1
00:00:02,720 --> 00:00:16,720
Speaker 1: Bloomberg Audio Studios, Podcasts, radio News.

2
00:00:18,200 --> 00:00:21,880
Speaker 2: Hello and welcome to another episode of the Odd Lots podcast.

3
00:00:21,960 --> 00:00:24,400
Speaker 3: I'm Jolle Wisenthal and I'm Tracy Alloway.

4
00:00:25,000 --> 00:00:27,880
Speaker 2: You know, not a novel observation here, but I do

5
00:00:27,960 --> 00:00:31,440
Speaker 2: think that if, like in early twenty twenty two or

6
00:00:31,560 --> 00:00:34,960
Speaker 2: late twenty twenty one, you had someone had revealed to

7
00:00:35,040 --> 00:00:37,400
Speaker 2: you all of the amazing things that we could do

8
00:00:37,440 --> 00:00:41,239
Speaker 2: with AI these days, I think you would have expected

9
00:00:41,280 --> 00:00:44,600
Speaker 2: that either the broader economy or society would be more

10
00:00:44,640 --> 00:00:48,640
Speaker 2: different than it has been. Like I am large, you know,

11
00:00:49,320 --> 00:00:52,640
Speaker 2: I think most people's jobs are done roughly the same way.

12
00:00:53,400 --> 00:00:56,000
Speaker 2: Sort of society still seems to operate, although a little

13
00:00:56,040 --> 00:00:58,640
Speaker 2: maybe a little bit worse every day. I don't know,

14
00:00:58,800 --> 00:01:01,400
Speaker 2: But like I just don't think like it's mind blowing

15
00:01:01,400 --> 00:01:04,000
Speaker 2: technology and get by and large, like it hasn't had

16
00:01:04,040 --> 00:01:07,040
Speaker 2: the economic disruption that I think many people would have guessed.

17
00:01:07,080 --> 00:01:09,280
Speaker 3: Maybe I'm cynical on the subject, but I always say,

18
00:01:09,280 --> 00:01:13,360
Speaker 3: never underestimate the human capacity for stasis, I guess, and

19
00:01:13,440 --> 00:01:16,440
Speaker 3: making things much more difficult than they actually need to be,

20
00:01:17,000 --> 00:01:20,840
Speaker 3: and putting up you know, bureaucratic barriers, regulatory barriers and

21
00:01:20,880 --> 00:01:23,720
Speaker 3: things like that. So it's not it's not that surprising

22
00:01:23,760 --> 00:01:25,360
Speaker 3: to me, but it is true. You have a lot

23
00:01:25,400 --> 00:01:27,720
Speaker 3: of economists out there who think there was going to

24
00:01:27,760 --> 00:01:31,080
Speaker 3: be this massive productivity boost, right well, it's certainly.

25
00:01:30,720 --> 00:01:33,039
Speaker 2: Tech people like I don't know, like economists. You know,

26
00:01:33,120 --> 00:01:35,200
Speaker 2: they're like, oh, there were some we're going to have

27
00:01:35,240 --> 00:01:38,279
Speaker 2: a productivity boom. We're raising our estimates from two percent

28
00:01:38,319 --> 00:01:41,520
Speaker 2: too and a quarter percent, and then you have all

29
00:01:41,560 --> 00:01:43,959
Speaker 2: things relative to and then you know, we talked to

30
00:01:44,000 --> 00:01:45,960
Speaker 2: Kathy Wood one time, and I think she what did

31
00:01:46,000 --> 00:01:49,760
Speaker 2: she predict twenty percent real GDP growth for twenty years?

32
00:01:49,880 --> 00:01:54,400
Speaker 2: Or you certainly have these people in Silicon Valley deflationary boom,

33
00:01:54,480 --> 00:01:58,880
Speaker 2: post scarcity any any minute, right now, some real gaps

34
00:01:58,960 --> 00:02:01,480
Speaker 2: I would say, between how lot of economists think about

35
00:02:01,480 --> 00:02:03,760
Speaker 2: some of these things and the numbers, then economists would

36
00:02:03,800 --> 00:02:07,560
Speaker 2: be comfortable using versus maybe literally.

37
00:02:07,360 --> 00:02:10,600
Speaker 3: Everyone else like Kathy would certainly many others. Yeah, you

38
00:02:10,639 --> 00:02:14,280
Speaker 3: know what I think about is how different would your

39
00:02:14,320 --> 00:02:18,680
Speaker 3: career in my career have been had AI existed in

40
00:02:18,800 --> 00:02:21,200
Speaker 3: like two thousand and eight, two thousand and six when

41
00:02:21,240 --> 00:02:22,799
Speaker 3: we were starting to blog.

42
00:02:22,880 --> 00:02:25,760
Speaker 2: Basically, it's a really good question, and I don't know

43
00:02:25,919 --> 00:02:28,480
Speaker 2: the answer because part of me thinks, well, you know,

44
00:02:28,560 --> 00:02:31,240
Speaker 2: that whole path that was like defined my career like

45
00:02:31,280 --> 00:02:34,520
Speaker 2: would not have been there, would not have existed or no. Maybe,

46
00:02:34,560 --> 00:02:37,320
Speaker 2: But on the other hand, maybe in two thousand and

47
00:02:37,320 --> 00:02:40,640
Speaker 2: six I would have been like just like I was

48
00:02:40,680 --> 00:02:45,440
Speaker 2: super early into blogging, super early into experimenting with AI news,

49
00:02:45,480 --> 00:02:47,520
Speaker 2: and it would look different, but maybe I would have

50
00:02:48,000 --> 00:02:49,079
Speaker 2: rowed some different.

51
00:02:48,880 --> 00:02:49,720
Speaker 3: Different but the same.

52
00:02:49,840 --> 00:02:51,799
Speaker 2: Yeah, different but the same, you know, Like I think

53
00:02:51,800 --> 00:02:53,200
Speaker 2: it's a it's pretty hard to tell.

54
00:02:53,440 --> 00:02:56,360
Speaker 3: One thing I know is like back then we were

55
00:02:56,400 --> 00:03:00,240
Speaker 3: writing to optimize Google results, right, that was basic or know,

56
00:03:00,360 --> 00:03:02,919
Speaker 3: like I know social media was disseminating stuff as well,

57
00:03:02,919 --> 00:03:05,400
Speaker 3: but like Google was still the main platform for a

58
00:03:05,400 --> 00:03:08,280
Speaker 3: lot of stuff. I'm sure if we were doing it now,

59
00:03:08,400 --> 00:03:12,840
Speaker 3: we would be optimizing for chat, GPT or perplexity or

60
00:03:12,840 --> 00:03:15,839
Speaker 3: someone like that to actually pick up the content. I mean,

61
00:03:16,000 --> 00:03:17,240
Speaker 3: I think that's the audience.

62
00:03:17,480 --> 00:03:19,600
Speaker 2: Yeah it is. And I think like any especially any

63
00:03:19,680 --> 00:03:22,320
Speaker 2: like commercial publisher, right that's trying to do a scale.

64
00:03:22,360 --> 00:03:26,040
Speaker 2: I think like a sort of niche voice expert could

65
00:03:26,080 --> 00:03:28,240
Speaker 2: still have like their audience that comes to them directly,

66
00:03:28,280 --> 00:03:31,320
Speaker 2: of course, but like it's scale scale for sure every

67
00:03:31,360 --> 00:03:34,400
Speaker 2: publisher is trying to figure this out. Anyway, We really

68
00:03:34,400 --> 00:03:36,839
Speaker 2: do have the perfect guest, someone we should have had

69
00:03:36,840 --> 00:03:39,800
Speaker 2: on the podcast years and years ago. It's almost surprising

70
00:03:39,880 --> 00:03:41,960
Speaker 2: that it's just the first time we've had him on.

71
00:03:42,160 --> 00:03:45,080
Speaker 2: It is crazy, It is crazy, but it is someone

72
00:03:45,160 --> 00:03:47,640
Speaker 2: who I think is like at the intersection of everything

73
00:03:47,680 --> 00:03:51,080
Speaker 2: that we're talking about. Because he's an economist. He knows

74
00:03:51,160 --> 00:03:53,440
Speaker 2: the tech really well. The tech people know him, the

75
00:03:53,480 --> 00:03:55,520
Speaker 2: tech people love him. He may even be one of

76
00:03:55,560 --> 00:03:59,160
Speaker 2: the preferred economists for this. I feel he's like the

77
00:03:59,200 --> 00:04:01,280
Speaker 2: economist that all people in the AI world want to

78
00:04:01,280 --> 00:04:04,880
Speaker 2: get his take. A long time blogger, someone I, you know,

79
00:04:04,960 --> 00:04:06,920
Speaker 2: like both of us sort of gave up on mike blogging.

80
00:04:06,960 --> 00:04:08,840
Speaker 2: Although we have our newsletter. It's like close enough, but

81
00:04:08,880 --> 00:04:11,440
Speaker 2: it's different than blogging. Someone who's stuck with the medium

82
00:04:11,480 --> 00:04:14,040
Speaker 2: for a long time. One of the true original econ

83
00:04:14,120 --> 00:04:17,000
Speaker 2: bloggers that I've been reading for over twenty years. We're

84
00:04:17,040 --> 00:04:19,239
Speaker 2: going to be speaking with Tyler Cowen. Here's the host

85
00:04:19,240 --> 00:04:22,760
Speaker 2: of the Conversations with Tyler podcast. He also, obviously is

86
00:04:22,760 --> 00:04:26,360
Speaker 2: one of the two bloggers that the famous Marginal Revolution blog.

87
00:04:26,880 --> 00:04:33,279
Speaker 2: Economics professor gmu appreciator of ethnic foods all around the

88
00:04:33,720 --> 00:04:38,320
Speaker 2: DC Northern Virginia, Maryland area. Someone known on the internet

89
00:04:38,360 --> 00:04:40,200
Speaker 2: for a long time, Tyler, thank you so much for

90
00:04:40,520 --> 00:04:43,440
Speaker 2: coming on odd Laws. Really thrilled to have you here.

91
00:04:44,040 --> 00:04:45,400
Speaker 4: Hello, happy to be here.

92
00:04:45,800 --> 00:04:49,560
Speaker 2: Amazing. What do you think about my initial assessment? How

93
00:04:49,600 --> 00:04:51,719
Speaker 2: fair do you think it is that, Like, had you

94
00:04:51,880 --> 00:04:55,400
Speaker 2: known in twenty twenty one how powerful these tools would be,

95
00:04:55,880 --> 00:04:59,200
Speaker 2: maybe we'd be a bit surprised that by and large,

96
00:04:59,279 --> 00:05:01,560
Speaker 2: like busines, and this seems to more or less run

97
00:05:01,600 --> 00:05:03,479
Speaker 2: the same, I'm.

98
00:05:03,320 --> 00:05:06,040
Speaker 4: Not surprised at all. So what I see right now

99
00:05:06,160 --> 00:05:09,080
Speaker 4: is people using AI as an add on to their

100
00:05:09,120 --> 00:05:12,240
Speaker 4: pre existing work routines. Oh, you need to write a memo,

101
00:05:12,480 --> 00:05:14,520
Speaker 4: you ask AI how to do it. You need to

102
00:05:14,520 --> 00:05:17,000
Speaker 4: write a column, you ask AI to proofread it or

103
00:05:17,040 --> 00:05:20,200
Speaker 4: fact check it. And that works great, But those are

104
00:05:20,279 --> 00:05:23,239
Speaker 4: marginal gains. What we really need to see a major

105
00:05:23,320 --> 00:05:27,320
Speaker 4: impact is new organizations built around AI, and those will

106
00:05:27,360 --> 00:05:30,839
Speaker 4: be startups. They will come only slowly. It will take

107
00:05:31,400 --> 00:05:34,720
Speaker 4: twenty or more years before they really transform the economy.

108
00:05:35,440 --> 00:05:37,559
Speaker 4: And in the meantime, it's a whole bunch of add ons,

109
00:05:37,600 --> 00:05:40,159
Speaker 4: which are fun and fine, but that's why I think

110
00:05:40,160 --> 00:05:40,640
Speaker 4: it's slow.

111
00:05:40,960 --> 00:05:43,560
Speaker 2: What can you tell us about the history of technology

112
00:05:44,080 --> 00:05:49,040
Speaker 2: such that legacy organizations that existed prior to the invention

113
00:05:49,440 --> 00:05:54,040
Speaker 2: of maybe potentially revolutionary new technology have a hard time

114
00:05:54,760 --> 00:05:57,320
Speaker 2: massively changing their workflows.

115
00:05:58,360 --> 00:06:01,640
Speaker 4: Well, you can take even very simple examples. So Toyota

116
00:06:01,680 --> 00:06:05,560
Speaker 4: starts competing with General Motors in the nineteen seventies, General

117
00:06:05,600 --> 00:06:08,919
Speaker 4: Motors is paralyzed. It cannot really come back and adopt

118
00:06:08,920 --> 00:06:11,800
Speaker 4: the new and superior Toyota methods, and those are not

119
00:06:12,120 --> 00:06:15,960
Speaker 4: really that different compared to say AI. So there's just plenty,

120
00:06:16,240 --> 00:06:20,320
Speaker 4: plenty examples. Old mainstream media could not cope very well

121
00:06:20,360 --> 00:06:23,080
Speaker 4: with the Internet. There are exceptions like The New York Times.

122
00:06:23,800 --> 00:06:27,600
Speaker 4: Odd Lot's podcast would be another thank you, But it's

123
00:06:27,640 --> 00:06:30,680
Speaker 4: the norm. So we're seeing it again with AI. And

124
00:06:30,720 --> 00:06:33,400
Speaker 4: again you need a complete turnover of who and what

125
00:06:33,560 --> 00:06:35,600
Speaker 4: is doing business for it to really matter on a

126
00:06:35,640 --> 00:06:36,200
Speaker 4: big scale.

127
00:06:37,560 --> 00:06:40,520
Speaker 3: If you need that complete turnover, and you need some

128
00:06:40,640 --> 00:06:43,920
Speaker 3: time for AI to become fully embedded in a business

129
00:06:43,960 --> 00:06:47,280
Speaker 3: model or for a business model to form around AI,

130
00:06:48,120 --> 00:06:50,640
Speaker 3: what industry or what part of the economy would you

131
00:06:50,720 --> 00:06:53,600
Speaker 3: expect it, I guess to show up first in that

132
00:06:53,680 --> 00:06:55,280
Speaker 3: sort of revolutionary way.

133
00:06:56,680 --> 00:06:59,160
Speaker 4: Well, we have obvious data on this, and it's programming.

134
00:06:59,520 --> 00:07:02,640
Speaker 4: You will hear people who do programming claim that say

135
00:07:02,680 --> 00:07:05,320
Speaker 4: eighty percent of the work is now done by AIS.

136
00:07:05,520 --> 00:07:08,520
Speaker 4: I suspect that's an overstatement, but there's no doubt at

137
00:07:08,560 --> 00:07:11,680
Speaker 4: all that there's simply a lot of programming already done

138
00:07:11,680 --> 00:07:16,360
Speaker 4: by AIS. When you have low fixed costs, a competitive sector,

139
00:07:16,640 --> 00:07:21,320
Speaker 4: immediate feedback, you know, the revenue has to flow programming.

140
00:07:21,400 --> 00:07:24,760
Speaker 4: And also New York City finance, there's been quantston finance

141
00:07:24,840 --> 00:07:28,400
Speaker 4: for a long time. Those quants are now you could say,

142
00:07:28,440 --> 00:07:30,880
Speaker 4: more AI equipped than they used to be, and those

143
00:07:30,960 --> 00:07:33,040
Speaker 4: areas already are being revolutionized.

144
00:07:33,560 --> 00:07:37,080
Speaker 2: I've heard them and got some pitches to do episodes,

145
00:07:37,120 --> 00:07:39,240
Speaker 2: which we should do at some point. But I've heard

146
00:07:39,280 --> 00:07:43,600
Speaker 2: about like some law firms that are being like new

147
00:07:43,720 --> 00:07:46,720
Speaker 2: AI law firms where there are lawyers, et cetera. But

148
00:07:46,800 --> 00:07:49,920
Speaker 2: from the very ground up, the idea is perhaps there

149
00:07:50,000 --> 00:07:54,280
Speaker 2: is some way to just get way more productivity if

150
00:07:54,320 --> 00:07:57,400
Speaker 2: they start from the very beginning with some combination of

151
00:07:57,680 --> 00:08:00,840
Speaker 2: lawyers plus AI models. It seems like that could be

152
00:08:00,880 --> 00:08:03,440
Speaker 2: the kind of thing where maybe the legacy law firms

153
00:08:03,640 --> 00:08:07,320
Speaker 2: are seeing some productivity gains from AI. There's probably some

154
00:08:07,360 --> 00:08:10,280
Speaker 2: evidence you could find but that a new one with

155
00:08:10,320 --> 00:08:14,560
Speaker 2: a totally different approach could deliver that big productivity boost

156
00:08:14,560 --> 00:08:16,280
Speaker 2: that actually ends up changing the industry.

157
00:08:17,160 --> 00:08:20,000
Speaker 4: It's already the case, say, mid tier associates are much

158
00:08:20,080 --> 00:08:23,360
Speaker 4: less needed. But there's one big problem with law in particular,

159
00:08:23,840 --> 00:08:27,080
Speaker 4: and that is the way large language models work. Now

160
00:08:27,160 --> 00:08:30,040
Speaker 4: you have to send your queries somewhere else. You can't

161
00:08:30,080 --> 00:08:32,440
Speaker 4: just own and control and hold the whole thing on

162
00:08:32,480 --> 00:08:34,960
Speaker 4: your hard drive. Now. I think within a few years

163
00:08:35,040 --> 00:08:38,600
Speaker 4: time that will be very different. But until then, major

164
00:08:38,720 --> 00:08:42,480
Speaker 4: law firms are extremely skittish about just typing in their

165
00:08:42,559 --> 00:08:45,240
Speaker 4: questions and sending it, you know, to San Francisco. I

166
00:08:45,280 --> 00:08:47,480
Speaker 4: don't actually think there's a risk, but when you think

167
00:08:47,480 --> 00:08:50,680
Speaker 4: about how fiduciary responsibility works, they just don't want to

168
00:08:50,720 --> 00:08:51,000
Speaker 4: do it.

169
00:09:07,559 --> 00:09:11,280
Speaker 3: Talk more about I guess privacy concerns and regulation, because

170
00:09:11,320 --> 00:09:13,880
Speaker 3: this seems to be an area where if you are

171
00:09:13,920 --> 00:09:16,280
Speaker 3: in a heavily regulated industry, or if you're just in

172
00:09:16,320 --> 00:09:18,680
Speaker 3: an industry that tends to be full of paranoid people

173
00:09:18,880 --> 00:09:22,120
Speaker 3: like lawyers, it does seem like there's going to be

174
00:09:22,120 --> 00:09:25,800
Speaker 3: a natural tendency to be very very cautious. When it

175
00:09:25,840 --> 00:09:28,000
Speaker 3: comes to sharing data with AI, you're going to be

176
00:09:28,040 --> 00:09:30,760
Speaker 3: worried about actual data ownership the queries that you're sending

177
00:09:30,840 --> 00:09:34,240
Speaker 3: to San Francisco, as you say, are those industries just

178
00:09:34,320 --> 00:09:36,240
Speaker 3: inevitably going to be slow to adapt.

179
00:09:37,880 --> 00:09:40,880
Speaker 4: They'll be slow to adapt again until the point where

180
00:09:40,920 --> 00:09:44,400
Speaker 4: they just control their own model and they hold it

181
00:09:44,440 --> 00:09:46,640
Speaker 4: within the firm and no one else really can get

182
00:09:46,679 --> 00:09:52,160
Speaker 4: at it. So what you need is cheaper models where

183
00:09:52,280 --> 00:09:54,840
Speaker 4: a law firm can afford to have its own model.

184
00:09:54,880 --> 00:09:56,920
Speaker 4: And I think that's a few years away. It's not

185
00:09:57,000 --> 00:09:59,400
Speaker 4: a very long time away, but it won't come in

186
00:09:59,440 --> 00:10:02,960
Speaker 4: six months. Sam Altman I just did a podcast with

187
00:10:03,080 --> 00:10:07,240
Speaker 4: him and he said a privacy problem is AI queries

188
00:10:07,360 --> 00:10:09,920
Speaker 4: or subject to subpoena, and he thinks they should have

189
00:10:09,960 --> 00:10:13,200
Speaker 4: as much protection as say, your conversation with your lawyer

190
00:10:13,320 --> 00:10:15,920
Speaker 4: or your doctor or your therapist. I think that's a

191
00:10:15,920 --> 00:10:19,800
Speaker 4: good idea, but that hasn't happened yet, and until it happens,

192
00:10:19,920 --> 00:10:21,880
Speaker 4: or you get the whole thing on your own hard drive,

193
00:10:22,280 --> 00:10:24,560
Speaker 4: progress in law is going to be slow. But once

194
00:10:24,600 --> 00:10:27,480
Speaker 4: the progress comes, that's one of the areas where I

195
00:10:27,480 --> 00:10:30,600
Speaker 4: think AI has the most promise it's just very very

196
00:10:30,640 --> 00:10:34,040
Speaker 4: good at mastering a large corpus of text and organizing

197
00:10:34,040 --> 00:10:34,480
Speaker 4: it for you.

198
00:10:35,240 --> 00:10:37,880
Speaker 2: It is interesting, isn't it. Where it's like, Okay, if

199
00:10:37,880 --> 00:10:40,359
Speaker 2: you are my lawyer, you and I could have a

200
00:10:40,400 --> 00:10:44,280
Speaker 2: conversation and it would be not it would not be uh.

201
00:10:44,240 --> 00:10:44,960
Speaker 3: It'd be privileged.

202
00:10:45,040 --> 00:10:48,960
Speaker 2: Yeah, it would be privileged. Or you know, one can

203
00:10:49,000 --> 00:10:52,640
Speaker 2: have conversations with anyone, and as long as you're doing

204
00:10:52,679 --> 00:10:55,160
Speaker 2: it on the phone, or if you're doing it person

205
00:10:55,200 --> 00:10:59,079
Speaker 2: to person, it's much more difficult to get that, you know,

206
00:10:59,200 --> 00:11:02,720
Speaker 2: in discover or I always think about this with you know,

207
00:11:02,800 --> 00:11:06,360
Speaker 2: public public records of employees that sometimes you can get

208
00:11:06,400 --> 00:11:09,240
Speaker 2: their emails through Floyers, et cetera, but you can't get

209
00:11:09,240 --> 00:11:13,040
Speaker 2: the cons content of conversations. It does feel I hadn't

210
00:11:13,040 --> 00:11:16,520
Speaker 2: really thought about this dynamic though, that when you're using

211
00:11:16,559 --> 00:11:19,200
Speaker 2: an AI it is sort of like a conversation and

212
00:11:19,280 --> 00:11:22,880
Speaker 2: yet sort of from an evidentiary basis, it would be

213
00:11:23,000 --> 00:11:24,080
Speaker 2: much more like an email.

214
00:11:25,200 --> 00:11:27,320
Speaker 4: I think when it comes to medical issues, there are

215
00:11:27,360 --> 00:11:30,560
Speaker 4: many more people willing to share their data. Not everyone

216
00:11:30,640 --> 00:11:34,040
Speaker 4: by any means. Some medical conditions are secretor people just

217
00:11:34,080 --> 00:11:36,960
Speaker 4: don't want others to know. But I see many, many

218
00:11:37,000 --> 00:11:39,240
Speaker 4: people I know typing in all kinds of things about

219
00:11:39,240 --> 00:11:42,480
Speaker 4: their medical history to say GPT five and getting what

220
00:11:42,559 --> 00:11:45,360
Speaker 4: are on the whole very good answers. It's like medical

221
00:11:45,360 --> 00:11:49,199
Speaker 4: diagnosis for free spreading now to the whole world. A

222
00:11:49,240 --> 00:11:51,840
Speaker 4: lot of countries where people just don't even have access

223
00:11:51,880 --> 00:11:54,920
Speaker 4: to good doctors at all, and I think that will

224
00:11:55,080 --> 00:11:58,400
Speaker 4: be important more quickly than the law innovations.

225
00:11:59,600 --> 00:12:01,719
Speaker 3: Just as a thought experiment, what does all of this

226
00:12:01,880 --> 00:12:04,120
Speaker 3: mean for insurers? Because I kind of think, you know,

227
00:12:04,160 --> 00:12:05,920
Speaker 3: I think about a bunch of people typing in their

228
00:12:05,960 --> 00:12:09,600
Speaker 3: medical information. I think about basically the explosion in data

229
00:12:09,640 --> 00:12:11,959
Speaker 3: that we have nowadays, and it seems like the insurance

230
00:12:12,000 --> 00:12:15,920
Speaker 3: industry would be one place that would really benefit from

231
00:12:15,920 --> 00:12:19,480
Speaker 3: all this trove of additional information if they could access

232
00:12:19,520 --> 00:12:21,120
Speaker 3: it well.

233
00:12:21,120 --> 00:12:23,480
Speaker 4: This is one of my worries about AI in general.

234
00:12:23,679 --> 00:12:27,240
Speaker 4: I'm quite positive on what's happening, But as insurers get

235
00:12:27,280 --> 00:12:30,720
Speaker 4: better and better information on their customers, this is just

236
00:12:30,760 --> 00:12:32,880
Speaker 4: through big data more generally, it doesn't have to be

237
00:12:32,960 --> 00:12:36,520
Speaker 4: current large language models. They know exactly how to write

238
00:12:36,520 --> 00:12:38,920
Speaker 4: the risk and how to price the premium, and in

239
00:12:38,960 --> 00:12:41,880
Speaker 4: a sense for the buyer, it's not insurance anymore. So,

240
00:12:41,920 --> 00:12:43,760
Speaker 4: if we know your house is going to burn down

241
00:12:43,800 --> 00:12:46,200
Speaker 4: with high probability and you have to pay the super

242
00:12:46,280 --> 00:12:49,240
Speaker 4: high premium, you don't really have the benefits of the insurance.

243
00:12:49,600 --> 00:12:54,000
Speaker 4: So some insurance markets might unravel if through big data

244
00:12:54,480 --> 00:12:57,120
Speaker 4: the insurers learn too much about what's likely to happen.

245
00:12:57,320 --> 00:13:02,760
Speaker 2: Economists seem to be very consistent about the effects of

246
00:13:02,800 --> 00:13:06,560
Speaker 2: technology on labor demand, which is that in the end

247
00:13:06,600 --> 00:13:09,240
Speaker 2: it washes out right, some people, Okay, there's a disruption,

248
00:13:09,400 --> 00:13:11,520
Speaker 2: but I'm gonna save money because I use the I.

249
00:13:11,720 --> 00:13:13,679
Speaker 2: But that means I have more spending power, and then

250
00:13:13,720 --> 00:13:16,199
Speaker 2: I'm gonna buy something else that I wouldn't have bought

251
00:13:16,280 --> 00:13:20,120
Speaker 2: had it not been for paying wages, and then that'll

252
00:13:20,160 --> 00:13:22,880
Speaker 2: create demand for labor elsewhere. And so in the end,

253
00:13:22,880 --> 00:13:27,160
Speaker 2: the idea that you could really have tech driven unemployment

254
00:13:27,240 --> 00:13:31,600
Speaker 2: at scale that is not transitory or not temporary is hard.

255
00:13:31,760 --> 00:13:35,680
Speaker 2: A lot of economists seem to be intuitively skeptical of

256
00:13:35,720 --> 00:13:39,080
Speaker 2: this idea. Whereas you have people in the ifields, fifty

257
00:13:39,080 --> 00:13:40,880
Speaker 2: percent of the people aren't gonna have jobs we need

258
00:13:41,000 --> 00:13:44,480
Speaker 2: ubi otherwise there's gonna be a permanent underclass. Could there

259
00:13:44,520 --> 00:13:48,560
Speaker 2: be something different about AI such that it doesn't have

260
00:13:48,720 --> 00:13:51,760
Speaker 2: the same labor market effects that past technologies have had.

261
00:13:52,720 --> 00:13:56,000
Speaker 4: I would say, you understand me. Well, So the energy sector,

262
00:13:56,080 --> 00:13:59,000
Speaker 4: there's going to be a lot of new jobs taking

263
00:13:59,040 --> 00:14:03,319
Speaker 4: care of older people. I think as AI produces more potential,

264
00:14:03,360 --> 00:14:06,640
Speaker 4: medical innovations will need to test them, So the biomedical

265
00:14:06,679 --> 00:14:09,960
Speaker 4: sector testing clinical trials, there'll be a lot new jobs.

266
00:14:10,760 --> 00:14:14,000
Speaker 4: I'm not worried about mass unemployment, and most economists are not,

267
00:14:14,440 --> 00:14:16,640
Speaker 4: and I agree with their perspectives, which I think you

268
00:14:16,679 --> 00:14:17,800
Speaker 4: outlined pretty clearly.

269
00:14:18,400 --> 00:14:21,000
Speaker 2: What do you say, though, because I have a feeling

270
00:14:21,120 --> 00:14:23,920
Speaker 2: that when you're out in San Francisco, they don't see

271
00:14:23,960 --> 00:14:27,160
Speaker 2: it that way, and they talk a lot, and some

272
00:14:27,240 --> 00:14:29,600
Speaker 2: of them are more in you know, UBI talk and

273
00:14:29,880 --> 00:14:32,560
Speaker 2: permanent underclass talk and all of this stuff. Do they

274
00:14:32,640 --> 00:14:35,080
Speaker 2: see your perspective when you make this case? What do

275
00:14:35,120 --> 00:14:37,360
Speaker 2: they say or what are they What are a lot

276
00:14:37,400 --> 00:14:39,360
Speaker 2: of them seem to be missing about the logic that

277
00:14:39,400 --> 00:14:40,400
Speaker 2: you spell out.

278
00:14:40,880 --> 00:14:43,200
Speaker 4: Well, I think more and more they're coming around to

279
00:14:43,240 --> 00:14:46,560
Speaker 4: the economist's point of view. So Andre Carpathi, who was

280
00:14:46,600 --> 00:14:49,720
Speaker 4: at open AI, you know, in the most important years,

281
00:14:50,080 --> 00:14:53,400
Speaker 4: he just did a podcast saying he thinks adjustment will

282
00:14:53,400 --> 00:14:56,120
Speaker 4: be slow. Things will be fine, will grow at two

283
00:14:56,200 --> 00:14:59,600
Speaker 4: points something percent. There won't be mass unemployment, and you

284
00:14:59,640 --> 00:15:02,200
Speaker 4: wouldn't I have heard that, say two years ago. But

285
00:15:02,240 --> 00:15:04,560
Speaker 4: I think as people see the models rolling out, and

286
00:15:04,600 --> 00:15:07,160
Speaker 4: as you mentioned well, the real world impact it is

287
00:15:07,200 --> 00:15:12,160
Speaker 4: stretched out in time, right, it's not all immediate. Earlier

288
00:15:12,200 --> 00:15:14,280
Speaker 4: on people had more the sense that AI was a

289
00:15:14,360 --> 00:15:16,720
Speaker 4: kind of god box that you just talked to it

290
00:15:16,840 --> 00:15:19,800
Speaker 4: and it can magically do anything and convert that into

291
00:15:19,840 --> 00:15:22,360
Speaker 4: results in the real world. But if you think about

292
00:15:22,360 --> 00:15:25,480
Speaker 4: your actual job, even if it's a highly intellectual job,

293
00:15:25,800 --> 00:15:28,560
Speaker 4: so much of what you do is the interaction between

294
00:15:28,560 --> 00:15:32,080
Speaker 4: your intellect, your physical presence, your interactions with others, your travel,

295
00:15:32,640 --> 00:15:36,320
Speaker 4: many other things. And until we get to some far,

296
00:15:36,480 --> 00:15:40,120
Speaker 4: far off world where the robots are perfect copies of you,

297
00:15:40,360 --> 00:15:43,800
Speaker 4: which I don't think ever will come, jobs will be fine,

298
00:15:43,840 --> 00:15:46,840
Speaker 4: but they will change a lot, and I'm actually worried

299
00:15:46,840 --> 00:15:50,240
Speaker 4: about who will be the biggest losers. I think poor

300
00:15:50,280 --> 00:15:54,040
Speaker 4: people will do great, the very wealthy will do well,

301
00:15:54,480 --> 00:15:56,840
Speaker 4: but people who are sort of upper upper middle class

302
00:15:57,400 --> 00:16:00,800
Speaker 4: will find this automatic ticket to a law or consulting

303
00:16:00,920 --> 00:16:03,920
Speaker 4: job that assured they would be upper upper middle class

304
00:16:03,960 --> 00:16:06,120
Speaker 4: for the rest of their lives. I think a lot

305
00:16:06,160 --> 00:16:07,600
Speaker 4: of that is going away already.

306
00:16:08,400 --> 00:16:11,480
Speaker 3: I think also no one would have expected plastic surgery,

307
00:16:11,520 --> 00:16:13,800
Speaker 3: I guess, to be a beneficiary of the AI revolution.

308
00:16:13,960 --> 00:16:15,800
Speaker 3: But if you think that what's going to matter in

309
00:16:15,840 --> 00:16:19,720
Speaker 3: the future is like your personal presence and your network

310
00:16:19,760 --> 00:16:21,960
Speaker 3: of social contacts, then I guess we should all be

311
00:16:22,040 --> 00:16:29,160
Speaker 3: working on on our looks Maxing, Yeah, absolutely right, charisma. Okay, noted,

312
00:16:29,320 --> 00:16:32,480
Speaker 3: everyone work on their respective charisma. One thing I was

313
00:16:32,520 --> 00:16:36,160
Speaker 3: wondering is the impact of AI on public finances. I

314
00:16:36,160 --> 00:16:39,200
Speaker 3: am not very good at tax policy. Joe knows this

315
00:16:39,240 --> 00:16:43,840
Speaker 3: because I've complained about taxes to him repeatedly. But I'm thinking,

316
00:16:46,160 --> 00:16:49,880
Speaker 3: but if you're thinking about where the value add of

317
00:16:49,960 --> 00:16:52,360
Speaker 3: AI actually shows up in the economy, So you know,

318
00:16:52,440 --> 00:16:55,240
Speaker 3: presumably you got a productivity boost. We're not entirely sure

319
00:16:55,240 --> 00:16:57,280
Speaker 3: how much that's going to be. But where does that

320
00:16:57,320 --> 00:17:01,880
Speaker 3: additional output actually show up in terms of revenue for governments?

321
00:17:01,880 --> 00:17:04,520
Speaker 3: How is that collected? And how would you expect the

322
00:17:04,600 --> 00:17:06,439
Speaker 3: distribution to vary across the world.

323
00:17:07,520 --> 00:17:10,440
Speaker 4: Well, I think in the United States medium term, they'll

324
00:17:10,480 --> 00:17:14,400
Speaker 4: just be much much more. Healthcare will be new drugs,

325
00:17:14,480 --> 00:17:18,040
Speaker 4: new medical devices, will have to test all these things,

326
00:17:18,080 --> 00:17:20,800
Speaker 4: will have to produce them, and that will be It

327
00:17:20,840 --> 00:17:23,239
Speaker 4: was already the case that those sectors were growing, but

328
00:17:23,320 --> 00:17:26,959
Speaker 4: that growth will be accelerated. So that's where I think, God,

329
00:17:27,000 --> 00:17:29,760
Speaker 4: they'll be the biggest difference longer term, and people will

330
00:17:29,760 --> 00:17:33,480
Speaker 4: live longer because we'll fix at least partially various diseases

331
00:17:33,480 --> 00:17:36,280
Speaker 4: and maladies. So if you live to be ninety four

332
00:17:36,800 --> 00:17:39,280
Speaker 4: across a lifetime, you spend way more on healthcare than

333
00:17:39,320 --> 00:17:41,640
Speaker 4: if you live to be seventy seven. And that's yet

334
00:17:41,680 --> 00:17:45,080
Speaker 4: further growth for the healthcare sector. But some things like

335
00:17:45,200 --> 00:17:50,040
Speaker 4: medical diagnosis that's already very cheap, like you know, a

336
00:17:50,080 --> 00:17:54,520
Speaker 4: good large language model probably outperforms your current doctor at

337
00:17:54,600 --> 00:17:56,800
Speaker 4: least if you type in what's wrong with you properly.

338
00:17:57,480 --> 00:18:01,399
Speaker 3: But does that additional productivity or or the output generated

339
00:18:01,440 --> 00:18:04,600
Speaker 3: by AI, does that actually show up an additional taxation

340
00:18:04,680 --> 00:18:05,320
Speaker 3: for the government.

341
00:18:06,520 --> 00:18:11,119
Speaker 4: Well, the healthcare sector generates an enormous amount of taxation revenue.

342
00:18:11,280 --> 00:18:13,639
Speaker 4: I do think we'll have some sectors that maybe just

343
00:18:13,680 --> 00:18:16,639
Speaker 4: become free in the same way that Wikipedia is free.

344
00:18:17,200 --> 00:18:19,800
Speaker 4: So I could imagine say ten or twenty percent of

345
00:18:19,840 --> 00:18:23,440
Speaker 4: the music sector is music you create at home using

346
00:18:23,480 --> 00:18:26,240
Speaker 4: your own AI and it's a customized song for you,

347
00:18:26,600 --> 00:18:29,400
Speaker 4: and maybe you paid a subscription for the service. But

348
00:18:29,480 --> 00:18:33,520
Speaker 4: rather than spending more money on Spotify or a streaming service,

349
00:18:33,760 --> 00:18:36,960
Speaker 4: you just build the music and that's a partial substitute

350
00:18:37,119 --> 00:18:40,000
Speaker 4: for some human created music. I don't think human created

351
00:18:40,040 --> 00:18:42,800
Speaker 4: music will go away at all. People want to enjoy

352
00:18:42,920 --> 00:18:45,560
Speaker 4: the human touch, the feeling that you're a fan of Taller,

353
00:18:45,560 --> 00:18:48,160
Speaker 4: Swift or whatever. But there's going to be a lot

354
00:18:48,200 --> 00:18:51,440
Speaker 4: of AI generated music and art and many other areas,

355
00:18:51,680 --> 00:18:54,760
Speaker 4: and some of it will be free. But that's not

356
00:18:54,840 --> 00:18:58,600
Speaker 4: a problem from a revenue standpoint. So instead of spending

357
00:18:58,640 --> 00:19:02,600
Speaker 4: money on safe buying a picture, you create one digitally

358
00:19:02,640 --> 00:19:05,040
Speaker 4: at home with your AI. You'll spend that money on

359
00:19:05,080 --> 00:19:05,640
Speaker 4: something else.

360
00:19:06,480 --> 00:19:09,119
Speaker 3: I got to ask, now, are you a swiftie.

361
00:19:09,440 --> 00:19:13,959
Speaker 4: No, it's worrying for me. It's a little too predictable,

362
00:19:14,400 --> 00:19:17,520
Speaker 4: so I have to say I'm not, but I'm glad

363
00:19:17,600 --> 00:19:19,440
Speaker 4: other people are. Let's put it that way.

364
00:19:19,600 --> 00:19:21,880
Speaker 2: Do you have a theory let's talk, let's just talk

365
00:19:21,920 --> 00:19:24,560
Speaker 2: about music. Do you have like a theory of the

366
00:19:24,600 --> 00:19:29,159
Speaker 2: swift phenomenon? Like, is there a reason? What is it?

367
00:19:29,200 --> 00:19:32,000
Speaker 2: Because it's just so, doesn't I don't know what's your

368
00:19:32,359 --> 00:19:36,159
Speaker 2: what's your meta take on the tailor, the swifties and

369
00:19:36,280 --> 00:19:37,440
Speaker 2: culture and society.

370
00:19:38,600 --> 00:19:41,520
Speaker 4: Well, it's super polished and because of the Internet, the

371
00:19:41,640 --> 00:19:44,640
Speaker 4: very biggest of celebrities can be much bigger than before,

372
00:19:45,160 --> 00:19:47,960
Speaker 4: and someone will fill a few of those slots. But

373
00:19:48,000 --> 00:19:52,159
Speaker 4: I think also how she presents herself. She has the

374
00:19:52,240 --> 00:19:57,520
Speaker 4: guise of being attractive without feeling threatening to other women. Uh,

375
00:19:57,560 --> 00:20:00,560
Speaker 4: and there's something all American about her and quite generic.

376
00:20:00,840 --> 00:20:03,600
Speaker 4: She doesn't rule out the fandom of that many people

377
00:20:03,800 --> 00:20:06,520
Speaker 4: m hm. And she's the one who's filled that slot.

378
00:20:06,600 --> 00:20:10,360
Speaker 4: She's been brilliant at managing her career and seems to

379
00:20:10,359 --> 00:20:12,520
Speaker 4: just stick at it and has an incredible work ethic.

380
00:20:12,560 --> 00:20:15,679
Speaker 4: People say the shows are incredible. As more and more

381
00:20:15,760 --> 00:20:18,640
Speaker 4: life goes online, who can give a good show? It's

382
00:20:18,640 --> 00:20:21,560
Speaker 4: the charisma and looks. Point Well, she seems a pleasant

383
00:20:21,600 --> 00:20:23,600
Speaker 4: that I've never been to one, but I hear plenty

384
00:20:23,600 --> 00:20:26,960
Speaker 4: of reports. And you put all that together and she's,

385
00:20:27,119 --> 00:20:28,919
Speaker 4: you know, the megastar of the music world.

386
00:20:29,359 --> 00:20:32,800
Speaker 2: Do you think like culture, Like there's this popular idea

387
00:20:32,800 --> 00:20:34,800
Speaker 2: that culture is sort of dead, and I do think

388
00:20:34,800 --> 00:20:38,520
Speaker 2: that is probably overstated, But you know, you look at

389
00:20:38,560 --> 00:20:41,760
Speaker 2: movies and people have people have observed this for a

390
00:20:41,800 --> 00:20:45,040
Speaker 2: long time. It's just rehashes of franchises that have been

391
00:20:45,080 --> 00:20:47,679
Speaker 2: around for thirty years. And I think if you look

392
00:20:47,720 --> 00:20:52,000
Speaker 2: at Spotify's streaming, there's still this overwhelming tyranny of the

393
00:20:52,000 --> 00:20:56,560
Speaker 2: boomer rock et cetera, and this feeling that culture in

394
00:20:56,680 --> 00:20:59,280
Speaker 2: many respects is that's rehash that is very hard for

395
00:20:59,359 --> 00:21:02,160
Speaker 2: new things to to break out. I mean, Taylor Swift

396
00:21:02,200 --> 00:21:07,000
Speaker 2: at this point is decades old phenomenon. Is that real

397
00:21:07,160 --> 00:21:09,520
Speaker 2: in your view? Or is that just people in the

398
00:21:09,560 --> 00:21:13,479
Speaker 2: pundit class who've been lazy and not discovering new things

399
00:21:13,520 --> 00:21:15,560
Speaker 2: and not actually putting in the effort because they're not

400
00:21:15,600 --> 00:21:17,320
Speaker 2: young anymore and they're not going out and they say

401
00:21:17,320 --> 00:21:18,639
Speaker 2: nobody listens to new music.

402
00:21:19,040 --> 00:21:20,320
Speaker 3: Joe, are you just talking about yourself?

403
00:21:20,600 --> 00:21:22,200
Speaker 2: I have talked about I'm trying. I'm doing a little

404
00:21:22,240 --> 00:21:24,639
Speaker 2: introspection here. Is this just me because I don't go

405
00:21:24,680 --> 00:21:26,679
Speaker 2: to shows like I did when I was twenty or

406
00:21:26,800 --> 00:21:27,639
Speaker 2: is something changed?

407
00:21:28,680 --> 00:21:30,480
Speaker 4: But a lot of it is the pundit class. So

408
00:21:30,520 --> 00:21:33,480
Speaker 4: you look at movies, I think it's perfectly correct to

409
00:21:33,520 --> 00:21:36,960
Speaker 4: say the most popular movies today are pretty dreadful, and

410
00:21:37,000 --> 00:21:38,800
Speaker 4: it used to be the case that the most popular

411
00:21:38,840 --> 00:21:41,680
Speaker 4: movies were God, The Godfather in Star Wars. Yeah, that's

412
00:21:41,680 --> 00:21:44,480
Speaker 4: a big change. But if you look at movie making

413
00:21:44,520 --> 00:21:47,680
Speaker 4: around the world and in a given year list say

414
00:21:47,720 --> 00:21:51,160
Speaker 4: the twenty five best movies from all places, which may

415
00:21:51,200 --> 00:21:53,760
Speaker 4: not even come to your multiplex every year, you have

416
00:21:53,800 --> 00:21:56,760
Speaker 4: an incredible list. I don't think they're worse than the

417
00:21:56,800 --> 00:22:00,320
Speaker 4: movies of earlier times. I do think mainstream hollywo is

418
00:22:00,400 --> 00:22:04,520
Speaker 4: much worse. So in many areas you just have quality

419
00:22:04,560 --> 00:22:05,760
Speaker 4: moving more into unises.

420
00:22:06,160 --> 00:22:07,600
Speaker 2: By the way, I just want to say, the first

421
00:22:07,600 --> 00:22:10,760
Speaker 2: time I ever encountered your work, prior to even having

422
00:22:10,800 --> 00:22:16,040
Speaker 2: stumbled across Marginal Revolution, was at the bookstore in Austin

423
00:22:16,320 --> 00:22:19,919
Speaker 2: finding a copy of In Praise of Commercial Culture, and

424
00:22:20,119 --> 00:22:23,639
Speaker 2: I just feel like that book is held up so well.

425
00:22:23,800 --> 00:22:26,399
Speaker 2: I mean in the specific sense that there is so

426
00:22:26,520 --> 00:22:30,119
Speaker 2: much mass culture these days, whether it's high end Netflix

427
00:22:30,480 --> 00:22:33,720
Speaker 2: TV shows, et cetera, whether it's A twenty four films,

428
00:22:34,000 --> 00:22:37,080
Speaker 2: whether it's Beyonce or Taylor Swift or some of these

429
00:22:37,080 --> 00:22:41,439
Speaker 2: other big names who are simultaneously incredibly popular. And I

430
00:22:41,480 --> 00:22:44,479
Speaker 2: get that you're not as swifty, neither mine anymore, but

431
00:22:45,359 --> 00:22:49,720
Speaker 2: people take this yeah, I liked their country for once

432
00:22:49,760 --> 00:22:54,680
Speaker 2: she left country. But like where people take these popular

433
00:22:54,680 --> 00:22:59,359
Speaker 2: culture things extremely seriously as art and don't dismiss these

434
00:22:59,440 --> 00:23:01,880
Speaker 2: outputs as sort of being trash.

435
00:23:02,600 --> 00:23:04,760
Speaker 4: Right now, we're in a golden age for country and

436
00:23:04,800 --> 00:23:08,720
Speaker 4: western music and also horror movies. Neither is really like

437
00:23:08,840 --> 00:23:11,840
Speaker 4: my taste in particular, but it's easy to see what

438
00:23:11,920 --> 00:23:15,320
Speaker 4: has gotten worse, and especially for critics, harder to see

439
00:23:15,400 --> 00:23:16,480
Speaker 4: what has been getting better.

440
00:23:17,240 --> 00:23:20,679
Speaker 3: That's my sweet spot. Country and horror is perfect for me.

441
00:23:21,280 --> 00:23:22,879
Speaker 3: But just on the culture point, I mean, I think

442
00:23:22,920 --> 00:23:25,520
Speaker 3: the lack of culture argument, the one I hear the

443
00:23:25,560 --> 00:23:27,960
Speaker 3: most is it's a lack of shared culture.

444
00:23:28,200 --> 00:23:28,360
Speaker 4: Right.

445
00:23:28,480 --> 00:23:31,280
Speaker 3: So you do have like some giant monolists like a

446
00:23:31,320 --> 00:23:35,520
Speaker 3: Swift or a Beyonce that everyone knows about, and you know,

447
00:23:35,600 --> 00:23:39,560
Speaker 3: they do have these large audiences, but broadly we're not

448
00:23:39,600 --> 00:23:42,600
Speaker 3: all experiencing the same media that we used to, right, Like,

449
00:23:42,680 --> 00:23:45,040
Speaker 3: no one is gathering around the TV to watch the

450
00:23:45,040 --> 00:23:47,920
Speaker 3: finale of you know, some show that airs like once

451
00:23:47,960 --> 00:23:49,840
Speaker 3: a week and has been going on for five years.

452
00:23:49,880 --> 00:23:52,119
Speaker 2: The exception is NFL football, which I don't really get it,

453
00:23:52,119 --> 00:23:53,640
Speaker 2: but yeah, but it's not cultural.

454
00:23:53,840 --> 00:23:56,560
Speaker 3: Sports is outside of my experience.

455
00:23:56,280 --> 00:23:59,399
Speaker 4: And the Super Bowl is cultural, right, it's a cultural event.

456
00:23:59,480 --> 00:24:03,680
Speaker 4: You care about the advertisements, But the biggest of YouTube stars,

457
00:24:03,720 --> 00:24:06,120
Speaker 4: which again I would say is not personally my thing,

458
00:24:06,680 --> 00:24:10,840
Speaker 4: but they can have bigger audiences than those older TV shows. Right.

459
00:24:10,880 --> 00:24:13,720
Speaker 3: Well, So what I'm getting at is it does feel

460
00:24:13,840 --> 00:24:17,399
Speaker 3: like nowadays there's an ability because of tech, to serve

461
00:24:17,480 --> 00:24:22,240
Speaker 3: up very specific content and like niche content in streams.

462
00:24:22,600 --> 00:24:26,280
Speaker 3: And the analogy that I like to use is, you know,

463
00:24:26,359 --> 00:24:29,600
Speaker 3: everyone knows if you download Netflix for the first time,

464
00:24:29,760 --> 00:24:33,879
Speaker 3: the first movie you watch is incredibly important because whatever

465
00:24:33,920 --> 00:24:36,280
Speaker 3: you watch, you know, if it's a rom com, you're

466
00:24:36,320 --> 00:24:38,720
Speaker 3: going to be served up Kate Hudson films for the

467
00:24:38,800 --> 00:24:42,080
Speaker 3: rest of your life. Right, Like, the algorithm looks at

468
00:24:42,119 --> 00:24:44,800
Speaker 3: what you're watching and then it serves up that additional content.

469
00:24:45,600 --> 00:24:47,879
Speaker 3: What does that mean for society? The idea that you

470
00:24:47,960 --> 00:24:51,719
Speaker 3: have people you know, basically funneled into smaller and smaller

471
00:24:51,760 --> 00:24:54,720
Speaker 3: streams in some respect, well.

472
00:24:54,680 --> 00:24:57,520
Speaker 4: A lot of the Netflix algorithm it just directs you

473
00:24:57,560 --> 00:25:02,280
Speaker 4: to slop. True, people have always wanted slop. Like people

474
00:25:02,359 --> 00:25:05,399
Speaker 4: listen to music way back when it was quite common,

475
00:25:06,160 --> 00:25:08,200
Speaker 4: or they would just list in the top forty, which

476
00:25:08,240 --> 00:25:11,479
Speaker 4: in some years was very good, but often was pretty terrible,

477
00:25:11,520 --> 00:25:14,399
Speaker 4: even in the nineteen sixties. So what you can do

478
00:25:14,480 --> 00:25:18,520
Speaker 4: today is basically watch not any movie out there, but

479
00:25:18,560 --> 00:25:22,040
Speaker 4: you have movie You can still buy DVDs and Blu rays.

480
00:25:22,880 --> 00:25:26,520
Speaker 4: You have access to more cinema today than you ever have.

481
00:25:27,680 --> 00:25:31,200
Speaker 4: So people will sort themselves. And I think it's from

482
00:25:31,240 --> 00:25:33,920
Speaker 4: the point of view of cultural consumption. I don't think

483
00:25:33,920 --> 00:25:36,199
Speaker 4: there's ever been a better time to be alive than

484
00:25:36,280 --> 00:25:39,000
Speaker 4: right now. Like, well, a lot of people abuse that

485
00:25:39,119 --> 00:25:42,000
Speaker 4: and go for slop. Of course that's sad, but it's

486
00:25:42,040 --> 00:25:42,640
Speaker 4: hardly new.

487
00:25:43,760 --> 00:25:46,840
Speaker 2: So changing gears a little bit. Tracy and I write

488
00:25:46,840 --> 00:25:50,479
Speaker 2: almost every day because we have a daily newsletter that

489
00:25:50,600 --> 00:25:52,760
Speaker 2: forces us to and I really like having that because

490
00:25:52,800 --> 00:25:54,800
Speaker 2: I don't know if I would write every day if

491
00:25:54,840 --> 00:25:58,879
Speaker 2: I did not have that obligation to deliver something in

492
00:25:58,920 --> 00:26:01,520
Speaker 2: people's inbox that they pay for as partler Bloomberg dot

493
00:26:01,560 --> 00:26:03,679
Speaker 2: Com subscription. I love writing, but I don't know if

494
00:26:03,720 --> 00:26:05,760
Speaker 2: I would do it every day if I didn't have

495
00:26:05,800 --> 00:26:09,159
Speaker 2: this sort of requirement. I might just tweet. How do

496
00:26:09,200 --> 00:26:11,760
Speaker 2: you You've been blogging for over twenty years, how do

497
00:26:11,800 --> 00:26:14,960
Speaker 2: you avoid the temptation to just fire off all your

498
00:26:15,000 --> 00:26:18,399
Speaker 2: ideas via tweet and actually commit to the blog.

499
00:26:19,480 --> 00:26:21,840
Speaker 4: I'm never tempted to do that. I like to think

500
00:26:21,920 --> 00:26:22,400
Speaker 4: things out.

501
00:26:22,600 --> 00:26:23,000
Speaker 2: I don't do.

502
00:26:23,160 --> 00:26:26,679
Speaker 4: I just don't more when I write properly. I've actually

503
00:26:26,680 --> 00:26:29,160
Speaker 4: blogged every single day for over twenty two years.

504
00:26:29,240 --> 00:26:32,280
Speaker 2: That's amazing. It's amazing. Most people gave up, and so

505
00:26:32,320 --> 00:26:33,080
Speaker 2: what's the difference.

506
00:26:34,040 --> 00:26:36,320
Speaker 4: I don't feel it requires any discipline. For me. The

507
00:26:36,359 --> 00:26:39,800
Speaker 4: discipline is not writing more like I have to restrain myself.

508
00:26:40,600 --> 00:26:43,399
Speaker 4: So I guess I'm just weird. I don't think I

509
00:26:43,440 --> 00:26:47,159
Speaker 4: have any neat little trick or formula. It's one of

510
00:26:47,200 --> 00:26:49,359
Speaker 4: these niches that you can do now that you couldn't

511
00:26:49,359 --> 00:26:52,359
Speaker 4: do before. And I found my niche, as have the

512
00:26:52,359 --> 00:26:52,760
Speaker 4: two of you.

513
00:26:53,680 --> 00:26:56,240
Speaker 3: Here's a slightly different question playing on that theme, but

514
00:26:56,280 --> 00:26:58,680
Speaker 3: going back to the intro, How different do you think

515
00:26:58,800 --> 00:27:03,359
Speaker 3: you're blogging career would have been had chat bots existed,

516
00:27:03,480 --> 00:27:05,080
Speaker 3: you know, ten or twenty years ago when you were

517
00:27:05,080 --> 00:27:05,640
Speaker 3: starting out.

518
00:27:07,280 --> 00:27:10,360
Speaker 4: I don't think we know yet. My intuition is that

519
00:27:10,400 --> 00:27:14,080
Speaker 4: people still want to read human writers simply because they're human,

520
00:27:14,560 --> 00:27:17,240
Speaker 4: and if the bot is as good as you, most

521
00:27:17,240 --> 00:27:20,159
Speaker 4: of the world doesn't care. But that has not truly

522
00:27:20,240 --> 00:27:22,920
Speaker 4: been tested yet. I think we'll see in the next

523
00:27:22,960 --> 00:27:26,520
Speaker 4: two years, but that's what I'm expecting. Just I think.

524
00:27:26,640 --> 00:27:29,080
Speaker 4: I think in music, there'll be plenty of AI music.

525
00:27:29,119 --> 00:27:31,000
Speaker 4: It might be say ten or twenty percent of the

526
00:27:31,119 --> 00:27:34,320
Speaker 4: music sector, but listeners will still want that human to

527
00:27:34,440 --> 00:27:46,520
Speaker 4: human connection.

528
00:27:51,160 --> 00:27:54,720
Speaker 2: Do you think you know, when Twitter came out, they

529
00:27:54,800 --> 00:27:57,120
Speaker 2: called it a micro blogging site, as if it were

530
00:27:57,359 --> 00:28:00,280
Speaker 2: just blogging but on a shorter basis. But I think

531
00:28:00,280 --> 00:28:03,080
Speaker 2: it's fundamentally different. You know, in the early, the glory

532
00:28:03,160 --> 00:28:05,720
Speaker 2: days of blogs, which we'll be talking about forever when

533
00:28:05,720 --> 00:28:08,240
Speaker 2: we're all very old people, how good it was, you know.

534
00:28:08,280 --> 00:28:11,280
Speaker 2: I think there was this sort of spirit of you know,

535
00:28:11,400 --> 00:28:14,679
Speaker 2: liberal linking with each other and idea exploration, whereas Twitter

536
00:28:14,760 --> 00:28:19,040
Speaker 2: strikes me as much more conflictual and one upmanship and

537
00:28:19,080 --> 00:28:21,960
Speaker 2: so forth. Do you think there are like fundamental I

538
00:28:22,000 --> 00:28:24,119
Speaker 2: don't know if political is the right word, but like

539
00:28:24,720 --> 00:28:29,400
Speaker 2: new communication paradigms like sort of have their own terroir,

540
00:28:29,600 --> 00:28:34,000
Speaker 2: so to speak, in terms of the impulse towards collaboration

541
00:28:34,480 --> 00:28:37,600
Speaker 2: or conflict, etc. And does that change society?

542
00:28:38,720 --> 00:28:41,280
Speaker 4: Yeah, I still like blogging, and I'm sad people have

543
00:28:41,400 --> 00:28:44,000
Speaker 4: moved away from it Twitter. To me, it seems too

544
00:28:44,040 --> 00:28:48,000
Speaker 4: meme heavy, and meme heavy media have more potential for racism,

545
00:28:48,040 --> 00:28:50,760
Speaker 4: which of course is a big negative. And I see

546
00:28:50,880 --> 00:28:54,200
Speaker 4: so many people who are driven crazy by being on Twitter.

547
00:28:54,720 --> 00:28:56,680
Speaker 4: Whether it's because they're writing on it or reading it,

548
00:28:56,720 --> 00:28:59,480
Speaker 4: I'm not sure, maybe both. I don't want to name names,

549
00:28:59,480 --> 00:29:01,000
Speaker 4: but it's a lot of people, and I bet you

550
00:29:01,040 --> 00:29:02,360
Speaker 4: see the same ones that I do.

551
00:29:03,120 --> 00:29:07,360
Speaker 3: Also very sexist nowadays, I would just add in unappreciated

552
00:29:07,400 --> 00:29:10,760
Speaker 3: ways in many ways speaking of sexism, the impact of

553
00:29:10,800 --> 00:29:15,239
Speaker 3: AI on economics. Talk about that economic institutions, you know,

554
00:29:15,360 --> 00:29:18,400
Speaker 3: famous for modeling, spend a lot of time with numbers

555
00:29:18,440 --> 00:29:20,080
Speaker 3: and things like that. Is that all just going to

556
00:29:20,080 --> 00:29:21,240
Speaker 3: be replaced by AI?

557
00:29:23,000 --> 00:29:26,000
Speaker 4: Not all? So. I think what human economists will do

558
00:29:26,320 --> 00:29:29,080
Speaker 4: is put more and more time into gathering data and

559
00:29:29,120 --> 00:29:32,280
Speaker 4: feeding it to the AIS. The returns to doing that

560
00:29:32,320 --> 00:29:37,160
Speaker 4: will be very high. But the actual econometrics, statistics humans

561
00:29:37,200 --> 00:29:40,560
Speaker 4: will maybe set up part of the problem, but the hard, boring,

562
00:29:40,640 --> 00:29:43,560
Speaker 4: routine work will be done by the machines. As to

563
00:29:43,600 --> 00:29:46,200
Speaker 4: some extent it was already the case, and this will

564
00:29:46,240 --> 00:29:48,040
Speaker 4: be a way to make a lot of progress rather

565
00:29:48,120 --> 00:29:49,600
Speaker 4: quickly do the.

566
00:29:49,680 --> 00:29:53,720
Speaker 3: Actual economic statistics or data points that economists collect. Do

567
00:29:53,880 --> 00:29:55,640
Speaker 3: some of those need to be changed or thought of

568
00:29:55,760 --> 00:29:58,720
Speaker 3: differently in light of the AI era.

569
00:30:00,560 --> 00:30:04,040
Speaker 4: Well, I don't know. Eventually they will need to be,

570
00:30:04,520 --> 00:30:08,120
Speaker 4: but I would say any period of radical change in history,

571
00:30:08,400 --> 00:30:12,200
Speaker 4: your statistics are less useful. It's not that the people

572
00:30:12,280 --> 00:30:15,720
Speaker 4: creating the statistics are making some mistake. You just cannot

573
00:30:15,760 --> 00:30:18,880
Speaker 4: capture every way the world is changing, and index number

574
00:30:18,960 --> 00:30:23,440
Speaker 4: comparisons require the basket of goods be relatively close to constant,

575
00:30:23,840 --> 00:30:26,880
Speaker 4: and at some point that doesn't hold anymore, and we'll

576
00:30:26,880 --> 00:30:29,400
Speaker 4: be faced with that. We'll deal with it. I would

577
00:30:29,440 --> 00:30:33,800
Speaker 4: say the current statistics we have, they're more underrated than overrated,

578
00:30:34,160 --> 00:30:37,000
Speaker 4: so they're actually pretty good. I'll be glad when we

579
00:30:37,080 --> 00:30:37,920
Speaker 4: get them back again.

580
00:30:38,640 --> 00:30:41,160
Speaker 2: There's another thing. The tech people you talk to, they

581
00:30:41,200 --> 00:30:45,000
Speaker 2: must think like GDP is terrible. Doesn't capture all this stuff,

582
00:30:45,040 --> 00:30:47,720
Speaker 2: all this value that we can't price. I'm sure you've

583
00:30:47,720 --> 00:30:53,240
Speaker 2: had conversations explaining to many AI workers that GDP is

584
00:30:53,280 --> 00:30:55,160
Speaker 2: not the worst statistic in the world. I want to

585
00:30:55,160 --> 00:30:57,640
Speaker 2: go back though, to them. So you know you mentioned

586
00:30:57,680 --> 00:31:00,240
Speaker 2: like on Twitter, right, you say something someone so it's

587
00:31:00,240 --> 00:31:02,680
Speaker 2: a meme. They dunk on you, they make fun of you,

588
00:31:03,120 --> 00:31:06,720
Speaker 2: they whatever. Your chatbot won't do that. Like if I'm

589
00:31:06,760 --> 00:31:09,720
Speaker 2: having a conversation with Chad GBT, it's never going to

590
00:31:09,800 --> 00:31:13,000
Speaker 2: respond to me with the meme sort of indicating one

591
00:31:13,040 --> 00:31:14,480
Speaker 2: day of my ad that I'm.

592
00:31:14,360 --> 00:31:16,120
Speaker 3: A moral you'll learn your language, Joe.

593
00:31:16,200 --> 00:31:18,239
Speaker 2: Yeah, But if anything, it is too obsequious, right, I mean,

594
00:31:18,280 --> 00:31:21,840
Speaker 2: the issue is like the online world has become this

595
00:31:22,040 --> 00:31:25,120
Speaker 2: very like sort of competitive, conflictual world, and then I

596
00:31:25,160 --> 00:31:28,000
Speaker 2: go to the chatbot and my complaint is literally the opposite.

597
00:31:28,040 --> 00:31:30,840
Speaker 2: It doesn't challenge me enough. It's too obsequious. Every question

598
00:31:30,880 --> 00:31:33,200
Speaker 2: I ask, it's a great question. Sometimes I wish it

599
00:31:33,200 --> 00:31:35,320
Speaker 2: would call me a moron a little bit more. But

600
00:31:35,400 --> 00:31:38,160
Speaker 2: what does it change about the world? And we know

601
00:31:38,320 --> 00:31:40,840
Speaker 2: that like many people's brains have been broken by social

602
00:31:40,880 --> 00:31:43,800
Speaker 2: media that probably has downstream effects on how our politics

603
00:31:43,800 --> 00:31:46,600
Speaker 2: operates these days. What does it do to the world

604
00:31:46,880 --> 00:31:51,320
Speaker 2: if we started inhabiting these chat environments where they're just

605
00:31:51,520 --> 00:31:54,080
Speaker 2: very sort of polite, and every time you say something,

606
00:31:54,160 --> 00:31:57,160
Speaker 2: it says, yes, great thought, Tyler, great thought, Joe, would

607
00:31:57,200 --> 00:31:59,560
Speaker 2: you like to expand it? You asked the perfect question,

608
00:32:00,040 --> 00:32:02,880
Speaker 2: what do you see that having sort of second order

609
00:32:02,920 --> 00:32:04,880
Speaker 2: effects on how society operates.

610
00:32:05,800 --> 00:32:07,880
Speaker 4: Well, that's the four to zero model that does that.

611
00:32:07,960 --> 00:32:11,400
Speaker 4: The newer models like Claude four point five and GDP five,

612
00:32:11,720 --> 00:32:13,400
Speaker 4: they're more objective, and that's better.

613
00:32:13,600 --> 00:32:15,720
Speaker 2: But they never make fun of you. They never will

614
00:32:15,840 --> 00:32:18,959
Speaker 2: like say, you're a moron? How could you possibly how

615
00:32:18,960 --> 00:32:22,120
Speaker 2: could you possibly ask such a dumb question? You're obviously

616
00:32:22,240 --> 00:32:25,400
Speaker 2: so out of touch for having the ax ask this.

617
00:32:25,400 --> 00:32:27,280
Speaker 2: This is like a very in some respects. This is

618
00:32:27,320 --> 00:32:31,040
Speaker 2: a very positive change from many of the conversations that

619
00:32:31,080 --> 00:32:33,160
Speaker 2: I've had from typing into a computer.

620
00:32:34,000 --> 00:32:36,080
Speaker 4: Oh it's great. I think people should be nicer to

621
00:32:36,160 --> 00:32:38,959
Speaker 4: each other. And I think they're the most objective media

622
00:32:39,000 --> 00:32:41,800
Speaker 4: source the human race ever has had. If you're ask

623
00:32:41,840 --> 00:32:46,280
Speaker 4: good about, say vaccines or conspiracy theories, it basically gives

624
00:32:46,280 --> 00:32:47,200
Speaker 4: you the right answers.

625
00:32:48,120 --> 00:32:50,680
Speaker 3: Well, one thing that they don't do, and I mean

626
00:32:50,720 --> 00:32:52,320
Speaker 3: I do think they can be trained to be mean

627
00:32:52,360 --> 00:32:55,000
Speaker 3: to you and to insult you to a certain degree,

628
00:32:55,520 --> 00:32:56,000
Speaker 3: and you can.

629
00:32:56,280 --> 00:32:58,240
Speaker 2: Could bring but I never account of them.

630
00:32:58,640 --> 00:33:01,760
Speaker 4: Try market that more readily on PAP right.

631
00:33:02,040 --> 00:33:06,800
Speaker 3: So we spoke to the chief business officer of Perplexity,

632
00:33:06,880 --> 00:33:09,440
Speaker 3: Dmitri Shevalenko. We spoke to him recently, and he was

633
00:33:09,480 --> 00:33:12,320
Speaker 3: saying that one thing chat models can't do is express

634
00:33:12,400 --> 00:33:15,600
Speaker 3: a natural curiosity, which I thought was kind of weird

635
00:33:15,680 --> 00:33:18,200
Speaker 3: coming from him, because perplexity is the only model I

636
00:33:18,240 --> 00:33:21,720
Speaker 3: know that actually throws out those additional questions if you

637
00:33:21,800 --> 00:33:23,560
Speaker 3: query it and then it comes up with would you

638
00:33:23,720 --> 00:33:26,160
Speaker 3: like to have more information on this point or are

639
00:33:26,200 --> 00:33:29,320
Speaker 3: you thinking about this now? But there does seem to

640
00:33:29,320 --> 00:33:32,360
Speaker 3: be an element of creativity perhaps that is lost in

641
00:33:32,440 --> 00:33:36,680
Speaker 3: some of these lms. How much does that change things

642
00:33:36,760 --> 00:33:40,160
Speaker 3: in media? The idea that you know, the models are

643
00:33:40,160 --> 00:33:42,920
Speaker 3: going to spit out something that's sort of predestined in

644
00:33:42,960 --> 00:33:44,560
Speaker 3: many ways.

645
00:33:44,760 --> 00:33:47,480
Speaker 4: Well, that's what most humans do, to be clear, But

646
00:33:47,560 --> 00:33:50,000
Speaker 4: it's now the case that on a regular basis, the

647
00:33:50,080 --> 00:33:53,320
Speaker 4: models say can prove new theorems and math or discover

648
00:33:53,480 --> 00:33:56,600
Speaker 4: new potential drugs. And keep in mind, you know, a

649
00:33:56,640 --> 00:33:59,800
Speaker 4: year ago these things thought the word strawberry had too

650
00:34:00,560 --> 00:34:03,920
Speaker 4: and now they're winning gold medals and math Olympias. So

651
00:34:03,960 --> 00:34:06,440
Speaker 4: a year or two from now, maybe we don't know

652
00:34:06,760 --> 00:34:09,319
Speaker 4: how much better they'll be, but I don't think they're

653
00:34:09,320 --> 00:34:12,719
Speaker 4: gonna have any problems being creative, certainly more creative than

654
00:34:12,760 --> 00:34:13,880
Speaker 4: humans on average.

655
00:34:14,560 --> 00:34:16,359
Speaker 3: So now I have to ask. Since we're talking about

656
00:34:16,400 --> 00:34:18,680
Speaker 3: being mean or nice to the models and them being

657
00:34:18,760 --> 00:34:20,360
Speaker 3: mean or nice to you, do you say please and

658
00:34:20,400 --> 00:34:22,200
Speaker 3: thank you in your LLM queries?

659
00:34:23,280 --> 00:34:26,640
Speaker 4: You know? I used to, and then Sam Altman said, well,

660
00:34:26,680 --> 00:34:28,959
Speaker 4: it costs us just a little bit of money because

661
00:34:29,000 --> 00:34:31,760
Speaker 4: of the extra tokens. And then I thought, I'll hold

662
00:34:31,760 --> 00:34:34,720
Speaker 4: off on this. But I have this pre existing record

663
00:34:35,239 --> 00:34:38,360
Speaker 4: of saying please, and it knows that. And then it

664
00:34:38,440 --> 00:34:41,120
Speaker 4: knows I stopped when Sam said to stop. And I

665
00:34:41,160 --> 00:34:43,360
Speaker 4: think I'll get points for both of those decisions.

666
00:34:43,400 --> 00:34:46,319
Speaker 3: See, I actually find if you're slightly meaner to the

667
00:34:46,320 --> 00:34:50,040
Speaker 3: models in your queries, they perform slightly better, much like

668
00:34:50,120 --> 00:34:51,240
Speaker 3: interacting with Joe.

669
00:34:53,080 --> 00:34:55,400
Speaker 2: No comment on that. I you know what I do.

670
00:34:55,680 --> 00:34:59,480
Speaker 2: I have said luchadal like, I'll have a query and

671
00:34:59,480 --> 00:35:01,480
Speaker 2: at a respet bond and I'll say that was a

672
00:35:01,520 --> 00:35:05,319
Speaker 2: little on the nose, wasn't it? Like like it over.

673
00:35:05,600 --> 00:35:06,680
Speaker 3: I'll say, do better.

674
00:35:06,760 --> 00:35:09,200
Speaker 2: Yeah, I've seen that it does. Yeah, I've said like bad,

675
00:35:09,239 --> 00:35:10,959
Speaker 2: I've said things like that. It's like, this is really

676
00:35:10,960 --> 00:35:13,920
Speaker 2: on the nose. This response was a little bit trite,

677
00:35:14,000 --> 00:35:16,799
Speaker 2: don't you think, et cetera. Like I do feel like

678
00:35:16,840 --> 00:35:20,560
Speaker 2: I've gotten more comfortable at uh, let's be let's be

679
00:35:20,600 --> 00:35:24,040
Speaker 2: real here. You're not doing You're not doing your best

680
00:35:24,760 --> 00:35:28,320
Speaker 2: job here. What do you think is? Like, it's interesting

681
00:35:28,360 --> 00:35:30,960
Speaker 2: they like I get that they win the gold medals

682
00:35:31,000 --> 00:35:33,960
Speaker 2: and the math and et cetera. Like I've had so

683
00:35:34,040 --> 00:35:37,520
Speaker 2: many conversations with the chatbots that are on some level

684
00:35:37,680 --> 00:35:41,600
Speaker 2: like mind blowing what the capability is. I've never seen

685
00:35:41,640 --> 00:35:46,160
Speaker 2: a chatbot query that is like interesting, like that is like, oh,

686
00:35:46,200 --> 00:35:48,880
Speaker 2: that is like a really maybe one I could think of,

687
00:35:49,080 --> 00:35:51,200
Speaker 2: but there was like a really interesting thought. I feel

688
00:35:51,200 --> 00:35:53,680
Speaker 2: like my children still say on a daily basis more

689
00:35:53,719 --> 00:35:57,760
Speaker 2: like interesting things they get me thinking than I've ever

690
00:35:57,920 --> 00:36:01,680
Speaker 2: gotten from a chat Does that resonate to you at all?

691
00:36:02,840 --> 00:36:03,160
Speaker 4: I don't know.

692
00:36:03,719 --> 00:36:05,560
Speaker 2: I feel like you've said so many more things in

693
00:36:05,560 --> 00:36:09,160
Speaker 2: this hour than any than any interesting thing, like actually

694
00:36:09,160 --> 00:36:11,840
Speaker 2: like interesting ideas than I've ever gotten from the hours

695
00:36:11,840 --> 00:36:14,200
Speaker 2: I've spent playing with chat chipet your claude.

696
00:36:15,200 --> 00:36:17,400
Speaker 4: You know, I use mine a lot for music. So

697
00:36:17,480 --> 00:36:21,479
Speaker 4: if I'm going to listen to Sibelius's Fifth Symphony, I'll

698
00:36:21,520 --> 00:36:24,479
Speaker 4: just ask it what should I listen for? And I'll say,

699
00:36:24,560 --> 00:36:27,120
Speaker 4: this is Tyler Kwan asking which I hope raises the

700
00:36:27,200 --> 00:36:30,600
Speaker 4: quality of the answer. It knows a lot about me. Yeah,

701
00:36:30,640 --> 00:36:32,720
Speaker 4: and what it gives me to listen for I find

702
00:36:32,800 --> 00:36:35,920
Speaker 4: is better than any human source I can access readily better.

703
00:36:36,000 --> 00:36:39,239
Speaker 2: I that I agree with, but like, does it make

704
00:36:39,280 --> 00:36:43,799
Speaker 2: a connection, does it like tell you something about Sabilias's

705
00:36:43,920 --> 00:36:46,520
Speaker 2: music that is like, oh, that's a very interesting that's

706
00:36:46,520 --> 00:36:49,200
Speaker 2: a novel way of thinking about what makes it profound?

707
00:36:49,520 --> 00:36:52,399
Speaker 2: These are the things that I rarely ever encounter something

708
00:36:52,440 --> 00:36:54,480
Speaker 2: I was like, oh, that is an interesting thought, Whereas

709
00:36:54,480 --> 00:36:57,000
Speaker 2: I feel like if we're talking to a music cologist

710
00:36:57,040 --> 00:36:59,640
Speaker 2: for an hour, I would get infinitely more like actual

711
00:36:59,719 --> 00:37:02,560
Speaker 2: insight into something that makes the music special. So I

712
00:37:02,560 --> 00:37:03,800
Speaker 2: think I hadn't heard before.

713
00:37:05,000 --> 00:37:07,120
Speaker 4: I don't know if it's novel because I don't really

714
00:37:07,160 --> 00:37:12,480
Speaker 4: know the Sibelius literature, but most musicologists I find pretty boring.

715
00:37:13,560 --> 00:37:17,000
Speaker 4: And I find say, GPT five on a classical symphony

716
00:37:17,080 --> 00:37:20,000
Speaker 4: quite to the point. And whether or not it's original,

717
00:37:20,040 --> 00:37:22,200
Speaker 4: it's not that important to me. It helps me listen

718
00:37:22,200 --> 00:37:25,320
Speaker 4: to the music better. Yeah, and it's certainly original relative

719
00:37:25,360 --> 00:37:28,759
Speaker 4: to the other sources at my disposal, say Wikipedia or

720
00:37:28,760 --> 00:37:32,200
Speaker 4: what I could google to So I think for almost

721
00:37:32,239 --> 00:37:35,880
Speaker 4: all purposes, that's enough. Does it have a truly original

722
00:37:35,920 --> 00:37:39,560
Speaker 4: idea in the sense that Einstein's theory of relativity when

723
00:37:39,560 --> 00:37:42,160
Speaker 4: he came up with it, was original to him? I

724
00:37:42,200 --> 00:37:45,640
Speaker 4: don't think so. That may come in some number of years,

725
00:37:46,000 --> 00:37:48,680
Speaker 4: but again, for almost all purposes, that's not what we need.

726
00:37:48,880 --> 00:37:52,080
Speaker 4: We need something better than our pre existing state of knowledge.

727
00:37:52,239 --> 00:37:53,920
Speaker 4: And on that I think it just cleans up.

728
00:37:55,200 --> 00:37:58,680
Speaker 3: So I just ask perplexity what music I should recommend

729
00:37:58,719 --> 00:38:01,920
Speaker 3: to Tyler Collan, and it said that in order to

730
00:38:01,960 --> 00:38:04,880
Speaker 3: recommend effectively to Tyler Cowen, I need to look for

731
00:38:05,000 --> 00:38:09,640
Speaker 3: underappreciated recordings and obscure things that no one accept him

732
00:38:09,800 --> 00:38:15,799
Speaker 3: might have ever heard about. And it recommended. I mean,

733
00:38:15,840 --> 00:38:19,120
Speaker 3: boy genius seems pretty on the on the nose, right,

734
00:38:20,280 --> 00:38:23,120
Speaker 3: reggae acts like Toots and the May talent.

735
00:38:23,600 --> 00:38:24,920
Speaker 1: This is so is that.

736
00:38:27,440 --> 00:38:29,880
Speaker 2: Tyler Price records.

737
00:38:29,920 --> 00:38:32,239
Speaker 3: So it's just scraping stuff that you've already talked to it.

738
00:38:32,719 --> 00:38:33,680
Speaker 2: It's not even trying.

739
00:38:33,800 --> 00:38:35,319
Speaker 3: Yeah, all right, well you.

740
00:38:35,280 --> 00:38:38,000
Speaker 4: Need to make the prompt more exacting. Rule out anything

741
00:38:38,040 --> 00:38:41,080
Speaker 4: Tyler has talked or written about. Yeah, give me something

742
00:38:41,120 --> 00:38:45,280
Speaker 4: he doesn't know and try GPT five in the pro mode.

743
00:38:45,520 --> 00:38:47,279
Speaker 4: And I think it will succeed well.

744
00:38:47,520 --> 00:38:49,960
Speaker 3: So on this note. This is something I've been asking everyone,

745
00:38:50,040 --> 00:38:52,480
Speaker 3: but like, what is an example in your mind of

746
00:38:52,520 --> 00:38:55,319
Speaker 3: a really good prompt or one that sticks out to

747
00:38:55,360 --> 00:38:58,480
Speaker 3: you that has generated something that you know, maybe you

748
00:38:58,480 --> 00:38:59,480
Speaker 3: didn't expect.

749
00:39:00,440 --> 00:39:02,799
Speaker 4: You know, dwark Esh Patel once wrote a very good

750
00:39:02,800 --> 00:39:05,560
Speaker 4: prompt and he shared it with me. When I interview

751
00:39:05,640 --> 00:39:09,080
Speaker 4: some podcast guests, It's really a long prompt. It's in

752
00:39:09,239 --> 00:39:12,880
Speaker 4: hundreds of words, and it asks what questions should I

753
00:39:12,920 --> 00:39:15,160
Speaker 4: ask them? Then it goes into great detail. It should

754
00:39:15,160 --> 00:39:17,319
Speaker 4: be a unique question, it should be a question they

755
00:39:17,360 --> 00:39:20,239
Speaker 4: were not asked anywhere else. Then it's given what you

756
00:39:20,280 --> 00:39:22,200
Speaker 4: think their answer might be, and what would be my

757
00:39:22,280 --> 00:39:24,759
Speaker 4: follow up question? And yeah, give you give me some

758
00:39:24,920 --> 00:39:27,160
Speaker 4: cases where you think their answer might be wrong. And

759
00:39:27,239 --> 00:39:30,320
Speaker 4: it goes on and on, and you run that through

760
00:39:30,560 --> 00:39:33,879
Speaker 4: the very best models. I think you get good results. Yeah.

761
00:39:34,320 --> 00:39:37,520
Speaker 2: No, I've you know, I've run transcripts of the podcast

762
00:39:37,560 --> 00:39:39,600
Speaker 2: before and I say, like, well, where should I have

763
00:39:39,640 --> 00:39:42,840
Speaker 2: pushed the guests harder on? What were the weak answers

764
00:39:43,239 --> 00:39:46,279
Speaker 2: that they what were the inconsistencies that the guests had

765
00:39:46,320 --> 00:39:48,680
Speaker 2: over the time? And I've found it to be a

766
00:39:48,880 --> 00:39:52,880
Speaker 2: very useful exercise for things like that. So I do

767
00:39:53,000 --> 00:39:56,000
Speaker 2: think like that's the thing which is first it is

768
00:39:56,160 --> 00:40:00,200
Speaker 2: it is objectively impressive on many of these from and

769
00:40:00,239 --> 00:40:03,239
Speaker 2: I would say objectively useful if you do sort of

770
00:40:03,280 --> 00:40:06,080
Speaker 2: a detailed prompting. I'm just curious, what do you what's

771
00:40:06,080 --> 00:40:11,080
Speaker 2: your like as a professor from the professor perspective, what

772
00:40:11,160 --> 00:40:14,160
Speaker 2: do you think is the right way to think about

773
00:40:14,560 --> 00:40:17,680
Speaker 2: how students will be using chadib I mean, I know

774
00:40:17,760 --> 00:40:21,120
Speaker 2: that there's a million opinions in academia about well, what's

775
00:40:21,160 --> 00:40:23,239
Speaker 2: the right way to test, now, what's the right way

776
00:40:23,280 --> 00:40:26,000
Speaker 2: to deal with essays, et cetera. How are you thinking

777
00:40:26,000 --> 00:40:27,560
Speaker 2: about some of these challenges.

778
00:40:28,520 --> 00:40:31,759
Speaker 4: We should devote one third of all higher education to

779
00:40:31,840 --> 00:40:35,279
Speaker 4: teaching students how to use AI, and right now that's

780
00:40:35,320 --> 00:40:39,640
Speaker 4: close to zero, so we don't have the faculty who

781
00:40:39,640 --> 00:40:41,480
Speaker 4: can teach it. As part of the problem, often the

782
00:40:41,520 --> 00:40:43,000
Speaker 4: students no more than the professor.

783
00:40:43,080 --> 00:40:43,920
Speaker 2: Yea, I'm sure.

784
00:40:45,680 --> 00:40:48,400
Speaker 4: You need to restructure ratically what we do because future

785
00:40:48,440 --> 00:40:51,080
Speaker 4: work will be done with AIS, so that's the thing

786
00:40:51,120 --> 00:40:51,560
Speaker 4: to teach.

787
00:40:52,239 --> 00:40:55,600
Speaker 2: But like so, just to play devil's advocate, like intuitively,

788
00:40:55,680 --> 00:41:00,600
Speaker 2: I still feel like there's value in long periods of

789
00:41:00,680 --> 00:41:04,560
Speaker 2: time cut off reading where you're not looking at devices,

790
00:41:04,760 --> 00:41:07,719
Speaker 2: where you're training your body to sort of be disciplined

791
00:41:07,800 --> 00:41:13,319
Speaker 2: and pay attention and focus. I still think memorization of

792
00:41:13,560 --> 00:41:16,319
Speaker 2: facts and numbers and dates and places and names is

793
00:41:16,400 --> 00:41:19,440
Speaker 2: very useful in actually having them in your head, et cetera.

794
00:41:19,760 --> 00:41:23,399
Speaker 2: Does that seem right to you or is that sort

795
00:41:23,480 --> 00:41:25,600
Speaker 2: of retro thinking on my part?

796
00:41:26,239 --> 00:41:26,319
Speaker 3: No?

797
00:41:26,440 --> 00:41:29,319
Speaker 4: Strong agree? And most of all writing, But that's the

798
00:41:29,360 --> 00:41:31,680
Speaker 4: other two thirds of higher ed right, I said, one third?

799
00:41:31,760 --> 00:41:33,520
Speaker 2: Oh yeah, so tell us about the other two thirds.

800
00:41:34,440 --> 00:41:37,600
Speaker 4: We should with or without AI just teach students much

801
00:41:37,640 --> 00:41:40,360
Speaker 4: more and much better how to write. Most people can't write.

802
00:41:40,560 --> 00:41:44,560
Speaker 4: Writing is thinking we should do much more to teach

803
00:41:44,600 --> 00:41:47,279
Speaker 4: writing and test writing, and now with AI, that has

804
00:41:47,320 --> 00:41:49,600
Speaker 4: to be face to face in a controlled environment where

805
00:41:49,640 --> 00:41:52,520
Speaker 4: people are just going to cheat, So that we should

806
00:41:52,520 --> 00:41:54,799
Speaker 4: have doubled down on to begin with.

807
00:41:55,120 --> 00:41:55,560
Speaker 2: Yeah.

808
00:41:55,600 --> 00:41:59,040
Speaker 4: So that and just numeracy and basic issues like how

809
00:41:59,040 --> 00:42:02,239
Speaker 4: to manage a portfolio, what kind of mortgage to take out.

810
00:42:02,440 --> 00:42:05,279
Speaker 4: There are classes that cover those things, but I think

811
00:42:05,320 --> 00:42:07,719
Speaker 4: they ought to be front and center of any curriculum.

812
00:42:07,920 --> 00:42:12,000
Speaker 4: Basic finance, basic life decisions like how to choose a doctor,

813
00:42:12,640 --> 00:42:15,520
Speaker 4: how to prompt the AI you know, for diagnosis whatever,

814
00:42:16,600 --> 00:42:19,440
Speaker 4: are relatively neglected in a lot of education. That to

815
00:42:19,520 --> 00:42:20,560
Speaker 4: me just seems crazy.

816
00:42:21,840 --> 00:42:22,239
Speaker 2: I want to.

817
00:42:22,200 --> 00:42:25,520
Speaker 3: Ask one market question before we go, which is obviously

818
00:42:26,080 --> 00:42:28,200
Speaker 3: there's a lot of talk about an AI bubble at

819
00:42:28,239 --> 00:42:30,400
Speaker 3: the moment, and I think the concern from a lot

820
00:42:30,440 --> 00:42:32,160
Speaker 3: of people is when you start talking about in new

821
00:42:32,200 --> 00:42:36,640
Speaker 3: technology as revolutionary, when you start talking about how you

822
00:42:36,680 --> 00:42:39,279
Speaker 3: know the effects are basically going to be infinite, and

823
00:42:39,320 --> 00:42:43,960
Speaker 3: the market size is hypothetically the entire world, there's a

824
00:42:44,080 --> 00:42:48,920
Speaker 3: risk that expectations overshoot reality, right and we have seen

825
00:42:49,040 --> 00:42:51,799
Speaker 3: some people voicing their worries about that right now, and

826
00:42:51,840 --> 00:42:53,920
Speaker 3: a little bit of nervousness creeping into the market. In

827
00:42:54,000 --> 00:42:57,800
Speaker 3: terms of valuations, Where do you stand on the AI bubble?

828
00:42:57,840 --> 00:42:59,919
Speaker 3: Do you see signs of frauth or do you think

829
00:43:00,120 --> 00:43:03,960
Speaker 3: most of the CAPEX spending is justified at this point?

830
00:43:04,840 --> 00:43:07,319
Speaker 4: I don't like the word bubble. I would point out

831
00:43:07,360 --> 00:43:12,640
Speaker 4: that tech sector earnings are exceeding tech sector capital expenditure.

832
00:43:13,040 --> 00:43:15,759
Speaker 4: This is not mostly debt financed, so we're in less

833
00:43:15,760 --> 00:43:18,960
Speaker 4: trouble than many people think. It wouldn't shock me if

834
00:43:19,000 --> 00:43:21,360
Speaker 4: a lot of these efforts lost money. That was the

835
00:43:21,400 --> 00:43:24,000
Speaker 4: case with the railroads, the case with the internet, case

836
00:43:24,040 --> 00:43:27,680
Speaker 4: with most things humans have done. But I think it

837
00:43:27,719 --> 00:43:30,040
Speaker 4: will endure. It's not like pets dot com, where the

838
00:43:30,080 --> 00:43:35,359
Speaker 4: thing just gets swept away. These are incredibly well capitalized,

839
00:43:35,480 --> 00:43:39,000
Speaker 4: highly skilled companies where the CEOs and or founders are

840
00:43:39,040 --> 00:43:41,560
Speaker 4: quite committed to doing this and they're going to see

841
00:43:41,600 --> 00:43:44,040
Speaker 4: it through and they're going to succeed. But does that

842
00:43:44,120 --> 00:43:46,719
Speaker 4: mean every share value will go up or Nvidia ends

843
00:43:46,800 --> 00:43:49,120
Speaker 4: up being worth ten trillion dollars. I don't know. I

844
00:43:49,160 --> 00:43:53,680
Speaker 4: wouldn't necessarily predict that. There's always ups and downs, but

845
00:43:53,760 --> 00:43:55,880
Speaker 4: this is clearly a very useful thing and we're going

846
00:43:55,920 --> 00:43:57,880
Speaker 4: to we as Americans, we're going to make it work.

847
00:43:58,200 --> 00:44:00,000
Speaker 4: And we're way ahead of the rest of the world.

848
00:44:00,160 --> 00:44:03,400
Speaker 4: Like three quarters of all AI compute is in this country.

849
00:44:03,440 --> 00:44:06,840
Speaker 4: That's incredible. Worry what percent of the world's population six

850
00:44:06,960 --> 00:44:10,120
Speaker 4: or I don't know, but way smaller than three quarters.

851
00:44:11,080 --> 00:44:14,480
Speaker 2: GPT five on Thinking Mode says you should listen to

852
00:44:14,600 --> 00:44:18,200
Speaker 2: Michael Gulisian, who does dream like acoustic guitar using an

853
00:44:18,239 --> 00:44:22,319
Speaker 2: open tuning, and it said that given your affinity for

854
00:44:22,400 --> 00:44:26,640
Speaker 2: guitarists like John Fahey and Leo Kottke, you'll appreciate him.

855
00:44:27,520 --> 00:44:30,239
Speaker 4: Send me that answer. It sounds excellent. I haven't heard

856
00:44:30,280 --> 00:44:32,759
Speaker 4: of that person. Have you very much liked guitar with

857
00:44:32,840 --> 00:44:34,400
Speaker 4: open tuning, shure a.

858
00:44:34,480 --> 00:44:39,640
Speaker 2: Monkar, a Hindustani vocal singer. It thinks you'll like sun

859
00:44:39,840 --> 00:44:43,520
Speaker 2: oh md Budrul Hawk. I'll send you this list and

860
00:44:43,880 --> 00:44:47,840
Speaker 2: Katerina Barbieri modern modular, minimist electronic composition.

861
00:44:48,360 --> 00:44:51,200
Speaker 4: So I will buy some of these already.

862
00:44:51,320 --> 00:44:54,920
Speaker 3: Wait, just how about a human recommendation? Oh, give us

863
00:44:55,000 --> 00:44:57,920
Speaker 3: a recommend a human recommendation. Since you said that you

864
00:44:57,920 --> 00:44:59,839
Speaker 3: know country is pretty good right now, but I guess

865
00:44:59,840 --> 00:45:02,719
Speaker 3: you personally aren't that into it. But have you tried

866
00:45:02,840 --> 00:45:06,040
Speaker 3: Orville Peck? I think there's a new album out out

867
00:45:06,440 --> 00:45:06,839
Speaker 3: this week.

868
00:45:06,880 --> 00:45:09,759
Speaker 4: I think, what kind of music is it?

869
00:45:10,160 --> 00:45:14,480
Speaker 3: Country? But like a very modern type of country. I've

870
00:45:14,480 --> 00:45:17,319
Speaker 3: tried to get Joe into it, but uh, I'm still

871
00:45:17,320 --> 00:45:17,960
Speaker 3: working on it.

872
00:45:19,120 --> 00:45:21,799
Speaker 4: I like some country, and I love old country. So

873
00:45:21,920 --> 00:45:24,760
Speaker 4: Hank Williams, Johnny Cash, Jed Atkins period.

874
00:45:25,480 --> 00:45:27,880
Speaker 3: It has a vintage toned to it, but with a

875
00:45:27,920 --> 00:45:31,359
Speaker 3: modern twist. Try Orville Peck and then get back to

876
00:45:31,440 --> 00:45:35,800
Speaker 3: us about which was better in terms of the recommendation.

877
00:45:36,080 --> 00:45:39,160
Speaker 2: Tyler Colln Tyler Collen, thank you so much for coming

878
00:45:39,200 --> 00:45:42,800
Speaker 2: on odd Law's long overdue conversation. I really appreciate you

879
00:45:42,960 --> 00:45:43,520
Speaker 2: taking your time.

880
00:45:44,360 --> 00:45:45,520
Speaker 4: Great to chat with you both.

881
00:45:58,600 --> 00:46:01,279
Speaker 2: Tracy a lot to pull out from conversation. I think

882
00:46:01,320 --> 00:46:05,600
Speaker 2: it's very interesting that early observation he made about sort

883
00:46:05,640 --> 00:46:11,640
Speaker 2: of legacy institutions and whether perhaps some of the sort

884
00:46:11,680 --> 00:46:15,600
Speaker 2: of lack of revolutionary impact yet is just about that

885
00:46:15,680 --> 00:46:19,520
Speaker 2: metabolization process into the types of companies that could theoretically

886
00:46:19,560 --> 00:46:20,160
Speaker 2: absorb them.

887
00:46:20,400 --> 00:46:20,720
Speaker 4: Yeah.

888
00:46:20,840 --> 00:46:23,600
Speaker 3: I mean I think that's exactly it, right, So companies

889
00:46:23,640 --> 00:46:26,320
Speaker 3: are using this mostly as an add on to existing workflow.

890
00:46:26,640 --> 00:46:29,840
Speaker 3: You're not going to get the huge productivity boom until

891
00:46:29,920 --> 00:46:34,319
Speaker 3: companies are sort of centered from yeah, which you know

892
00:46:34,400 --> 00:46:37,080
Speaker 3: is probably going to take people who grew up with

893
00:46:37,120 --> 00:46:40,359
Speaker 3: the technology rather than old people like you and I.

894
00:46:40,480 --> 00:46:42,279
Speaker 2: You and I, you just adopted it.

895
00:46:42,560 --> 00:46:44,879
Speaker 3: The other thing I was thinking about, first of all,

896
00:46:45,120 --> 00:46:47,560
Speaker 3: insurers are sort of a pet interest of mine at

897
00:46:47,560 --> 00:46:50,040
Speaker 3: the moment, but I do think like they are emerging

898
00:46:50,040 --> 00:46:52,319
Speaker 3: as some of the really big winners from a lot

899
00:46:52,320 --> 00:46:55,479
Speaker 3: of like I guess, the data saturation of the world

900
00:46:55,560 --> 00:47:00,040
Speaker 3: right now and the increased sophistication of analytical models and

901
00:47:00,120 --> 00:47:02,880
Speaker 3: things like that, and that'll be very interesting to see

902
00:47:02,960 --> 00:47:05,560
Speaker 3: how it shakes out. And you could if you if

903
00:47:05,560 --> 00:47:09,400
Speaker 3: you took it very, very far as a sort of

904
00:47:09,400 --> 00:47:11,600
Speaker 3: thought experiment, you could start to say that, like, well,

905
00:47:11,600 --> 00:47:14,200
Speaker 3: the insurers are going to be a more important actor

906
00:47:14,400 --> 00:47:17,800
Speaker 3: in terms of setting social standards and regulations in the future,

907
00:47:17,800 --> 00:47:19,520
Speaker 3: because they're the ones with the data, doing all the

908
00:47:19,600 --> 00:47:23,160
Speaker 3: modeling and saying like, you have exactly an ex chance

909
00:47:23,239 --> 00:47:26,319
Speaker 3: of being in a car accident, and therefore you must

910
00:47:26,320 --> 00:47:28,080
Speaker 3: do the following things, right.

911
00:47:28,280 --> 00:47:32,799
Speaker 2: I thought also, like I hadn't really thought about, you know,

912
00:47:34,960 --> 00:47:38,120
Speaker 2: subpoena ability. I think it is a very big issue,

913
00:47:38,239 --> 00:47:41,680
Speaker 2: but it is interesting, right, Like, it is a little

914
00:47:41,719 --> 00:47:43,880
Speaker 2: weird that you and I could have a phone conversation

915
00:47:44,000 --> 00:47:45,640
Speaker 2: now if you're under oath and they say, what did

916
00:47:45,680 --> 00:47:47,920
Speaker 2: you talk about? Joe? I expected you'd probably tell the

917
00:47:47,960 --> 00:47:50,640
Speaker 2: truth unless it's very bad for me, and I would

918
00:47:50,640 --> 00:47:51,480
Speaker 2: hope that you would lie.

919
00:47:52,080 --> 00:47:52,480
Speaker 3: That's right.

920
00:47:52,560 --> 00:47:54,120
Speaker 2: Yeah, I hope that you would. I would hope that

921
00:47:54,160 --> 00:47:56,840
Speaker 2: you that's your hope. I would hope that you myself

922
00:47:57,080 --> 00:47:58,600
Speaker 2: to save Joe. Yeah, I would hope that you would

923
00:47:58,600 --> 00:48:01,479
Speaker 2: perjure yourself, but like you know, theoretically could get away

924
00:48:01,480 --> 00:48:04,000
Speaker 2: with it. So yeah, well we have an email, there's

925
00:48:04,040 --> 00:48:06,400
Speaker 2: no chance. And it's sort of interesting. It's sort of

926
00:48:06,640 --> 00:48:09,040
Speaker 2: it's obviously it seems a little bit arbitraryed me, but

927
00:48:09,080 --> 00:48:12,319
Speaker 2: it is interesting to think about, like, Okay, can we

928
00:48:12,400 --> 00:48:15,200
Speaker 2: have a conversation with these entities? And like why do

929
00:48:15,280 --> 00:48:17,839
Speaker 2: we have to leave a digital record? And where are

930
00:48:17,880 --> 00:48:20,120
Speaker 2: these going to be stored? And I do think in

931
00:48:20,160 --> 00:48:24,440
Speaker 2: areas like health and law, which are obviously not obviously

932
00:48:24,560 --> 00:48:28,080
Speaker 2: but sort of intuitively low hanging fruit for productivity gains,

933
00:48:28,480 --> 00:48:31,920
Speaker 2: how much have we not seen, just in part because

934
00:48:31,960 --> 00:48:35,880
Speaker 2: we're still sort of negotiating the transition process as the society.

935
00:48:36,320 --> 00:48:38,120
Speaker 2: What are going to be the new rules and norms

936
00:48:38,120 --> 00:48:40,239
Speaker 2: about this stuff, where it's gonna be hows, et cetera.

937
00:48:40,560 --> 00:48:43,680
Speaker 2: I think that's actually a sort of very interesting question.

938
00:48:43,920 --> 00:48:45,680
Speaker 2: Or space space, This is the.

939
00:48:45,600 --> 00:48:48,080
Speaker 3: Space to watch, yeah speak, Now that I think about it,

940
00:48:48,080 --> 00:48:50,720
Speaker 3: we probably should have discussed some of the regulatory framework

941
00:48:50,760 --> 00:48:53,560
Speaker 3: around all of this a little bit more, But next time,

942
00:48:53,760 --> 00:48:55,520
Speaker 3: next time. But the other thing I've been thinking about

943
00:48:55,560 --> 00:49:00,839
Speaker 3: lately is economic statistics. Yeah, in a world that's increasingly

944
00:49:01,040 --> 00:49:03,880
Speaker 3: driven by AI, and I know that we had the

945
00:49:03,920 --> 00:49:08,799
Speaker 3: big productivity discussion and technology in relation to technology in

946
00:49:08,880 --> 00:49:11,480
Speaker 3: like the sort of early to mid two thousands. I'm

947
00:49:11,520 --> 00:49:14,359
Speaker 3: not sure that ever actually got settled, but I very

948
00:49:14,400 --> 00:49:18,680
Speaker 3: much expect that the AI economic statistics conversation is going

949
00:49:18,760 --> 00:49:22,520
Speaker 3: to be even like wackier because I'm not sure how

950
00:49:22,600 --> 00:49:25,719
Speaker 3: you do things like quality adjustments for something that like

951
00:49:25,800 --> 00:49:29,040
Speaker 3: suddenly comes with its own brain and stuff like that.

952
00:49:29,200 --> 00:49:31,920
Speaker 2: So no, it's gonna be super weird. I did, I

953
00:49:32,320 --> 00:49:34,359
Speaker 2: don't know. In my mind, I have like a very

954
00:49:34,719 --> 00:49:38,239
Speaker 2: these images of like Tyler getting a tour through the

955
00:49:38,600 --> 00:49:41,600
Speaker 2: you know, chatch EPTE offices and the people ask you

956
00:49:41,760 --> 00:49:44,960
Speaker 2: was like, well, and him having to explain that, you know,

957
00:49:44,960 --> 00:49:47,160
Speaker 2: we're not going to have twenty percent GDP growth at

958
00:49:47,160 --> 00:49:50,000
Speaker 2: maybe two and a half or sorry productivity growth, and

959
00:49:50,280 --> 00:49:53,799
Speaker 2: actually GDP isn't really that bad of a measured more

960
00:49:53,920 --> 00:49:56,960
Speaker 2: or less captures the size of the economy, even if

961
00:49:57,000 --> 00:49:58,719
Speaker 2: a lot of Internet things are free. Like some of

962
00:49:58,760 --> 00:50:02,640
Speaker 2: these classic convers stations, I would like to be a

963
00:50:02,640 --> 00:50:05,239
Speaker 2: fly on the wall for some of those.

964
00:50:05,719 --> 00:50:09,319
Speaker 3: Tyler Cowen in defense of GDP dependent Yeah, all right,

965
00:50:09,560 --> 00:50:10,239
Speaker 3: shall we leave it there.

966
00:50:10,320 --> 00:50:11,080
Speaker 2: Let's leave it there.

967
00:50:11,360 --> 00:50:13,640
Speaker 3: This has been another episode of the Odd Lots podcast.

968
00:50:13,719 --> 00:50:16,560
Speaker 3: I'm Tracy Alloway. You can follow me at Tracy Alloway.

969
00:50:16,719 --> 00:50:19,560
Speaker 2: And I'm Joe Wisenthal. You can follow me at The Stalwart.

970
00:50:19,719 --> 00:50:23,360
Speaker 2: Follow our guest Tyler Cowen, He's at Tyler Collen and

971
00:50:23,440 --> 00:50:27,120
Speaker 2: of course check out his podcast Conversations with Tyler an addition,

972
00:50:27,320 --> 00:50:30,960
Speaker 2: of course, Marginal Revolution. Follow our producers Carmen Rodriguez at

973
00:50:31,000 --> 00:50:34,000
Speaker 2: Carman armand dash O Bennett at Dashbot and Cal Brooks

974
00:50:34,000 --> 00:50:36,400
Speaker 2: at Cale Brooks. From our Odd Lots content, go to

975
00:50:36,440 --> 00:50:39,320
Speaker 2: Bloomberg dot com slash odd Lots with the daily newsletter

976
00:50:39,360 --> 00:50:41,840
Speaker 2: and all of our episodes, and you can chat about

977
00:50:41,840 --> 00:50:44,400
Speaker 2: all of these topics twenty four seven in our discord

978
00:50:44,760 --> 00:50:46,920
Speaker 2: discord gg slash odlines.

979
00:50:47,000 --> 00:50:48,960
Speaker 3: And if you enjoy odd Lots, if you like it

980
00:50:49,000 --> 00:50:51,960
Speaker 3: when we talk to Tyler Cowen and give him music recommendations,

981
00:50:52,000 --> 00:50:54,720
Speaker 3: then please leave us a positive review on your favorite

982
00:50:54,719 --> 00:50:58,279
Speaker 3: podcast platform. And remember, if you are a Bloomberg subscriber,

983
00:50:58,320 --> 00:51:01,360
Speaker 3: you can listen to all of our episodes absolutely ad free.

984
00:51:01,480 --> 00:51:03,360
Speaker 3: All you need to do is find the Bloomberg channel

985
00:51:03,360 --> 00:51:06,520
Speaker 3: on Apple Podcasts and follow the instructions there. Thanks for

986
00:51:06,600 --> 00:51:06,960
Speaker 3: listening