1
00:00:02,720 --> 00:00:13,960
Speaker 1: Bloomberg Audio Studios, Podcasts, radio News.

2
00:00:18,360 --> 00:00:19,639
Speaker 2: Hello, Odd Laws listeners.

3
00:00:19,680 --> 00:00:21,959
Speaker 3: I'm Joe Wisenthal and I'm Tracy Alloway.

4
00:00:22,120 --> 00:00:24,360
Speaker 4: Tracy, we were down at south By Southwest recently.

5
00:00:24,520 --> 00:00:25,239
Speaker 5: Yeah, that's right.

6
00:00:25,320 --> 00:00:27,520
Speaker 6: I didn't get to have any barbecue in Austin, but

7
00:00:27,560 --> 00:00:29,160
Speaker 6: we did have some really good Mexican food.

8
00:00:29,200 --> 00:00:32,080
Speaker 4: We did have really good Mexican food, and we recorded

9
00:00:32,080 --> 00:00:34,040
Speaker 4: a really fun episode of Our Parts.

10
00:00:34,200 --> 00:00:34,320
Speaker 3: Right.

11
00:00:34,400 --> 00:00:37,040
Speaker 6: More importantly, we recorded a good episode that's.

12
00:00:36,960 --> 00:00:38,840
Speaker 4: Right, so check it out, listeners. We did a live

13
00:00:38,960 --> 00:00:43,360
Speaker 4: episode of the podcast talking about sort of the prospect

14
00:00:43,440 --> 00:00:46,880
Speaker 4: of mass white collar displacement due to AI, as well

15
00:00:46,920 --> 00:00:49,800
Speaker 4: as the politics and how politicians should be thinking about

16
00:00:49,800 --> 00:00:52,800
Speaker 4: this and how voters are thinking about this. Our guests

17
00:00:52,800 --> 00:00:56,440
Speaker 4: were David Shore, founder of Blue Rose Research, and Burne Hobart,

18
00:00:56,440 --> 00:00:59,840
Speaker 4: who writes the excellent diff newsletter and the general partner

19
00:01:00,160 --> 00:01:03,440
Speaker 4: Anomaly Fund. Take a listen. Thanks everyone for coming out

20
00:01:03,480 --> 00:01:04,679
Speaker 4: on a cold Sunday morning.

21
00:01:04,800 --> 00:01:05,640
Speaker 2: Yeah, hello, and.

22
00:01:05,600 --> 00:01:08,319
Speaker 6: Welcome to a live recording of the Odd Loots podcast.

23
00:01:08,360 --> 00:01:10,720
Speaker 6: This is definitely going to be the most uplifting of

24
00:01:10,760 --> 00:01:13,960
Speaker 6: all the conversations that have happened this weekend.

25
00:01:14,040 --> 00:01:15,880
Speaker 2: Right, I think so, I think it really will be

26
00:01:15,920 --> 00:01:17,000
Speaker 2: we are everyone.

27
00:01:16,680 --> 00:01:20,440
Speaker 6: Here is going to leave feeling really good about the future.

28
00:01:21,080 --> 00:01:22,880
Speaker 4: All right, well let's kick it off. So we have

29
00:01:23,160 --> 00:01:25,320
Speaker 4: two great guests. We are going to be speaking with

30
00:01:25,440 --> 00:01:28,520
Speaker 4: David Shore. He is the founder of blue Rose Research,

31
00:01:28,560 --> 00:01:32,360
Speaker 4: political consultancy pollster knows a lot about AI. And we

32
00:01:32,440 --> 00:01:35,880
Speaker 4: have Burn Hobart, the founder of The disc a great

33
00:01:35,920 --> 00:01:38,479
Speaker 4: newsletter that everyone should read, and a general partner at

34
00:01:38,520 --> 00:01:41,120
Speaker 4: Anomaly Funds. And we're going to talk about all things

35
00:01:41,200 --> 00:01:44,280
Speaker 4: AI and job loss potential and the politics of it

36
00:01:44,319 --> 00:01:47,440
Speaker 4: and so forth. And so David and Burn, thank you

37
00:01:47,480 --> 00:01:49,760
Speaker 4: so much for joining us on stage here.

38
00:01:50,040 --> 00:01:50,720
Speaker 2: Great to be here.

39
00:01:51,000 --> 00:01:51,320
Speaker 7: Thank you.

40
00:01:51,640 --> 00:01:56,440
Speaker 4: So let's just start on this question. David, you're pretty

41
00:01:56,640 --> 00:01:59,960
Speaker 4: like this is happening now. The economy is going to

42
00:02:00,000 --> 00:02:03,600
Speaker 4: look radically different, maybe even a year eighteen months from now,

43
00:02:03,920 --> 00:02:06,160
Speaker 4: and you're trying to wake politicians up. This is happening

44
00:02:06,240 --> 00:02:08,640
Speaker 4: right now. But like, tell us what's happening right now

45
00:02:08,720 --> 00:02:09,680
Speaker 4: or what's about to happen.

46
00:02:10,000 --> 00:02:12,519
Speaker 7: Yeah, you know, I'm not an AI expert, and I

47
00:02:12,840 --> 00:02:14,760
Speaker 7: don't want to claim I one, but you know I would.

48
00:02:14,760 --> 00:02:17,840
Speaker 7: I'll say is that I use AI a lot, I

49
00:02:17,960 --> 00:02:20,080
Speaker 7: use cloud code a lot. I think in December, I

50
00:02:20,120 --> 00:02:22,840
Speaker 7: spent maybe like fifteen percent of my pre tax income

51
00:02:22,880 --> 00:02:26,840
Speaker 7: on cloud code overage VIS and I think there's a

52
00:02:26,880 --> 00:02:29,720
Speaker 7: real disconnect among my friends between you know, people who

53
00:02:29,800 --> 00:02:32,800
Speaker 7: use cloud code and people who don't, where I could

54
00:02:32,800 --> 00:02:35,960
Speaker 7: really feel the extent to which I can do so

55
00:02:36,080 --> 00:02:39,160
Speaker 7: much more now versus a month ago and a month

56
00:02:39,200 --> 00:02:42,840
Speaker 7: before that. Like, the scale at which these things are

57
00:02:42,840 --> 00:02:45,480
Speaker 7: getting better and the speed at which things are getting

58
00:02:45,480 --> 00:02:48,440
Speaker 7: better is really jarring. And I think you might not

59
00:02:48,680 --> 00:02:50,960
Speaker 7: notice that if you're just using chatchpt, Like you know,

60
00:02:51,040 --> 00:02:52,840
Speaker 7: chatchapt is like a little better than it was a

61
00:02:52,880 --> 00:02:56,760
Speaker 7: year ago. But again, it's very hard to make predictions

62
00:02:56,760 --> 00:02:59,880
Speaker 7: about the future. But I do think that if things

63
00:03:00,080 --> 00:03:03,639
Speaker 7: continue to improve on this scale, then things could get

64
00:03:03,880 --> 00:03:05,800
Speaker 7: really weird, really quickly and burn.

65
00:03:05,919 --> 00:03:07,320
Speaker 6: One of the reasons we wanted to talk to you

66
00:03:07,440 --> 00:03:09,959
Speaker 6: is because you're sort of at the intersection of technology

67
00:03:10,000 --> 00:03:13,800
Speaker 6: and finance. If I look at the valuations of big

68
00:03:13,840 --> 00:03:15,880
Speaker 6: tech at the moment and a bunch of the AI companies,

69
00:03:16,240 --> 00:03:19,200
Speaker 6: they basically suggest that they're going to take over the

70
00:03:19,320 --> 00:03:22,840
Speaker 6: entire economy. When you see those numbers, what do they

71
00:03:22,840 --> 00:03:25,040
Speaker 6: suggest to you about the future of labor?

72
00:03:25,240 --> 00:03:27,560
Speaker 8: So I think right now, it's definitely true that we're

73
00:03:27,919 --> 00:03:30,040
Speaker 8: I think you can think of this coherent category of

74
00:03:30,120 --> 00:03:32,519
Speaker 8: this is an AI company, or this is a non

75
00:03:32,720 --> 00:03:35,560
Speaker 8: this was a non AI company, but they're incorporating AI

76
00:03:35,680 --> 00:03:38,160
Speaker 8: and they have these distribution advantages, so they'll probably do

77
00:03:38,320 --> 00:03:40,880
Speaker 8: well at that. But I think that it's a mistake

78
00:03:40,920 --> 00:03:44,240
Speaker 8: to think that this is a durable, discrete category, in

79
00:03:44,280 --> 00:03:45,880
Speaker 8: the same way that you could refer to a lot

80
00:03:45,920 --> 00:03:48,320
Speaker 8: of companies as electricity companies. And if it were nineteen

81
00:03:48,360 --> 00:03:49,880
Speaker 8: twenty five and you're trying to figure out which stock

82
00:03:49,960 --> 00:03:52,360
Speaker 8: to buy and you're just really really bullish on electricity,

83
00:03:52,560 --> 00:03:56,000
Speaker 8: then you would really really care that RCA is exposed

84
00:03:56,040 --> 00:03:58,760
Speaker 8: to electricity, and I don't know us deal is mostly not.

85
00:03:59,280 --> 00:04:02,360
Speaker 8: But eventually every company becomes an electricity company, just in

86
00:04:02,360 --> 00:04:04,800
Speaker 8: the sense of you would run a very different business

87
00:04:04,840 --> 00:04:07,240
Speaker 8: if the lights did not turn on, and so it

88
00:04:07,440 --> 00:04:09,760
Speaker 8: like gets subsumed by the rest of the economy. And

89
00:04:09,960 --> 00:04:11,840
Speaker 8: you see that with software too, where there are just

90
00:04:11,880 --> 00:04:15,000
Speaker 8: a lot more companies that have software developers and are

91
00:04:15,040 --> 00:04:18,560
Speaker 8: writing software for internal use, and they're not software companies

92
00:04:18,560 --> 00:04:21,200
Speaker 8: per se. You know, their restaurants or like tractor companies

93
00:04:21,279 --> 00:04:21,599
Speaker 8: or whatever.

94
00:04:21,640 --> 00:04:22,960
Speaker 2: But they have that element.

95
00:04:23,080 --> 00:04:25,400
Speaker 8: And so I think that's part of what you see

96
00:04:25,440 --> 00:04:28,520
Speaker 8: just early in the role out of any general purpose technology,

97
00:04:28,600 --> 00:04:31,560
Speaker 8: is that you have a lot of really narrow specific bets,

98
00:04:31,560 --> 00:04:34,440
Speaker 8: and then over time the impact gets so widely distributed

99
00:04:34,480 --> 00:04:36,160
Speaker 8: that it's very hard to trace and you have to

100
00:04:36,240 --> 00:04:37,719
Speaker 8: kind of go back and look at the history and

101
00:04:37,760 --> 00:04:39,560
Speaker 8: look at things like, Okay, the rise of the car

102
00:04:40,000 --> 00:04:41,400
Speaker 8: leads to the rise of the suburb, but it also

103
00:04:41,480 --> 00:04:43,320
Speaker 8: leads to the rise at the grocery store because or

104
00:04:43,360 --> 00:04:45,760
Speaker 8: like the supermarket, where you can have a much larger

105
00:04:45,800 --> 00:04:48,400
Speaker 8: selection and that means lower labor costs per units sold,

106
00:04:48,440 --> 00:04:51,280
Speaker 8: and that means lower cost overall. And that works if

107
00:04:51,320 --> 00:04:53,880
Speaker 8: people are not walking to get their groceries. It doesn't

108
00:04:53,920 --> 00:04:55,760
Speaker 8: work if people are walking and it's like a daily

109
00:04:55,920 --> 00:04:57,640
Speaker 8: you know, stop on the way home from work or something.

110
00:04:58,000 --> 00:05:00,320
Speaker 8: So we will probably see a lot of those weird

111
00:05:00,440 --> 00:05:02,400
Speaker 8: kinds of outcomes. Like I think, if you're thinking about

112
00:05:02,600 --> 00:05:05,680
Speaker 8: the risk to a given career, I think for most

113
00:05:05,680 --> 00:05:08,240
Speaker 8: of the careers, people worry about the average, like the

114
00:05:08,279 --> 00:05:11,960
Speaker 8: mean compensation goes up, the median compensation of people who

115
00:05:11,960 --> 00:05:14,039
Speaker 8: are in that industry right now, the median compensation they

116
00:05:14,120 --> 00:05:16,479
Speaker 8: get from being in that industry probably goes down where

117
00:05:16,480 --> 00:05:17,600
Speaker 8: a lot of people get washed out.

118
00:05:17,600 --> 00:05:18,400
Speaker 2: But this happened before.

119
00:05:18,520 --> 00:05:21,880
Speaker 8: Like the spreadsheet did not eliminate investment banker or accountant

120
00:05:21,920 --> 00:05:23,640
Speaker 8: as a job. It actually made it a more lucrative job,

121
00:05:23,640 --> 00:05:25,440
Speaker 8: but it also made it a more measurable one where

122
00:05:25,839 --> 00:05:28,719
Speaker 8: you just can't slack off the way that you perhaps

123
00:05:28,800 --> 00:05:30,600
Speaker 8: used to be able to somewhat slack off in some

124
00:05:30,760 --> 00:05:33,600
Speaker 8: of the white collar professions. It's just a lot easier,

125
00:05:33,680 --> 00:05:35,479
Speaker 8: Like the expectation for output is so high.

126
00:05:35,560 --> 00:05:37,799
Speaker 4: I'm on Twitter all day and a lot of people

127
00:05:37,839 --> 00:05:40,680
Speaker 4: are tweeting, and I'm like, it seems like a lot

128
00:05:40,680 --> 00:05:42,400
Speaker 4: of people are slacking off these days.

129
00:05:42,400 --> 00:05:43,279
Speaker 2: I'm like, how do you have time?

130
00:05:43,320 --> 00:05:45,960
Speaker 4: I have time because I'm like a journalist, and I said,

131
00:05:46,040 --> 00:05:48,200
Speaker 4: you know, I create words professionally.

132
00:05:48,279 --> 00:05:50,400
Speaker 6: Some would argue that you do not in fact have

133
00:05:50,480 --> 00:05:50,919
Speaker 6: time to.

134
00:05:50,960 --> 00:05:56,120
Speaker 4: Know that that's true, But David, so, Okay, there's been

135
00:05:56,120 --> 00:05:59,560
Speaker 4: a lot of progress, no question, there are new harnesses

136
00:05:59,760 --> 00:06:03,240
Speaker 4: for AI models that increase all of our capabilities and

137
00:06:03,279 --> 00:06:06,760
Speaker 4: so forth. That's different than we all know.

138
00:06:06,839 --> 00:06:07,320
Speaker 2: That's true.

139
00:06:07,640 --> 00:06:10,839
Speaker 4: That's different than like job wipe out in a significant way.

140
00:06:10,880 --> 00:06:13,920
Speaker 4: What is it about to you that you think, yes,

141
00:06:14,040 --> 00:06:17,320
Speaker 4: there is progress, but progress on the scale of this

142
00:06:17,400 --> 00:06:19,560
Speaker 4: is an imminent thing that we have to be talking

143
00:06:19,600 --> 00:06:22,320
Speaker 4: about that could really reshape white color labor.

144
00:06:22,920 --> 00:06:25,960
Speaker 7: The analogy that I think about a lot is COVID,

145
00:06:26,640 --> 00:06:30,400
Speaker 7: because the thing about AI progress is just that it's really,

146
00:06:30,760 --> 00:06:34,640
Speaker 7: in many ways exponential, where the amount of time that

147
00:06:34,680 --> 00:06:38,240
Speaker 7: an AI can operate autonomously without a human really has

148
00:06:38,279 --> 00:06:41,400
Speaker 7: been doubling every one hundred and twelve days or so

149
00:06:41,800 --> 00:06:43,960
Speaker 7: for the past six years. And you know, when I

150
00:06:44,000 --> 00:06:47,880
Speaker 7: think about COVID, this was a thing that nobody saw

151
00:06:47,960 --> 00:06:51,080
Speaker 7: coming and then it happened really fast. And then I

152
00:06:51,120 --> 00:06:54,200
Speaker 7: think the political system was very reactive, and I think

153
00:06:54,200 --> 00:06:57,279
Speaker 7: that a lot of quietly, a lot of Democrats wish

154
00:06:57,360 --> 00:06:59,560
Speaker 7: that they had handled things a little bit differently. But

155
00:06:59,600 --> 00:07:02,360
Speaker 7: what's cool is it unlike COVID, you know, we really

156
00:07:02,400 --> 00:07:05,640
Speaker 7: can see this coming. All of the warning signs are blinking.

157
00:07:06,120 --> 00:07:07,720
Speaker 7: And you know, the other point I want to make

158
00:07:07,720 --> 00:07:10,080
Speaker 7: here is, you know, I personally think that there's a

159
00:07:10,120 --> 00:07:13,880
Speaker 7: lot of potential for large scale job loss, particularly white collar.

160
00:07:13,960 --> 00:07:16,080
Speaker 7: But you know, there are a lot of truck drivers,

161
00:07:16,080 --> 00:07:18,040
Speaker 7: there are a lot of Uber drivers, you know, all

162
00:07:18,120 --> 00:07:21,240
Speaker 7: kinds of people could lose their jobs, and this could

163
00:07:21,240 --> 00:07:24,200
Speaker 7: all happen very quickly. I think the more important point, though,

164
00:07:24,240 --> 00:07:27,440
Speaker 7: is that the American people see this and are really

165
00:07:27,520 --> 00:07:30,880
Speaker 7: quite worried. You know, when you ask people how likely

166
00:07:30,880 --> 00:07:33,520
Speaker 7: do you think it is that in the next five

167
00:07:33,600 --> 00:07:36,400
Speaker 7: years there might be large scale job loss because of AI,

168
00:07:36,960 --> 00:07:39,600
Speaker 7: seventy percent of the population says that it's either very

169
00:07:39,680 --> 00:07:42,920
Speaker 7: likely or somewhat likely. And so that's really the main

170
00:07:43,080 --> 00:07:46,560
Speaker 7: point I'd make there. Is the reason politically why politicians

171
00:07:46,560 --> 00:07:49,720
Speaker 7: should act now rather than waiting until there's a problem

172
00:07:49,800 --> 00:07:53,080
Speaker 7: is you know, one, the American people are already worried

173
00:07:53,120 --> 00:07:57,360
Speaker 7: about this, and two, once it's already happening, it will

174
00:07:57,360 --> 00:08:00,120
Speaker 7: be too late for our political system to respond. So

175
00:08:00,120 --> 00:08:01,840
Speaker 7: I think it's important to try to get ahead.

176
00:08:01,560 --> 00:08:04,320
Speaker 6: Of definitely want to get more into the politics. But

177
00:08:04,520 --> 00:08:06,680
Speaker 6: Burn you brought up something that seems to be kind

178
00:08:06,720 --> 00:08:09,880
Speaker 6: of standard in these conversations, which is we've been here before,

179
00:08:10,080 --> 00:08:13,800
Speaker 6: right we all, well, not literally went through the Industrial Revolution,

180
00:08:13,920 --> 00:08:16,040
Speaker 6: but that's something that happened, and we went through the

181
00:08:16,040 --> 00:08:20,640
Speaker 6: Internet boom, and the economy and society adapted more or less.

182
00:08:20,680 --> 00:08:24,360
Speaker 6: But when you ask people, now, what the alternative professions

183
00:08:24,400 --> 00:08:27,720
Speaker 6: are for white collar workers. We haven't been able to

184
00:08:27,760 --> 00:08:31,120
Speaker 6: really get like a slate of possibilities. I know it's

185
00:08:31,120 --> 00:08:33,439
Speaker 6: hard to imagine the future, but you know, if you're

186
00:08:33,480 --> 00:08:36,000
Speaker 6: an insurance broker and you have been for twenty years,

187
00:08:36,040 --> 00:08:38,280
Speaker 6: what are you going to be doing in the new economy?

188
00:08:38,360 --> 00:08:40,120
Speaker 5: Like, what are the new jobs that are coming down

189
00:08:40,160 --> 00:08:40,559
Speaker 5: the line?

190
00:08:41,320 --> 00:08:43,120
Speaker 8: Yeah, that is actually a tough question, Like one of

191
00:08:43,120 --> 00:08:45,559
Speaker 8: them is just like temporarily. I think there are a

192
00:08:45,600 --> 00:08:48,760
Speaker 8: lot of jobs that are basically either human who's required

193
00:08:48,760 --> 00:08:50,559
Speaker 8: to be in the loop for regulatory reasons, Like doctors

194
00:08:50,559 --> 00:08:54,880
Speaker 8: are incredibly rapid adopters of AI tools, and the supply

195
00:08:54,960 --> 00:08:58,640
Speaker 8: of doctors is sort of artificially constrained, And so if

196
00:08:58,640 --> 00:09:01,000
Speaker 8: you decrease the percentage there time that they spend on

197
00:09:01,200 --> 00:09:03,960
Speaker 8: admin tasks and you decrease the rate of mistakes that

198
00:09:03,960 --> 00:09:07,480
Speaker 8: they make, you basically get the equivalent of manufacturing more doctors.

199
00:09:07,880 --> 00:09:11,120
Speaker 8: And in cases like healthcare, there is effectively unlimited demand,

200
00:09:11,200 --> 00:09:13,520
Speaker 8: Like there's yet to be an economy where people don't

201
00:09:13,520 --> 00:09:15,960
Speaker 8: spend more of their marginal dollar on help.

202
00:09:16,000 --> 00:09:17,360
Speaker 5: We're all going to be healthcare workers.

203
00:09:17,559 --> 00:09:19,800
Speaker 8: I think that's actually like that sector probably will grow,

204
00:09:20,240 --> 00:09:23,160
Speaker 8: and you know, that's it's kind of glumm to say, Okay,

205
00:09:23,200 --> 00:09:25,440
Speaker 8: some some white collar workers are going to kind of

206
00:09:25,800 --> 00:09:28,840
Speaker 8: move downscale in terms of status, where they will probably

207
00:09:28,840 --> 00:09:29,599
Speaker 8: have jobs.

208
00:09:29,320 --> 00:09:30,160
Speaker 2: That sound less cool.

209
00:09:30,160 --> 00:09:32,760
Speaker 8: But if overall output is high enough and if the

210
00:09:32,800 --> 00:09:35,600
Speaker 8: returns on capital are high enough that people like the

211
00:09:35,600 --> 00:09:39,200
Speaker 8: economy starts shifting more in criminal production into just building

212
00:09:39,280 --> 00:09:41,760
Speaker 8: data centers, that does mean that if you are the

213
00:09:41,800 --> 00:09:44,320
Speaker 8: complement to a data center, like you, you are a

214
00:09:44,360 --> 00:09:47,480
Speaker 8: necessary component of this entire supply chain. Your bargaining power

215
00:09:47,520 --> 00:09:48,960
Speaker 8: is a lot stronger because there's just a lot more

216
00:09:48,960 --> 00:09:52,240
Speaker 8: capital that you're adjacent to. If you are completely substitutable

217
00:09:52,320 --> 00:09:54,760
Speaker 8: by the data centers, then you're in a tough spot.

218
00:09:54,760 --> 00:09:58,400
Speaker 8: But we always find that models have these really spikey abilities,

219
00:09:58,440 --> 00:10:01,920
Speaker 8: like they're they're superhuman in some respects and in other

220
00:10:02,000 --> 00:10:03,960
Speaker 8: respects they are, you know, in terms of things like

221
00:10:04,040 --> 00:10:06,360
Speaker 8: math ability, like they are there beyond the point where

222
00:10:06,360 --> 00:10:09,120
Speaker 8: I could reliably distinguish between two models and say, well,

223
00:10:09,120 --> 00:10:10,760
Speaker 8: this one's really good at math and this is okay

224
00:10:10,800 --> 00:10:14,079
Speaker 8: at math. But in other domains they do just kind

225
00:10:14,120 --> 00:10:16,200
Speaker 8: of fall flat because they don't have this comprehensive world

226
00:10:16,200 --> 00:10:17,880
Speaker 8: model and the reason for that is that they're trained

227
00:10:17,880 --> 00:10:22,320
Speaker 8: on text, and text actually skews towards areas that are

228
00:10:22,679 --> 00:10:25,160
Speaker 8: uncertain and open to debate. So I use the term

229
00:10:25,200 --> 00:10:27,480
Speaker 8: the maybe sphere, which is like, if you imagine there's

230
00:10:27,520 --> 00:10:30,040
Speaker 8: like this bedrock of facts that are so obvious that

231
00:10:30,080 --> 00:10:32,120
Speaker 8: nobody ever bothers to write them down, and then there's

232
00:10:32,160 --> 00:10:34,480
Speaker 8: this infinite space of questions that are so weird that

233
00:10:34,520 --> 00:10:36,880
Speaker 8: it's almost certain they don't have an answer. There's like

234
00:10:36,920 --> 00:10:40,280
Speaker 8: this little narrow layer, like an atmosphere where it's worth

235
00:10:40,360 --> 00:10:42,280
Speaker 8: asking a question and you might get an answer. So

236
00:10:42,280 --> 00:10:44,080
Speaker 8: they have a really good world model for the parts

237
00:10:44,080 --> 00:10:45,839
Speaker 8: of the world that we're either not sure about or

238
00:10:45,920 --> 00:10:48,120
Speaker 8: that we've codified in textbooks, and then a really bad

239
00:10:48,160 --> 00:10:50,000
Speaker 8: world model for a lot of the obvious stuff. And

240
00:10:50,040 --> 00:10:51,840
Speaker 8: so it would be kind of weird to be like

241
00:10:52,040 --> 00:10:54,760
Speaker 8: to say, like, my job right now is to say

242
00:10:54,840 --> 00:10:58,560
Speaker 8: extremely obvious things to this superhuman intelligence. I don't even

243
00:10:58,600 --> 00:11:02,600
Speaker 8: know what job title historically that might correspond to, like

244
00:11:02,679 --> 00:11:05,760
Speaker 8: sort of a servant for you know, a brilliant person

245
00:11:05,840 --> 00:11:09,280
Speaker 8: who's also like, you know, absent minded professor. But yeah,

246
00:11:09,360 --> 00:11:12,040
Speaker 8: I think we'll have a lot more manservant for absent

247
00:11:12,080 --> 00:11:14,280
Speaker 8: minded professor professions, and that's pure the.

248
00:11:14,280 --> 00:11:34,040
Speaker 4: Future manservant for absent minded professors. It sounds fine, David.

249
00:11:34,320 --> 00:11:37,880
Speaker 4: You have a firm and you hire people. Has the

250
00:11:38,000 --> 00:11:41,679
Speaker 4: nature of your hiring changed? Are you hiring for different

251
00:11:41,760 --> 00:11:44,920
Speaker 4: types of roles than you would have a few years

252
00:11:44,960 --> 00:11:48,679
Speaker 4: ago or something like that given the change in technology?

253
00:11:48,960 --> 00:11:53,640
Speaker 7: Absolutely, you know, I think just to give a simple example, like,

254
00:11:53,679 --> 00:11:55,680
Speaker 7: we used to have lots of copy editors to write

255
00:11:55,720 --> 00:11:58,400
Speaker 7: pulling questions or to write messages. And you know the

256
00:11:58,440 --> 00:12:02,600
Speaker 7: reality is that now AIS are generally better than people

257
00:12:02,640 --> 00:12:04,319
Speaker 7: at doing that, not uniformly, but we have.

258
00:12:04,280 --> 00:12:06,760
Speaker 4: A lot of good at writing poll Like can AI

259
00:12:07,480 --> 00:12:10,320
Speaker 4: you want to find interesting polling questions, right, not the

260
00:12:10,440 --> 00:12:15,160
Speaker 4: obvious stuff? Is AI good at finding non obvious polling

261
00:12:15,240 --> 00:12:17,560
Speaker 4: questions that are non correlated to things so that you

262
00:12:17,559 --> 00:12:18,960
Speaker 4: can actually get signal from them?

263
00:12:19,120 --> 00:12:21,400
Speaker 7: Well, you know, I don't want to talk too specifically

264
00:12:21,440 --> 00:12:24,200
Speaker 7: about that, but I will say that there's just tons

265
00:12:24,240 --> 00:12:27,200
Speaker 7: of copy edending tasks that I think, frankly.

266
00:12:27,000 --> 00:12:28,640
Speaker 2: What do you want to talk specifically? Yeah, I know

267
00:12:29,120 --> 00:12:31,239
Speaker 2: that means this is the question I should be asking.

268
00:12:31,080 --> 00:12:32,560
Speaker 7: No, no, no, But you know, look, there's a lot

269
00:12:32,559 --> 00:12:34,480
Speaker 7: of stuff like translation. There's a lot of stuff like

270
00:12:34,520 --> 00:12:37,920
Speaker 7: copy edding. I think the big shift for us is

271
00:12:38,000 --> 00:12:41,720
Speaker 7: that we're focusing a lot more on person centric jobs,

272
00:12:41,760 --> 00:12:44,400
Speaker 7: and you know, now you can do a lot more

273
00:12:44,480 --> 00:12:47,400
Speaker 7: engineering than before. So I think there's definitely been a

274
00:12:47,400 --> 00:12:49,880
Speaker 7: big job shift, you know, in terms of how we've

275
00:12:49,880 --> 00:12:50,319
Speaker 7: been hiring.

276
00:12:50,440 --> 00:12:53,120
Speaker 6: Absolutely, maybe we can do a little bit of content

277
00:12:53,200 --> 00:12:56,240
Speaker 6: creation naval gazing here. So I imagine this is probably

278
00:12:56,280 --> 00:12:57,840
Speaker 6: of interest to a lot of people at south By

279
00:12:57,880 --> 00:13:01,440
Speaker 6: Southwest as well. But Burne, you right, a newsletter on

280
00:13:01,480 --> 00:13:03,600
Speaker 6: a daily basis, Joe and I do as well.

281
00:13:03,840 --> 00:13:06,559
Speaker 5: How are you using AI just in your sort of

282
00:13:06,640 --> 00:13:07,120
Speaker 5: day to day?

283
00:13:07,480 --> 00:13:10,160
Speaker 8: So I use it a ton for research, and one

284
00:13:10,160 --> 00:13:13,320
Speaker 8: of the specific use cases is asking questions like does

285
00:13:13,320 --> 00:13:16,199
Speaker 8: this Joe Land or Hey, I'm making a statement, you know,

286
00:13:16,360 --> 00:13:19,000
Speaker 8: kind of narrow technical statement about a domain that I'm

287
00:13:19,040 --> 00:13:21,520
Speaker 8: reasonably familiar with, but I'm not an expert on You

288
00:13:21,559 --> 00:13:22,280
Speaker 8: are an expert on this.

289
00:13:22,360 --> 00:13:23,400
Speaker 2: Tell me what I'm getting wrong.

290
00:13:23,720 --> 00:13:25,200
Speaker 8: And one of the reasons for that is just a

291
00:13:25,200 --> 00:13:28,080
Speaker 8: lot of my readers are software engineers and are eager

292
00:13:28,120 --> 00:13:30,599
Speaker 8: to tell you when you're wrong, extremely eager, Like I

293
00:13:30,960 --> 00:13:34,880
Speaker 8: sort of measure people's ability as software engineers, in particular

294
00:13:34,880 --> 00:13:36,400
Speaker 8: based on how quickly they get an email from them

295
00:13:36,400 --> 00:13:38,440
Speaker 8: that there's a typo. Because the really good ones, like

296
00:13:38,840 --> 00:13:40,720
Speaker 8: one of the skills that they have is just looking

297
00:13:40,760 --> 00:13:42,680
Speaker 8: at a lot of text and immediately seeing what's wrong.

298
00:13:42,880 --> 00:13:44,840
Speaker 8: Now that skill is kind of obsolete right now ellms

299
00:13:44,840 --> 00:13:47,439
Speaker 8: are better at it, but still the mindset of I'm

300
00:13:47,440 --> 00:13:49,320
Speaker 8: going to look at this and kind of absorb something

301
00:13:49,320 --> 00:13:51,800
Speaker 8: coherent and then to look for any little issue.

302
00:13:51,440 --> 00:13:52,960
Speaker 2: With it, that's still quite valuable.

303
00:13:52,960 --> 00:13:55,080
Speaker 8: In fact, since there's more code being produced than ever

304
00:13:55,120 --> 00:13:57,559
Speaker 8: before by a huge margin, it's really really valuable to

305
00:13:57,559 --> 00:13:59,440
Speaker 8: be able to read it flum you understand what it's

306
00:13:59,440 --> 00:14:00,200
Speaker 8: supposed to be doing.

307
00:14:00,800 --> 00:14:04,080
Speaker 2: So I do. I use AI a lot for research.

308
00:14:04,200 --> 00:14:07,080
Speaker 8: I have only in the last few months have I

309
00:14:07,120 --> 00:14:11,079
Speaker 8: actually gotten ideas for things to write, specifically from chat

310
00:14:11,120 --> 00:14:14,079
Speaker 8: GPT or only in the last few months that has

311
00:14:14,120 --> 00:14:16,760
Speaker 8: it made some original points I would not have thought

312
00:14:16,760 --> 00:14:19,000
Speaker 8: of that. We're just like clever insights and not just

313
00:14:19,360 --> 00:14:21,280
Speaker 8: it's reciting a fact that I did not happen to know.

314
00:14:22,480 --> 00:14:25,080
Speaker 4: David talk to us a little bit more so this

315
00:14:25,200 --> 00:14:27,600
Speaker 4: question of progress, right we all know it's getting better.

316
00:14:27,920 --> 00:14:29,640
Speaker 2: That's obvious, But the more.

317
00:14:29,480 --> 00:14:32,520
Speaker 4: Interesting question is is it getting better at a pace

318
00:14:32,600 --> 00:14:37,880
Speaker 4: faster than what people had previously anticipated and talk to

319
00:14:37,920 --> 00:14:40,520
Speaker 4: us about, like, Okay, where we are now with the

320
00:14:40,560 --> 00:14:43,600
Speaker 4: technology and where the people that you were talking to,

321
00:14:43,640 --> 00:14:45,880
Speaker 4: and maybe I'll throw this to both of you where

322
00:14:45,880 --> 00:14:47,720
Speaker 4: would they have said we would be in?

323
00:14:48,440 --> 00:14:50,240
Speaker 2: Is it March March twenty twenty five.

324
00:14:51,160 --> 00:14:54,760
Speaker 7: Yeah. I think that the big surprise of the last

325
00:14:54,840 --> 00:14:56,960
Speaker 7: year has been, you know, the rise of vibe coding,

326
00:14:57,000 --> 00:14:59,480
Speaker 7: the rise of tools like clod code. If you just

327
00:14:59,600 --> 00:15:02,960
Speaker 7: went back a year ago, I think that nobody really

328
00:15:03,080 --> 00:15:06,360
Speaker 7: expected the extent to which these things would be able

329
00:15:06,400 --> 00:15:09,880
Speaker 7: to do large scale, complex autonomous coding problems. You know,

330
00:15:09,880 --> 00:15:12,080
Speaker 7: I think in a lot of ways these models have

331
00:15:12,160 --> 00:15:16,240
Speaker 7: become useful faster than they've become smarter, and I don't

332
00:15:16,240 --> 00:15:18,760
Speaker 7: think that that's something that people expected probably. You know.

333
00:15:18,800 --> 00:15:21,280
Speaker 7: What's interesting is that every year there are a bunch

334
00:15:21,320 --> 00:15:24,480
Speaker 7: of AI experts who then go and make predictions, and

335
00:15:24,600 --> 00:15:27,960
Speaker 7: one of the biggest surprises has been the revenue growth

336
00:15:28,040 --> 00:15:30,160
Speaker 7: of these companies, which I think is in many ways

337
00:15:30,200 --> 00:15:34,120
Speaker 7: the most important benchmark. I think anthropics revenue last year

338
00:15:34,240 --> 00:15:37,600
Speaker 7: was something like two x experts who already I think

339
00:15:37,640 --> 00:15:41,040
Speaker 7: were quite aipilled predicted, and so I think that's probably

340
00:15:41,320 --> 00:15:44,440
Speaker 7: the biggest surprise. And you know, what I would say,

341
00:15:44,920 --> 00:15:48,960
Speaker 7: just on the broader transition questions, is just that tools

342
00:15:49,000 --> 00:15:52,040
Speaker 7: like LMS have been now adopted faster. You know, there

343
00:15:52,040 --> 00:15:53,640
Speaker 7: are a bunch of graphs that are like how long

344
00:15:53,680 --> 00:15:56,560
Speaker 7: did it take to implement radio or electricity or the internet?

345
00:15:56,880 --> 00:16:00,160
Speaker 7: And this is much faster than any of those things,

346
00:16:00,240 --> 00:16:03,400
Speaker 7: And so you know, I really worry, like when we

347
00:16:03,440 --> 00:16:07,840
Speaker 7: talk about past transitions of like telephone switch operators or

348
00:16:07,880 --> 00:16:11,640
Speaker 7: factory workers. You know, all of these things happened over

349
00:16:11,720 --> 00:16:15,200
Speaker 7: an extended period of time and only impacted certain sectors

350
00:16:15,200 --> 00:16:18,760
Speaker 7: of the economy, and this technology really threatens to upend

351
00:16:18,840 --> 00:16:22,520
Speaker 7: every single job at the same time, you know, at

352
00:16:22,560 --> 00:16:25,880
Speaker 7: a point when you know people's overall views of the

353
00:16:26,120 --> 00:16:28,960
Speaker 7: economy are not good, and so I really worry just

354
00:16:29,040 --> 00:16:32,440
Speaker 7: politically the extent to which our political system can handle this.

355
00:16:33,320 --> 00:16:35,560
Speaker 8: It is true that just to like, when you look

356
00:16:35,560 --> 00:16:38,240
Speaker 8: at these measures of how broaday technology is adopted, I

357
00:16:38,240 --> 00:16:40,240
Speaker 8: think the thing that's really hard to measure is at

358
00:16:40,240 --> 00:16:42,160
Speaker 8: what point does it become just part of your baseline

359
00:16:42,160 --> 00:16:45,920
Speaker 8: expectation and at what point are the impacts so obvious

360
00:16:45,960 --> 00:16:48,840
Speaker 8: that they're unspoken. So if you look at electrification, like

361
00:16:48,920 --> 00:16:51,560
Speaker 8: I think anyone who talks about AI is just required

362
00:16:51,600 --> 00:16:53,240
Speaker 8: to say that it took like half a century to

363
00:16:53,240 --> 00:16:56,200
Speaker 8: go from the first electrified factory to most US factories

364
00:16:56,200 --> 00:16:59,000
Speaker 8: being electrified. But it's and the reason for them, there's

365
00:16:59,000 --> 00:17:01,000
Speaker 8: a lot of fun economic history on this is that

366
00:17:01,200 --> 00:17:02,640
Speaker 8: you have to run your factory in a different way.

367
00:17:02,640 --> 00:17:04,400
Speaker 8: You actually have to build a different kind of building,

368
00:17:04,840 --> 00:17:07,600
Speaker 8: and you also have to finance it a different kind

369
00:17:07,640 --> 00:17:08,919
Speaker 8: of way, or you can finance it a different head

370
00:17:08,920 --> 00:17:11,000
Speaker 8: of way. So if you have a factory that has

371
00:17:11,000 --> 00:17:13,600
Speaker 8: some kind of mechanical power source, you tend to expand

372
00:17:13,640 --> 00:17:17,520
Speaker 8: your business in these discrete, factory sized increments. And what

373
00:17:17,560 --> 00:17:19,399
Speaker 8: that means is that to your investors, you pay out

374
00:17:19,440 --> 00:17:21,520
Speaker 8: most of your earnings as dividends because there's nothing to

375
00:17:21,560 --> 00:17:23,440
Speaker 8: do with retained earnings. Like if you're going to grow,

376
00:17:23,480 --> 00:17:24,760
Speaker 8: you want to issue a bunch of stock, you want

377
00:17:24,760 --> 00:17:26,439
Speaker 8: to issue a bunch of bonds and then grow in

378
00:17:26,520 --> 00:17:30,360
Speaker 8: like one shot. But once you have electrified factories where

379
00:17:30,400 --> 00:17:32,920
Speaker 8: they can expand, they can just add another assembly line

380
00:17:33,000 --> 00:17:34,800
Speaker 8: or they can upgrade this machine to a newer machine,

381
00:17:34,840 --> 00:17:37,840
Speaker 8: et cetera. They can actually expand more organically and incrementally,

382
00:17:37,960 --> 00:17:39,760
Speaker 8: and so that's when you start to see dividend payout

383
00:17:39,800 --> 00:17:41,640
Speaker 8: ratios come down, and that's when you start to see

384
00:17:41,840 --> 00:17:44,440
Speaker 8: the growth company as a concept emerge. Like if you

385
00:17:44,720 --> 00:17:47,720
Speaker 8: read investor accounts from the nineteen twenties and they're talking

386
00:17:47,760 --> 00:17:49,200
Speaker 8: about the stock market, one of the weird things is

387
00:17:49,240 --> 00:17:52,200
Speaker 8: they're very fixated on whether the stock is above or

388
00:17:52,200 --> 00:17:54,679
Speaker 8: below one hundred dollars a share, because that was the

389
00:17:54,720 --> 00:17:56,879
Speaker 8: par value. That was just like the standard value for

390
00:17:56,920 --> 00:17:58,400
Speaker 8: the stock, and it was supposed to be the book value,

391
00:17:58,400 --> 00:18:00,679
Speaker 8: et cetera. Then people kind of viewed equity is just

392
00:18:00,760 --> 00:18:03,399
Speaker 8: the most junior claimant, like the way you view the

393
00:18:03,400 --> 00:18:06,080
Speaker 8: equity slice of a CDO or something like that. And

394
00:18:06,560 --> 00:18:09,560
Speaker 8: now we view equities completely differently, and so we deploy

395
00:18:09,640 --> 00:18:12,719
Speaker 8: money completely differently. Like it would be incomprehensible to an

396
00:18:12,800 --> 00:18:16,040
Speaker 8: investor circa nineteen twenty six to say, I'm going to

397
00:18:16,520 --> 00:18:20,000
Speaker 8: invest in this company that has no operations. It's just people,

398
00:18:20,440 --> 00:18:23,520
Speaker 8: and I'm going to put the actual money into this business,

399
00:18:23,520 --> 00:18:25,000
Speaker 8: but most of it will be owned by these people.

400
00:18:25,320 --> 00:18:27,119
Speaker 8: And then the business, you know, my stock could be

401
00:18:27,160 --> 00:18:29,560
Speaker 8: worth fifty times as much in a few years if

402
00:18:29,560 --> 00:18:32,080
Speaker 8: things go perfectly like that wouldn't make any sense to them.

403
00:18:32,119 --> 00:18:33,960
Speaker 8: The idea of a stock going from par value to

404
00:18:34,000 --> 00:18:35,440
Speaker 8: fifty times par value.

405
00:18:35,200 --> 00:18:35,960
Speaker 2: It's just insane.

406
00:18:36,480 --> 00:18:41,760
Speaker 8: So but that change actually was part of I think

407
00:18:41,760 --> 00:18:45,120
Speaker 8: it was like causal and in both directions. With America

408
00:18:45,200 --> 00:18:47,879
Speaker 8: just having a really flexible financial system, we also have

409
00:18:47,920 --> 00:18:50,240
Speaker 8: a really flexible labor market relative to every other country,

410
00:18:50,240 --> 00:18:52,560
Speaker 8: which means we will be patient zero for like all

411
00:18:52,600 --> 00:18:54,320
Speaker 8: the job law stuff like it'll hit us before it

412
00:18:54,359 --> 00:18:56,680
Speaker 8: hits anyone else, and it'll hit us harder than anyone else.

413
00:18:57,080 --> 00:18:59,240
Speaker 8: On the other hand, one of the nque things about

414
00:18:59,240 --> 00:19:01,960
Speaker 8: this particular neral purpose technology rolled out is that it

415
00:19:02,040 --> 00:19:05,440
Speaker 8: is happening much much faster than before. But the thing

416
00:19:05,440 --> 00:19:08,679
Speaker 8: that slightly offsets that is that the specific technology actually

417
00:19:08,680 --> 00:19:11,480
Speaker 8: gives you access to information and cognition such that you

418
00:19:11,520 --> 00:19:14,199
Speaker 8: can ask chat shept like here's my job, like I've

419
00:19:14,200 --> 00:19:15,600
Speaker 8: been selling insurance for twenty years.

420
00:19:16,119 --> 00:19:18,600
Speaker 2: How should I reskill? Like what should I do? And

421
00:19:18,640 --> 00:19:20,000
Speaker 2: it will ask follow up questions.

422
00:19:20,040 --> 00:19:22,800
Speaker 8: It'll ask, Okay, well you know what do what other

423
00:19:22,840 --> 00:19:25,040
Speaker 8: problems do the companies and people you work with have

424
00:19:25,560 --> 00:19:27,600
Speaker 8: or you know, let's break down the skills that you have,

425
00:19:27,680 --> 00:19:30,960
Speaker 8: what makes you a particularly good or bad insurance person,

426
00:19:31,160 --> 00:19:33,280
Speaker 8: and then what are the other jobs that might be

427
00:19:33,359 --> 00:19:35,439
Speaker 8: less ai exposed that you could do or you know,

428
00:19:35,440 --> 00:19:36,960
Speaker 8: as the models get smarter, you can just sell it. Hey,

429
00:19:37,000 --> 00:19:38,639
Speaker 8: I want you to invent a new job for me,

430
00:19:38,760 --> 00:19:38,960
Speaker 8: like a.

431
00:19:39,040 --> 00:19:41,760
Speaker 2: Bespoke like I got job I was born for. No,

432
00:19:42,040 --> 00:19:43,760
Speaker 2: you already have that job. That's right.

433
00:19:44,320 --> 00:19:45,480
Speaker 5: Podcasters are safe.

434
00:19:45,560 --> 00:19:47,639
Speaker 6: I guess we can ask questions, but I mean it

435
00:19:47,680 --> 00:19:50,200
Speaker 6: does seem kind of dystopian. When the upside is well,

436
00:19:50,240 --> 00:19:53,440
Speaker 6: you can ask the chat GPT what your alternative job

437
00:19:53,520 --> 00:19:53,880
Speaker 6: is and.

438
00:19:53,880 --> 00:19:56,520
Speaker 8: What this always feel dystopian at the time, Like if

439
00:19:56,560 --> 00:19:59,399
Speaker 8: you told someone two hundred years ago, hey, it's going

440
00:19:59,440 --> 00:20:01,919
Speaker 8: to be completely viable for you to inherit your father's

441
00:20:01,960 --> 00:20:04,400
Speaker 8: farm and work on that farm and then give that

442
00:20:04,440 --> 00:20:07,000
Speaker 8: farm to your son. And if you said, you know,

443
00:20:07,040 --> 00:20:09,320
Speaker 8: not only is that economically non viable, but you're also

444
00:20:09,320 --> 00:20:10,960
Speaker 8: not going to be so fixated on is your son

445
00:20:11,080 --> 00:20:13,119
Speaker 8: or your daughter? And also you're probably gonna move to

446
00:20:13,160 --> 00:20:15,679
Speaker 8: a city you'll be surrounded by complete strangers. You'll have

447
00:20:15,720 --> 00:20:20,000
Speaker 8: a job like in this loud noisy, very uncomfortable building,

448
00:20:20,240 --> 00:20:23,280
Speaker 8: like messing around with you know whatever. These physical things

449
00:20:23,320 --> 00:20:25,360
Speaker 8: you're manufacturing are like a lot of people would say

450
00:20:25,359 --> 00:20:26,919
Speaker 8: that's that's kind of nightmarish, and in fact a lot

451
00:20:26,920 --> 00:20:29,520
Speaker 8: of contemporary observers did talk about that being kind of nightmarish.

452
00:20:29,600 --> 00:20:32,640
Speaker 8: But it did end up working out reasonably well over time.

453
00:20:32,720 --> 00:20:35,440
Speaker 8: It was just new and weird and completely incomprehensible from

454
00:20:35,560 --> 00:20:36,680
Speaker 8: the original standpoint.

455
00:20:36,880 --> 00:20:38,920
Speaker 6: Let me ask the question in a slightly different way,

456
00:20:38,960 --> 00:20:42,639
Speaker 6: which is like who captures the productivity gains here? And

457
00:20:42,680 --> 00:20:45,600
Speaker 6: how are they distributed? Because I can imagine a situation

458
00:20:45,720 --> 00:20:48,119
Speaker 6: where we all have jobs. There are certain tasks in

459
00:20:48,160 --> 00:20:50,719
Speaker 6: our job that we don't necessarily like doing, and if

460
00:20:50,720 --> 00:20:53,320
Speaker 6: we can use AI to do them more efficiently, then

461
00:20:53,320 --> 00:20:56,600
Speaker 6: maybe that's great for us. But history is full of

462
00:20:56,600 --> 00:20:59,239
Speaker 6: examples of new technology that is pitched as you know,

463
00:20:59,720 --> 00:21:02,600
Speaker 6: a productivity enhancer like email is going to let you

464
00:21:02,680 --> 00:21:05,200
Speaker 6: do things faster, and then it turns out that actually

465
00:21:05,359 --> 00:21:07,880
Speaker 6: email means we have to apply to emails twenty four

466
00:21:07,920 --> 00:21:09,800
Speaker 6: hours a day and we just get more volume and

467
00:21:09,840 --> 00:21:11,800
Speaker 6: it actually makes us more miserable.

468
00:21:12,080 --> 00:21:14,960
Speaker 5: Who captures the efficiency gains here?

469
00:21:15,119 --> 00:21:16,040
Speaker 2: People who don't.

470
00:21:15,800 --> 00:21:20,080
Speaker 8: Feel miserable when they can produce more economic output but

471
00:21:20,280 --> 00:21:22,600
Speaker 8: have to pay different kinds of attention or put in

472
00:21:22,600 --> 00:21:25,600
Speaker 8: more effort. It is true that these things have negative

473
00:21:25,600 --> 00:21:28,200
Speaker 8: side effects, but like when the cost of communication goes down,

474
00:21:28,480 --> 00:21:30,480
Speaker 8: the amount of coordination you can do goes way up,

475
00:21:30,600 --> 00:21:33,359
Speaker 8: and you can coordinate in different kinds of in different ways.

476
00:21:33,720 --> 00:21:36,000
Speaker 8: And I think this also illustrates that it's very hard

477
00:21:36,000 --> 00:21:38,639
Speaker 8: to predict the demand for service sector work because a

478
00:21:38,680 --> 00:21:40,240
Speaker 8: lot of it is so meta, like a lot of

479
00:21:40,359 --> 00:21:42,480
Speaker 8: is interacting with other parts of the service sector. So

480
00:21:42,560 --> 00:21:44,600
Speaker 8: like Excel is kind of the tried example. But there's

481
00:21:44,600 --> 00:21:47,680
Speaker 8: also word processing where you could think, okay, word processing

482
00:21:47,800 --> 00:21:50,879
Speaker 8: makes it easier for lawyers to just quickly draft contracts

483
00:21:50,920 --> 00:21:53,240
Speaker 8: and so we'll need fewer lawyers, But actually it meant

484
00:21:53,280 --> 00:21:55,119
Speaker 8: they could turn a two page contract into a fifty

485
00:21:55,119 --> 00:21:59,280
Speaker 8: page contract, and then you need more lawyers to handle that.

486
00:22:00,000 --> 00:22:02,600
Speaker 6: So did I tell you I met someone in Connecticut

487
00:22:02,680 --> 00:22:05,200
Speaker 6: who was training to be a copy editor. He started

488
00:22:05,240 --> 00:22:07,840
Speaker 6: two years ago, and I just thought, my god, what

489
00:22:08,040 --> 00:22:08,679
Speaker 6: bad timing.

490
00:22:09,160 --> 00:22:11,160
Speaker 4: I've heard a few other stories, like I heard someone

491
00:22:11,200 --> 00:22:14,119
Speaker 4: recently they like went to like a coding boot camp

492
00:22:14,200 --> 00:22:17,879
Speaker 4: like six months ago and I'm like, that's an awkward time, right, Yeah, David,

493
00:22:17,960 --> 00:22:20,840
Speaker 4: in your polling, who legs AI the most and who

494
00:22:20,880 --> 00:22:21,960
Speaker 4: legs AI the least?

495
00:22:22,400 --> 00:22:24,560
Speaker 7: Yeah, there's a pretty clear coalition.

496
00:22:24,640 --> 00:22:24,840
Speaker 2: You know.

497
00:22:24,880 --> 00:22:27,600
Speaker 7: The actual levels depend a lot on how you ask it,

498
00:22:27,680 --> 00:22:31,080
Speaker 7: But the main demographic split is that young people like

499
00:22:31,160 --> 00:22:34,280
Speaker 7: AI a lot more than older people, than men more

500
00:22:34,320 --> 00:22:37,840
Speaker 7: than women, which is probably unsurprising. And then I think interestingly,

501
00:22:38,119 --> 00:22:42,719
Speaker 7: generally educated people have much more positive views they ironically,

502
00:22:42,760 --> 00:22:45,120
Speaker 7: despite all the talk of the white collar job loss,

503
00:22:44,920 --> 00:22:47,560
Speaker 7: it's the people with degrees who I think are the

504
00:22:47,560 --> 00:22:49,920
Speaker 7: most optimistic. And then working class people are a lot

505
00:22:50,320 --> 00:22:54,280
Speaker 7: less optimistic. And then finally, after all of that, generally speaking,

506
00:22:54,440 --> 00:22:57,080
Speaker 7: not white people are more pessimistic about AI, and black

507
00:22:57,119 --> 00:23:00,359
Speaker 7: and Latino voters are generally more optimistic. You know, the

508
00:23:00,440 --> 00:23:03,560
Speaker 7: Mississippi Delta, for example, actually has the highest rate of

509
00:23:03,600 --> 00:23:05,680
Speaker 7: folks who are excited about AI.

510
00:23:06,720 --> 00:23:07,360
Speaker 5: That's interesting.

511
00:23:07,720 --> 00:23:11,240
Speaker 6: Explain the difference in attitudes between the educated and more

512
00:23:11,359 --> 00:23:14,159
Speaker 6: like working class. Is that just because the working classes

513
00:23:14,200 --> 00:23:17,600
Speaker 6: I guess maybe historically more used to getting screwed over

514
00:23:17,720 --> 00:23:19,080
Speaker 6: for lack of a better word.

515
00:23:19,160 --> 00:23:21,879
Speaker 7: Well, I think that is exactly it. That you know,

516
00:23:21,920 --> 00:23:25,679
Speaker 7: the way what I'll just he painted the story of well,

517
00:23:26,320 --> 00:23:28,040
Speaker 7: you know, the jobs are going to change, but there's

518
00:23:28,080 --> 00:23:29,520
Speaker 7: going to be tons of growth, There's going to be

519
00:23:29,680 --> 00:23:32,439
Speaker 7: new jobs. And I think it's just really worth saying

520
00:23:32,520 --> 00:23:36,280
Speaker 7: that voters are extremely skeptical of this claim. Like if

521
00:23:36,280 --> 00:23:38,439
Speaker 7: you go and you say, oh, how much do you

522
00:23:38,480 --> 00:23:41,000
Speaker 7: trust the statement that AI is going to create lots

523
00:23:41,000 --> 00:23:43,880
Speaker 7: of new jobs? I think it's something like minus forty.

524
00:23:44,240 --> 00:23:47,800
Speaker 7: The reality is that, you know, voters are extremely negative

525
00:23:48,000 --> 00:23:51,359
Speaker 7: about the economy right now. It's really impossible to overstate

526
00:23:51,440 --> 00:23:54,280
Speaker 7: where something like two thirds of the public thinks the

527
00:23:54,280 --> 00:23:57,199
Speaker 7: economy is rigged, only thirty five percent of the public

528
00:23:57,440 --> 00:24:02,320
Speaker 7: think feels that they're financially secure. And in that context

529
00:24:02,359 --> 00:24:06,480
Speaker 7: where people are genuinely quite angry, they are extremely skeptical

530
00:24:06,840 --> 00:24:09,480
Speaker 7: of the claim that things are just going to be okay.

531
00:24:10,000 --> 00:24:13,840
Speaker 7: And obviously, you know, working class people remember the decline

532
00:24:13,840 --> 00:24:17,960
Speaker 7: and manufacturing there like basically every economic big economic shift

533
00:24:18,000 --> 00:24:21,240
Speaker 7: has had winners and losers, and the winners generally have

534
00:24:21,359 --> 00:24:24,159
Speaker 7: been either the top one or the top ten percent.

535
00:24:24,280 --> 00:24:27,800
Speaker 7: The economists argue about that, but other people have lost.

536
00:24:27,840 --> 00:24:30,119
Speaker 7: And so you know, the main point I want to

537
00:24:30,119 --> 00:24:33,159
Speaker 7: make is just I think the picture that BRNN is

538
00:24:33,680 --> 00:24:37,119
Speaker 7: painting isn't going to be allowed to happen because the

539
00:24:37,160 --> 00:24:38,639
Speaker 7: public has a say.

540
00:24:39,000 --> 00:24:40,280
Speaker 2: This is a really important point.

541
00:24:40,520 --> 00:24:43,520
Speaker 4: The public will sort of it's almost impossible to talk

542
00:24:43,560 --> 00:24:47,439
Speaker 4: about the future without knowing what the politics are going

543
00:24:47,480 --> 00:24:49,760
Speaker 4: to be. Let me ask you, So you work with Democrats,

544
00:24:49,800 --> 00:24:53,119
Speaker 4: And when I think about the Democrats and AI, I

545
00:24:53,119 --> 00:24:56,160
Speaker 4: think there's a few different things. There's a pretty big

546
00:24:56,160 --> 00:24:59,520
Speaker 4: contingent on the sort of capital l left that thinks

547
00:24:59,520 --> 00:25:02,639
Speaker 4: it's all that thinks it's like this is theronose again,

548
00:25:03,119 --> 00:25:07,960
Speaker 4: this is NFTs, this is completely economic sustainable. Then there's

549
00:25:08,080 --> 00:25:10,920
Speaker 4: sort of like the anti data center people. They don't

550
00:25:10,920 --> 00:25:13,640
Speaker 4: want them in the backyard, they don't want them around period.

551
00:25:14,080 --> 00:25:17,080
Speaker 4: Then there's sort of like what's been building in the

552
00:25:17,080 --> 00:25:19,199
Speaker 4: Democratic Party for a while, just the sort of like

553
00:25:19,440 --> 00:25:22,680
Speaker 4: anti big tech, and it's sort of the anti oligarchs

554
00:25:22,800 --> 00:25:25,600
Speaker 4: stuff like that, Like is there anyone that you talk

555
00:25:25,720 --> 00:25:30,080
Speaker 4: to who's most I'm trying to think of the execut

556
00:25:30,160 --> 00:25:35,439
Speaker 4: rum taking it very seriously as an important technology that

557
00:25:35,800 --> 00:25:38,239
Speaker 4: is going to evolve, Like, is there anyone you talk

558
00:25:38,320 --> 00:25:41,600
Speaker 4: to is like, no, this actually works, this is real,

559
00:25:41,800 --> 00:25:44,359
Speaker 4: This could be a productivity gains, et cetera. Or is

560
00:25:44,400 --> 00:25:47,640
Speaker 4: it almost just various flavors of political negativity.

561
00:25:48,680 --> 00:25:51,199
Speaker 7: Well, I think the backdrop is that this is a

562
00:25:51,280 --> 00:25:54,440
Speaker 7: very new political issue. You know, even since last year,

563
00:25:54,560 --> 00:25:57,560
Speaker 7: the share of voters who care about AI has increased

564
00:25:57,600 --> 00:25:59,800
Speaker 7: more than any of the other thirty nine issues that

565
00:25:59,800 --> 00:26:04,080
Speaker 7: were packing and so politicians obviously are catching up. Usually

566
00:26:04,240 --> 00:26:06,840
Speaker 7: politicians are pretty reactive. But you know what I will

567
00:26:06,840 --> 00:26:09,440
Speaker 7: say is I think that Democrats aren't a much better

568
00:26:09,520 --> 00:26:12,520
Speaker 7: position to capitalize on this than Republicans are, just for

569
00:26:12,560 --> 00:26:15,919
Speaker 7: the basic reason that Republicans have really painted themselves in

570
00:26:16,000 --> 00:26:18,399
Speaker 7: a corner on this. You know, Donald Trump is on

571
00:26:18,480 --> 00:26:21,280
Speaker 7: tape saying that AI is going to create tons of

572
00:26:21,359 --> 00:26:23,760
Speaker 7: new jobs, that job loss isn't going to happen. Jade

573
00:26:23,840 --> 00:26:26,119
Speaker 7: Vance gave the speech where he was like, oh, we

574
00:26:26,200 --> 00:26:29,119
Speaker 7: will never regulate AI. And so I think, you know,

575
00:26:29,200 --> 00:26:32,679
Speaker 7: just for my conversations, I'm seeing Democratic politicians care a

576
00:26:32,720 --> 00:26:34,880
Speaker 7: lot more about this than they did six months ago.

577
00:26:34,960 --> 00:26:37,679
Speaker 7: And I think that folks are kind of ambling about

578
00:26:37,720 --> 00:26:39,439
Speaker 7: to figure out what the right way to respond is.

579
00:26:40,000 --> 00:26:42,640
Speaker 6: Wait, say more about why you think AI has become

580
00:26:42,680 --> 00:26:45,919
Speaker 6: a concern so quickly, because you know, everyone in this

581
00:26:46,000 --> 00:26:48,760
Speaker 6: room has probably played around with chat, GPT and things

582
00:26:48,800 --> 00:26:51,040
Speaker 6: like that, but I would imagine for a big chunk

583
00:26:51,040 --> 00:26:54,720
Speaker 6: of the population it hasn't necessarily impacted their day to

584
00:26:54,800 --> 00:26:58,040
Speaker 6: day lives just yet. So it's kind of surprising to

585
00:26:58,080 --> 00:27:01,760
Speaker 6: see AI concerns rise the ranks worries so fast.

586
00:27:02,080 --> 00:27:04,439
Speaker 7: Well, I do think people will see the writing on

587
00:27:04,640 --> 00:27:07,840
Speaker 7: the wall. You know that basically, as it stands, something

588
00:27:07,920 --> 00:27:10,320
Speaker 7: like sixty percent of the public has used these tools.

589
00:27:10,400 --> 00:27:13,080
Speaker 7: Thirteen percent of the public uses them every day, And

590
00:27:14,000 --> 00:27:17,560
Speaker 7: I think that people really underestimate the extent to which

591
00:27:17,600 --> 00:27:21,440
Speaker 7: the public is concerned, or I think these tools really

592
00:27:21,480 --> 00:27:25,000
Speaker 7: are being rolled out quite widely across a whole host

593
00:27:25,000 --> 00:27:25,920
Speaker 7: of different sectors.

594
00:27:26,160 --> 00:27:26,359
Speaker 2: You know.

595
00:27:26,400 --> 00:27:28,760
Speaker 7: I was talking this weekend to you know, a medical

596
00:27:28,800 --> 00:27:30,280
Speaker 7: tech who was like, oh, yeah, I know the AI

597
00:27:30,480 --> 00:27:32,080
Speaker 7: is being deployed, like you know, she was in a

598
00:27:32,160 --> 00:27:34,520
Speaker 7: rural hospital in Montana and even them, she was like,

599
00:27:34,560 --> 00:27:37,119
Speaker 7: oh yeah, no, AI is being deployed quite widely. And

600
00:27:37,160 --> 00:27:40,359
Speaker 7: I think that again, this is happening in the context

601
00:27:40,400 --> 00:27:44,240
Speaker 7: of voters feeling extremely negatively about the economy, and so

602
00:27:44,359 --> 00:27:48,000
Speaker 7: whenever they see the prospect of a large scale revolution

603
00:27:48,160 --> 00:27:49,880
Speaker 7: and how all of their jobs are going to happen.

604
00:27:49,960 --> 00:27:51,840
Speaker 7: I think that they're very concerned they're going to be screwed.

605
00:27:52,400 --> 00:27:56,040
Speaker 4: Burn you're plugged into the opposite side of the aisle,

606
00:27:56,119 --> 00:27:59,919
Speaker 4: I believe, yes, And on there there's some interesting cleavages

607
00:28:00,119 --> 00:28:02,720
Speaker 4: as well, because obviously we have the sort of the

608
00:28:02,800 --> 00:28:08,719
Speaker 4: tech right that's very enthusiastic the progress and acceleration and

609
00:28:08,760 --> 00:28:11,560
Speaker 4: so forth, and then you have the sort of obvious

610
00:28:11,640 --> 00:28:15,120
Speaker 4: like populist coalition. You have like politicians talking about how

611
00:28:15,200 --> 00:28:17,760
Speaker 4: awful it would be if you know, we ever had

612
00:28:17,760 --> 00:28:21,720
Speaker 4: self driving trucks and how terrible that would be for drivers.

613
00:28:22,280 --> 00:28:25,679
Speaker 4: How do you see the tectonic plates moving around on

614
00:28:25,720 --> 00:28:26,879
Speaker 4: the other side.

615
00:28:27,160 --> 00:28:29,960
Speaker 8: Well, I like AI, so I'm very pessimistic on the

616
00:28:29,960 --> 00:28:32,280
Speaker 8: political dimension. And I think part of it is that

617
00:28:32,840 --> 00:28:35,320
Speaker 8: when you ask people about AI, when you're making it salient,

618
00:28:35,400 --> 00:28:36,920
Speaker 8: they have one set of views, but if you look

619
00:28:36,920 --> 00:28:38,920
Speaker 8: at their behavior, they have a different set of views.

620
00:28:39,200 --> 00:28:41,640
Speaker 8: And I think this is like a broader point about

621
00:28:41,680 --> 00:28:44,160
Speaker 8: the large scale deployments of technologies is that they do

622
00:28:44,320 --> 00:28:48,800
Speaker 8: increase measured income and wealth inequality, but they decrease consumption inequality.

623
00:28:49,000 --> 00:28:51,320
Speaker 8: So things like flying on a plane, it's just a

624
00:28:51,400 --> 00:28:54,600
Speaker 8: much more attainable, affordable thing. There were some recent stats

625
00:28:54,640 --> 00:28:57,880
Speaker 8: on how DoorDash usage is heaviest among lower income people.

626
00:28:58,360 --> 00:29:00,600
Speaker 8: And so you just have act to a lot of

627
00:29:00,600 --> 00:29:02,280
Speaker 8: things where it used to be that either nobody had

628
00:29:02,280 --> 00:29:04,240
Speaker 8: it or only very wealthy people had it and actually

629
00:29:04,280 --> 00:29:06,280
Speaker 8: chut GPTs. It is the kind of thing where if

630
00:29:06,280 --> 00:29:08,360
Speaker 8: it didn't exist and I were much much wealthier, I

631
00:29:08,400 --> 00:29:09,800
Speaker 8: would just hire people to do that kind of thing.

632
00:29:09,840 --> 00:29:13,239
Speaker 8: I would just ask them weird questions about history, or

633
00:29:13,320 --> 00:29:15,720
Speaker 8: just send them off on little research tasks, and I

634
00:29:15,720 --> 00:29:18,080
Speaker 8: wouldn't feel at all bad if they get back to me,

635
00:29:18,160 --> 00:29:20,080
Speaker 8: you know, a week later, and present me with this

636
00:29:20,120 --> 00:29:21,760
Speaker 8: report and I say, oh, I actually change my mind.

637
00:29:21,760 --> 00:29:23,960
Speaker 8: I don't care about that anymore. Just but here's the

638
00:29:23,960 --> 00:29:25,760
Speaker 8: next one. But like you can, you can do that

639
00:29:25,800 --> 00:29:29,880
Speaker 8: with an LM. But also older demographics nominally don't like AI,

640
00:29:30,000 --> 00:29:31,600
Speaker 8: but they spend a lot of time on Facebook, which

641
00:29:31,600 --> 00:29:34,040
Speaker 8: means they actually do really like AI. They like AI

642
00:29:34,200 --> 00:29:37,880
Speaker 8: recommendation engines, They are very tolerant of AI recommended ads,

643
00:29:37,960 --> 00:29:41,200
Speaker 8: They love AI generated images and AI generated texts. They

644
00:29:41,200 --> 00:29:43,280
Speaker 8: feel much more comfortable on a site where an AI

645
00:29:43,360 --> 00:29:45,640
Speaker 8: is actually way to tell them what the comment should say,

646
00:29:46,000 --> 00:29:47,440
Speaker 8: and they don't have to come up with a comment,

647
00:29:47,520 --> 00:29:50,880
Speaker 8: Like people actually love AI from a consumption perspective and

648
00:29:51,000 --> 00:29:54,280
Speaker 8: hate it from like the outside abstract perspective. So the

649
00:29:54,320 --> 00:29:56,479
Speaker 8: thing that makes me more optimistic is just the sailings

650
00:29:56,520 --> 00:30:00,000
Speaker 8: of these different issues fluctuates over time. Maybe AI deployment

651
00:30:00,040 --> 00:30:01,960
Speaker 8: it is so fast that becomes just part of the

652
00:30:01,960 --> 00:30:05,440
Speaker 8: background noise, like the internal combustion engine or electricity or

653
00:30:05,440 --> 00:30:07,360
Speaker 8: even the Internet, where like the Internet is not a

654
00:30:07,360 --> 00:30:11,080
Speaker 8: campaign issue right now. It was sort of an issue

655
00:30:11,160 --> 00:30:13,560
Speaker 8: in two thousand and it's been I guess you know,

656
00:30:13,720 --> 00:30:17,360
Speaker 8: there were internety issues in twenty sixteen and twenty twenty,

657
00:30:17,760 --> 00:30:20,800
Speaker 8: but it's becoming less and less salient. We just don't

658
00:30:20,800 --> 00:30:22,280
Speaker 8: think about it like are you pro or anti?

659
00:30:22,480 --> 00:30:25,240
Speaker 2: Like it just is. And I think that with a

660
00:30:25,240 --> 00:30:26,600
Speaker 2: lot of these invisible.

661
00:30:26,160 --> 00:30:29,400
Speaker 8: Productivity gains from AI, or you know, less salient productivity

662
00:30:29,400 --> 00:30:31,800
Speaker 8: gains from AI, we're in a better world where people

663
00:30:31,840 --> 00:30:36,240
Speaker 8: are not thinking AI is and is only the homework

664
00:30:36,440 --> 00:30:39,880
Speaker 8: cheating and plagiarism, like deniable plagiarism machine, plus the thing

665
00:30:39,880 --> 00:30:43,280
Speaker 8: that's taking away my best potential job.

666
00:30:58,920 --> 00:31:01,960
Speaker 6: Do either of you have a FAE historical analogy? For

667
00:31:02,040 --> 00:31:04,680
Speaker 6: AI right now in terms of understanding it from a

668
00:31:04,720 --> 00:31:06,320
Speaker 6: political or social perspective.

669
00:31:06,920 --> 00:31:08,479
Speaker 7: You know, for me, as I said before, I think

670
00:31:08,520 --> 00:31:11,160
Speaker 7: it's COVID where I think that this is going to

671
00:31:11,640 --> 00:31:15,680
Speaker 7: hit very quickly. People get obsessed about, you know, exactly

672
00:31:15,720 --> 00:31:18,360
Speaker 7: when it will happen or exactly how it will happen.

673
00:31:18,520 --> 00:31:21,320
Speaker 7: But I think voters hate change. I think that that's

674
00:31:21,320 --> 00:31:24,400
Speaker 7: one of the most underrated status quo bias of politics

675
00:31:24,520 --> 00:31:26,480
Speaker 7: is very strong. If you look at who are the

676
00:31:26,520 --> 00:31:28,960
Speaker 7: most popular politicians in the country, it's always, you know,

677
00:31:29,000 --> 00:31:33,040
Speaker 7: the governors who do absolutely nothing, you know, underrated political fact.

678
00:31:33,360 --> 00:31:35,720
Speaker 7: So I think if you're talking about a world where

679
00:31:36,200 --> 00:31:40,160
Speaker 7: every single job is being simultaneously transformed and there are

680
00:31:40,200 --> 00:31:42,880
Speaker 7: tons of winners or losers, like you know, obviously people

681
00:31:42,960 --> 00:31:45,240
Speaker 7: get hung up on this question of will everyone lose

682
00:31:45,280 --> 00:31:48,080
Speaker 7: their job because of AI. But if like three percent

683
00:31:48,280 --> 00:31:50,560
Speaker 7: of people lose their job because of AI, it's just

684
00:31:50,600 --> 00:31:52,719
Speaker 7: going to be the biggest issue in the world.

685
00:31:52,840 --> 00:31:53,080
Speaker 2: You know.

686
00:31:53,440 --> 00:31:57,440
Speaker 7: Whenever you have diffuse benefits and concentrated losers, that's just

687
00:31:57,480 --> 00:32:01,640
Speaker 7: like a public choice recipe for chaos. And so I

688
00:32:01,680 --> 00:32:05,640
Speaker 7: really liked the COVID analogy because COVID happened basically all

689
00:32:05,680 --> 00:32:08,440
Speaker 7: at once, and then our political system was scrambled and

690
00:32:08,560 --> 00:32:11,320
Speaker 7: new coalitions were created, and it was very difficult to respond.

691
00:32:11,720 --> 00:32:13,800
Speaker 7: And I think whenever you go back, as we talked

692
00:32:13,840 --> 00:32:16,360
Speaker 7: about before, it's really hard to think of an economic

693
00:32:16,360 --> 00:32:17,760
Speaker 7: transition that happened that quickly.

694
00:32:19,040 --> 00:32:21,320
Speaker 8: Yeah, I think I think the COVID analogy might be

695
00:32:21,360 --> 00:32:24,560
Speaker 8: revealing in another way, which is like the coalitions completely

696
00:32:24,720 --> 00:32:28,600
Speaker 8: shifted multiple times. It was like in January, it's only

697
00:32:29,120 --> 00:32:33,640
Speaker 8: the weird online, extremely online right wing anonymous people, and

698
00:32:33,680 --> 00:32:35,840
Speaker 8: then a handful of rationalists people who say this is

699
00:32:35,880 --> 00:32:38,080
Speaker 8: a really big deal and it's like we need to

700
00:32:38,080 --> 00:32:41,320
Speaker 8: shut down air travel from China, and like the for

701
00:32:41,400 --> 00:32:44,280
Speaker 8: the EA types, for the rationalist types, that was more

702
00:32:44,360 --> 00:32:47,520
Speaker 8: like this is a prudent response to a potential existential threat.

703
00:32:47,560 --> 00:32:49,560
Speaker 2: And then I think for a lot of people.

704
00:32:49,280 --> 00:32:50,920
Speaker 8: On the far right, it was like, this is a

705
00:32:50,960 --> 00:32:53,080
Speaker 8: way to make China look bad and also to have

706
00:32:53,160 --> 00:32:55,280
Speaker 8: fewer flights to and from China, and therefore we want

707
00:32:55,280 --> 00:32:55,640
Speaker 8: to do it.

708
00:32:56,440 --> 00:32:58,520
Speaker 2: And then you know, Trump, I think, was very.

709
00:32:58,640 --> 00:33:00,440
Speaker 8: Just tied to what the S and P was telling

710
00:33:00,520 --> 00:33:02,120
Speaker 8: him about whether this was a good thing or a

711
00:33:02,160 --> 00:33:04,360
Speaker 8: bad thing. Like you could kind of see it, like

712
00:33:04,800 --> 00:33:07,720
Speaker 8: I think if he'd had a live ticker telling him

713
00:33:07,840 --> 00:33:09,880
Speaker 8: how the market was reacting to some of his early

714
00:33:09,920 --> 00:33:11,960
Speaker 8: COVID speeches, we would have had a completely different COVID

715
00:33:12,000 --> 00:33:13,920
Speaker 8: policy of the country. And then we kind of did

716
00:33:13,920 --> 00:33:17,520
Speaker 8: this flip where now then the right became the more

717
00:33:17,560 --> 00:33:20,280
Speaker 8: COVID libertarian. You know, just let it rip. You know,

718
00:33:20,280 --> 00:33:21,720
Speaker 8: we're all going to be immune to this at some

719
00:33:21,800 --> 00:33:24,200
Speaker 8: point faction, which was really weird to me because the

720
00:33:24,280 --> 00:33:26,640
Speaker 8: right skills older and older people were more at risk.

721
00:33:27,040 --> 00:33:29,720
Speaker 8: But I guess it goes back to salience and information environment,

722
00:33:29,760 --> 00:33:32,560
Speaker 8: and you are if you're in the cohort where your sedentary,

723
00:33:32,560 --> 00:33:35,120
Speaker 8: you're elderly, and you're sedentary because we're spending a lot

724
00:33:35,160 --> 00:33:37,000
Speaker 8: of time watching TV. If what you're watching on TV

725
00:33:37,080 --> 00:33:39,080
Speaker 8: is Fox, then maybe you just get a very different

726
00:33:39,160 --> 00:33:42,200
Speaker 8: view of the world. So I think the coalitions probably

727
00:33:42,240 --> 00:33:44,360
Speaker 8: fluctuate a lot. I don't think there's actually any I

728
00:33:44,360 --> 00:33:46,720
Speaker 8: don't think either major party is a good home for

729
00:33:46,960 --> 00:33:50,560
Speaker 8: the pro growth slash abundance faction, Like they can be

730
00:33:51,120 --> 00:33:54,360
Speaker 8: a minority among Democrats or minority among Republicans, and we'll

731
00:33:54,360 --> 00:33:55,920
Speaker 8: have to sell out in some ways to have any

732
00:33:55,960 --> 00:34:00,000
Speaker 8: kind of influence whatsoever. So in that sense, again pretty pessimistic.

733
00:34:00,080 --> 00:34:02,520
Speaker 8: But when you were talking about the issues that changed

734
00:34:02,520 --> 00:34:05,040
Speaker 8: in Saliens, I think in the data you've had, it

735
00:34:05,040 --> 00:34:07,280
Speaker 8: was like a year ago it was trade and everyone

736
00:34:07,360 --> 00:34:09,960
Speaker 8: was obsessed with trade, and then trade became a huge

737
00:34:10,000 --> 00:34:12,960
Speaker 8: thing and was like the main producer of headlines for

738
00:34:12,960 --> 00:34:13,799
Speaker 8: a while, and.

739
00:34:13,760 --> 00:34:14,560
Speaker 2: Then it kind of faded.

740
00:34:15,040 --> 00:34:17,160
Speaker 8: So and even though like the situation is still very

741
00:34:17,200 --> 00:34:18,960
Speaker 8: messed up in terms of global trade, in fact, it

742
00:34:19,040 --> 00:34:20,640
Speaker 8: is more messed up than it was a few weeks

743
00:34:20,640 --> 00:34:23,440
Speaker 8: ago in terms of the oil trade. So in that sense,

744
00:34:23,719 --> 00:34:26,080
Speaker 8: trade is like the day to day like are you

745
00:34:26,120 --> 00:34:27,399
Speaker 8: better off than you were a week ago?

746
00:34:27,560 --> 00:34:27,840
Speaker 2: Question?

747
00:34:28,160 --> 00:34:31,120
Speaker 8: Trade is still the most salient issue. I mean, maybe

748
00:34:31,320 --> 00:34:33,319
Speaker 8: if there's a new new model really settle change things,

749
00:34:33,320 --> 00:34:35,279
Speaker 8: but for now that's the case.

750
00:34:35,320 --> 00:34:36,840
Speaker 2: So yeah, it's all pretty valatile.

751
00:34:37,320 --> 00:34:40,120
Speaker 8: In terms of the technology that I think maps most

752
00:34:40,160 --> 00:34:43,400
Speaker 8: closely to this, I think the transistor and integrated circuits

753
00:34:43,440 --> 00:34:46,200
Speaker 8: are probably probably the closest mapping like one in the

754
00:34:46,200 --> 00:34:48,520
Speaker 8: literal sense of there are a way to make everything

755
00:34:48,520 --> 00:34:50,719
Speaker 8: in your life a little bit smarter, And it is

756
00:34:50,760 --> 00:34:53,600
Speaker 8: just astonishing to think of how many little idiots savant

757
00:34:53,680 --> 00:34:56,840
Speaker 8: devices we have in our homes that can do little

758
00:34:56,880 --> 00:34:59,360
Speaker 8: calculations and computations and have a nice little interface, And

759
00:34:59,400 --> 00:35:02,480
Speaker 8: how that stuff got so cheap that it basically became free,

760
00:35:02,520 --> 00:35:04,560
Speaker 8: Like it would not make sense to think about can

761
00:35:04,600 --> 00:35:07,160
Speaker 8: we build a microwave that has entirely mechanical controls with

762
00:35:07,200 --> 00:35:08,560
Speaker 8: no transistors inside of it?

763
00:35:08,640 --> 00:35:09,719
Speaker 2: Like why would you do that?

764
00:35:10,680 --> 00:35:13,680
Speaker 4: Speaking of politics, like, David, do you think it's plausible.

765
00:35:14,600 --> 00:35:17,640
Speaker 4: Let's say the president were for it, a future president

766
00:35:17,760 --> 00:35:20,000
Speaker 4: or for it. Someone puts up a bill and says

767
00:35:20,239 --> 00:35:23,280
Speaker 4: we're going to ban ai new Ai data center construction.

768
00:35:23,400 --> 00:35:25,320
Speaker 2: Could you see a world where that actually passes?

769
00:35:25,560 --> 00:35:28,239
Speaker 4: Because I feel like that's like that seems crazy, but

770
00:35:28,360 --> 00:35:31,239
Speaker 4: also more I think about it, it almost seems like

771
00:35:31,280 --> 00:35:31,960
Speaker 4: it could pass.

772
00:35:33,120 --> 00:35:35,719
Speaker 7: Yeah, I could see all kinds of things passing. We're

773
00:35:35,719 --> 00:35:37,560
Speaker 7: going to we will probably have divided government, and so

774
00:35:37,640 --> 00:35:40,320
Speaker 7: it's always good. Yeah that against any specific thing happening

775
00:35:40,360 --> 00:35:41,120
Speaker 7: in that situation.

776
00:35:41,320 --> 00:35:44,240
Speaker 4: But like we just you know, there's it's very popular,

777
00:35:44,320 --> 00:35:48,440
Speaker 4: for example, to ban investors from owning homes for example,

778
00:35:48,480 --> 00:35:51,680
Speaker 4: which is both parties seem to be like very into

779
00:35:51,719 --> 00:35:55,560
Speaker 4: that idea that corporations shouldn't own large swaths of single

780
00:35:55,600 --> 00:35:59,440
Speaker 4: family homes. Totally seems to cut across both parties. The

781
00:35:59,520 --> 00:36:02,759
Speaker 4: anti data center thing strikes me as similar, where it

782
00:36:02,800 --> 00:36:04,840
Speaker 4: does not seem like a right or left thing. It

783
00:36:04,880 --> 00:36:09,080
Speaker 4: seems like a very broad, populist TikTok politics sort of thing.

784
00:36:09,400 --> 00:36:09,760
Speaker 2: Yeah.

785
00:36:09,800 --> 00:36:11,520
Speaker 7: You know what I'll say about the polling on data

786
00:36:11,520 --> 00:36:14,840
Speaker 7: centers is it's true that if you pull you know,

787
00:36:14,960 --> 00:36:17,600
Speaker 7: do you want a data center in your neighborhood, people

788
00:36:17,600 --> 00:36:20,040
Speaker 7: say no. But the flip side is if you add

789
00:36:20,200 --> 00:36:22,560
Speaker 7: literally anything to it, if you're like, what if it's

790
00:36:22,560 --> 00:36:24,640
Speaker 7: built by clean energy, or what if it lowered your

791
00:36:24,640 --> 00:36:28,000
Speaker 7: tax bills by ten percent, then suddenly it becomes really popular.

792
00:36:28,320 --> 00:36:31,520
Speaker 7: I've seen a lot of politicians like kind of grab

793
00:36:31,640 --> 00:36:33,920
Speaker 7: for the data center thing because this is just like

794
00:36:34,000 --> 00:36:36,960
Speaker 7: a big scary issue, and the data center stuff is

795
00:36:37,000 --> 00:36:40,000
Speaker 7: just like a really legible, oh Land use in this

796
00:36:40,160 --> 00:36:43,480
Speaker 7: and that. But I think it would be a mistake,

797
00:36:43,680 --> 00:36:46,959
Speaker 7: not not because I care that much personally about whether

798
00:36:47,000 --> 00:36:49,680
Speaker 7: we ban data centers or not, but really just because

799
00:36:49,719 --> 00:36:52,839
Speaker 7: I think the public really wants something else. Like if

800
00:36:52,840 --> 00:36:56,960
Speaker 7: we ban data centers tomorrow, that really doesn't change the

801
00:36:57,040 --> 00:37:00,120
Speaker 7: fact that voters think that the economy is rigged, or

802
00:37:00,160 --> 00:37:02,600
Speaker 7: that they're very scared about the future. And you know,

803
00:37:02,680 --> 00:37:06,759
Speaker 7: in our testing, you know, we've generally tested dozens of

804
00:37:06,760 --> 00:37:09,719
Speaker 7: different ways to talk about AI, and you know, the

805
00:37:09,800 --> 00:37:13,040
Speaker 7: data center stuff I think just doesn't move the public

806
00:37:13,360 --> 00:37:17,000
Speaker 7: very much. While it does I think is actually just

807
00:37:17,080 --> 00:37:20,840
Speaker 7: being a lot more radical talking about job guarantees or

808
00:37:20,880 --> 00:37:25,239
Speaker 7: income guarantees or eviction protections. Like I think that there's

809
00:37:25,280 --> 00:37:30,640
Speaker 7: just an enormous amount of fear and the political system

810
00:37:30,840 --> 00:37:33,600
Speaker 7: should try to address it. And it's not just should,

811
00:37:34,120 --> 00:37:36,279
Speaker 7: I think it will, you know, because we do live

812
00:37:36,280 --> 00:37:37,080
Speaker 7: in a democracy.

813
00:37:37,480 --> 00:37:38,680
Speaker 5: Actually, burn that reminds me.

814
00:37:38,800 --> 00:37:40,080
Speaker 6: One of the things we hear a lot when it

815
00:37:40,080 --> 00:37:42,600
Speaker 6: comes to AI is this idea of electricity as a

816
00:37:42,640 --> 00:37:45,160
Speaker 6: limiting factor. Right, everyone talks about how that's the big

817
00:37:45,200 --> 00:37:49,319
Speaker 6: constraint when you yourself think about adoption of the technology

818
00:37:49,400 --> 00:37:52,760
Speaker 6: and it's spread across America and I guess around the world,

819
00:37:53,120 --> 00:37:55,840
Speaker 6: how much are you actually thinking about things like data

820
00:37:55,840 --> 00:37:57,640
Speaker 6: centers and electricity consumption.

821
00:37:58,080 --> 00:38:00,520
Speaker 8: Well, I'm thinking less about those right now now, just

822
00:38:00,560 --> 00:38:04,279
Speaker 8: because everyone who is in that supply chain has all

823
00:38:04,360 --> 00:38:06,799
Speaker 8: their capacity booked out so far in the future, Like

824
00:38:06,880 --> 00:38:09,560
Speaker 8: new supply is pretty and elastic, Like I wouldn't be

825
00:38:10,040 --> 00:38:14,400
Speaker 8: entirely surprised to see the AI companies just guaranteeing that

826
00:38:14,440 --> 00:38:16,879
Speaker 8: they will take delivery on you know, something from seam

827
00:38:16,920 --> 00:38:19,560
Speaker 8: as Energy or ge in you know, the year twenty

828
00:38:19,560 --> 00:38:21,480
Speaker 8: thirty two or whatever, just so that they break round

829
00:38:21,480 --> 00:38:24,880
Speaker 8: on new manufacturing capacity. But I kind of view that

830
00:38:25,040 --> 00:38:28,239
Speaker 8: because it's such a long lag thing, it's kind of

831
00:38:28,239 --> 00:38:31,719
Speaker 8: a given. I think that the actual constraints are more

832
00:38:31,880 --> 00:38:36,040
Speaker 8: on the internal organization side. Just you know, do you

833
00:38:36,080 --> 00:38:39,360
Speaker 8: even have an org chart if you have AI everywhere

834
00:38:39,400 --> 00:38:42,160
Speaker 8: that's always routing messages to whoever needs them, Like, could you

835
00:38:42,200 --> 00:38:45,320
Speaker 8: just have there's one person who's the CEO and everyone

836
00:38:45,360 --> 00:38:48,200
Speaker 8: else is an individual contributor and you AI if I

837
00:38:48,280 --> 00:38:50,799
Speaker 8: all mental management, probably not, because what you act. The

838
00:38:50,800 --> 00:38:53,080
Speaker 8: other limiting factor is just liability, is that you do

839
00:38:53,120 --> 00:38:54,440
Speaker 8: want a human in the loop. You know, I was

840
00:38:54,480 --> 00:38:56,120
Speaker 8: joking with this earlier, but it is true. Like the

841
00:38:56,160 --> 00:38:58,439
Speaker 8: person who's in an ideal position right now, as someone

842
00:38:58,440 --> 00:39:01,239
Speaker 8: who is in a regulated job where there is some

843
00:39:01,360 --> 00:39:04,200
Speaker 8: kind of trade association that limits new entrants into that job,

844
00:39:04,239 --> 00:39:07,000
Speaker 8: but there's access demand for the services they provide, then

845
00:39:07,040 --> 00:39:09,800
Speaker 8: AI can increase their output per hour, Like these people

846
00:39:09,880 --> 00:39:13,640
Speaker 8: will be minting money and they will continue to do

847
00:39:13,680 --> 00:39:16,000
Speaker 8: so as long as they can limit supply inflows. And

848
00:39:16,040 --> 00:39:17,920
Speaker 8: so we might end up with as a more kind

849
00:39:17,960 --> 00:39:22,319
Speaker 8: of guildfied economy where we have limits on who specifically

850
00:39:22,560 --> 00:39:24,400
Speaker 8: not who can do the job, but who can actually

851
00:39:24,440 --> 00:39:26,360
Speaker 8: you know, stamp it and say I'm taking credit for this,

852
00:39:26,400 --> 00:39:28,440
Speaker 8: and you consume me if it goes wrong, because that's

853
00:39:28,480 --> 00:39:30,279
Speaker 8: actually a really valuable thing. And it's still like, you

854
00:39:30,280 --> 00:39:32,600
Speaker 8: can't really sue a data center. You know, you can

855
00:39:32,680 --> 00:39:34,319
Speaker 8: try to sue open AI, but they can probably afford

856
00:39:34,360 --> 00:39:35,120
Speaker 8: better lawyers.

857
00:39:34,840 --> 00:39:35,320
Speaker 2: Than you do.

858
00:39:35,400 --> 00:39:37,920
Speaker 8: You actually, it's probably more economically efficient to sue me

859
00:39:38,160 --> 00:39:40,239
Speaker 8: for something chat TPT told me to do that I

860
00:39:40,239 --> 00:39:43,799
Speaker 8: didn't double check by asking Claude, and so we'll, like

861
00:39:44,160 --> 00:39:46,560
Speaker 8: it is weird to think that one of our economic purposes,

862
00:39:46,640 --> 00:39:49,480
Speaker 8: as you know, living and sold human beings, is to

863
00:39:49,520 --> 00:39:52,279
Speaker 8: be an easy target for a lawsuit, but it is

864
00:39:52,280 --> 00:39:55,280
Speaker 8: actually something economically valuable that we can do. And because

865
00:39:55,280 --> 00:39:58,520
Speaker 8: we all have different risk preferences and risk tolerances for things,

866
00:39:58,560 --> 00:39:59,960
Speaker 8: it actually means that we would get this sort of

867
00:40:00,360 --> 00:40:02,759
Speaker 8: uneven deployment of AI where there would be people who

868
00:40:02,800 --> 00:40:06,120
Speaker 8: just demonstrate that you can go way too far and

869
00:40:06,160 --> 00:40:07,960
Speaker 8: that you know, if you decide that you're going to

870
00:40:07,960 --> 00:40:10,359
Speaker 8: be the CPA who does taxes for you know, does

871
00:40:10,560 --> 00:40:13,480
Speaker 8: on thy ten forties a day, you're probably the first

872
00:40:13,480 --> 00:40:16,839
Speaker 8: CPA to go to prison specifically for something that an

873
00:40:16,960 --> 00:40:18,120
Speaker 8: LLLM told you to do.

874
00:40:18,440 --> 00:40:19,560
Speaker 2: If that hasn't happened yet.

875
00:40:19,800 --> 00:40:22,120
Speaker 8: But I think that that kind of model where you

876
00:40:22,160 --> 00:40:24,759
Speaker 8: want to human in the loop because so many structures

877
00:40:24,800 --> 00:40:27,640
Speaker 8: that we have economically and socially, just assume there's a

878
00:40:27,680 --> 00:40:29,640
Speaker 8: person that will probably stick around.

879
00:40:30,200 --> 00:40:33,440
Speaker 4: You write a lot about finance as well as technology,

880
00:40:33,440 --> 00:40:35,120
Speaker 4: and I've been thinking a lot about the future of

881
00:40:35,160 --> 00:40:36,840
Speaker 4: finance and they I have. One of the things that

882
00:40:36,960 --> 00:40:39,560
Speaker 4: comes up is we did a really good episode with

883
00:40:39,600 --> 00:40:42,160
Speaker 4: the CEO pn C Bank. We're talking a little bit

884
00:40:42,160 --> 00:40:46,000
Speaker 4: about AI and lending, and he said, you know, if

885
00:40:46,040 --> 00:40:48,160
Speaker 4: you deny alone to someone, you have to have a

886
00:40:48,239 --> 00:40:50,520
Speaker 4: reason for it. And there are various laws that you know,

887
00:40:50,640 --> 00:40:54,240
Speaker 4: anti discrimination laws and stuff, So if someone has denied credit,

888
00:40:54,480 --> 00:40:57,000
Speaker 4: you have to be able to explain why. One of

889
00:40:57,040 --> 00:40:59,319
Speaker 4: the things with AI is that it can make very

890
00:40:59,360 --> 00:41:01,520
Speaker 4: good decisions, but it's hard to interpret and the AI

891
00:41:01,680 --> 00:41:05,400
Speaker 4: often can't explain how it arrived at a conclusion, and

892
00:41:05,480 --> 00:41:07,759
Speaker 4: I'm curious, like from a fun just thinking about the

893
00:41:07,800 --> 00:41:10,600
Speaker 4: finance industry, how much using that it's going to be

894
00:41:10,640 --> 00:41:13,520
Speaker 4: a limiting constraint on the degree to which AI disrupts

895
00:41:13,520 --> 00:41:17,839
Speaker 4: the industry, the fact that a lot of things need

896
00:41:17,880 --> 00:41:19,560
Speaker 4: to be articulable in English.

897
00:41:19,640 --> 00:41:23,239
Speaker 8: Basically, well, I think AI, like humans, is bad at

898
00:41:23,239 --> 00:41:25,359
Speaker 8: knowing why it actually did things, and really really good

899
00:41:25,400 --> 00:41:27,480
Speaker 8: at explaining why whatever it did was the right thing

900
00:41:27,520 --> 00:41:31,359
Speaker 8: to do, So it probably I mean, I'm not the

901
00:41:31,400 --> 00:41:33,000
Speaker 8: head of PNC, so I don't know for sure, but

902
00:41:33,040 --> 00:41:35,440
Speaker 8: I would imagine it actually makes it easier to just

903
00:41:35,480 --> 00:41:39,480
Speaker 8: come up with some very solid sounding rationalization for anything,

904
00:41:39,480 --> 00:41:42,480
Speaker 8: Like they're just really good at rationalizing, so I would

905
00:41:42,480 --> 00:41:45,279
Speaker 8: be less concerned with that. Like, I actually think it's

906
00:41:45,480 --> 00:41:49,200
Speaker 8: nice to have more open, accessible kind of reasoning, like

907
00:41:49,200 --> 00:41:51,600
Speaker 8: you can actually read the reasoning traces. And one of

908
00:41:51,600 --> 00:41:53,160
Speaker 8: the facts I think it has on finance is that

909
00:41:53,560 --> 00:41:56,520
Speaker 8: people who are coming up with recommendations like make this loan,

910
00:41:56,560 --> 00:41:58,799
Speaker 8: don't make that loan, it will make a lot of

911
00:41:58,800 --> 00:42:01,759
Speaker 8: sense to have much more detailed records of their entire

912
00:42:01,800 --> 00:42:04,680
Speaker 8: thought process, So you might have them, you know, working

913
00:42:04,719 --> 00:42:07,560
Speaker 8: in data bricks, notebooks or something equivalent to that specifically,

914
00:42:07,600 --> 00:42:09,640
Speaker 8: so you could see, Okay, first they asked this question,

915
00:42:09,760 --> 00:42:11,279
Speaker 8: and then they went down this rabbit hole. Then they

916
00:42:11,320 --> 00:42:13,280
Speaker 8: decided it's irrelevant, and then they went to this question,

917
00:42:13,560 --> 00:42:15,040
Speaker 8: and then they spent some time on it, et cetera.

918
00:42:15,160 --> 00:42:18,080
Speaker 8: Because you if you have just here's the question and

919
00:42:18,120 --> 00:42:21,200
Speaker 8: here's the polished answer someone, Yeah, with an edited you're

920
00:42:21,200 --> 00:42:23,919
Speaker 8: missing a lot of the intermediate layers, and so it's

921
00:42:23,920 --> 00:42:26,080
Speaker 8: hard to train a model that can trace through that

922
00:42:26,160 --> 00:42:28,040
Speaker 8: and reproduce it. Now, if you have a smart enough model,

923
00:42:28,239 --> 00:42:31,600
Speaker 8: it's basically implicitly reproducing all of that reasoning on its own.

924
00:42:31,920 --> 00:42:34,240
Speaker 8: But that means that it does worse when the reasoning

925
00:42:34,280 --> 00:42:37,080
Speaker 8: is really really clever and the person didn't shoot their work,

926
00:42:37,239 --> 00:42:39,239
Speaker 8: and a lot of clever people just figure things out

927
00:42:39,280 --> 00:42:41,080
Speaker 8: and then they're already bored with the problem, so they

928
00:42:41,080 --> 00:42:42,279
Speaker 8: don't want to tell you how they did it, and

929
00:42:42,320 --> 00:42:43,279
Speaker 8: they move on to the next thing.

930
00:42:44,000 --> 00:42:46,799
Speaker 6: I did have like a home DIY project recently, and

931
00:42:46,880 --> 00:42:48,880
Speaker 6: I was asking chat Shept how to do it, and

932
00:42:48,920 --> 00:42:52,200
Speaker 6: it basically told me to defy gravity and could not

933
00:42:52,480 --> 00:42:55,919
Speaker 6: explain why it had time I'll tell you more about

934
00:42:55,920 --> 00:42:58,040
Speaker 6: it later because it's a long story. But anyway, if

935
00:42:58,080 --> 00:43:01,320
Speaker 6: we do a little bit of like political polling naval

936
00:43:01,360 --> 00:43:04,040
Speaker 6: gazing for a second, David, when I think about AI

937
00:43:04,560 --> 00:43:06,600
Speaker 6: and some of what you do, I think it could

938
00:43:06,640 --> 00:43:08,959
Speaker 6: be very helpful to you because you can identify even

939
00:43:08,960 --> 00:43:11,799
Speaker 6: more granular issues for the population. You can come up

940
00:43:11,840 --> 00:43:15,319
Speaker 6: with like even better polling questions. But then I think

941
00:43:15,320 --> 00:43:18,839
Speaker 6: that everyone's going to have access to basically the same technology,

942
00:43:18,960 --> 00:43:22,520
Speaker 6: and so the worst case outcome is probably going to

943
00:43:22,520 --> 00:43:24,880
Speaker 6: come to fruition, which is we're just going to get

944
00:43:24,920 --> 00:43:29,040
Speaker 6: more identity and sort of cultural grievance politics. How do

945
00:43:29,080 --> 00:43:32,440
Speaker 6: you see that going, Like, does political campaigning get smarter

946
00:43:32,600 --> 00:43:36,200
Speaker 6: with AI or do we just kind of descend more

947
00:43:36,200 --> 00:43:37,600
Speaker 6: into culture politics.

948
00:43:38,200 --> 00:43:41,320
Speaker 7: Well, obviously it's very hard to make predictions about the future,

949
00:43:41,320 --> 00:43:44,120
Speaker 7: which I guess I've said already today, but it's easy

950
00:43:44,120 --> 00:43:47,760
Speaker 7: to imagine really bad things happening. Deep fakes, the inability

951
00:43:47,760 --> 00:43:50,040
Speaker 7: to track what's true and what's not, you know, the

952
00:43:50,080 --> 00:43:53,719
Speaker 7: ability now of lots of people to make content that's persuasive,

953
00:43:53,800 --> 00:43:56,720
Speaker 7: that argues for things that previously weren't within the domain.

954
00:43:57,280 --> 00:43:59,839
Speaker 7: But you know, I do think in this discussion, people

955
00:44:00,040 --> 00:44:04,360
Speaker 7: kind of underestimate how dysfunctional the status quo is, where

956
00:44:04,480 --> 00:44:07,120
Speaker 7: if you look at social media, for example, something like

957
00:44:07,520 --> 00:44:10,440
Speaker 7: five percent of the public is responsible for a majority

958
00:44:10,560 --> 00:44:15,120
Speaker 7: of social media content, which is crazy, you know, and

959
00:44:15,360 --> 00:44:18,920
Speaker 7: I think that right now because content is expensive to produce.

960
00:44:19,280 --> 00:44:22,560
Speaker 7: You know, if you are an influencer or a writer

961
00:44:22,640 --> 00:44:27,080
Speaker 7: about politics, your economic incentives are really to focus on

962
00:44:27,560 --> 00:44:30,560
Speaker 7: the five percent of the public that is consuming way

963
00:44:30,680 --> 00:44:34,000
Speaker 7: way more political content than everyone else. And really basically

964
00:44:34,120 --> 00:44:36,360
Speaker 7: no one in the political spectrum right now is making

965
00:44:36,400 --> 00:44:39,080
Speaker 7: content focused on regular people who don't care that much

966
00:44:39,080 --> 00:44:39,840
Speaker 7: about politics.

967
00:44:40,000 --> 00:44:40,160
Speaker 2: You know.

968
00:44:40,239 --> 00:44:42,799
Speaker 7: Just to give an example, someone I know had a

969
00:44:42,960 --> 00:44:46,040
Speaker 7: panel of TikTok users where he you know, recorded their

970
00:44:46,080 --> 00:44:47,600
Speaker 7: phones and it was like two hundred people and he

971
00:44:47,640 --> 00:44:49,320
Speaker 7: was paying them to do that and he could see

972
00:44:49,480 --> 00:44:52,600
Speaker 7: who was watching what. And after Charlie Kirk got shot,

973
00:44:52,680 --> 00:44:55,280
Speaker 7: there was one person who was responsible for a majority

974
00:44:55,320 --> 00:44:57,759
Speaker 7: of the Charlie Kirk videos. Just the day he was

975
00:44:57,800 --> 00:45:02,400
Speaker 7: like really swiping and watching hundreds of Charlie Kirk assassination videos.

976
00:45:02,800 --> 00:45:05,080
Speaker 7: And you know, if you are a content producer, that

977
00:45:05,200 --> 00:45:08,719
Speaker 7: is currently your incentive. And so my point is really

978
00:45:08,800 --> 00:45:11,600
Speaker 7: the status quo is really quite bad. And I think

979
00:45:11,640 --> 00:45:14,040
Speaker 7: that if you look at who these people are, they

980
00:45:14,080 --> 00:45:16,520
Speaker 7: tend to be quite anxious, they tend to be quite neurotic.

981
00:45:16,560 --> 00:45:19,600
Speaker 7: Like right now, the attention game really pushes you toward

982
00:45:19,920 --> 00:45:23,440
Speaker 7: being more negative, while the persuasion game of like how

983
00:45:23,440 --> 00:45:25,480
Speaker 7: do you actually get someone to change their mind really

984
00:45:25,520 --> 00:45:27,359
Speaker 7: pushes you in the opposite direction. You know, every time

985
00:45:27,360 --> 00:45:29,160
Speaker 7: I've done a poll, every time I've done a test,

986
00:45:29,440 --> 00:45:32,560
Speaker 7: it's generally said, you should be more positive, you should

987
00:45:32,560 --> 00:45:35,919
Speaker 7: focus more on regular things that affect people's lives. And

988
00:45:35,960 --> 00:45:39,040
Speaker 7: you know, the reason why that doesn't happen are the

989
00:45:39,080 --> 00:45:42,000
Speaker 7: incentives of the actual content creators. And so I could see,

990
00:45:42,239 --> 00:45:45,880
Speaker 7: you know, lowering the costs of producing content and kind

991
00:45:45,920 --> 00:45:49,040
Speaker 7: of broadening out who is able to make content. It

992
00:45:49,080 --> 00:45:51,560
Speaker 7: could be good, but also, you know, to be clear,

993
00:45:51,600 --> 00:45:54,480
Speaker 7: there's a lot of ways it could be bad. We're

994
00:45:54,520 --> 00:45:57,200
Speaker 7: just entering, I think, a totally different world where it's

995
00:45:57,239 --> 00:45:58,200
Speaker 7: hard to predict anything.

996
00:45:58,680 --> 00:46:03,760
Speaker 4: It is weird how much politics, discourse and ideas truly

997
00:46:03,760 --> 00:46:07,600
Speaker 4: seemed disconnected from anything that affects people, many people on

998
00:46:07,680 --> 00:46:09,239
Speaker 4: a day to day life. I mean, this has been

999
00:46:09,239 --> 00:46:13,040
Speaker 4: brought up before. Why has no politician really made a

1000
00:46:13,080 --> 00:46:16,440
Speaker 4: big issue about like getting rid of like spam texts

1001
00:46:16,560 --> 00:46:19,400
Speaker 4: or something like. We all find it incredibly annoying and

1002
00:46:20,080 --> 00:46:21,680
Speaker 4: it's almost unimaginable.

1003
00:46:21,840 --> 00:46:23,719
Speaker 6: I'm sure you could find some one issue voters on

1004
00:46:23,800 --> 00:46:24,719
Speaker 6: spam text.

1005
00:46:24,520 --> 00:46:26,520
Speaker 4: Right, But like, that would be great if someone actually like,

1006
00:46:26,560 --> 00:46:30,000
Speaker 4: let's take this seriously as a thing that annoys everybody,

1007
00:46:30,080 --> 00:46:32,040
Speaker 4: and let's try to make a push. And yet it

1008
00:46:32,040 --> 00:46:33,000
Speaker 4: doesn't happen, does it.

1009
00:46:33,360 --> 00:46:36,480
Speaker 7: Yeah. The stat I really like on this is if

1010
00:46:36,520 --> 00:46:39,000
Speaker 7: you ask people what issue do you care about the most,

1011
00:46:39,040 --> 00:46:42,319
Speaker 7: it's cost of living by an enormous margin. But the

1012
00:46:42,360 --> 00:46:44,680
Speaker 7: flip side is that if you look at the zero

1013
00:46:44,680 --> 00:46:47,000
Speaker 7: point seven percent of the population that has donated to

1014
00:46:47,040 --> 00:46:49,480
Speaker 7: a democratic campaign in the last year, then cost of

1015
00:46:49,520 --> 00:46:51,600
Speaker 7: living goes from number one to like number five, and

1016
00:46:51,640 --> 00:46:54,680
Speaker 7: like climate changes on top, and so I think, you know,

1017
00:46:54,719 --> 00:46:57,480
Speaker 7: if you look at like I think the status quo

1018
00:46:57,600 --> 00:46:59,840
Speaker 7: is that politicians right now, if you look at what

1019
00:46:59,840 --> 00:47:02,359
Speaker 7: they talk about, it's much more explained by what their

1020
00:47:02,400 --> 00:47:05,680
Speaker 7: donors care about than you know, what the public cares about.

1021
00:47:06,239 --> 00:47:09,279
Speaker 7: And that's really pretty pretty bad. I think a lot

1022
00:47:09,680 --> 00:47:12,560
Speaker 7: an enormous source of our political dysfunction right now is

1023
00:47:12,600 --> 00:47:14,759
Speaker 7: really that people on both sides aren't listening enough to

1024
00:47:14,800 --> 00:47:16,040
Speaker 7: what regular people care about.

1025
00:47:16,440 --> 00:47:18,879
Speaker 6: Do either of you have a good read on the

1026
00:47:18,920 --> 00:47:23,359
Speaker 6: policies you would expect to be actually politically viable if

1027
00:47:23,400 --> 00:47:25,560
Speaker 6: we do start to get I'm not going to say

1028
00:47:25,560 --> 00:47:28,080
Speaker 6: an AI doom case scenario, because I know you disagree

1029
00:47:28,120 --> 00:47:31,719
Speaker 6: on that, but a painful AI adjustment process.

1030
00:47:32,160 --> 00:47:34,799
Speaker 7: I think that the public is much more radical on

1031
00:47:34,840 --> 00:47:38,360
Speaker 7: this issue than people think me personally. I'm not usually

1032
00:47:38,400 --> 00:47:40,640
Speaker 7: the person who goes out and says, ah, the people

1033
00:47:40,760 --> 00:47:43,960
Speaker 7: crave radical policy change. But we really are in a

1034
00:47:44,040 --> 00:47:47,319
Speaker 7: very radical time right now. If you ask voters, you know,

1035
00:47:47,360 --> 00:47:50,200
Speaker 7: do you support price controls, the answer is yes by

1036
00:47:50,200 --> 00:47:51,879
Speaker 7: two to one, which is not something that was true

1037
00:47:51,920 --> 00:47:55,000
Speaker 7: five years ago. You know, when we've done tests. You know,

1038
00:47:55,080 --> 00:47:57,399
Speaker 7: we had one test where we had a really quite

1039
00:47:57,480 --> 00:48:00,000
Speaker 7: radical thing where we're like, oh, we're going to guarantee

1040
00:48:00,040 --> 00:48:02,120
Speaker 7: your income up to one hundred and fifty thousand dollars,

1041
00:48:02,120 --> 00:48:03,440
Speaker 7: We're going to make sure that you have a job,

1042
00:48:03,480 --> 00:48:05,600
Speaker 7: we're going to make sure that you won't be evicted.

1043
00:48:05,960 --> 00:48:08,680
Speaker 7: And it tested better than like ninety eight percent of

1044
00:48:08,719 --> 00:48:11,600
Speaker 7: the clips that were made that were made by democratic

1045
00:48:11,680 --> 00:48:14,760
Speaker 7: professionals that specific policy, which I think is quite radical.

1046
00:48:14,760 --> 00:48:17,760
Speaker 7: I think is something like plus thirty and plus fifteen

1047
00:48:17,800 --> 00:48:20,839
Speaker 7: among Trump voters. And so that's That's just the thing

1048
00:48:20,920 --> 00:48:23,000
Speaker 7: I want to say is, you know, right now everyone

1049
00:48:23,080 --> 00:48:28,400
Speaker 7: has this discussion about timelines and policy specifics, and you know,

1050
00:48:28,440 --> 00:48:30,560
Speaker 7: the public is in a much more radical place than

1051
00:48:30,560 --> 00:48:33,720
Speaker 7: politicians or commentators are. So that's I think the big

1052
00:48:33,760 --> 00:48:36,400
Speaker 7: thing I'd say is, I think it's the most underpriced issue,

1053
00:48:36,960 --> 00:48:40,400
Speaker 7: the very best testing topic, better than populism, better than

1054
00:48:40,440 --> 00:48:43,399
Speaker 7: AI is AI populism, And I think I think that's

1055
00:48:43,440 --> 00:48:46,360
Speaker 7: the direction things are going to go in these thoughts.

1056
00:48:46,640 --> 00:48:47,520
Speaker 2: This is ominous.

1057
00:48:47,920 --> 00:48:50,080
Speaker 8: I do want to return to to the points that

1058
00:48:50,120 --> 00:48:53,920
Speaker 8: you had made on deep fakes and on how everyone

1059
00:48:53,920 --> 00:48:56,359
Speaker 8: in media is writing for the tiny minority of people

1060
00:48:56,400 --> 00:48:58,480
Speaker 8: who really care about or everyone in political media is

1061
00:48:58,480 --> 00:49:00,799
Speaker 8: writing for the tiny minority that cares about politic and one.

1062
00:49:00,800 --> 00:49:03,000
Speaker 8: I actually, I do think deep fakes are a net

1063
00:49:03,040 --> 00:49:08,719
Speaker 8: positive development in the specific sense. That is great. I've

1064
00:49:08,760 --> 00:49:10,880
Speaker 8: been writing about this one for a long time. My

1065
00:49:11,080 --> 00:49:14,759
Speaker 8: view is that it is always possible to create manipulative

1066
00:49:15,160 --> 00:49:19,080
Speaker 8: cuts of some media. It's always possible to say, Okay,

1067
00:49:19,120 --> 00:49:21,640
Speaker 8: there have been you know, fifty videos recorded in the

1068
00:49:21,719 --> 00:49:23,520
Speaker 8: last day of some egregious happening.

1069
00:49:23,600 --> 00:49:24,080
Speaker 2: Here's the one.

1070
00:49:24,160 --> 00:49:26,400
Speaker 8: We're choosing to make a new story. That's always been

1071
00:49:26,440 --> 00:49:29,800
Speaker 8: possible with sufficient resources, and now it's possible for everyone.

1072
00:49:29,920 --> 00:49:32,400
Speaker 2: And so to the extent that when you choose.

1073
00:49:32,120 --> 00:49:34,120
Speaker 8: What narrative, if you're in a position to choose what

1074
00:49:34,239 --> 00:49:36,880
Speaker 8: narrative is promoted, it's I think of it as kind

1075
00:49:36,880 --> 00:49:39,280
Speaker 8: of the modern It's like it is the de facto

1076
00:49:39,280 --> 00:49:43,319
Speaker 8: electoral college. Is that people opt into viewing certain kinds

1077
00:49:43,360 --> 00:49:45,839
Speaker 8: of media that will then shape their views to make

1078
00:49:45,880 --> 00:49:48,880
Speaker 8: them more more correlated with that kind of medium. So

1079
00:49:49,120 --> 00:49:51,600
Speaker 8: we're kind of doing is democratizing democracy. We're saying anyone

1080
00:49:51,640 --> 00:49:53,600
Speaker 8: can make a misleading video, like you don't have to

1081
00:49:53,640 --> 00:49:56,520
Speaker 8: watch twenty hours of footage of this person talking to

1082
00:49:56,560 --> 00:49:57,359
Speaker 8: find the one gaff.

1083
00:49:57,560 --> 00:49:58,399
Speaker 2: You can just fake it.

1084
00:49:58,560 --> 00:50:00,400
Speaker 8: And the fake is actually like for the fake to

1085
00:50:00,440 --> 00:50:03,440
Speaker 8: be a plausible, you know, opinion shifting fake, it has

1086
00:50:03,480 --> 00:50:07,400
Speaker 8: to be something fairly similar to reality. It can just

1087
00:50:07,440 --> 00:50:09,720
Speaker 8: be like Donald Trump is actually an alien.

1088
00:50:09,840 --> 00:50:10,560
Speaker 2: It has to be.

1089
00:50:10,520 --> 00:50:13,440
Speaker 8: Something like, you know, Donald Trump took a bribe in

1090
00:50:13,480 --> 00:50:15,799
Speaker 8: cash from a Katari business matter or something like that.

1091
00:50:15,840 --> 00:50:18,239
Speaker 8: But then if you can only make fakes that are

1092
00:50:18,280 --> 00:50:21,000
Speaker 8: actually kind of plausible, then they just exist in this

1093
00:50:21,200 --> 00:50:21,959
Speaker 8: space that is.

1094
00:50:21,880 --> 00:50:23,360
Speaker 2: Not that far from reality.

1095
00:50:23,719 --> 00:50:26,600
Speaker 8: And similarly, the big viral videos, they are often a

1096
00:50:26,880 --> 00:50:29,440
Speaker 8: sample from reality, like they are a sample from reality,

1097
00:50:29,760 --> 00:50:33,240
Speaker 8: but they are something where you can magnify the perceived

1098
00:50:33,280 --> 00:50:35,600
Speaker 8: frequency of an event if there's just wall to wall

1099
00:50:35,640 --> 00:50:37,719
Speaker 8: coverage of that event on video. So that was one

1100
00:50:37,760 --> 00:50:39,680
Speaker 8: point was I do think that it actually just makes

1101
00:50:39,760 --> 00:50:42,279
Speaker 8: us a less a culture that is less likely to

1102
00:50:42,400 --> 00:50:44,840
Speaker 8: make up our minds on important issues by watching a

1103
00:50:45,040 --> 00:50:47,800
Speaker 8: you know, grainy, shaky twelve seconds of footage of something.

1104
00:50:48,000 --> 00:50:48,680
Speaker 2: I think that's good.

1105
00:50:48,719 --> 00:50:50,640
Speaker 8: I think we should read more and watch less. But

1106
00:50:50,680 --> 00:50:53,120
Speaker 8: I also think that when the you know, the other

1107
00:50:53,160 --> 00:50:56,960
Speaker 8: piece of AI, just the the targeting recommendation algorithm, there's

1108
00:50:57,200 --> 00:51:00,759
Speaker 8: a huge chunk of AI spending and AI's impacts. That

1109
00:51:01,360 --> 00:51:04,600
Speaker 8: probably does mean that it's potentially more likely that someone

1110
00:51:04,640 --> 00:51:07,760
Speaker 8: could make a career out of something like agitating against

1111
00:51:07,760 --> 00:51:09,479
Speaker 8: spam texs. And we did actually see, you know, click

1112
00:51:09,480 --> 00:51:10,520
Speaker 8: to cancel is kind of.

1113
00:51:10,440 --> 00:51:12,760
Speaker 2: Moving alos, so you know, there's little.

1114
00:51:12,640 --> 00:51:14,120
Speaker 8: Quality of life things like click to cancel is the

1115
00:51:14,200 --> 00:51:15,759
Speaker 8: kind of thing a mayor would be really proud of,

1116
00:51:15,800 --> 00:51:17,239
Speaker 8: but it has to be done at more of a

1117
00:51:17,280 --> 00:51:22,160
Speaker 8: federal level. But maybe if it's possible to actually target

1118
00:51:22,320 --> 00:51:25,279
Speaker 8: whatever population niche like really wants to bring back pay

1119
00:51:25,320 --> 00:51:28,080
Speaker 8: toilets because they think that banning pay toilets creates all

1120
00:51:28,120 --> 00:51:30,520
Speaker 8: kinds of perverse economic outcomes, which it does. You can

1121
00:51:30,520 --> 00:51:33,040
Speaker 8: find those people, you can mobilize them, and you can

1122
00:51:33,080 --> 00:51:35,479
Speaker 8: get them to you know, if there is like some

1123
00:51:35,719 --> 00:51:38,640
Speaker 8: very close race somewhere in the house race for example,

1124
00:51:38,960 --> 00:51:41,440
Speaker 8: and one of the politicians happens to support this particular

1125
00:51:41,520 --> 00:51:43,239
Speaker 8: pet issue, it could be like the crypto people and

1126
00:51:43,280 --> 00:51:46,040
Speaker 8: just money bomb the person who happens to also support

1127
00:51:46,040 --> 00:51:48,120
Speaker 8: the thing that they like, So we could actually have

1128
00:51:48,160 --> 00:51:51,600
Speaker 8: That's another way we could potentially see AI democratizing democracy further,

1129
00:51:51,800 --> 00:51:54,200
Speaker 8: is that you can actually coordinate interest groups for people

1130
00:51:54,200 --> 00:51:58,359
Speaker 8: who are just less politics brained but still actually care

1131
00:51:58,360 --> 00:52:00,640
Speaker 8: about problems that should be solved through blood process.

1132
00:52:01,040 --> 00:52:03,240
Speaker 7: I just wanted to jump in quickly, you know, because

1133
00:52:03,560 --> 00:52:05,920
Speaker 7: you know you said that Brent said that this is ominous,

1134
00:52:06,080 --> 00:52:08,480
Speaker 7: and you know, I want to push back on that

1135
00:52:08,560 --> 00:52:10,840
Speaker 7: a little bit. There's an Ed Glazer quote that I

1136
00:52:10,880 --> 00:52:14,200
Speaker 7: really like, which is, you know, everyone wants macroeconomics to

1137
00:52:14,719 --> 00:52:19,360
Speaker 7: have micro foundations, but micro foundations itself should be micro founded.

1138
00:52:19,640 --> 00:52:22,960
Speaker 7: That we are not going to experience any of the

1139
00:52:23,040 --> 00:52:26,840
Speaker 7: potential productivity gains or or only experience a fraction of

1140
00:52:26,880 --> 00:52:31,160
Speaker 7: the productivity gains unless we can give voters some sense

1141
00:52:31,160 --> 00:52:34,719
Speaker 7: of security. That public is really crying out that they

1142
00:52:34,760 --> 00:52:36,919
Speaker 7: want economic security and that they want to be able

1143
00:52:36,920 --> 00:52:40,000
Speaker 7: to look out at the next five years without fear.

1144
00:52:40,320 --> 00:52:43,759
Speaker 7: And if you can provide a new social contract, if

1145
00:52:43,800 --> 00:52:47,280
Speaker 7: you can actually provide economic security at a high level,

1146
00:52:47,680 --> 00:52:51,000
Speaker 7: then you can actually have all of these like sector

1147
00:52:51,000 --> 00:52:55,080
Speaker 7: specific shifts to have higher productivity. But if politicians don't

1148
00:52:55,120 --> 00:52:58,040
Speaker 7: advance a vision like that, then we're just going to

1149
00:52:58,120 --> 00:53:03,440
Speaker 7: collapse into byzantine series of sector specific you know, regulations

1150
00:53:03,480 --> 00:53:05,799
Speaker 7: and guilds, which I think you don't want either. And

1151
00:53:05,880 --> 00:53:08,720
Speaker 7: so I think, you know, the libertarians should really choose,

1152
00:53:08,880 --> 00:53:11,560
Speaker 7: you know, what world they want. You know, either we

1153
00:53:11,600 --> 00:53:15,400
Speaker 7: can have a large scale solution that protects people's incomes

1154
00:53:15,440 --> 00:53:18,600
Speaker 7: and prevents there from being losers, or we can just

1155
00:53:18,719 --> 00:53:21,759
Speaker 7: kind of have this giant negative sum fight playing out

1156
00:53:21,840 --> 00:53:25,400
Speaker 7: in every single sector simultaneously. And I'll also say, you know,

1157
00:53:25,600 --> 00:53:29,120
Speaker 7: we're not really in the ideal political circumstances, you know,

1158
00:53:29,239 --> 00:53:31,879
Speaker 7: for that latter thing to happen, and so I think

1159
00:53:31,880 --> 00:53:33,120
Speaker 7: they really could get quite ugly.

1160
00:53:33,480 --> 00:53:35,799
Speaker 4: David and burn thank you so much. That was a

1161
00:53:35,800 --> 00:53:37,799
Speaker 4: fascinating chip. We really are going to have to do

1162
00:53:37,840 --> 00:53:40,040
Speaker 4: another hour on what deep fakes are doing with us.

1163
00:53:40,480 --> 00:53:42,800
Speaker 4: Thank you all for joining and every great rest of

1164
00:53:42,800 --> 00:53:50,160
Speaker 4: your day.

1165
00:53:55,760 --> 00:53:59,080
Speaker 6: That was our conversation with David Shore and Burn Hobart

1166
00:53:59,120 --> 00:54:02,360
Speaker 6: recorded a lot. I have at south By Southwest in Austin.

1167
00:54:02,719 --> 00:54:05,960
Speaker 6: I'm Tracy Alloway. You can follow me at Tracy Alloway.

1168
00:54:05,840 --> 00:54:08,760
Speaker 4: And I'm Jill Wisenthal. You can follow me at the Stalwart.

1169
00:54:08,840 --> 00:54:11,719
Speaker 4: Follow our guest Burne Hobart He's at Burne Hobart and

1170
00:54:11,800 --> 00:54:15,680
Speaker 4: David Shore at David Shore. Follow our producers Carmen Rodriguez

1171
00:54:15,719 --> 00:54:18,799
Speaker 4: at Carmen armand dash O Bennett at Dashbot and Kelbrooks

1172
00:54:18,840 --> 00:54:21,279
Speaker 4: at Kelbrooks. And for more odd Laws content, go to

1173
00:54:21,280 --> 00:54:24,200
Speaker 4: Bloomberg dot com slash odd Lots with the daily newsletter

1174
00:54:24,239 --> 00:54:26,399
Speaker 4: and all of our episodes, and you can chat about

1175
00:54:26,440 --> 00:54:28,359
Speaker 4: all of these topics twenty four to seven in our

1176
00:54:28,480 --> 00:54:32,279
Speaker 4: discord Discord dot gg slash odd Lots.

1177
00:54:31,960 --> 00:54:33,440
Speaker 3: And if you enjoy odd lots.

1178
00:54:33,480 --> 00:54:35,800
Speaker 6: If you like it when we record these live episodes,

1179
00:54:35,840 --> 00:54:38,160
Speaker 6: then please leave us a positive review on your favorite

1180
00:54:38,160 --> 00:54:41,600
Speaker 6: podcast platform. And remember, if you are a Bloomberg subscriber,

1181
00:54:41,640 --> 00:54:44,680
Speaker 6: you can listen to all of our episodes absolutely add free.

1182
00:54:44,840 --> 00:54:47,080
Speaker 6: All you need to do is find the Bloomberg channel

1183
00:54:47,120 --> 00:54:49,600
Speaker 6: on Apple Podcasts and follow the instructions there.

1184
00:54:49,960 --> 00:55:07,320
Speaker 3: Thanks for listening

1185
00:55:11,719 --> 00:55:16,040
Speaker 8: Behind it