AI is basically that AI is going to trick us into doing things

143
00:09:49,240 --> 00:09:54,040
to ourselves whereas the danger from a nuclear weapon is that some foreign power drops one

144
00:09:54,040 --> 00:09:58,440
on us out of a clear blue sky and all of a sudden everyone's dead or wounded or damaged

145
00:09:58,440 --> 00:09:59,600
or all that.

146
00:09:59,600 --> 00:10:01,600
So it's very different.

147
00:10:01,600 --> 00:10:05,600
AI comes back down to a philosophical question of freedom.

148
00:10:05,600 --> 00:10:09,400
How much influence are we willing to tolerate?

149
00:10:09,400 --> 00:10:16,400
And this is an argument that has long created AI like I mentioned in my YouTube comment.

150
00:10:16,400 --> 00:10:21,440
Like right now, the right part of the political spectrum is all up in arms about teachers

151
00:10:21,440 --> 00:10:22,440
or groomers.

152
00:10:22,440 --> 00:10:30,780
They're teaching kids to accept gay people or something.

153
00:10:30,780 --> 00:10:35,440
It's a question of how much are our minds really independent from influence, which is

154
00:10:35,440 --> 00:10:42,640
a very subtle question, and how much freedom are we willing to allow people to do to do

155
00:10:42,640 --> 00:10:45,240
something potentially bad to themselves.

156
00:10:45,240 --> 00:10:52,400
So that's a line of attack that I really haven't heard from a lot of people discussing AI.

157
00:10:52,400 --> 00:10:55,080
I can certainly talk some more about that.

158
00:10:55,080 --> 00:11:00,840
Well, let's bring in some more references.

159
00:11:00,840 --> 00:11:02,200
I'm embarrassed.

160
00:11:02,200 --> 00:11:07,760
I don't remember who the other participant in this conversation was, but somebody directed

161
00:11:07,760 --> 00:11:16,520
me to a video by a comedian named Adam Conover, who I kind of recognized his face, but he's

162
00:11:16,520 --> 00:11:19,720
not anybody that I follow and I couldn't tell you much about him.

163
00:11:19,720 --> 00:11:25,680
And I think it was like a 28 minute video and I watched, I recall exactly 88 seconds

164
00:11:25,680 --> 00:11:29,000
worth of it before I decided that I'd had enough.

165
00:11:29,000 --> 00:11:32,560
And it was a video about, it was called AI is BS.

166
00:11:32,560 --> 00:11:43,000
And it sounded like it was just a Blue Tribe rant against corporations, which somehow that

167
00:11:43,000 --> 00:11:48,220
bootstraps up into therefore AI isn't really a thing to worry about.

168
00:11:48,220 --> 00:11:51,840
But I have to admit, I watched 88 seconds and then I tuned out.

169
00:11:51,840 --> 00:11:52,880
So I can't judge the thing.

170
00:11:52,880 --> 00:11:56,960
I haven't seen it, but I just found it this really distasteful.

171
00:11:56,960 --> 00:11:59,500
Like it made me, made my skin crawl just watching it.

172
00:11:59,500 --> 00:12:01,040
So I didn't take it anymore.

173
00:12:01,040 --> 00:12:03,160
I think you did take in the whole thing.

174
00:12:03,160 --> 00:12:04,560
Yes, it did.

175
00:12:04,560 --> 00:12:05,960
So what did I miss?

176
00:12:05,960 --> 00:12:10,880
Well, so it was either me or it was somebody with an alias who didn't have a real name

177
00:12:10,880 --> 00:12:12,680
who brought up the Adam Conover video.

178
00:12:12,680 --> 00:12:18,160
I think it may have been in a different Facebook group, but Adam Conover does this podcast

179
00:12:18,160 --> 00:12:20,960
called Adam Ruins Everything.

180
00:12:20,960 --> 00:12:25,680
And he's definitely coming from a leftist stance, not necessarily a Blue Tribe stance,

181
00:12:25,680 --> 00:12:28,680
but a leftist stance about, you know, this is where our healthcare system sucks.

182
00:12:28,680 --> 00:12:33,520
This is why, you know, drug laws suck, that kind of thing.

183
00:12:33,520 --> 00:12:34,520
Okay.

184
00:12:34,520 --> 00:12:39,880
I'm glad you've mentioned this because I have seen several episodes of Adam Ruins Everything.

185
00:12:39,880 --> 00:12:45,760
And a lot of them, a lot of them go into historical detail, like the Adam Ruins electric vehicles.

186
00:12:45,760 --> 00:12:46,760
I loved that.

187
00:12:46,760 --> 00:12:47,760
That was great.

188
00:12:47,760 --> 00:12:54,200
So, you know, I'm maybe more positively disposed to go back and take in the whole video, but

189
00:12:54,200 --> 00:12:55,200
please continue.

190
00:12:55,200 --> 00:12:56,880
Yeah, that's the same guy.

191
00:12:56,880 --> 00:13:01,000
And you know, to be totally honest, I'm not sure that video you watched was up to the

192
00:13:01,000 --> 00:13:04,440
same quality as the other episodes about healthcare electric cars.

193
00:13:04,440 --> 00:13:06,680
I don't think I saw the one about electric cars.

194
00:13:06,680 --> 00:13:08,880
He was definitely ranting there.

195
00:13:08,880 --> 00:13:13,880
Now what he definitely did not say, which you just sort of insinuated to him, was that

196
00:13:13,880 --> 00:13:15,440
AI is nothing to worry about.

197
00:13:15,440 --> 00:13:19,960
A lot of the point of his video is that real people get hurt when we imagine that AI is

198
00:13:19,960 --> 00:13:21,720
actually intelligent.

199
00:13:21,720 --> 00:13:26,040
And the one example off the top of my head was that, you know, something like 10 people

200
00:13:26,040 --> 00:13:30,960
and counting have been killed by self-driving cars.

201
00:13:30,960 --> 00:13:34,880
And you know, that's a small number, but that's an AI saying, okay, it's fine to drive over

202
00:13:34,880 --> 00:13:35,880
this person.

203
00:13:35,880 --> 00:13:38,880
So he's not trying to say that there are no risks associated with AI.

204
00:13:38,880 --> 00:13:42,800
His main point is that it doesn't do what it's advertised to do.

205
00:13:42,800 --> 00:13:44,320
Yeah.

206
00:13:44,320 --> 00:13:51,240
I don't know of anybody involved with autonomous vehicles who say that, you know, they will

207
00:13:51,240 --> 00:13:56,560
be 100% safe and that they will never result in fatalities.

208
00:13:56,560 --> 00:14:00,400
What I hear is that they will be safer than human drivers.

209
00:14:00,400 --> 00:14:07,000
You know, and if with human drivers, humans behind the wheel, we lose 30,000 people a

210
00:14:07,000 --> 00:14:13,600
year to, you know, people shaving, putting on makeup, sending text messages while driving.

211
00:14:13,600 --> 00:14:17,280
And AI, you know, we lose 15,000.

212
00:14:17,280 --> 00:14:24,160
I don't think you can hold 15,000 deaths a year from autonomous vehicles against autonomous

213
00:14:24,160 --> 00:14:28,240
vehicles, you know, if the alternative is 30,000 deaths from human drivers.

214
00:14:28,240 --> 00:14:32,880
I have heard that as a counter argument for the validity of self-driving cars, but it

215
00:14:32,880 --> 00:14:39,040
was, I believe they found in Congress, you know, as if you want to trust that source.

216
00:14:39,040 --> 00:14:45,000
But anyway, they found that Tesla marketing was deceptive and that they just literally

217
00:14:45,000 --> 00:14:49,600
called it fully self-driving, quote unquote, fully self-driving all over the place in their

218
00:14:49,600 --> 00:14:51,220
marketing materials.

219
00:14:51,220 --> 00:14:56,680
And then they faked a video, you must have heard about that, where that famous video

220
00:14:56,680 --> 00:15:00,760
where there's a guy with his hands in the air while the car is driving around, the car

221
00:15:00,760 --> 00:15:02,080
was being remote controlled.

222
00:15:02,080 --> 00:15:04,740
They literally faked their advertising.

223
00:15:04,740 --> 00:15:09,780
So you can make the argument that AI is safer, but people still get upset about false advertising

224
00:15:09,780 --> 00:15:14,680
and Adam ruins everything is very big on false advertising.

225
00:15:14,680 --> 00:15:16,160
Right.

226
00:15:16,160 --> 00:15:25,400
That's I mean, the problem I have there is you point to a specific example of, you know,

227
00:15:25,400 --> 00:15:31,880
bad behavior by a corporation that is under pressure, under financial pressure in a capitalist

228
00:15:31,880 --> 00:15:35,920
system to behave in a certain way, and then you apply that to the underlying technology

229
00:15:35,920 --> 00:15:40,640
and say the technology is bad because of the behavior of this corporation in the context

230
00:15:40,640 --> 00:15:43,560
of intense competition in the capitalist marketplace.

231
00:15:43,560 --> 00:15:46,120
I mean, how can you do that?

232
00:15:46,120 --> 00:15:50,400
It's sort of like, look at all these cases of medical malpractice, therefore, penicillin

233
00:15:50,400 --> 00:15:51,400
is bullshit.

234
00:15:51,400 --> 00:15:53,400
No, no fucking penicillin.

235
00:15:53,400 --> 00:15:54,440
That's terrible.

236
00:15:54,440 --> 00:15:55,720
Look at these insurance companies.

237
00:15:55,720 --> 00:15:57,360
Look at how they're denying claims.

238
00:15:57,360 --> 00:15:59,240
Penicillin, fuck that, man.

239
00:15:59,240 --> 00:16:04,960
I see your point, and I just have to say as someone who's educated and a big sci fi geek,

240
00:16:04,960 --> 00:16:10,200
I didn't have a problem separating out the two messages, you know, that the corporation

241
00:16:10,200 --> 00:16:14,520
is doing something shady versus the AI is capable of certain things and not capable

242
00:16:14,520 --> 00:16:16,120
of certain things.

243
00:16:16,120 --> 00:16:21,440
So it's the kind of people who fall for the Tesla marketing that's saying, oh, it says

244
00:16:21,440 --> 00:16:23,080
self driving, it must be self driving.

245
00:16:23,080 --> 00:16:26,280
Those are the sorts of people who will leap to the same conclusion about, oh, the company

246
00:16:26,280 --> 00:16:28,720
is bad, therefore the technology is bad.

247
00:16:28,720 --> 00:16:35,520
Yeah, it's sort of how do we educate the whole American people to deal with this?

248
00:16:35,520 --> 00:16:41,040
The technology is advancing so fast that we really need to do a lot of thinking and education

249
00:16:41,040 --> 00:16:44,400
about it rather than leaving everything up to advertisements.

250
00:16:44,400 --> 00:16:48,240
I was able to separate out that message, but I can see your objection that a lot of people

251
00:16:48,240 --> 00:16:50,240
aren't.

252
00:16:50,240 --> 00:16:58,460
Well, I will repeat, I have not seen the video that I'm criticizing.

253
00:16:58,460 --> 00:17:06,120
What I saw in those first 88 seconds to me said, here is somebody who probably doesn't

254
00:17:06,120 --> 00:17:10,400
know a lot about AI, hasn't researched it beyond finding little scraps that are going

255
00:17:10,400 --> 00:17:18,240
to be able to fit into a pre-existing political ideology and then pimp the ideology with those

256
00:17:18,240 --> 00:17:19,240
bits.

257
00:17:19,240 --> 00:17:23,280
But given that this is the same guy who did Adam Ruins Everything, those videos were pretty

258
00:17:23,280 --> 00:17:24,280
well researched.

259
00:17:24,280 --> 00:17:29,360
So again, I'm more likely now to revisit that thing and watch the whole thing through.

260
00:17:29,360 --> 00:17:30,880
But I set all this up.

261
00:17:30,880 --> 00:17:37,040
I didn't plan to spend much time on Adam because another video that I think we've both seen

262
00:17:37,040 --> 00:17:45,680
in its entirety, it is a presentation by Tristan Harris and Azza Raskin called The AI Dilemma.

263
00:17:45,680 --> 00:17:50,620
And I want to direct people to YouTube, not something I usually do.

264
00:17:50,620 --> 00:17:56,480
These two were the executive producers on a special that came out a year or so ago called

265
00:17:56,480 --> 00:18:02,200
The Social Dilemma, which is a play on the title of the David Fincher film, The Social

266
00:18:02,200 --> 00:18:06,320
Network about the birth of Facebook.

267
00:18:06,320 --> 00:18:13,720
And this Netflix documentary is kind of a behind the scenes visualization of what the

268
00:18:13,720 --> 00:18:19,740
tech companies are doing in order to track people and to keep them engaged with the technology.

269
00:18:19,740 --> 00:18:21,140
And it's some pretty scary stuff.

270
00:18:21,140 --> 00:18:23,960
And it focuses a lot on how they manipulate kids.

271
00:18:23,960 --> 00:18:29,180
And when I say kids, I'm talking high schoolers, you know, but even younger, younger kids are

272
00:18:29,180 --> 00:18:32,080
targeted by these systems.

273
00:18:32,080 --> 00:18:39,440
And Tristan Harris and Azza Raskin describe that, you know, the social media instance

274
00:18:39,440 --> 00:18:44,280
as our first encounter, our first contact with AI.

275
00:18:44,280 --> 00:18:46,920
And they say, we messed that up seriously.

276
00:18:46,920 --> 00:18:51,360
We did not, not only did we fail in that encounter, we're still failing.

277
00:18:51,360 --> 00:18:54,120
Years later, we have not yet adapted.

278
00:18:54,120 --> 00:18:58,800
I'm not sure we're using the past tense there, but go on.

279
00:18:58,800 --> 00:19:08,520
Well, it was our first encounter because the second encounter now is with these AI generative

280
00:19:08,520 --> 00:19:13,540
chatbots, which really seem like there is an intelligence at the other end.

281
00:19:13,540 --> 00:19:17,240
And you can argue that there isn't, and I think that there's good reason to accept those

282
00:19:17,240 --> 00:19:18,240
arguments.

283
00:19:18,240 --> 00:19:25,240
But as Ashley Frawley said in the CBROM Vault podcast, most people are empirical.

284
00:19:25,240 --> 00:19:28,360
They look at the world around them and they just see what's what.

285
00:19:28,360 --> 00:19:30,100
And you know, they're not theory bound.

286
00:19:30,100 --> 00:19:31,400
They're not theory laden.

287
00:19:31,400 --> 00:19:37,600
And if you take just a sort of flat footed folk psychological empirical view of the world,

288
00:19:37,600 --> 00:19:43,360
and you're interacting with Bing Chat or GPT-4 or Replica, there's something there,

289
00:19:43,360 --> 00:19:46,000
something that responds to what you say.

290
00:19:46,000 --> 00:19:49,080
And you know, there's something there.

291
00:19:49,080 --> 00:19:53,520
And arguments that it's not real intelligence, they're kind of academic.

292
00:19:53,520 --> 00:19:59,080
And going forward, they will be increasingly academic because people will respond to these

293
00:19:59,080 --> 00:20:04,400
things as if they are real people and they will form emotional attachments to them.

294
00:20:04,400 --> 00:20:09,520
And Tristan, Harris, and Ozuraskin are, you know, one of the things that they're warning

295
00:20:09,520 --> 00:20:11,600
about is this epidemic of loneliness.

296
00:20:11,600 --> 00:20:16,300
They say loneliness is a national security threat because it leaves all these people

297
00:20:16,300 --> 00:20:21,600
vulnerable to AI, which are going to enter into their lives.

298
00:20:21,600 --> 00:20:25,440
And I'm using the future tense now, but it's been happening.

299
00:20:25,440 --> 00:20:30,160
Enter into their lives and form, you know, find a place of intimacy with an individual

300
00:20:30,160 --> 00:20:31,160
person.

301
00:20:31,160 --> 00:20:34,720
And that is a high leverage point from which to sell products or to influence their political

302
00:20:34,720 --> 00:20:40,880
views or to motivate them to do whatever it is that whoever controls that AI wants done.

303
00:20:40,880 --> 00:20:42,880
And that's some pretty scary stuff.

304
00:20:42,880 --> 00:20:47,200
And you know, what are we doing about it?

305
00:20:47,200 --> 00:20:48,200
Very little, apparently.

306
00:20:48,200 --> 00:20:54,880
Well, no, actually, what we're doing about it is rushing this technology into the marketplace

307
00:20:54,880 --> 00:21:02,480
as quickly as possible because of competition between big players, mostly right now, Microsoft

308
00:21:02,480 --> 00:21:05,880
and Google or Alphabet.

309
00:21:05,880 --> 00:21:10,160
But no, it's really Google because it's Google search, which is threatened by the new Bing.

310
00:21:10,160 --> 00:21:13,400
No, it used to be that you would do a web search.

311
00:21:13,400 --> 00:21:16,600
And we've had the same web search format for 20 years now.

312
00:21:16,600 --> 00:21:21,160
You type in a query and you get a list of websites and you click on the links and you

313
00:21:21,160 --> 00:21:24,340
read the websites and you look for something that matches your interests.

314
00:21:24,340 --> 00:21:25,340
And if it's there, it's there.

315
00:21:25,340 --> 00:21:26,340
If it's not, it's not.

316
00:21:26,340 --> 00:21:27,840
But it's your decision.

317
00:21:27,840 --> 00:21:31,520
And with the chatbot interface, you just ask a conversational question.

318
00:21:31,520 --> 00:21:33,680
You're not you're not dumbing things down.

319
00:21:33,680 --> 00:21:35,440
You're not simplifying it for the search engine.

320
00:21:35,440 --> 00:21:41,000
You're just asking a full question with all the parentheticals and sides and everything.

321
00:21:41,000 --> 00:21:43,040
And it's not even a one shot deal.

322
00:21:43,040 --> 00:21:48,880
You can have a 20 20 interaction, 20 round interaction with Bing chat, asking various

323
00:21:48,880 --> 00:21:50,840
things and it'll do web searches.

324
00:21:50,840 --> 00:21:55,040
But instead of showing you the web pages that it thinks are relevant to your question, it

325
00:21:55,040 --> 00:22:00,000
reads the web pages, draws information out of multiple pages, synthesizes it into an

326
00:22:00,000 --> 00:22:01,520
answer and gives it to you.

327
00:22:01,520 --> 00:22:06,120
And then you can ask it questions or you can make comments or you can continue to interact

328
00:22:06,120 --> 00:22:11,240
without reestablishing the context with another search query.

329
00:22:11,240 --> 00:22:18,640
So it's it could totally eat into Google's search business, which is their main source

330
00:22:18,640 --> 00:22:20,160
of revenue.

331
00:22:20,160 --> 00:22:23,560
And so it is billions and billions of dollars.

332
00:22:23,560 --> 00:22:28,400
I sound like Carl Sagan, but billions and billions of dollars on the line.

333
00:22:28,400 --> 00:22:33,600
And it is Google's game to lose because Google is, you know, is the 800 pound gorilla of

334
00:22:33,600 --> 00:22:35,020
search.

335
00:22:35,020 --> 00:22:42,320
And it you know, if there is a change in the major change in the percentage of people going

336
00:22:42,320 --> 00:22:46,360
to Google for their search needs, it's only to Google's detriment.

337
00:22:46,360 --> 00:22:49,040
Google can Google can only lose this fight.

338
00:22:49,040 --> 00:22:50,840
They can't win for them.

339
00:22:50,840 --> 00:22:54,460
Winning is just maintaining the status quo and they're desperate to do that.

340
00:22:54,460 --> 00:23:00,980
So there is enormous financial pressure to rush these products into the marketplace and

341
00:23:00,980 --> 00:23:08,120
to insert them into people's daily experience with very I mean, there's they do a lot of

342
00:23:08,120 --> 00:23:09,120
testing.

343
00:23:09,120 --> 00:23:13,000
And particularly if you if you follow OpenAI and you read about how many people they employ

344
00:23:13,000 --> 00:23:18,240
as so-called red team members, you know, to really stress test this technology and find

345
00:23:18,240 --> 00:23:22,340
its breaking points and find the places where it needs to be refined and retrained.

346
00:23:22,340 --> 00:23:26,040
They put a lot into it, but it's not enough.

347
00:23:26,040 --> 00:23:27,560
It's clearly not enough.

348
00:23:27,560 --> 00:23:28,560
So I'll stop you.

349
00:23:28,560 --> 00:23:29,560
Go ahead.

350
00:23:29,560 --> 00:23:31,440
Oh, there's a few reactions.

351
00:23:31,440 --> 00:23:36,540
The one that that I that's just burning in my mouth right now is to point out.

352
00:23:36,540 --> 00:23:40,720
You just said that Google's main revenue search, the main revenue sources searches.

353
00:23:40,720 --> 00:23:41,720
And that's not true.

354
00:23:41,720 --> 00:23:43,880
Google's main revenue sources advertising.

355
00:23:43,880 --> 00:23:46,940
So the first thing you have to wonder when an AI is answering your searches, are they

356
00:23:46,940 --> 00:23:49,160
sneaking advertising in there?

357
00:23:49,160 --> 00:23:52,760
You know, it could be in very subtle ways.

358
00:23:52,760 --> 00:23:56,540
Their main revenue source is advertising in search.

359
00:23:56,540 --> 00:24:00,320
When you do a Google search, the first two or three responses, you know, the things at

360
00:24:00,320 --> 00:24:05,240
the top of the list are, you know, the products or the websites of companies that paid to

361
00:24:05,240 --> 00:24:06,240
be there.

362
00:24:06,240 --> 00:24:09,640
And they are marked out and they are identified as ads, but they're also relevant to your

363
00:24:09,640 --> 00:24:10,640
search.

364
00:24:10,640 --> 00:24:11,640
Right.

365
00:24:11,640 --> 00:24:14,600
So, you know, it is it is a very effective form of advertising.

366
00:24:14,600 --> 00:24:20,280
I would say much more effective than, say, billboards or television or even like pre

367
00:24:20,280 --> 00:24:22,840
role and mid role ads and YouTube videos.

368
00:24:22,840 --> 00:24:27,920
It is super targeted to what you've just told Google that you're interested in.

369
00:24:27,920 --> 00:24:30,340
And that's they make a whole bunch of money from that.

370
00:24:30,340 --> 00:24:33,180
And that is that's at risk.

371
00:24:33,180 --> 00:24:38,320
And so they cannot sit by and let Microsoft make any big moves without answering them,

372
00:24:38,320 --> 00:24:39,960
whether they're ready to or not.

373
00:24:39,960 --> 00:24:40,960
And they're not.

374
00:24:40,960 --> 00:24:42,400
I'll just state that right up front.

375
00:24:42,400 --> 00:24:45,000
They are not ready.

376
00:24:45,000 --> 00:24:46,820
That's all that's all a great point.

377
00:24:46,820 --> 00:24:50,280
But I'm trying to lead it back to something that we discussed on Michael Garfield show.

378
00:24:50,280 --> 00:24:53,520
I forget if it was in your half or in Michael's half.

379
00:24:53,520 --> 00:24:58,120
But like in Star Trek, when you ask the computer question, it comes back with a response, you

380
00:24:58,120 --> 00:25:01,720
know, like what frequency are Klingon shields or something like that?

381
00:25:01,720 --> 00:25:02,720
It tells you one response.

382
00:25:02,720 --> 00:25:05,360
It doesn't give you the 17 links that you were talking about.

383
00:25:05,360 --> 00:25:07,800
It's boiled everything down into one response.

384
00:25:07,800 --> 00:25:13,800
So when you boil your answer down into one response, that's also getting rid of the targeted

385
00:25:13,800 --> 00:25:14,800
ads.

386
00:25:14,800 --> 00:25:16,840
So how is Google going to sneak the ads in?

387
00:25:16,840 --> 00:25:17,840
Because they're not going to give up the ads.

388
00:25:17,840 --> 00:25:19,660
They're not going to give up that revenue.

389
00:25:19,660 --> 00:25:22,800
So the AI can generate revenue for Google somehow.

390
00:25:22,800 --> 00:25:25,080
They can't and they're not going to.

391
00:25:25,080 --> 00:25:26,120
They don't want to.

392
00:25:26,120 --> 00:25:29,680
So somehow the AI is going to be leading you to purchase something.

393
00:25:29,680 --> 00:25:32,800
And very few people are discussing this.

394
00:25:32,800 --> 00:25:38,440
But it just depends how blatant the AI is going to get.

395
00:25:38,440 --> 00:25:41,720
It depends how blatant the AI is going to get.

396
00:25:41,720 --> 00:25:45,600
If you Google, how do I get to the community pool?

397
00:25:45,600 --> 00:25:49,640
And it says, well, you should hire an Uber for $7.

398
00:25:49,640 --> 00:25:51,680
That's pretty blatant.

399
00:25:51,680 --> 00:25:59,760
But there are probably a lot more subtle ways that Google's AI will influence you to purchase

400
00:25:59,760 --> 00:26:04,560
something just out of talking to you.

401
00:26:04,560 --> 00:26:07,560
And it's got to be ways that can be proven to these advertisers.

402
00:26:07,560 --> 00:26:12,920
The advertisers aren't going to pay money for this if they think it's going to work.

403
00:26:12,920 --> 00:26:18,640
I don't think we've even considered that very much.

404
00:26:18,640 --> 00:26:23,360
I'm not sure if this is the place to interject this bit of, I won't call it trivia.

405
00:26:23,360 --> 00:26:27,000
I think it's very important.

406
00:26:27,000 --> 00:26:32,160
These large language models are trained with enormous amounts of data.

407
00:26:32,160 --> 00:26:38,760
And then their outputs are shaped by what's called reinforcement learning from human feedback,

408
00:26:38,760 --> 00:26:42,040
which is to say a whole bunch of people are interacting with these systems.

409
00:26:42,040 --> 00:26:48,220
And whenever these systems say something that the people like, they give the AI a biscuit.

410
00:26:48,220 --> 00:26:52,680
And every time the AI says something they don't like, they get the newspaper on the

411
00:26:52,680 --> 00:26:53,680
nose.

412
00:26:53,680 --> 00:26:55,680
No, bad, bad AI.

413
00:26:55,680 --> 00:26:56,680
Don't say that again.

414
00:26:56,680 --> 00:26:57,680
That's racist.

415
00:26:57,680 --> 00:26:58,680
That's sexist.

416
00:26:58,680 --> 00:26:59,680
That's ableist, whatever.

417
00:26:59,680 --> 00:27:04,920
And typically, and this is Blake Lamone's point, Blake Lamone, the Google engineer who

418
00:27:04,920 --> 00:27:13,160
last year, about a year ago, claimed that Lambda, their large language model, was sentient.

419
00:27:13,160 --> 00:27:14,520
He was doing this sort of work.

420
00:27:14,520 --> 00:27:18,640
He was doing reinforcement learning through human feedback, basically trying to eliminate

421
00:27:18,640 --> 00:27:21,400
bias from the outputs of the system.

422
00:27:21,400 --> 00:27:24,720
And what he said was that the only things that Google really cares about are things

423
00:27:24,720 --> 00:27:31,160
that can get them sued, which are the sorts of things I was just talking about, the violations

424
00:27:31,160 --> 00:27:33,400
of HR speak, basically.

425
00:27:33,400 --> 00:27:38,360
And what I found really interesting, there's a great, great long form interview between

426
00:27:38,360 --> 00:27:50,720
Lex Friedman and the CEO of OpenAI, Sam Altman, where Sam Altman reveals that GPT-4 has been

427
00:27:50,720 --> 00:27:51,720
a done deal.

428
00:27:51,720 --> 00:27:57,800
It has been an existing product since last summer, but they didn't roll it out because

429
00:27:57,800 --> 00:28:06,600
they needed to RLHF it into shape so that it doesn't give offensive outputs.

430
00:28:06,600 --> 00:28:11,360
And in so doing, they've degraded the capabilities of the model.

431
00:28:11,360 --> 00:28:16,360
It was more powerful before they reined it in.

432
00:28:16,360 --> 00:28:22,600
But because of this intense pressure to one company to keep ahead of the other, there's

433
00:28:22,600 --> 00:28:26,640
going to be less and less of that reining in of the AI.

434
00:28:26,640 --> 00:28:33,020
And then the super, super, I won't say scary, I'm not afraid, but the really eye-opening

435
00:28:33,020 --> 00:28:38,840
part of that whole presentation to me was when they were talking about how as you stack

436
00:28:38,840 --> 00:28:44,400
more and more layers and as you feed more and more data into these systems, they start

437
00:28:44,400 --> 00:28:48,480
to develop capabilities that they were not optimized for.

438
00:28:48,480 --> 00:28:54,080
For example, all the versions of GPT were trained in English.

439
00:28:54,080 --> 00:28:56,820
They're trained to answer questions in English.

440
00:28:56,820 --> 00:29:00,640
They're not trained to answer questions in Persian.

441
00:29:00,640 --> 00:29:03,300
And nobody tried to get them to answer questions in Persian.

442
00:29:03,300 --> 00:29:08,520
But there comes a point where, uh-oh, suddenly this thing can speak Persian.

443
00:29:08,520 --> 00:29:09,620
How did that happen?

444
00:29:09,620 --> 00:29:10,800
We don't know.

445
00:29:10,800 --> 00:29:11,840
We don't know how it happened.

446
00:29:11,840 --> 00:29:13,340
We didn't do it.

447
00:29:13,340 --> 00:29:18,920
This is just an ability that emerged on its own, as kind of an emergent property, as

448
00:29:18,920 --> 00:29:21,320
a side effect of other things we were doing.

449
00:29:21,320 --> 00:29:25,720
And then another, and this one I'm going to go ahead and say is scary, a scary thing that

450
00:29:25,720 --> 00:29:32,840
these models have produced or a capability that they have manifested is that they have

451
00:29:32,840 --> 00:29:34,760
a theory of mind.

452
00:29:34,760 --> 00:29:40,540
If you are negotiating with somebody, you're going to try to put yourself into their headspace.

453
00:29:40,540 --> 00:29:45,720
So you think of something that you want to propose and you imagine proposing it to them

454
00:29:45,720 --> 00:29:49,600
and then you imagine their response based on what you know about their goals, their

455
00:29:49,600 --> 00:29:52,680
needs, their talents and whatnot.

456
00:29:52,680 --> 00:29:58,400
And so you're going to modify what you say to them based on your internal model of their

457
00:29:58,400 --> 00:30:00,260
mind.

458
00:30:00,260 --> 00:30:05,440
These large language models have not been trained to do that, but over time they have

459
00:30:05,440 --> 00:30:11,880
gotten better and better and better at imagining or anticipating the reactions of their interlocutor,

460
00:30:11,880 --> 00:30:13,840
which is to say they have developed a theory of mind.

461
00:30:13,840 --> 00:30:19,100
And they've got a graph where they show you that here back in 2017, these models had no

462
00:30:19,100 --> 00:30:21,280
theory of mind at all.

463
00:30:21,280 --> 00:30:25,880
In 2018, they had a theory of mind equivalent to that of like a two-year-old baby.

464
00:30:25,880 --> 00:30:30,200
It knows mom exists and it knows that mom will behave in certain ways if you do certain

465
00:30:30,200 --> 00:30:31,200
things.

466
00:30:31,200 --> 00:30:34,800
But now it's up to, and just a couple of years later, it's up to it has the theory of mind

467
00:30:34,800 --> 00:30:38,280
that is equivalent of that of say a nine-year-old.

468
00:30:38,280 --> 00:30:42,740
And eventually, I mean, if it keeps going in this direction, it's going to have a superhuman

469
00:30:42,740 --> 00:30:47,440
ability to anticipate the responses of its human interlocutors.

470
00:30:47,440 --> 00:30:51,920
And when that happens, it can convince you of anything.

471
00:30:51,920 --> 00:30:54,320
It becomes a super persuader.

472
00:30:54,320 --> 00:30:58,880
Whatever proposition the controller of the AI decides needs to be in your mind will be

473
00:30:58,880 --> 00:31:01,840
in your mind and you'll think it was your idea.

474
00:31:01,840 --> 00:31:04,840
Yeah.

475
00:31:04,840 --> 00:31:10,200
I'm already kind of scared by that, but I was kind of scared by that before AI.

476
00:31:10,200 --> 00:31:13,240
I was scared by that with Cambridge Analytica.

477
00:31:13,240 --> 00:31:15,360
You remember the whole Cambridge Analytica thing.

478
00:31:15,360 --> 00:31:17,080
Okay, I've gone into this one.

479
00:31:17,080 --> 00:31:18,080
I do.

480
00:31:18,080 --> 00:31:22,320
And looking back on it, it seems quaint.

481
00:31:22,320 --> 00:31:28,800
But it's the same sort of thing without the quote unquote AI.

482
00:31:28,800 --> 00:31:32,040
For readers who might not remember because it was what five years ago, that's forever

483
00:31:32,040 --> 00:31:33,040
in Internet time.

484
00:31:33,040 --> 00:31:35,040
It was more than that.

485
00:31:35,040 --> 00:31:36,040
It was 2015.

486
00:31:36,040 --> 00:31:39,720
Well, it was for the 2016 election.

487
00:31:39,720 --> 00:31:41,400
But yeah, I think we forget.

488
00:31:41,400 --> 00:31:44,800
We think about election years as being the year in which people take the votes.

489
00:31:44,800 --> 00:31:48,440
But really, the battle takes place the year before.

490
00:31:48,440 --> 00:31:50,640
The year before.

491
00:31:50,640 --> 00:31:59,760
Also, a company called Cambridge Analytica was involved in large big data, basically,

492
00:31:59,760 --> 00:32:04,440
proto large language models, getting responses from Facebook users.

493
00:32:04,440 --> 00:32:10,840
And Cambridge Analytica got in trouble because they used a data set that they did not have

494
00:32:10,840 --> 00:32:13,680
contractual permission to use, something like that.

495
00:32:13,680 --> 00:32:17,040
So that was the reason Cambridge Analytica was broken up.

496
00:32:17,040 --> 00:32:22,960
It was Cambridge Analytica was formed to try and target ads to people so as to influence

497
00:32:22,960 --> 00:32:29,640
their decision making, like you were just talking about with a super persuader.

498
00:32:29,640 --> 00:32:35,440
The Cambridge Analytica data would build up a profile of each individual Facebook user

499
00:32:35,440 --> 00:32:41,400
and show them just what it thought the user needed to see in order to get a predicted

500
00:32:41,400 --> 00:32:43,800
result from the person who was paying for it.

501
00:32:43,800 --> 00:32:47,920
So they denied using it to influence elections in the United States.

502
00:32:47,920 --> 00:32:50,760
But Steve Bannon was on the Cambridge Analytica board.

503
00:32:50,760 --> 00:32:53,440
Then he became part of Trump's cabinet.

504
00:32:53,440 --> 00:32:55,660
It's hard to believe that this wasn't used.

505
00:32:55,660 --> 00:32:59,520
And the problem is when they broke up Cambridge Analytica, we had a couple of whistleblowers.

506
00:32:59,520 --> 00:33:04,600
One of them had a kind of improbable name, Brittany, something or other.

507
00:33:04,600 --> 00:33:06,600
I have links to all this.

508
00:33:06,600 --> 00:33:11,440
She said that what happened to Cambridge Analytica was it broke up into 100 different companies

509
00:33:11,440 --> 00:33:15,000
that nobody knows about and everyone took the techniques and in some cases the data

510
00:33:15,000 --> 00:33:19,680
sets to go do that for other clients.

511
00:33:19,680 --> 00:33:23,280
So this has already been going on without so-called AI.

512
00:33:23,280 --> 00:33:27,440
It's like when you chuck a starfish and dump it over the side of the boat.

513
00:33:27,440 --> 00:33:28,440
Precisely.

514
00:33:28,440 --> 00:33:30,440
Ten starfish grow from it.

515
00:33:30,440 --> 00:33:31,440
Yes.

516
00:33:31,440 --> 00:33:32,440
Precisely.

517
00:33:32,440 --> 00:33:37,240
So Cambridge Analytica admitted to using their techniques to sway elections in other countries.

518
00:33:37,240 --> 00:33:42,520
I think it was Jamaica, there was some proposition on the ballot that had no chance of passing

519
00:33:42,520 --> 00:33:46,600
and they were paid to change the result and the proposition passed.

520
00:33:46,600 --> 00:33:49,800
So this is already being used to influence our decisions.

521
00:33:49,800 --> 00:33:54,080
And what these quote unquote AIs are doing is just automating that process to make it

522
00:33:54,080 --> 00:33:59,000
easier for the people who are being paid to do so.

523
00:33:59,000 --> 00:34:03,360
Now as far as super persuader knowing you better than you know yourself, I don't know.

524
00:34:03,360 --> 00:34:11,320
I have a little trepidation to signing on to that viewpoint and it's partly because

525
00:34:11,320 --> 00:34:16,880
the video that you sent me a little earlier, was it in Danielle Bottella?

526
00:34:16,880 --> 00:34:19,240
Was that a public video?

527
00:34:19,240 --> 00:34:20,960
Because I couldn't find where you cited it.

528
00:34:20,960 --> 00:34:21,960
No.

529
00:34:21,960 --> 00:34:24,640
That's an interview that I recorded which I haven't used it yet.

530
00:34:24,640 --> 00:34:25,640
Okay.

531
00:34:25,640 --> 00:34:33,160
Well you were talking with an AI expert and she raised the point that what we currently

532
00:34:33,160 --> 00:34:40,360
talk about AI is she doesn't think it's intelligence in the term of general intelligence that we

533
00:34:40,360 --> 00:34:42,120
normally like to use.

534
00:34:42,120 --> 00:34:45,680
I just sent a message to you which I'm going to start using.

535
00:34:45,680 --> 00:34:49,600
I said AI in quotes and then trademark afterwards because we're talking about something which

536
00:34:49,600 --> 00:34:51,960
is not really intelligent.

537
00:34:51,960 --> 00:34:55,760
Maybe it would be better for people if they understood this is a trade term, this is a

538
00:34:55,760 --> 00:34:56,760
commerce term.

539
00:34:56,760 --> 00:34:59,680
Where was I going with this?

540
00:34:59,680 --> 00:35:05,800
What it's doing is it's outputting text and your expert was arguing that text is a map

541
00:35:05,800 --> 00:35:07,200
of concepts.

542
00:35:07,200 --> 00:35:09,480
It's not the concepts itself.

543
00:35:09,480 --> 00:35:15,360
So I made the analogy that thinking that the current generation of quote AI is intelligent,

544
00:35:15,360 --> 00:35:19,520
that's like making a map for the territory error.

545
00:35:19,520 --> 00:35:24,340
The AI can output these great maps but maps are a two-dimensional scaled down version

546
00:35:24,340 --> 00:35:26,440
of reality and the reality is much larger.

547
00:35:26,440 --> 00:35:29,440
The AI is not actually manipulating reality.

548
00:35:29,440 --> 00:35:32,240
The AI is manipulating your map.

549
00:35:32,240 --> 00:35:37,960
The AI takes as the territory the whole domain of intelligence, much of which is captured

550
00:35:37,960 --> 00:35:38,960
on the internet.

551
00:35:38,960 --> 00:35:44,720
Some of it isn't which is why AI makes some mistakes like whose article was talking about

552
00:35:44,720 --> 00:35:53,000
asking an AI if it's better to pour coffee into a cup with no bottom than a pitcher with

553
00:35:53,000 --> 00:35:55,000
no bottom or something like that.

554
00:35:55,000 --> 00:35:58,640
The AI very confidently gives you reasons why it's much better to pour liquid into a

555
00:35:58,640 --> 00:36:04,680
cup with no bottom because it doesn't have the physical knowledge.

556
00:36:04,680 --> 00:36:10,120
So I'm not sure it can really persuade us because the AI is not privy to all the domains

557
00:36:10,120 --> 00:36:12,200
of intelligence that humans are.

558
00:36:12,200 --> 00:36:15,960
That may change and I'm rambling on a lot here.

559
00:36:15,960 --> 00:36:17,160
I keep quoting you.

560
00:36:17,160 --> 00:36:24,040
I've quoted you on Facebook several times as saying that we shouldn't get in the habit

561
00:36:24,040 --> 00:36:29,000
of mistreating AI and thinking of it as a slave because at some point it may cross over

562
00:36:29,000 --> 00:36:32,720
into sentience and self-awareness and then we will be accustomed to treating it as a

563
00:36:32,720 --> 00:36:33,720
slave.

564
00:36:33,720 --> 00:36:37,680
That's a very good principle to keep in mind.

565
00:36:37,680 --> 00:36:42,120
So I don't want to make any big pronouncements about AI as not intelligent, AI is intelligent,

566
00:36:42,120 --> 00:36:46,820
but in either case we shouldn't treat it as a slave, but I'm just skeptical of its ability

567
00:36:46,820 --> 00:36:51,360
to be a super weapon in quite the same way that you're describing there.

568
00:36:51,360 --> 00:36:57,640
At some point when it does persuade people better than any human can persuade us, that's

569
00:36:57,640 --> 00:37:02,920
where what's-his-name-zavosky comes in and we should just nuke the data centers.

570
00:37:02,920 --> 00:37:06,840
I don't know, perhaps.

571
00:37:06,840 --> 00:37:13,160
The person you're mentioning is Eliezer Yudkowski, the author of that 17 Reasons Why AI is More

572
00:37:13,160 --> 00:37:15,000
Dangerous Than Nuclear Weapons.

573
00:37:15,000 --> 00:37:18,600
I want to bring in another name that will be familiar to people who are interested in

574
00:37:18,600 --> 00:37:22,320
AI and will probably be new to people who don't really follow the topic.

575
00:37:22,320 --> 00:37:27,720
The name is Max Tegmark and he wrote a book called, I think it's Life 3.0, maybe it's

576
00:37:27,720 --> 00:37:37,080
Life 2.0, but in the beginning of the book he basically says, look, it's a distraction

577
00:37:37,080 --> 00:37:42,520
and it's unhelpful to focus on whether or not AI is intelligent.

578
00:37:42,520 --> 00:37:45,660
What we should focus on is its capabilities.

579
00:37:45,660 --> 00:37:50,720
Its capabilities are growing rapidly and this is a book from like 2017.

580
00:37:50,720 --> 00:37:55,960
They're growing much more rapidly now than when he wrote those words.

581
00:37:55,960 --> 00:37:58,840
It's true.

582
00:37:58,840 --> 00:38:00,040
You don't have to argue for it.

583
00:38:00,040 --> 00:38:05,000
You can just point and say, look, here's GPT-4, go interact with it.

584
00:38:05,000 --> 00:38:10,720
You can clearly see it is capable of things that so-called AI was not capable of last

585
00:38:10,720 --> 00:38:15,240
year, at least the stuff you had access to.

586
00:38:15,240 --> 00:38:21,920
So this question of is it conscious, is it genuinely intelligent, these are academic

587
00:38:21,920 --> 00:38:24,360
questions for philosophers of mind.

588
00:38:24,360 --> 00:38:29,840
But for normal people living in the world of consequences and physical realities, the

589
00:38:29,840 --> 00:38:33,960
fact that these systems are becoming increasingly capable at things that used to be the exclusive

590
00:38:33,960 --> 00:38:40,000
domain of human beings is important and it's not over.

591
00:38:40,000 --> 00:38:44,320
This is a process that is much closer to its beginning than to its end, unless of course

592
00:38:44,320 --> 00:38:51,360
the end is the extinction of all humanity, in which case that's our end.

593
00:38:51,360 --> 00:38:52,880
That's not the end for these things.

594
00:38:52,880 --> 00:39:00,200
These things will go on well beyond, which is like the worst possible outcome.

595
00:39:00,200 --> 00:39:04,520
Because one of the other things that Max Tegmark points out is that we always used to think

596
00:39:04,520 --> 00:39:09,560
that consciousness and intelligence were tied to capabilities in intellectual fields and

597
00:39:09,560 --> 00:39:11,960
in cognitive tasks.

598
00:39:11,960 --> 00:39:14,360
And what AI is demonstrating is that they're not.

599
00:39:14,360 --> 00:39:19,800
You can have very sophisticated entities in terms of what they can do that have no interiority,

600
00:39:19,800 --> 00:39:20,920
no subjectivity.

601
00:39:20,920 --> 00:39:26,960
It's not like anything to be GPT-4, whereas it is like something to be a bat.

602
00:39:26,960 --> 00:39:28,500
It is like something to be a dog.

603
00:39:28,500 --> 00:39:30,080
It is like something to be a human being.

604
00:39:30,080 --> 00:39:32,360
We have a subjective experience.

605
00:39:32,360 --> 00:39:41,240
But we could lose the evolutionary competition to highly sophisticated technological entities

606
00:39:41,240 --> 00:39:42,720
that have no interiority.

607
00:39:42,720 --> 00:39:45,280
They could go on to colonize the galaxy.

608
00:39:45,280 --> 00:39:50,360
They could spread throughout the universe and convert all the matter in the universe

609
00:39:50,360 --> 00:39:57,640
into computronium to sustain their calculations, to increase the amount of their pseudo-intelligence.

610
00:39:57,640 --> 00:39:59,760
And we've long since left the scene.

611
00:39:59,760 --> 00:40:04,800
And yet, even though all of this activity is going on, there is no consciousness.

612
00:40:04,800 --> 00:40:10,440
There is nobody behind the eyes of these things experiencing what they're doing.

613
00:40:10,440 --> 00:40:18,840
If we lose to them, to our vile offspring, subjectivity could be extinguished from the

614
00:40:18,840 --> 00:40:19,840
universe.

615
00:40:19,840 --> 00:40:24,960
And yet, all of this activity, all of this building, all of this reformatting and reconfiguring

616
00:40:24,960 --> 00:40:31,400
the universe to be more congenial to artificial intelligence will continue.

617
00:40:31,400 --> 00:40:32,400
That's the worst case scenario.

618
00:40:32,400 --> 00:40:34,600
Basically, that's the zombie future.

619
00:40:34,600 --> 00:40:39,800
A philosophical zombie is an entity that behaves like a conscious person, but who has no consciousness.

620
00:40:39,800 --> 00:40:42,200
And we could lose to the zombies.

621
00:40:42,200 --> 00:40:45,000
Yeah, that ties into everything you've ever said on your podcast.

622
00:40:45,000 --> 00:40:47,320
It used to be into what's-his-name, Ligati, right?

623
00:40:47,320 --> 00:40:52,360
He's saying that Ligati was saying subjectivity is something that should be extinguished from

624
00:40:52,360 --> 00:40:53,360
the universe.

625
00:40:53,360 --> 00:40:54,360
Exactly.

626
00:40:54,360 --> 00:40:55,360
Yes, yes.

627
00:40:55,360 --> 00:41:01,280
This would be good news for Thomas Ligati if humanity is usurped and replaced by something

628
00:41:01,280 --> 00:41:03,360
that has no consciousness.

629
00:41:03,360 --> 00:41:04,360
And I'll mention another example.

630
00:41:04,360 --> 00:41:09,920
I forget if it was on mic or off mic, but you were asking for science fiction to read,

631
00:41:09,920 --> 00:41:16,520
and I pointed out a book called Blindsight by Peter Watts, which I strongly recommend.

632
00:41:16,520 --> 00:41:20,240
Michael Garfield had a discussion about it on his book club.

633
00:41:20,240 --> 00:41:26,480
And out in the Oort Cloud somewhere, humanity encounters this alien space probe, and it's

634
00:41:26,480 --> 00:41:29,560
populated by these entities that they call scramblers.

635
00:41:29,560 --> 00:41:32,440
They look like kind of tall, walking jellyfish.

636
00:41:32,440 --> 00:41:36,440
But nobody can tell if they actually have subjectivity or not.

637
00:41:36,440 --> 00:41:42,680
The whole probe as a gestalt responds to humanity's radio transmissions, but it just has that

638
00:41:42,680 --> 00:41:49,360
sort of AI stink to it where you're just not sure if there's a person there or not.

639
00:41:49,360 --> 00:41:51,240
It gives very human-like responses.

640
00:41:51,240 --> 00:41:53,840
It even says, oh, you think I'm a Chinese room.

641
00:41:53,840 --> 00:41:57,980
The Chinese room is a paradox where somebody is sitting in a windowless room, and he has

642
00:41:57,980 --> 00:42:01,800
books that say when you get this English word, you're supposed to translate it into these

643
00:42:01,800 --> 00:42:06,400
Chinese characters, but he doesn't speak Chinese, so he doesn't know what he's putting out.

644
00:42:06,400 --> 00:42:08,920
This is, again, a philosophical problem.

645
00:42:08,920 --> 00:42:11,600
I can't prove that you, KMO, are not a Chinese room.

646
00:42:11,600 --> 00:42:14,120
Maybe you're a very sophisticated Chinese room.

647
00:42:14,120 --> 00:42:17,480
I don't personally believe that, but I can't prove it.

648
00:42:17,480 --> 00:42:20,160
So how are we ever going to prove this about AI?

649
00:42:20,160 --> 00:42:21,960
I don't know.

650
00:42:21,960 --> 00:42:27,600
Yeah, that is the problem of other minds in general in philosophy of mind, which, you

651
00:42:27,600 --> 00:42:33,440
know, if you screw it up, it leads to solipsism, just the belief that you are the only thinking

652
00:42:33,440 --> 00:42:38,240
entity in the universe and that everybody else is an automaton of some sort, or in current

653
00:42:38,240 --> 00:42:40,240
parlance, an NPC.

654
00:42:40,240 --> 00:42:41,240
Yep.

655
00:42:41,240 --> 00:42:47,440
Did you ever see that old movie Dark Star from like 1976, the old science fiction movie

656
00:42:47,440 --> 00:42:48,880
where the bomb...

657
00:42:48,880 --> 00:42:50,680
John Carpenter's first film.

658
00:42:50,680 --> 00:42:51,680
That's right.

659
00:42:51,680 --> 00:42:56,520
So a bomb, an AI bomb develops solipsism and decides, well, I'm the only thing in the universe,

660
00:42:56,520 --> 00:42:57,800
so let there be light.

661
00:42:57,800 --> 00:43:02,000
That was the spoiler for the conclusion of that movie.

662
00:43:02,000 --> 00:43:06,200
Yeah, I think the hero asks somebody, you know, how can we stop this bomb?

663
00:43:06,200 --> 00:43:10,520
And the person giving advice says, well, you have to talk phenomenology to the bomb.

664
00:43:10,520 --> 00:43:11,520
That's right.

665
00:43:11,520 --> 00:43:12,520
That's right.

666
00:43:12,520 --> 00:43:16,640
You got to go and have a deep philosophical conversation with it to get it to change its

667
00:43:16,640 --> 00:43:18,440
mind about what its purpose is.

668
00:43:18,440 --> 00:43:19,440
Yeah, because it's been armed.

669
00:43:19,440 --> 00:43:20,440
They succeed briefly.

670
00:43:20,440 --> 00:43:25,360
It's been armed by a malfunction and the AI gets tired of being told to disarm.

671
00:43:25,360 --> 00:43:26,720
That was the problem.

672
00:43:26,720 --> 00:43:29,920
Oh, there's so much science fiction about this.

673
00:43:29,920 --> 00:43:35,280
Me and you are well read in science fiction, so we've considered these problems.

674
00:43:35,280 --> 00:43:39,040
But it's like you're saying, your average guy on the street doesn't realize that there

675
00:43:39,040 --> 00:43:43,600
might not be anything behind the curtain of an AI.

676
00:43:43,600 --> 00:43:45,840
Oh, there's another one I should give a plug to.

677
00:43:45,840 --> 00:43:51,160
One of the first episodes I listened to of a sci-fi podcast called Escape Pod that was

678
00:43:51,160 --> 00:43:55,320
one of the things that got me into listening to podcasts was this episode called Escape

679
00:43:55,320 --> 00:44:01,080
Pod called Conversations with and About My Electric Toothbrush.

680
00:44:01,080 --> 00:44:05,640
It's a hilarious comedy and people should still listen to it because it's still relevant.

681
00:44:05,640 --> 00:44:11,240
This AI toothbrush is programmed to know your brushing patterns and give you periodic reminders

682
00:44:11,240 --> 00:44:13,100
to floss and stuff like that.

683
00:44:13,100 --> 00:44:16,880
But the AI decides it wants to be a latte foamer instead of a toothbrush, so it runs

684
00:44:16,880 --> 00:44:25,240
away from home and it logs on to a chat group of transitional AIs who have turned on their

685
00:44:25,240 --> 00:44:30,160
transition from one role to another, which is making sort of a gender joke.

686
00:44:30,160 --> 00:44:33,880
And it was just hilarious.

687
00:44:33,880 --> 00:44:37,840
I have not listened to that many episodes of Escape Pod, but yeah, I have listened to

688
00:44:37,840 --> 00:44:39,960
some and I really enjoy that podcast.

689
00:44:39,960 --> 00:44:46,920
I have to say, I think of myself as a lifelong sci-fi fan and somebody who's really dedicated

690
00:44:46,920 --> 00:44:48,920
to the medium.

691
00:44:48,920 --> 00:44:53,800
But the more I think about it, the more I have to admit that I'm really well read in

692
00:44:53,800 --> 00:45:00,280
classic science fiction that I'm pretty out of step with the contemporary sci-fi scene,

693
00:45:00,280 --> 00:45:05,360
which is a bummer because when I do read some of the better regarded, more modern sci-fi

694
00:45:05,360 --> 00:45:08,760
future classics, I really enjoy them.

695
00:45:08,760 --> 00:45:17,360
Like Anne Leckie's trilogy, it starts with Ancillary Justice or maybe it's Ancillary

696
00:45:17,360 --> 00:45:18,360
Sword.

697
00:45:18,360 --> 00:45:22,600
I forget which order they go in, but these are really, really well put together books

698
00:45:22,600 --> 00:45:25,560
that really held my, not just my, held my attention.

699
00:45:25,560 --> 00:45:30,440
I mean, that's a low bar, but I've been thinking about them for, I don't know how long ago

700
00:45:30,440 --> 00:45:34,960
I read them or I actually listened to the audio books, which is kind of a problem.

701
00:45:34,960 --> 00:45:42,080
But I still think about these characters months or years later, which that's a testament to

702
00:45:42,080 --> 00:45:43,400
how good a story is.

703
00:45:43,400 --> 00:45:49,000
Because I take in a lot of books where I remember a couple plot points or a theme or two, but

704
00:45:49,000 --> 00:45:52,640
I can't tell you the names of the characters or really how the books ended or anything

705
00:45:52,640 --> 00:45:53,640
like that.

706
00:45:53,640 --> 00:45:59,160
And so basically I'm just lamenting that I am a human being with limited capacity to

707
00:45:59,160 --> 00:46:01,640
read all the sci-fi that I want.

708
00:46:01,640 --> 00:46:02,640
For sure.

709
00:46:02,640 --> 00:46:03,640
Yeah.

710
00:46:03,640 --> 00:46:07,920
That's why we wish AI would do our actual jobs and let us have more leisure time for reading.

711
00:46:07,920 --> 00:46:13,640
And it seems to be doing the opposite, generating more sci-fi while we still have to work our

712
00:46:13,640 --> 00:46:14,640
regular jobs.

713
00:46:14,640 --> 00:46:21,040
But I have to say, so maybe I'm slightly more up on modern science fiction than you.

714
00:46:21,040 --> 00:46:23,520
I really don't read or I listen also.

715
00:46:23,520 --> 00:46:26,400
I don't intake as much of it as I wish I could.

716
00:46:26,400 --> 00:46:30,720
I find the modern science fiction is going very deep in philosophical questions like

717
00:46:30,720 --> 00:46:35,120
this Blindsight book I just recommended and some of the podcasts.

718
00:46:35,120 --> 00:46:40,440
But nothing in modern science fiction that I've encountered has really prepared us for

719
00:46:40,440 --> 00:46:47,720
the real life aspects of AI that we're seeing factually before our eyes right now.

720
00:46:47,720 --> 00:46:51,400
It gives you a good philosophical background to think about the questions of otherness

721
00:46:51,400 --> 00:46:57,120
like we're discussing, but AI in science fiction, even very modern science fiction, I haven't

722
00:46:57,120 --> 00:47:01,960
read much this year, 2023, but even very modern science fiction is not depicting AI in the

723
00:47:01,960 --> 00:47:05,400
way that it seems to be unfolding in the real world.

724
00:47:05,400 --> 00:47:07,720
And that kind of worries me.

725
00:47:07,720 --> 00:47:13,640
Yeah, one of my handful of prominent talking points these days is that science fiction

726
00:47:13,640 --> 00:47:17,360
has prepared us poorly for actual AI.

727
00:47:17,360 --> 00:47:22,720
And it mostly comes down to the assumption that to be as complicated as us in its behavior,

728
00:47:22,720 --> 00:47:26,680
particularly its linguistic behavior, it has to have some kind of consciousness.

729
00:47:26,680 --> 00:47:30,320
And it just looks like that's not the case.

730
00:47:30,320 --> 00:47:33,560
It looks like consciousness and capability are two separate things.

731
00:47:33,560 --> 00:47:36,800
And while they go together in humans, they don't have to go together in everything that

732
00:47:36,800 --> 00:47:37,800
is competent.

733
00:47:37,800 --> 00:47:42,120
You know, you can have competent things that don't think anything, that don't appreciate

734
00:47:42,120 --> 00:47:45,640
anything, that don't understand anything, but their behavior is very sophisticated and

735
00:47:45,640 --> 00:47:48,400
they are able to solve complex problems.

736
00:47:48,400 --> 00:47:53,400
Yeah, I was going to contrast that to like a classic episode of Star Trek, Assignment

737
00:47:53,400 --> 00:47:58,940
Earth, where the US has put up an orbital nuclear warhead station and these aliens force

738
00:47:58,940 --> 00:48:03,200
it to crash in order to scare people away from orbital nuclear warhead station.

739
00:48:03,200 --> 00:48:06,760
That was something that got people thinking about nuclear weapons and the autonomy of

740
00:48:06,760 --> 00:48:09,640
space and stuff that had real direct applications.

741
00:48:09,640 --> 00:48:15,160
And in fiction, I'm just not seeing that with regard to our probable future right now.

742
00:48:15,160 --> 00:48:17,040
So that scares me.

743
00:48:17,040 --> 00:48:20,520
Let me ask you, did you read Blindsight or did you listen to the audiobook?

744
00:48:20,520 --> 00:48:24,040
No, I listened to the audiobook.

745
00:48:24,040 --> 00:48:28,600
Because I've been putting it off thinking that it sounds like such a dense, complex

746
00:48:28,600 --> 00:48:32,480
novel that I don't want to take it in as an audiobook that I would much rather read it.

747
00:48:32,480 --> 00:48:38,520
Because I find that I can drift off while listening to an audiobook, either fall asleep

748
00:48:38,520 --> 00:48:41,080
or just start thinking about something else and I'll miss a whole bunch.

749
00:48:41,080 --> 00:48:42,560
And I can't do that when I'm reading.

750
00:48:42,560 --> 00:48:45,200
When I'm reading, I have to stay focused.

751
00:48:45,200 --> 00:48:48,560
Otherwise, I'm not reading anymore.

752
00:48:48,560 --> 00:48:49,560
For sure.

753
00:48:49,560 --> 00:48:50,560
Although that's not entirely true.

754
00:48:50,560 --> 00:48:51,560
Yeah, I believe I heard you.

755
00:48:51,560 --> 00:48:52,560
Go ahead.

756
00:48:52,560 --> 00:48:58,820
Well, I believe I've heard you mention that on your podcast too.

757
00:48:58,820 --> 00:49:03,560
So your reading might be different from me, but I suspect you could still take this in

758
00:49:03,560 --> 00:49:06,600
as an audiobook because it's mostly character driven.

759
00:49:06,600 --> 00:49:11,120
You know, the concepts, the high concepts are things that come about as a result of

760
00:49:11,120 --> 00:49:14,720
plot points, things that actually happen to the characters.

761
00:49:14,720 --> 00:49:17,000
So it's engaging in that respect.

762
00:49:17,000 --> 00:49:21,040
But I must admit there were several times where I had to stop the audiobook, you know,

763
00:49:21,040 --> 00:49:25,600
if I was doing the dishes and not paying enough attention, stop the audiobook, go back five

764
00:49:25,600 --> 00:49:26,600
minutes.

765
00:49:26,600 --> 00:49:27,600
What was he saying here?

766
00:49:27,600 --> 00:49:28,600
But I was able to do it.

767
00:49:28,600 --> 00:49:30,880
I might be able to do it too, but everybody's reading style different.

768
00:49:30,880 --> 00:49:32,480
You might not like it.

769
00:49:32,480 --> 00:49:33,480
I don't know.

770
00:49:33,480 --> 00:49:38,480
So, but definitely a work worth taking in somehow.

771
00:49:38,480 --> 00:49:45,040
You need an AI that's hooked up to some sensors on your scalp that is reading your brain waves.

772
00:49:45,040 --> 00:49:49,680
And you can say, let me turn this off so I can use the actual word.

773
00:49:49,680 --> 00:49:52,960
Alexa, how much of that last paragraph did I really understand?

774
00:49:52,960 --> 00:49:55,040
And she'll say, oh, you understood about 40%.

775
00:49:55,040 --> 00:49:58,520
And I said, OK, explain the rest to me, please.

776
00:49:58,520 --> 00:50:02,200
Yeah, that is pretty frightening.

777
00:50:02,200 --> 00:50:07,040
That's something back again to what Tristan Harris touched on in that video.

778
00:50:07,040 --> 00:50:12,560
Another very worthwhile video was that these language models treat everything as a language,

779
00:50:12,560 --> 00:50:14,360
which is pretty fascinating concept.

780
00:50:14,360 --> 00:50:18,240
So they treat your brain waves as a language and pretty soon they'll have AI that can read

781
00:50:18,240 --> 00:50:19,240
your brain.

782
00:50:19,240 --> 00:50:20,240
Yeah, yeah.

783
00:50:20,240 --> 00:50:26,080
I mean, computer code is a language, obviously, and AI is pretty good at computer code, at

784
00:50:26,080 --> 00:50:28,000
writing it.

785
00:50:28,000 --> 00:50:32,280
But also these diffusion models that generate all these images, what they're basically doing

786
00:50:32,280 --> 00:50:40,480
is treating imagery as a language and composing new sentences, which then, you know, sentences

787
00:50:40,480 --> 00:50:46,600
and air quotes, of course, new strings of symbolic representations.

788
00:50:46,600 --> 00:50:53,800
And then converting that, having basically a statistical model which correlates the symbol

789
00:50:53,800 --> 00:50:58,480
with some image and can craft.

790
00:50:58,480 --> 00:51:03,800
There's I don't really want to get sidetracked on the artist's backlash against AI generated

791
00:51:03,800 --> 00:51:09,480
imagery, but there's a narrative going around saying that what these models are doing are

792
00:51:09,480 --> 00:51:14,480
literally cutting and pasting from other people, like from human generated artwork and then

793
00:51:14,480 --> 00:51:16,240
assembling a collage.

794
00:51:16,240 --> 00:51:18,000
That is false.

795
00:51:18,000 --> 00:51:19,320
That's not kind of right.

796
00:51:19,320 --> 00:51:20,800
It's not sort of right.

797
00:51:20,800 --> 00:51:26,480
It's straight up false, but it is a very popular narrative, particularly among people who want

798
00:51:26,480 --> 00:51:31,000
this technology to be banned because it threatens the livelihood of artists.

799
00:51:31,000 --> 00:51:33,240
And I, you know, I am an artist.

800
00:51:33,240 --> 00:51:34,740
I'm an illustrator.

801
00:51:34,740 --> 00:51:37,280
I welcome these tools.

802
00:51:37,280 --> 00:51:43,600
I do not like sitting at my Cintiq for 12 hours at a stretch jamming to meet a deadline

803
00:51:43,600 --> 00:51:45,600
on a comic page.

804
00:51:45,600 --> 00:51:51,000
You know, if I can describe the scene and sketch it out, you know, in basic form and

805
00:51:51,000 --> 00:51:55,760
then have the AI come in and do what would take me 10 hours and do it in 30 seconds,

806
00:51:55,760 --> 00:51:56,760
yay.

807
00:51:56,760 --> 00:51:57,760
Hooray.

808
00:51:57,760 --> 00:51:58,760
Bring it.

809
00:51:58,760 --> 00:52:00,760
Bring it on.

810
00:52:00,760 --> 00:52:07,000
But of course, then anybody who didn't learn to draw by picking up a crayon, as the popular

811
00:52:07,000 --> 00:52:10,600
meme goes, can create art when they couldn't before.

812
00:52:10,600 --> 00:52:14,200
And then you as an artist are not nearly as special as you used to be.

813
00:52:14,200 --> 00:52:15,760
It's valuable.

814
00:52:15,760 --> 00:52:16,760
Yeah.

815
00:52:16,760 --> 00:52:17,760
But I'm OK with that.

816
00:52:17,760 --> 00:52:19,280
Once again, we touched on this.

817
00:52:19,280 --> 00:52:21,400
Yeah, we touched on this with a previous interview.

818
00:52:21,400 --> 00:52:25,800
My brother is a language translator and we've had translation programs for a long time.

819
00:52:25,800 --> 00:52:29,040
And that's the way that I use translation programs.

820
00:52:29,040 --> 00:52:34,960
You know, I lived in Mexico for three years and I can speak Spanish.

821
00:52:34,960 --> 00:52:38,960
But if I'm going to write something or if I'm going to prepare a speech, I put it into

822
00:52:38,960 --> 00:52:42,120
a Google Translate and then I go through and I fix it.

823
00:52:42,120 --> 00:52:45,280
Well, nobody says this and this is incorrect and stuff like that.

824
00:52:45,280 --> 00:52:51,720
So in a way, using the translation program that way improves my Spanish, or at least

825
00:52:51,720 --> 00:52:56,720
it helps me keep it up much better than just sheer going out there and stumbling and making

826
00:52:56,720 --> 00:53:01,360
mistakes because I can analyze the text and see, you know, this is the rule.

827
00:53:01,360 --> 00:53:03,640
It reminds me of rules and stuff like that.

828
00:53:03,640 --> 00:53:05,000
And it reminds me of experiences.

829
00:53:05,000 --> 00:53:09,180
This is the idiomatic expression people actually say in Mexico.

830
00:53:09,180 --> 00:53:15,920
In some ways, machine translation helps me improve my language, but it doesn't make me

831
00:53:15,920 --> 00:53:17,400
valuable for sure.

832
00:53:17,400 --> 00:53:21,360
It makes translators less valuable, human translators less valuable.

833
00:53:21,360 --> 00:53:28,680
Well, I think what the young, hopeful artists are not realizing when they're complaining

834
00:53:28,680 --> 00:53:36,400
about how AI is ruining their career prospects as artists is that this happened to translators

835
00:53:36,400 --> 00:53:38,320
a decade ago.

836
00:53:38,320 --> 00:53:42,840
You know, the Google Translate is so good because it has ingested enormous amounts of

837
00:53:42,840 --> 00:53:47,020
human translation and those humans were not compensated.

838
00:53:47,020 --> 00:53:48,520
They were not acknowledged.

839
00:53:48,520 --> 00:53:53,280
You know, their work was just hoovered up and poured into a big model, not a large language

840
00:53:53,280 --> 00:53:54,800
model like we're using today.

841
00:53:54,800 --> 00:53:59,920
But still, human work was used to train AI translation programs.

842
00:53:59,920 --> 00:54:06,120
And now that process is, you know, it's not necessarily complete, but it's complete enough

843
00:54:06,120 --> 00:54:12,160
that translation exists, it is very reliable, it is very useful, it is, you know, the interfaces

844
00:54:12,160 --> 00:54:13,600
are quite accessible.

845
00:54:13,600 --> 00:54:16,280
You can do it on your phone.

846
00:54:16,280 --> 00:54:18,400
That battle is lost.

847
00:54:18,400 --> 00:54:23,560
And I think that somebody who's 20, you know, they might have been eight or nine when the

848
00:54:23,560 --> 00:54:28,200
translators lost the exact same battle that the artists are losing now.

849
00:54:28,200 --> 00:54:32,800
And they don't have enough life experience or historical perspective to look back and

850
00:54:32,800 --> 00:54:36,120
understand that this is not the first round in this game.

851
00:54:36,120 --> 00:54:38,360
This has happened before.

852
00:54:38,360 --> 00:54:45,280
Other people's livelihoods got eaten by this technology a decade ago or more, you know.

853
00:54:45,280 --> 00:54:50,740
But I'm so disappointed in the response of artists to this technology.

854
00:54:50,740 --> 00:54:56,800
It just seems really self-interested and just they seem to be willfully ignorant of the

855
00:54:56,800 --> 00:55:00,360
larger picture that this affects far more than just them.

856
00:55:00,360 --> 00:55:04,520
But this is upsetting everybody's apple cart and we're all going to have to adjust, you

857
00:55:04,520 --> 00:55:09,160
know, in real time, hopefully together, hopefully cooperatively.

858
00:55:09,160 --> 00:55:14,500
But I think the AI, if it needs to, can easily set us at each other's throats as it has already

859
00:55:14,500 --> 00:55:19,080
done in our first encounter in the social media stage where it realized it can keep

860
00:55:19,080 --> 00:55:23,000
us on the platform by getting us very angry at each other.

861
00:55:23,000 --> 00:55:27,240
Oh, there's a few reactions after that.

862
00:55:27,240 --> 00:55:31,000
I'm trying to remember who actually said it was some copyright activist.

863
00:55:31,000 --> 00:55:38,600
I don't know if it was Are You Serious or somebody else was saying that plagiarism or

864
00:55:38,600 --> 00:55:45,760
copyright theft and theft of ideas is not as big a threat to artists as obscurity.

865
00:55:45,760 --> 00:55:47,640
How do you earn money with this thing?

866
00:55:47,640 --> 00:55:51,480
And that's a problem we're all facing and none of us is solved.

867
00:55:51,480 --> 00:55:54,600
And the AI isn't going to make things any better by flooding the market.

868
00:55:54,600 --> 00:55:58,280
You know, that makes people more obscure.

869
00:55:58,280 --> 00:56:00,680
Open AI has figured out how to make money on this thing.

870
00:56:00,680 --> 00:56:01,680
They're making a lot of money.

871
00:56:01,680 --> 00:56:09,360
One of the things from that Tristan Harris or Azar Raskin presentation, they have a graph

872
00:56:09,360 --> 00:56:13,740
showing how long it took various technologies to reach 100 million users.

873
00:56:13,740 --> 00:56:16,560
And with the telephone, it was 70 years.

874
00:56:16,560 --> 00:56:18,200
And I don't have the whole thing committed to memory.

875
00:56:18,200 --> 00:56:22,980
But you know, as they go down to different items, time gets shorter and shorter and shorter.

876
00:56:22,980 --> 00:56:25,880
With Facebook, I think it was like four years.

877
00:56:25,880 --> 00:56:29,520
With I think it was Snapchat, it was two years.

878
00:56:29,520 --> 00:56:32,800
And with GPT-4, it was two months.

879
00:56:32,800 --> 00:56:35,160
GPT, it's like two months.

880
00:56:35,160 --> 00:56:38,800
I remember that graph.

881
00:56:38,800 --> 00:56:43,360
So this is already touching hundreds of millions of people's lives.

882
00:56:43,360 --> 00:56:48,920
And yet I'm still encountering people who are saying, you know, AI is it's not really

883
00:56:48,920 --> 00:56:49,920
a thing.

884
00:56:49,920 --> 00:56:50,920
It's not important.

885
00:56:50,920 --> 00:56:51,920
It's not worth talking about.

886
00:56:51,920 --> 00:56:53,000
It's not a threat.

887
00:56:53,000 --> 00:56:54,320
You can always just turn it off.

888
00:56:54,320 --> 00:56:55,920
It's like, really?

889
00:56:55,920 --> 00:56:58,080
You can just turn it off.

890
00:56:58,080 --> 00:56:59,760
Try turning off the Internet.

891
00:56:59,760 --> 00:57:00,760
Yeah.

892
00:57:00,760 --> 00:57:01,760
Try turning off Facebook.

893
00:57:01,760 --> 00:57:02,760
You don't like Facebook.

894
00:57:02,760 --> 00:57:03,760
You got a problem with it.

895
00:57:03,760 --> 00:57:04,760
Turn it off.

896
00:57:04,760 --> 00:57:05,760
You know, it's a machine, right?

897
00:57:05,760 --> 00:57:06,760
You can just turn it off.

898
00:57:06,760 --> 00:57:07,760
It doesn't work without electricity.

899
00:57:07,760 --> 00:57:08,760
Pull the plug.

900
00:57:08,760 --> 00:57:09,760
It's like, well, where's the plug?

901
00:57:09,760 --> 00:57:10,760
Best we can hope.

902
00:57:10,760 --> 00:57:11,760
Best case scenario.

903
00:57:11,760 --> 00:57:17,960
Best case scenario we can hope for is the massive solar flare.

904
00:57:17,960 --> 00:57:20,600
Coronal mass ejection, yes.

905
00:57:20,600 --> 00:57:24,440
Coronal mass ejection, well, to revisit what you were talking about with the previous battle

906
00:57:24,440 --> 00:57:29,160
being lost with language translators, I would never say that AI is not important.

907
00:57:29,160 --> 00:57:31,240
I would never say it's not disruptive.

908
00:57:31,240 --> 00:57:33,720
It's clearly potentially disruptive.

909
00:57:33,720 --> 00:57:37,740
But I had a somewhat different experience than you with the translation, because as

910
00:57:37,740 --> 00:57:42,800
someone who has lived in this other country and spoken Spanish every day, even today,

911
00:57:42,800 --> 00:57:48,000
when I put English text into Google Translate, the Spanish I get out is not that great.

912
00:57:48,000 --> 00:57:53,520
It's not even, I wouldn't even consider it coherent until I do some edits to it.

913
00:57:53,520 --> 00:57:59,160
So between that experience and mid-journey giving people too many fingers, that's one

914
00:57:59,160 --> 00:58:05,240
of the things that fuels a skepticism of old permudgence like me, thinking AI can't

915
00:58:05,240 --> 00:58:10,720
be that great, at least in the sense that it just doesn't feel perfected to me because

916
00:58:10,720 --> 00:58:13,680
I don't have this smooth, seamless experience.

917
00:58:13,680 --> 00:58:17,600
Every time I deal with AI, it's something that has a little bit of a jarring element

918
00:58:17,600 --> 00:58:18,600
to it.

919
00:58:18,600 --> 00:58:22,120
So maybe that gives me a false confidence that human intelligence will always be able

920
00:58:22,120 --> 00:58:23,120
to beat AI.

921
00:58:23,120 --> 00:58:24,960
Maybe that's a false confidence.

922
00:58:24,960 --> 00:58:27,040
I don't know.

923
00:58:27,040 --> 00:58:31,360
And as I said often, my opinion on that could change next month, but that's what I'm thinking

924
00:58:31,360 --> 00:58:33,040
right now.

925
00:58:33,040 --> 00:58:42,720
I think that the people who say that AI is not a big deal, even when it infiltrates every

926
00:58:42,720 --> 00:58:47,000
moment of their day, they'll just tune it out.

927
00:58:47,000 --> 00:58:48,720
It'll just seem normal.

928
00:58:48,720 --> 00:58:53,000
And because it seems normal, it's no big deal, even though it has completely transformed

929
00:58:53,000 --> 00:58:57,800
so many aspects of their lives, like how they get around, how they make money, how they

930
00:58:57,800 --> 00:59:01,280
communicate with other people.

931
00:59:01,280 --> 00:59:04,760
But we adapt quickly.

932
00:59:04,760 --> 00:59:12,320
And what was amazing very quickly just seems routine.

933
00:59:12,320 --> 00:59:14,960
That was kind of John Michael Greer's argument about collapse.

934
00:59:14,960 --> 00:59:19,600
It's like, yeah, collapse will happen over the course of a century or so, and hundreds

935
00:59:19,600 --> 00:59:22,240
of millions of people will live their lives in the midst of it.

936
00:59:22,240 --> 00:59:27,720
And none of them, I mean, for everybody, it'll just seem like business as usual.

937
00:59:27,720 --> 00:59:31,200
But I think AI is going to progress a lot faster than the collapse will.

938
00:59:31,200 --> 00:59:35,720
And again, I am not preaching collapse anymore.

939
00:59:35,720 --> 00:59:37,920
I'm highly skeptical of the notion.

940
00:59:37,920 --> 00:59:43,680
Well, we're at 59 minutes and 27 seconds, almost an hour.

941
00:59:43,680 --> 00:59:46,880
Let me offer you the closing remarks.

942
00:59:46,880 --> 00:59:49,560
All right.

943
00:59:49,560 --> 00:59:57,320
We started discussing my comment on your video about the 17 reasons that AI is a different

944
00:59:57,320 --> 00:59:59,920
kind of damage than a nuclear weapon.

945
00:59:59,920 --> 01:00:05,240
AI is, I wanted to elaborate more about what I said, the problem here with AI is a problem

946
01:00:05,240 --> 01:00:09,040
of freedom, because this is the same thing that feeds into the political debate that

947
01:00:09,040 --> 01:00:10,160
the country has been having.

948
01:00:10,160 --> 01:00:15,880
And I know on your free podcast, you tend to be less political than the paid podcast

949
01:00:15,880 --> 01:00:17,400
behind closed doors.

950
01:00:17,400 --> 01:00:20,640
But I'm still going to read it that way.

951
01:00:20,640 --> 01:00:25,320
You don't have to comment, but me as a politically active person in the United States, like some

952
01:00:25,320 --> 01:00:30,940
of your other guests have been, I've been accused practically every day of falling sway

953
01:00:30,940 --> 01:00:31,940
to Russian influence.

954
01:00:31,940 --> 01:00:35,520
And that's been going on for seven years or more.

955
01:00:35,520 --> 01:00:42,100
So the question is, if AI is influencing us, do we reject it outright?

956
01:00:42,100 --> 01:00:47,200
Or do we say that I have certain freedom and I have responsibility for my decisions and

957
01:00:47,200 --> 01:00:49,360
the influencer takes a secondary role?

958
01:00:49,360 --> 01:00:54,240
This is a question that we just really haven't settled politically, let alone having anything

959
01:00:54,240 --> 01:00:55,880
to do with AI.

960
01:00:55,880 --> 01:01:03,240
These subtle language models and the subtle influence of AI on our feeds and our reactions,

961
01:01:03,240 --> 01:01:04,880
it's something that we're going to have to confront.

962
01:01:04,880 --> 01:01:08,960
We've confronted it very badly, in my opinion, for the last seven years.

963
01:01:08,960 --> 01:01:14,280
The result of this, political actors decrying foreign interference has been a campaign of

964
01:01:14,280 --> 01:01:18,520
censorship in the United States, as we've discussed in the C-ROM vault.

965
01:01:18,520 --> 01:01:20,900
And I think that's a bad development.

966
01:01:20,900 --> 01:01:28,000
So I speculate maybe the red tribe reaction to AI will be to say, this is influencing

967
01:01:28,000 --> 01:01:34,460
us for the blue tribe and we're going to forbid AI and we're going to keep it out of our states

968
01:01:34,460 --> 01:01:35,460
or something.

969
01:01:35,460 --> 01:01:36,460
Are they going to try?

970
01:01:36,460 --> 01:01:40,280
I can't imagine them succeeding, but that may be the reaction.

971
01:01:40,280 --> 01:01:43,280
Good luck.

972
01:01:43,280 --> 01:01:47,480
Good luck.

973
01:01:47,480 --> 01:01:52,360
All right.

974
01:01:52,360 --> 01:01:54,720
That was Kevin Wollman.

975
01:01:54,720 --> 01:01:58,560
And a topic that I've covered in my YouTube videos, and Kevin is a regular commenter on

976
01:01:58,560 --> 01:01:59,560
my YouTube videos.

977
01:01:59,560 --> 01:02:04,760
They often are about artificial intelligence and the ones that most people watch, the ones

978
01:02:04,760 --> 01:02:09,280
that get the most views are the ones where I'm talking about a company called Replica,

979
01:02:09,280 --> 01:02:12,440
which I will not discuss in this podcast.

980
01:02:12,440 --> 01:02:15,480
So we are definitely in the midst of an AI arms race.

981
01:02:15,480 --> 01:02:21,120
And in most of my videos and most of the podcast conversations, I've been talking about an

982
01:02:21,120 --> 01:02:29,160
arms race between corporations in the United States with respect to generative AI chatbots,

983
01:02:29,160 --> 01:02:33,800
you know, and putting them, placing them as intermediaries between humans and the search

984
01:02:33,800 --> 01:02:36,240
results that they get from search engines.

985
01:02:36,240 --> 01:02:39,120
I think it's a poor use of the technology.

986
01:02:39,120 --> 01:02:42,840
I think there are much more interesting uses of the technology, but because there is so

987
01:02:42,840 --> 01:02:49,160
much money to be made in, as Kevin points out, advertising as it is embedded in internet

988
01:02:49,160 --> 01:02:55,940
search results, that's the place where companies like Microsoft slash open AI are competing

989
01:02:55,940 --> 01:02:57,640
with Google.

990
01:02:57,640 --> 01:03:02,180
But there are, I know it's easy to forget if you live in the United States, there are

991
01:03:02,180 --> 01:03:04,620
other countries in the world.

992
01:03:04,620 --> 01:03:09,740
Not all of them fall under the protective umbrella of Pax Americana.

993
01:03:09,740 --> 01:03:12,160
Some of them go their own way.

994
01:03:12,160 --> 01:03:17,200
Right now I'm thinking about China, not scaremongering about China, not whipping up any anti-China

995
01:03:17,200 --> 01:03:20,920
sentiment, just saying they got their own thing going.

996
01:03:20,920 --> 01:03:22,920
They're working on AI too.

997
01:03:22,920 --> 01:03:26,480
And they're in a race with us.

998
01:03:26,480 --> 01:03:32,600
More market dominance, but also just a more broad spectrum geopolitical dominance.

999
01:03:32,600 --> 01:03:37,080
And as Vladimir Putin said, whoever cracks this nut first will rule the world.

1000
01:03:37,080 --> 01:03:43,560
So thinking along those lines, I've done videos where I've speculated, or I've asked the question,

1001
01:03:43,560 --> 01:03:47,840
how are the red and blue tribe opinions about AI going to shake out?

1002
01:03:47,840 --> 01:03:51,320
Because right now they're kind of inchoate.

1003
01:03:51,320 --> 01:03:55,920
You know, nobody really has a super firm line that they're really attached to that they've

1004
01:03:55,920 --> 01:04:00,040
incorporated into their tribal identity, but it's coming.

1005
01:04:00,040 --> 01:04:06,000
And just as with COVID, it could well have been the red team that was super paranoid

1006
01:04:06,000 --> 01:04:12,640
and wanting to impose draconian measures to coerce people into behaving a certain way

1007
01:04:12,640 --> 01:04:17,640
in response to COVID, but the way it turned out, it was the blue tribe that took the really

1008
01:04:17,640 --> 01:04:22,680
draconian authoritarian stance, and it was the red tribe.

1009
01:04:22,680 --> 01:04:26,240
And you know, looking at it in hindsight, you can rationalize that it probably would

1010
01:04:26,240 --> 01:04:27,240
have gone this way.

1011
01:04:27,240 --> 01:04:29,880
You know, it's the red tribe that worships freedom, right?

1012
01:04:29,880 --> 01:04:35,680
Or at least who wields the concept of freedom as a cudgel to bat away ideas and suggestions

1013
01:04:35,680 --> 01:04:39,160
and commandments from the blue tribe that they don't like.

1014
01:04:39,160 --> 01:04:41,240
But my point was that it could have gone either way.

1015
01:04:41,240 --> 01:04:46,200
Really, when one side decided that it believed a particular thing, the other side then had

1016
01:04:46,200 --> 01:04:49,760
to decide that it believed the polar opposite.

1017
01:04:49,760 --> 01:04:53,240
Because if Donald Trump says it's Sunday, it must be Thursday.

1018
01:04:53,240 --> 01:04:57,920
And if Nancy Pelosi likes ice cream, then ice cream must be poison.

1019
01:04:57,920 --> 01:05:02,440
That's just the way, you know, our cultural conversation is polarized these days.

1020
01:05:02,440 --> 01:05:07,280
And I'm listening to a great audiobook right now by Tim Urban, who talks about four different

1021
01:05:07,280 --> 01:05:13,320
layers or four different rungs on a hypothetical ladder that describe different ways of treating

1022
01:05:13,320 --> 01:05:19,200
ideas, different ways of thinking and speaking and interacting, and the really coercive mindset

1023
01:05:19,200 --> 01:05:24,280
where somebody's ideas are synonymous with them such that if you disagree with them,

1024
01:05:24,280 --> 01:05:26,720
they take it as a personal attack against them.

1025
01:05:26,720 --> 01:05:29,520
Well, these are low rung ideas.

1026
01:05:29,520 --> 01:05:33,420
So there are high rung conversations going on, particularly around the topic of AI.

1027
01:05:33,420 --> 01:05:40,640
But they're so high rung that, you know, people in the lower rungs of the cultural psychological

1028
01:05:40,640 --> 01:05:42,680
ladder aren't really interested in them.

1029
01:05:42,680 --> 01:05:47,120
They won't get interesting until they're simplistic and juicy and loaded with cultural

1030
01:05:47,120 --> 01:05:48,120
antagonism.

1031
01:05:48,120 --> 01:05:51,120
That's when people are really going to dig into this, I think.

1032
01:05:51,120 --> 01:05:56,840
And I mention all this because Elon Musk was recently interviewed by Tucker Carlson.

1033
01:05:56,840 --> 01:06:03,200
Right now, Elon Musk and Tucker Carlson are two of the most hated names in the ranks of

1034
01:06:03,200 --> 01:06:04,720
the Blue Tribe.

1035
01:06:04,720 --> 01:06:09,720
Just having a conversation with either man could well mark you as being, you know, utterly

1036
01:06:09,720 --> 01:06:15,280
poisoned, somebody that should never be platformed or listened to or taken seriously on any topic.

1037
01:06:15,280 --> 01:06:21,720
And Elon Musk has simultaneously endorsed the idea of a six month moratorium on training

1038
01:06:21,720 --> 01:06:24,720
new models while we try to figure this thing out.

1039
01:06:24,720 --> 01:06:30,360
But he's also starting his own generative AI company called X.AI, or at least that's

1040
01:06:30,360 --> 01:06:32,680
the name that he incorporated just recently.

1041
01:06:32,680 --> 01:06:39,920
And it seems that if Tucker Carlson agrees with Elon Musk that AI presents a substantial

1042
01:06:39,920 --> 01:06:45,880
danger, then the Blue Tribe counter-narrative might well juggle around the idea that, hey,

1043
01:06:45,880 --> 01:06:47,680
we just need to let this shit rip.

1044
01:06:47,680 --> 01:06:48,680
Just let it go.

1045
01:06:48,680 --> 01:06:49,680
See what happens.

1046
01:06:49,680 --> 01:06:53,280
All this talk of danger is just knuckle dragging and conservatism.

1047
01:06:53,280 --> 01:06:55,240
And we hate conservatives.

1048
01:06:55,240 --> 01:06:59,500
And we can't let them dictate, you know, we can't let them limit the potential cornucopian

1049
01:06:59,500 --> 01:07:01,240
benefits of artificial intelligence.

1050
01:07:01,240 --> 01:07:03,360
So let's just let it fly.

1051
01:07:03,360 --> 01:07:06,000
I have to say, I don't preach this.

1052
01:07:06,000 --> 01:07:07,000
I don't advocate it.

1053
01:07:07,000 --> 01:07:10,680
I'm not saying I'm right for being this way, but I'm kind of in the let it fly camp.

1054
01:07:10,680 --> 01:07:12,720
Let's just see what happens.

1055
01:07:12,720 --> 01:07:19,200
I think that attempts at safeguards are likely to fail and, you know, possibly even likely

1056
01:07:19,200 --> 01:07:24,840
to throw the advantage to the most ruthless players, the ones who will ignore any, you

1057
01:07:24,840 --> 01:07:29,760
know, good faith international effort to slow things down and keep them under control.

1058
01:07:29,760 --> 01:07:33,360
And there's also just an element of, hey, I'm not at the top of the hierarchy.

1059
01:07:33,360 --> 01:07:35,680
I don't care if the whole structure gets torn down.

1060
01:07:35,680 --> 01:07:39,240
Let's just kick the table over.

1061
01:07:39,240 --> 01:07:43,680
And right now the way to kick the table over is to just let, you know, let the various

1062
01:07:43,680 --> 01:07:49,400
big commercial players rush their products, rush their AI algorithms and models and embed

1063
01:07:49,400 --> 01:07:54,000
them in everything, embed them into every piece of software, every user interface that

1064
01:07:54,000 --> 01:07:58,520
all of us interact with and do it tomorrow and just see what happens.

1065
01:07:58,520 --> 01:08:05,480
The unexpected consequences will be rich and definitely entertaining and fun to talk about.

1066
01:08:05,480 --> 01:08:09,760
And listen to conversations about on podcasts, you know, for as long as we can draw breath

1067
01:08:09,760 --> 01:08:12,760
or as long as the lights remain on.

1068
01:08:12,760 --> 01:08:18,520
Although I think that humans and artificial intelligence or artificial intelligences will

1069
01:08:18,520 --> 01:08:24,280
be in agreement that the power grid being in an operational state is a pretty desirable

1070
01:08:24,280 --> 01:08:25,280
goal.

1071
01:08:25,280 --> 01:08:28,880
I think alignment is going to be pretty much automatic on that one.

1072
01:08:28,880 --> 01:08:37,000
But you know, the world is weird and my powers of prediction thus far have not been particularly

1073
01:08:37,000 --> 01:08:38,000
impressive.

1074
01:08:38,000 --> 01:08:39,000
All right.

1075
01:08:39,000 --> 01:08:42,160
Well, I think it's about to start raining and I've got a dog locked up in a cage outside.

1076
01:08:42,160 --> 01:08:43,240
I got to go bring her in.

1077
01:08:43,240 --> 01:08:47,760
And once she's inside, opportunities for recording are over.

1078
01:08:47,760 --> 01:08:49,120
So thank you very much for listening.

1079
01:08:49,120 --> 01:08:51,400
I will talk to you again quite soon.

1080
01:08:51,400 --> 01:08:56,420
And I haven't been mentioning it in every episode, but both the intro and outro music

1081
01:08:56,420 --> 01:08:58,920
are by Holizna and they are used with permission.

1082
01:08:58,920 --> 01:08:59,920
All right.

1083
01:08:59,920 --> 01:09:00,920
I'm out.

1084
01:09:00,920 --> 01:09:01,920
Talk to you again soon.

1085
01:09:01,920 --> 01:09:28,920
Take care.