1
00:00:04,519 --> 00:00:12,680
Speaker 1: Welcome to Tech Stuff, a production from iHeartRadio. Today we

2
00:00:12,720 --> 00:00:15,680
Speaker 1: are witnessed to one of those rare moments in history,

3
00:00:16,079 --> 00:00:19,279
Speaker 1: the rise of an innovative technology with the potential to

4
00:00:19,440 --> 00:00:24,160
Speaker 1: radically transform business and society forever. That technology, of course,

5
00:00:24,640 --> 00:00:28,200
Speaker 1: is artificial intelligence, and it's the central focus for this

6
00:00:28,360 --> 00:00:32,360
Speaker 1: new season of Smart Talks with IBM. Join hosts from

7
00:00:32,400 --> 00:00:36,120
Speaker 1: your favorite Pushkin podcasts as a talk with industry experts

8
00:00:36,120 --> 00:00:39,760
Speaker 1: and leaders to explore how businesses can integrate AI into

9
00:00:39,760 --> 00:00:43,080
Speaker 1: their workflows and help drive real change in this new

10
00:00:43,200 --> 00:00:46,879
Speaker 1: era of AI. And of course, host Malcolm Gladwell will

11
00:00:46,880 --> 00:00:49,200
Speaker 1: be there to guide you through the season and throw

12
00:00:49,280 --> 00:00:52,040
Speaker 1: in his two cents as well. Look out for new

13
00:00:52,080 --> 00:00:54,960
Speaker 1: episodes of Smart Talks with IBM every other week on

14
00:00:55,000 --> 00:00:59,240
Speaker 1: the iHeartRadio app, Apple Podcasts, wherever you get your podcasts

15
00:00:59,440 --> 00:01:03,680
Speaker 1: and learn more are at IBM dot com slash smart Talks.

16
00:01:06,680 --> 00:01:12,840
Speaker 1: All Right, Welcome everybody, you guys excited, Here we go.

17
00:01:13,160 --> 00:01:16,640
Speaker 2: Hello, Hello, Welcome to Smart Talks with IBM, a podcast

18
00:01:16,680 --> 00:01:22,560
Speaker 2: from Pushkin Industries, iHeartRadio and IBM. I'm Malcolm Gladwell. This season,

19
00:01:22,600 --> 00:01:26,600
Speaker 2: we're continuing our conversations with new creators visionaries who are

20
00:01:26,640 --> 00:01:30,160
Speaker 2: creatively applying technology and business to drive change, but with

21
00:01:30,240 --> 00:01:34,560
Speaker 2: a focus on the transformative power of artificial intelligence and

22
00:01:34,600 --> 00:01:37,479
Speaker 2: what it means to leverage AI as a game changing

23
00:01:37,560 --> 00:01:42,240
Speaker 2: multiplier for your business. Today's episode is a bit different.

24
00:01:42,560 --> 00:01:46,480
Speaker 2: I was recently joined on stage by Dario Gill for

25
00:01:46,520 --> 00:01:48,880
Speaker 2: a conversation in front of a live audience at the

26
00:01:48,920 --> 00:01:53,000
Speaker 2: iHeartMedia headquarters in Manhattan. Dario is the senior vice president

27
00:01:53,240 --> 00:01:56,760
Speaker 2: and director of IBM Research, one of the world's largest

28
00:01:56,840 --> 00:02:01,160
Speaker 2: and most influential corporate research labs. We discussed the rise

29
00:02:01,160 --> 00:02:04,640
Speaker 2: of generative AI, what it means for business and society.

30
00:02:04,960 --> 00:02:08,480
Speaker 2: He also explained how organizations that leverage AI to create

31
00:02:08,560 --> 00:02:13,239
Speaker 2: value will dominate in the near future. Okay, let's get

32
00:02:13,280 --> 00:02:19,840
Speaker 2: on to the conversation. Hello everyone, welcome, and I'm here

33
00:02:19,880 --> 00:02:23,560
Speaker 2: with doctor Dario Gil and I wanted to say before

34
00:02:23,560 --> 00:02:25,160
Speaker 2: we get started. This is something I said backstage that

35
00:02:25,639 --> 00:02:31,920
Speaker 2: I feel very guilty today because you're the you know,

36
00:02:32,680 --> 00:02:34,720
Speaker 2: you know, arguably one of the most important figures in

37
00:02:35,080 --> 00:02:37,799
Speaker 2: AI research in the world, and we have taken you

38
00:02:37,880 --> 00:02:42,160
Speaker 2: away from your job for a morning. It's like, if

39
00:02:42,520 --> 00:02:46,359
Speaker 2: you know, Oppenheimer's wife in nineteen forty four said let's

40
00:02:46,680 --> 00:02:50,440
Speaker 2: go and have a little getaway in the Bahamas. It's

41
00:02:50,520 --> 00:02:52,320
Speaker 2: that kind of thing. You know, What do you say

42
00:02:52,320 --> 00:02:55,400
Speaker 2: to your wife, I can't We have got to work

43
00:02:55,440 --> 00:02:57,680
Speaker 2: on this thing I can't tell you about. She's like

44
00:02:57,760 --> 00:03:00,160
Speaker 2: getting me out of Los Alamos. No, So I do

45
00:03:00,200 --> 00:03:05,040
Speaker 2: feel guilty. We've set back AI research by by about

46
00:03:05,080 --> 00:03:10,080
Speaker 2: four hours here. But I wanted to you've been up

47
00:03:10,080 --> 00:03:13,320
Speaker 2: with with ibo for twenty years, twenty years this summer.

48
00:03:13,680 --> 00:03:15,280
Speaker 2: So and how old were you when you Not to

49
00:03:15,320 --> 00:03:17,040
Speaker 2: give away your age, but you were how old when

50
00:03:17,080 --> 00:03:17,560
Speaker 2: you started?

51
00:03:17,840 --> 00:03:18,560
Speaker 3: I was twenty eight?

52
00:03:18,639 --> 00:03:20,360
Speaker 2: Okay, yeah, So I want to go back to your

53
00:03:20,400 --> 00:03:23,280
Speaker 2: twenty eight year old self. Now, if I asked you

54
00:03:23,360 --> 00:03:26,960
Speaker 2: about artificial intelligence, I asked twenty eight year old Dario,

55
00:03:27,840 --> 00:03:31,080
Speaker 2: what does the future hold for AI? How quickly will

56
00:03:31,520 --> 00:03:35,000
Speaker 2: this new technology transform our world? Et cetera, et cetera.

57
00:03:35,080 --> 00:03:37,080
Speaker 2: What would twenty eight year old Darigo have said?

58
00:03:37,520 --> 00:03:39,760
Speaker 4: Well, I think the first thing is that even though

59
00:03:39,800 --> 00:03:42,160
Speaker 4: AI as a feel has been with us for a

60
00:03:42,200 --> 00:03:45,280
Speaker 4: long time since the mid nineteen fifties, at that time,

61
00:03:45,680 --> 00:03:49,120
Speaker 4: AI was not a very polite word to say, meaning

62
00:03:49,200 --> 00:03:52,600
Speaker 4: within the scientific community, people didn't use sort of that

63
00:03:52,720 --> 00:03:54,800
Speaker 4: term that would have said things like, you know, maybe

64
00:03:54,800 --> 00:03:58,640
Speaker 4: I do things relate to machine learning, right, or statistical

65
00:03:58,680 --> 00:04:00,760
Speaker 4: techniques in terms of classific fires and so on.

66
00:04:01,120 --> 00:04:03,320
Speaker 3: But AI had a mixed.

67
00:04:03,000 --> 00:04:06,000
Speaker 4: Reputation, right, it had gone through different cycles of hype,

68
00:04:06,440 --> 00:04:10,240
Speaker 4: and it's also if moments of you know, a lot

69
00:04:10,280 --> 00:04:15,040
Speaker 4: of negativity towards it because of lack of success. And

70
00:04:15,120 --> 00:04:16,560
Speaker 4: so I think that would be the first thing we

71
00:04:16,640 --> 00:04:19,120
Speaker 4: probably say, like AI is like what is that?

72
00:04:19,240 --> 00:04:21,960
Speaker 3: Like, you know, respectable scientists are not working.

73
00:04:21,720 --> 00:04:25,239
Speaker 4: On AI the fin as such, and that really changed

74
00:04:25,279 --> 00:04:27,640
Speaker 4: over the last fifteen years. Only, right, I would say,

75
00:04:27,640 --> 00:04:30,560
Speaker 4: with the advent of deep learning over the last decade

76
00:04:30,720 --> 00:04:33,679
Speaker 4: is when that re enter again the lexicon of saying

77
00:04:33,680 --> 00:04:36,839
Speaker 4: AI and that that was a legitimate thing to work on.

78
00:04:36,960 --> 00:04:38,440
Speaker 4: So I would say that that's the first thing I

79
00:04:38,440 --> 00:04:40,400
Speaker 4: think we would have noticed a contrast twenty years ago.

80
00:04:40,640 --> 00:04:44,559
Speaker 2: Yeah, So what point in your twenty year tenure at

81
00:04:44,760 --> 00:04:48,520
Speaker 2: IBM would you say you kind of snapped into present

82
00:04:49,160 --> 00:04:50,440
Speaker 2: kind of wow mode.

83
00:04:51,800 --> 00:04:59,680
Speaker 4: I would say in a late two thousands, when IBM

84
00:05:00,520 --> 00:05:05,800
Speaker 4: was working on the Jeopardy project and just seeing the

85
00:05:05,839 --> 00:05:09,279
Speaker 4: demonstrations of what could be done in question answering.

86
00:05:09,720 --> 00:05:12,880
Speaker 2: It's literally Jeopardy is this crucial moment in the history

87
00:05:12,920 --> 00:05:14,680
Speaker 2: of Yeah.

88
00:05:14,160 --> 00:05:16,520
Speaker 4: You know, there had been a long and wonderful history

89
00:05:17,240 --> 00:05:21,000
Speaker 4: inside IBM on AI. So so for example, like you know,

90
00:05:21,080 --> 00:05:23,920
Speaker 4: in terms of like these grand challenges, at the very

91
00:05:23,960 --> 00:05:26,839
Speaker 4: beginning of the field founding, which is this famous Dartmouth

92
00:05:26,880 --> 00:05:31,200
Speaker 4: conference that actually IBM sponsored uh TO to create, there

93
00:05:31,240 --> 00:05:35,800
Speaker 4: was an IBM and there called Nathaniel Rochester and there

94
00:05:35,800 --> 00:05:38,640
Speaker 4: were a few others who right after that they started

95
00:05:38,640 --> 00:05:41,800
Speaker 4: thinking about demonstrations of this field. And then for example,

96
00:05:41,800 --> 00:05:45,039
Speaker 4: they created the first you know game to play checkers

97
00:05:45,320 --> 00:05:48,640
Speaker 4: and to demonstrate that you could do machine learning on that.

98
00:05:49,320 --> 00:05:51,960
Speaker 4: Obviously we saw later in the nineties like chess that

99
00:05:52,040 --> 00:05:54,640
Speaker 4: was a very famous example of that, deep Blue with

100
00:05:54,680 --> 00:05:57,960
Speaker 4: Deep Blue right and playing with Caspar and then but

101
00:05:58,040 --> 00:06:01,120
Speaker 4: I think the moment that was really the ones felt like,

102
00:06:01,240 --> 00:06:03,440
Speaker 4: you know, kind of like brute force anticipating, sort of

103
00:06:03,480 --> 00:06:06,200
Speaker 4: like moves ahead. But this aspect of dealing with language

104
00:06:06,200 --> 00:06:10,560
Speaker 4: and question answering felt different, and I think for for

105
00:06:10,680 --> 00:06:12,919
Speaker 4: us internally and many others, was when a moment of

106
00:06:13,000 --> 00:06:16,000
Speaker 4: saying like wow, you know, what are the possibilities here?

107
00:06:16,440 --> 00:06:19,000
Speaker 4: And then soon after that connected to the sort of

108
00:06:19,040 --> 00:06:22,520
Speaker 4: advancements in computing and with deep learning. The last decade

109
00:06:22,600 --> 00:06:24,520
Speaker 4: has just been an all out, you know, sort of

110
00:06:24,520 --> 00:06:26,960
Speaker 4: like front of advancements and that, and I just continued

111
00:06:27,000 --> 00:06:28,960
Speaker 4: to be more and more impressed. And the last few

112
00:06:29,040 --> 00:06:30,280
Speaker 4: years have been remarkable too.

113
00:06:30,480 --> 00:06:34,960
Speaker 2: Yeah, So I ask you three quick conceptual questions before

114
00:06:34,960 --> 00:06:36,760
Speaker 2: we dig into it, just so I sort of get

115
00:06:36,760 --> 00:06:41,040
Speaker 2: a we all get a feel for the shape of AI.

116
00:06:41,839 --> 00:06:45,880
Speaker 2: Question Number one is where are we in the evolution

117
00:06:46,000 --> 00:06:50,040
Speaker 2: of this? So you know the obvious question. We we're

118
00:06:50,040 --> 00:06:52,640
Speaker 2: all suddenly aware of it, we're talking about it. Can

119
00:06:52,680 --> 00:06:54,760
Speaker 2: you give us an analogy about where we are in

120
00:06:54,800 --> 00:06:57,479
Speaker 2: the kind of likely evolution of this is a technology?

121
00:06:59,640 --> 00:07:03,320
Speaker 4: So I think we're on a significant inflection point that

122
00:07:03,760 --> 00:07:07,840
Speaker 4: it feels the equivalent of the first browsers when they

123
00:07:07,880 --> 00:07:12,080
Speaker 4: appear and people imagine the possibilities of the Internet or

124
00:07:12,120 --> 00:07:16,040
Speaker 4: more imagined experience the internet. The Internet had been around,

125
00:07:16,200 --> 00:07:18,560
Speaker 4: right for quite a few decades. AI has been around

126
00:07:18,840 --> 00:07:21,520
Speaker 4: for many decades. I think the moment we find ourselves

127
00:07:21,560 --> 00:07:24,679
Speaker 4: is that people can touch it, and they can. Before

128
00:07:24,680 --> 00:07:26,840
Speaker 4: they were I systems that were like behind the scenes,

129
00:07:26,880 --> 00:07:31,040
Speaker 4: like your search results or translation systems, but they didn't

130
00:07:31,040 --> 00:07:32,960
Speaker 4: have the experience of like, this is what it feels

131
00:07:32,960 --> 00:07:34,160
Speaker 4: like to interact with this thing.

132
00:07:34,840 --> 00:07:35,960
Speaker 3: So that's what I mean.

133
00:07:36,000 --> 00:07:38,080
Speaker 4: I think maybe that analogy of the browser is appropriate

134
00:07:38,160 --> 00:07:40,360
Speaker 4: because it's all of a sudden, it's like whoa, you know,

135
00:07:40,560 --> 00:07:44,160
Speaker 4: these network of machines and content can be distributed and

136
00:07:44,200 --> 00:07:46,840
Speaker 4: everybody can self publish, and there was a moment that

137
00:07:46,880 --> 00:07:49,000
Speaker 4: we all remember that, and I think that that is

138
00:07:49,080 --> 00:07:51,560
Speaker 4: what the world has experience over the last nine months

139
00:07:51,640 --> 00:07:55,360
Speaker 4: or so on. So but fundamentally, also what is important

140
00:07:55,520 --> 00:07:57,680
Speaker 4: is that this moment is where the ease of the

141
00:07:57,760 --> 00:08:00,640
Speaker 4: number of people that can build and u use AI

142
00:08:01,280 --> 00:08:06,240
Speaker 4: has skyrocketed. So over the last decade, you know, technology

143
00:08:06,280 --> 00:08:10,440
Speaker 4: firms that had large research teams could build AI that

144
00:08:10,560 --> 00:08:14,000
Speaker 4: worked really well, honestly, but when you went down into

145
00:08:14,160 --> 00:08:16,880
Speaker 4: say hey, can everybody use it? Can a data science

146
00:08:16,880 --> 00:08:20,320
Speaker 4: team in a bank, you know, go and develop these applications.

147
00:08:19,920 --> 00:08:22,240
Speaker 3: It was like more complicated. Some could do it, but

148
00:08:22,280 --> 00:08:24,120
Speaker 3: it was more the barrier of entry was high.

149
00:08:24,440 --> 00:08:27,800
Speaker 4: Now it's very different because of foundation models and the

150
00:08:27,840 --> 00:08:29,160
Speaker 4: implications that that has.

151
00:08:28,960 --> 00:08:32,640
Speaker 2: For at the moment where the technology is being democratized.

152
00:08:32,000 --> 00:08:37,480
Speaker 4: In demarketized, frankly, it works better for classes of problems

153
00:08:37,480 --> 00:08:40,120
Speaker 4: like programming and other things, is really incredibly impressive what

154
00:08:40,160 --> 00:08:42,520
Speaker 4: it can do. So the accuracy and the performance of

155
00:08:42,520 --> 00:08:45,440
Speaker 4: it is much better, and the ease of use and

156
00:08:45,480 --> 00:08:47,720
Speaker 4: the number of use cases we can pursue it much bigger,

157
00:08:47,720 --> 00:08:49,440
Speaker 4: So that democratization is a big difference.

158
00:08:49,480 --> 00:08:51,480
Speaker 2: But when you say, when you make it an analogy to

159
00:08:51,520 --> 00:08:55,600
Speaker 2: the first browsers, if you if we do another one

160
00:08:55,640 --> 00:08:58,360
Speaker 2: of these time travel questions back at the beginning of

161
00:08:58,360 --> 00:09:02,800
Speaker 2: the first browsers, to say, many of the potential uses

162
00:09:02,920 --> 00:09:05,880
Speaker 2: of the Internet and such we hadn't even begun. We

163
00:09:05,880 --> 00:09:08,600
Speaker 2: couldn't even anticipate, right, right, So we're at the point

164
00:09:08,600 --> 00:09:11,640
Speaker 2: where the future direction is largely unpredictable.

165
00:09:11,920 --> 00:09:14,719
Speaker 4: Yeah, I think that that is right, because it's such

166
00:09:14,720 --> 00:09:19,080
Speaker 4: a horizontal technology that the intersection of the horizontal capability,

167
00:09:19,160 --> 00:09:23,400
Speaker 4: which is about expanding our productivity and tasks that we

168
00:09:23,440 --> 00:09:26,280
Speaker 4: wouldn't be able to do efficiently without it, has to

169
00:09:26,360 --> 00:09:29,160
Speaker 4: marry now the use cases that reflect the diversity of

170
00:09:29,200 --> 00:09:32,240
Speaker 4: human experience, our institutional diversity. So as more and more

171
00:09:32,280 --> 00:09:34,920
Speaker 4: institutions said, you know, I'm focused on agriculture, you know,

172
00:09:35,040 --> 00:09:38,160
Speaker 4: to be able to improve seeds, you know, in these

173
00:09:38,200 --> 00:09:41,200
Speaker 4: kinds of environments. They'll find their own context in which

174
00:09:41,240 --> 00:09:43,040
Speaker 4: that matters that the creators of a I did not

175
00:09:43,040 --> 00:09:46,000
Speaker 4: anticipate at the beginning. So I think that that is

176
00:09:46,040 --> 00:09:48,480
Speaker 4: then the fruit of surprises will be like why I

177
00:09:48,480 --> 00:09:49,680
Speaker 4: wouldn't even think that it could be.

178
00:09:49,720 --> 00:09:50,199
Speaker 3: Used for that?

179
00:09:50,480 --> 00:09:52,679
Speaker 5: And also clever people will create.

180
00:09:52,400 --> 00:09:55,400
Speaker 4: New business models as associated with that, like it happened

181
00:09:55,440 --> 00:09:58,120
Speaker 4: with the Internet of course as well, and that will

182
00:09:58,120 --> 00:10:01,280
Speaker 4: be its own source of transformation, unchanged in its own right.

183
00:10:01,360 --> 00:10:03,559
Speaker 5: So I think all of that is yet to unfold.

184
00:10:03,679 --> 00:10:03,839
Speaker 2: Right.

185
00:10:03,880 --> 00:10:06,640
Speaker 4: What we're seeing is this catalyst moment of technology that

186
00:10:06,679 --> 00:10:08,680
Speaker 4: works well enough and it can be democratized.

187
00:10:09,600 --> 00:10:11,559
Speaker 2: What next sort of conceptual question?

188
00:10:12,200 --> 00:10:12,360
Speaker 3: You know?

189
00:10:12,640 --> 00:10:19,000
Speaker 2: We can loosely understand or categorize innovations in terms of

190
00:10:19,040 --> 00:10:23,360
Speaker 2: their impact on the kind of balance of power between

191
00:10:23,679 --> 00:10:28,720
Speaker 2: haves and have nots. Some innovations, you know, obviously favor

192
00:10:28,760 --> 00:10:32,920
Speaker 2: those who already have will make the rich richer. Some

193
00:10:32,920 --> 00:10:35,360
Speaker 2: some it's a rising tie the lift cell boats and

194
00:10:35,400 --> 00:10:39,640
Speaker 2: some bias in the other direction, they close the gap between.

195
00:10:40,840 --> 00:10:44,080
Speaker 2: Is it possible to say to predict which of those

196
00:10:44,120 --> 00:10:46,160
Speaker 2: three categories AI might fall into.

197
00:10:47,160 --> 00:10:48,880
Speaker 3: It's a great question, you know.

198
00:10:49,000 --> 00:10:53,880
Speaker 4: A first observation I would make on your first two categories.

199
00:10:53,920 --> 00:10:57,720
Speaker 5: Is that it will be both likely be true that

200
00:10:57,880 --> 00:10:58,680
Speaker 5: the use.

201
00:10:58,520 --> 00:11:00,839
Speaker 4: Of AI will be highly demarketized, meaning the number of

202
00:11:00,880 --> 00:11:04,239
Speaker 4: people that have access to its power to make improvements

203
00:11:04,240 --> 00:11:07,119
Speaker 4: in terms of efficiency and so on will be fairly universal,

204
00:11:07,920 --> 00:11:11,319
Speaker 4: and that the ones who are able to create AI

205
00:11:12,080 --> 00:11:15,319
Speaker 4: UH may be quite concentrated. So if you look at

206
00:11:15,320 --> 00:11:18,920
Speaker 4: it from the lens of who creates wealth and value

207
00:11:19,280 --> 00:11:23,280
Speaker 4: over sustained periods of time, particularly it's saying a context

208
00:11:23,360 --> 00:11:26,840
Speaker 4: like business, I think just being a user of AI

209
00:11:26,960 --> 00:11:31,160
Speaker 4: technology is an insufficient strategy and UH. And the reason

210
00:11:31,240 --> 00:11:33,120
Speaker 4: for that is, like, yes, you will get the immediate

211
00:11:33,120 --> 00:11:36,080
Speaker 4: productivity boost of like just making API calls and you

212
00:11:36,120 --> 00:11:38,760
Speaker 4: know that would be a new baseline for everybody, but

213
00:11:39,200 --> 00:11:42,959
Speaker 4: you're not accruing value in terms of representing your data

214
00:11:43,040 --> 00:11:44,640
Speaker 4: inside the AI in way that it.

215
00:11:44,679 --> 00:11:46,599
Speaker 3: Gives you a sustainable competitive advantage.

216
00:11:47,000 --> 00:11:49,360
Speaker 4: So I always try to tell people is don't just

217
00:11:49,400 --> 00:11:52,200
Speaker 4: be an AI user, be an you know, AI value creator.

218
00:11:52,679 --> 00:11:55,040
Speaker 4: And I think that that will have a lot of

219
00:11:55,120 --> 00:11:58,360
Speaker 4: consequences in terms of the haves and have nots as

220
00:11:58,400 --> 00:12:01,000
Speaker 4: an example, and that will apply both to institutions and

221
00:12:01,040 --> 00:12:04,360
Speaker 4: regions and countries, et cetera. So I think it would

222
00:12:04,360 --> 00:12:07,920
Speaker 4: be kind of a mistake, right to just develop strategies

223
00:12:07,920 --> 00:12:09,200
Speaker 4: that are just about.

224
00:12:09,040 --> 00:12:13,199
Speaker 2: Usage, but to come back to that question from them

225
00:12:13,200 --> 00:12:16,400
Speaker 2: and to give you a specific suppose I'm a I'm

226
00:12:16,400 --> 00:12:21,199
Speaker 2: an industrial farmer in Iowa with ten million dollars of

227
00:12:21,240 --> 00:12:24,880
Speaker 2: equipment and move and I'm comparing it to a subsistence

228
00:12:24,960 --> 00:12:29,040
Speaker 2: farmer somewhere in the developing world who's got a cell phone.

229
00:12:29,840 --> 00:12:34,559
Speaker 2: Over the next five years, who's whose well being rises

230
00:12:34,600 --> 00:12:36,000
Speaker 2: by a greater amount.

231
00:12:37,080 --> 00:12:39,800
Speaker 4: Yeah, I think, I mean, it's a it's a good question,

232
00:12:39,840 --> 00:12:41,480
Speaker 4: but it might be hard to do a one to

233
00:12:41,600 --> 00:12:44,160
Speaker 4: one sort of like attribution to just one variable in

234
00:12:44,200 --> 00:12:48,400
Speaker 4: this case, which is AI. But again, provided that you

235
00:12:48,520 --> 00:12:51,360
Speaker 4: have access to a phone, right and some kind to

236
00:12:51,600 --> 00:12:54,000
Speaker 4: you know, be able to be connected. I do think

237
00:12:54,080 --> 00:12:57,080
Speaker 4: so for example, in that context we've developed, we don't

238
00:12:57,120 --> 00:12:59,600
Speaker 4: work with NASA as an example, to build you spatial

239
00:12:59,640 --> 00:13:03,080
Speaker 4: models using some of these new techniques, and I think

240
00:13:03,120 --> 00:13:06,079
Speaker 4: for example, or ability to do flood prediction, I'll tell

241
00:13:06,080 --> 00:13:08,559
Speaker 4: you an advantage of why it would be a democratization

242
00:13:08,640 --> 00:13:11,680
Speaker 4: force in that context. Before to build a flowed model

243
00:13:12,200 --> 00:13:15,640
Speaker 4: based on satellite imagery was actually so.

244
00:13:15,520 --> 00:13:17,840
Speaker 3: Onerous and so complicated and difficult.

245
00:13:17,400 --> 00:13:19,880
Speaker 4: That you would just target to very specific regions and

246
00:13:19,920 --> 00:13:23,240
Speaker 4: then obviously countries prioritize their own right. But what we've

247
00:13:23,280 --> 00:13:26,040
Speaker 4: demonstrated is actually you can extend that technique to have

248
00:13:26,280 --> 00:13:29,000
Speaker 4: global coverage around that. So in that context, I would

249
00:13:29,040 --> 00:13:32,000
Speaker 4: say it's a four stores democratization that everybody sort of

250
00:13:32,000 --> 00:13:34,840
Speaker 4: would have access if you have some connectivity, as.

251
00:13:34,679 --> 00:13:37,640
Speaker 2: Today, Iowa farmer might have a flood model. The guy

252
00:13:37,720 --> 00:13:40,480
Speaker 2: in the developing world definitely didn't, and now he's a

253
00:13:40,520 --> 00:13:41,200
Speaker 2: shot at getting one.

254
00:13:41,240 --> 00:13:42,679
Speaker 3: Yeah, but now it has a shot of getting one.

255
00:13:42,720 --> 00:13:44,560
Speaker 4: So there's aspects of it that so long as we

256
00:13:44,640 --> 00:13:47,600
Speaker 4: provide connectivity and access to it, that they can be

257
00:13:47,760 --> 00:13:49,040
Speaker 4: democratization forces.

258
00:13:49,160 --> 00:13:49,640
Speaker 3: But I'll give you.

259
00:13:49,640 --> 00:13:53,320
Speaker 4: Another example that that can be quite concerning, which is language. Right,

260
00:13:53,440 --> 00:13:57,559
Speaker 4: So there's so much language in the you know, in English,

261
00:13:58,160 --> 00:14:01,400
Speaker 4: and there is sort of like the reinforcement loop that

262
00:14:01,520 --> 00:14:03,960
Speaker 4: happens that the more you concentrate because it has obvious

263
00:14:03,960 --> 00:14:07,559
Speaker 4: benefits for global communication and standardization, the more you can

264
00:14:07,679 --> 00:14:11,200
Speaker 4: enrich like base AI models based on that capability. If

265
00:14:11,200 --> 00:14:15,319
Speaker 4: you have very resource cars languages, you tend to develop

266
00:14:15,400 --> 00:14:18,840
Speaker 4: less powerful AI with those languages and so on, So

267
00:14:19,000 --> 00:14:23,080
Speaker 4: one has to actually worry and focus on the ability

268
00:14:23,440 --> 00:14:27,160
Speaker 4: to actually represent that in that case is language as

269
00:14:27,160 --> 00:14:29,920
Speaker 4: a piece of culture. Also in the AI sets that

270
00:14:30,080 --> 00:14:33,240
Speaker 4: everybody can benefit from it too. So there's a lot

271
00:14:33,320 --> 00:14:36,600
Speaker 4: of considerations in terms of equity about the data and

272
00:14:36,680 --> 00:14:39,920
Speaker 4: the data sets that we accrue and what problems are

273
00:14:39,920 --> 00:14:42,120
Speaker 4: we trying to solve. I mean, you mentioned agriculture or

274
00:14:42,120 --> 00:14:44,640
Speaker 4: healthcare and so on. If we only solve problems that

275
00:14:44,640 --> 00:14:47,240
Speaker 4: are related to marketing as an example, that would be

276
00:14:47,280 --> 00:14:49,960
Speaker 4: a less rich world in terms of opportunity that if

277
00:14:49,960 --> 00:14:52,680
Speaker 4: we incorporate many many other broader server problems.

278
00:14:53,680 --> 00:14:55,600
Speaker 2: Who do you think what do you think are the

279
00:14:55,640 --> 00:15:01,040
Speaker 2: biggest impediments to the adoption of a as you would

280
00:15:01,120 --> 00:15:03,520
Speaker 2: like as you think aire to be adopted. I mean,

281
00:15:03,560 --> 00:15:05,400
Speaker 2: if you look, what are the sticking points that.

282
00:15:05,400 --> 00:15:08,240
Speaker 4: You would look In the end, I'm going to give

283
00:15:08,280 --> 00:15:10,840
Speaker 4: a non technological answer as a first one has to

284
00:15:10,840 --> 00:15:13,760
Speaker 4: do with workflow, right, So even if the technology is

285
00:15:13,880 --> 00:15:19,000
Speaker 4: very capable, the organizational change inside a company to incorporate

286
00:15:19,000 --> 00:15:21,720
Speaker 4: into the natural workflow of people or how we work

287
00:15:22,320 --> 00:15:24,800
Speaker 4: is It's a lesson we have learned over the last

288
00:15:24,800 --> 00:15:26,320
Speaker 4: decade is hugely important.

289
00:15:26,920 --> 00:15:29,920
Speaker 3: So there's a lot of design considerations.

290
00:15:29,960 --> 00:15:33,000
Speaker 4: There's a lot of how do people want to work right,

291
00:15:33,240 --> 00:15:34,280
Speaker 4: how do they work today?

292
00:15:34,320 --> 00:15:36,400
Speaker 3: And what is the natural entry points for AI? So

293
00:15:36,440 --> 00:15:37,760
Speaker 3: that's like number one.

294
00:15:38,000 --> 00:15:40,760
Speaker 4: And then the second one is, you know, for the

295
00:15:40,800 --> 00:15:45,120
Speaker 4: broad value creation aspect of it is the understanding inside

296
00:15:45,120 --> 00:15:49,480
Speaker 4: the companies of how you have to curate and create

297
00:15:49,560 --> 00:15:53,000
Speaker 4: data to combine it with external data says that you

298
00:15:53,040 --> 00:15:56,280
Speaker 4: can have powerful AI models that actually fit your need.

299
00:15:56,760 --> 00:16:00,400
Speaker 4: And that aspect of what it takes to actually and

300
00:16:00,440 --> 00:16:05,800
Speaker 4: curate the data for this modern AI, it's still working progress, right.

301
00:16:06,120 --> 00:16:08,640
Speaker 4: I think part of the problem that happens very often

302
00:16:08,680 --> 00:16:11,960
Speaker 4: when I talk to institutions is that they say yah, yeah, yeah, yeah,

303
00:16:11,960 --> 00:16:14,520
Speaker 4: I'm doing it. I've been doing it for a long time.

304
00:16:15,240 --> 00:16:17,640
Speaker 4: And the reality is that that answer can sometimes be

305
00:16:17,680 --> 00:16:19,760
Speaker 4: a little of a cop out, right, is like, I

306
00:16:19,800 --> 00:16:22,400
Speaker 4: know you were doing machine learning, you were doing some

307
00:16:22,480 --> 00:16:23,359
Speaker 4: of these things.

308
00:16:23,080 --> 00:16:25,560
Speaker 5: But actually the leader's version of AI.

309
00:16:25,480 --> 00:16:28,200
Speaker 4: What was happening with foundation models, Not only is it

310
00:16:28,320 --> 00:16:31,680
Speaker 4: very new, it's very hard to do and honestly, if

311
00:16:31,720 --> 00:16:34,440
Speaker 4: you haven't been you know, assembling very large teams and

312
00:16:34,480 --> 00:16:36,960
Speaker 4: spending hundreds of millions of dollars of compute, and so

313
00:16:37,120 --> 00:16:39,880
Speaker 4: you're probably not doing it right. You're doing something else

314
00:16:39,920 --> 00:16:42,560
Speaker 4: that is in the broad category. And I think the

315
00:16:42,720 --> 00:16:45,480
Speaker 4: lessons about what it means to make this transition to

316
00:16:45,560 --> 00:16:48,400
Speaker 4: this new wave is still in early phases of understanding.

317
00:16:48,560 --> 00:16:50,160
Speaker 2: So what would you say? I want to give you

318
00:16:50,200 --> 00:16:53,720
Speaker 2: a couple of examples of people with kind of real

319
00:16:53,720 --> 00:16:56,800
Speaker 2: work in real world positions of responsibility. Imagine I'm sitting

320
00:16:56,880 --> 00:16:59,720
Speaker 2: right here, So imagine that I am the president of

321
00:16:59,720 --> 00:17:02,120
Speaker 2: a small liberal arts college and I come to you

322
00:17:02,120 --> 00:17:04,240
Speaker 2: and I say, Dario, I keep hearing about the AI

323
00:17:04,920 --> 00:17:07,680
Speaker 2: my college has. You know, I don't make it. You know,

324
00:17:07,800 --> 00:17:11,040
Speaker 2: I'm I'm I'm making this much money. If that every

325
00:17:11,080 --> 00:17:15,320
Speaker 2: year my enrollments declining, I feel like this maybe is

326
00:17:15,359 --> 00:17:18,720
Speaker 2: an opportunity. What is the opportunity for me? What would

327
00:17:18,720 --> 00:17:19,520
Speaker 2: you say?

328
00:17:20,640 --> 00:17:22,960
Speaker 3: So it's probably in a couple of segments around that.

329
00:17:23,200 --> 00:17:24,840
Speaker 3: Right one has.

330
00:17:24,680 --> 00:17:27,480
Speaker 4: To do is well, what is the implications of this

331
00:17:27,600 --> 00:17:31,919
Speaker 4: technology inside the institution itself instead of the college, and

332
00:17:31,960 --> 00:17:35,800
Speaker 4: how we operate and can we improve for example, efficiency,

333
00:17:35,960 --> 00:17:39,160
Speaker 4: like if you have in very low levels of sort

334
00:17:39,200 --> 00:17:41,959
Speaker 4: of margin to be able to reinvest, is you know

335
00:17:42,160 --> 00:17:46,440
Speaker 4: you run it, You run you know, infrastructure, you run

336
00:17:46,480 --> 00:17:49,000
Speaker 4: many things inside the college. What are the opportunities to

337
00:17:49,000 --> 00:17:52,800
Speaker 4: increase the productivity or automate and drive savings such that

338
00:17:52,880 --> 00:17:55,159
Speaker 4: you can reinvest that money so into the mission of

339
00:17:55,240 --> 00:17:56,840
Speaker 4: education right as an example.

340
00:17:56,520 --> 00:17:58,240
Speaker 2: So number one is operational efficiency.

341
00:17:58,240 --> 00:18:01,399
Speaker 4: Operational efficiencyage is is a big one. I think the

342
00:18:01,400 --> 00:18:03,560
Speaker 4: second one is within the context of the college, there's

343
00:18:03,560 --> 00:18:06,600
Speaker 4: implications for the educational mission on its own, right, how

344
00:18:06,640 --> 00:18:08,400
Speaker 4: will you know how.

345
00:18:08,359 --> 00:18:09,919
Speaker 3: Does a curriculum need to evolve or not?

346
00:18:10,359 --> 00:18:12,840
Speaker 4: What are acceptable use policies or of some of these

347
00:18:12,840 --> 00:18:15,119
Speaker 4: AI I think we've all read a lot about like

348
00:18:15,119 --> 00:18:17,520
Speaker 4: what can happen in terms of exams and so on,

349
00:18:17,560 --> 00:18:19,639
Speaker 4: and cheating and not cheating, or what are the actually

350
00:18:19,720 --> 00:18:22,480
Speaker 4: positive elements of it in terms of how curriculum should

351
00:18:22,480 --> 00:18:25,840
Speaker 4: be developed and professions sustain around that. And then there's

352
00:18:25,840 --> 00:18:28,879
Speaker 4: another third dimension, which is the outdoor oriented element of it,

353
00:18:28,920 --> 00:18:32,760
Speaker 4: which is like prospect students right, so, which is frankly speaking,

354
00:18:32,800 --> 00:18:34,960
Speaker 4: a big use case that is happening right now, which

355
00:18:35,000 --> 00:18:37,440
Speaker 4: in the broader industry is called customer care or client

356
00:18:37,480 --> 00:18:39,840
Speaker 4: care or citizen care. So in this question will be education,

357
00:18:40,000 --> 00:18:42,600
Speaker 4: like you know, hey are you reaching the right students

358
00:18:43,280 --> 00:18:46,200
Speaker 4: around that that may apply to the college. How can

359
00:18:46,240 --> 00:18:48,679
Speaker 4: you create them, for example, an environment to interact with

360
00:18:48,680 --> 00:18:50,720
Speaker 4: the college and answering questions that could be a chat

361
00:18:50,760 --> 00:18:53,800
Speaker 4: bought or something like that to learn about it, and personalization.

362
00:18:54,200 --> 00:18:55,320
Speaker 3: So I would say there.

363
00:18:55,240 --> 00:18:57,360
Speaker 4: Is like at least three lenses with which I would

364
00:18:57,359 --> 00:18:59,959
Speaker 4: give advice right the positive.

365
00:19:00,040 --> 00:19:04,239
Speaker 2: Second, because it's really interesting. So I really can't as

366
00:19:04,280 --> 00:19:07,720
Speaker 2: sign an essay anymore? Can I?

367
00:19:07,720 --> 00:19:09,000
Speaker 3: Can I sign an essay?

368
00:19:09,160 --> 00:19:12,000
Speaker 2: Can I say, rend me a research paper and come

369
00:19:12,040 --> 00:19:13,840
Speaker 2: back to being three? Can I do that anymore?

370
00:19:13,880 --> 00:19:14,520
Speaker 3: I think you can?

371
00:19:14,640 --> 00:19:15,400
Speaker 2: How do I do that?

372
00:19:15,560 --> 00:19:16,240
Speaker 3: And then you can that.

373
00:19:16,560 --> 00:19:21,040
Speaker 4: Look, there's there's two questions around that. I think that

374
00:19:21,320 --> 00:19:22,640
Speaker 4: if one goes.

375
00:19:22,520 --> 00:19:24,440
Speaker 5: And explains in the context like what is it?

376
00:19:24,560 --> 00:19:25,160
Speaker 3: Why are we here?

377
00:19:25,160 --> 00:19:27,159
Speaker 5: Why are in this class? What is the purpose of this?

378
00:19:28,040 --> 00:19:31,560
Speaker 4: And and one starts with assuming like an element of

379
00:19:31,600 --> 00:19:33,760
Speaker 4: like decency and people are people are there like to

380
00:19:33,840 --> 00:19:36,200
Speaker 4: learn and so on, and you just give it disclaimer. Look,

381
00:19:36,400 --> 00:19:38,959
Speaker 4: I know that one option you have is like just

382
00:19:39,000 --> 00:19:41,200
Speaker 4: you know, put the essay question and click go and

383
00:19:41,240 --> 00:19:43,200
Speaker 4: like and give an answer. You know, but that is

384
00:19:43,240 --> 00:19:45,359
Speaker 4: not why we're here, and that is not the intent

385
00:19:45,400 --> 00:19:46,960
Speaker 4: of what we're trying to do. So first I would

386
00:19:47,000 --> 00:19:51,320
Speaker 4: start with the sort of like the norms of intent

387
00:19:51,440 --> 00:19:54,600
Speaker 4: and decency and appeal to those as step number one.

388
00:19:55,119 --> 00:19:57,240
Speaker 4: Then we all know that there will be a distribution

389
00:19:57,320 --> 00:19:59,359
Speaker 4: of use cases of people like that will come in

390
00:19:59,359 --> 00:20:01,199
Speaker 4: one year and come one or the other and do that.

391
00:20:01,760 --> 00:20:04,280
Speaker 4: And so for a subset of that, you know, I

392
00:20:04,280 --> 00:20:06,240
Speaker 4: think the technology is going to have all in such

393
00:20:06,240 --> 00:20:08,679
Speaker 4: a way that we will have more and more of

394
00:20:08,680 --> 00:20:11,440
Speaker 4: the ability to discern right, you know, when that has

395
00:20:11,480 --> 00:20:14,240
Speaker 4: been AI generated right and uncreated.

396
00:20:14,400 --> 00:20:15,960
Speaker 3: It won't be perfect, right.

397
00:20:15,840 --> 00:20:18,399
Speaker 4: But there's some elements that you can imagine in putting

398
00:20:18,400 --> 00:20:20,399
Speaker 4: the essay and you say, hey, this is likely to

399
00:20:20,440 --> 00:20:23,560
Speaker 4: be generated right around that. And for example, one way

400
00:20:23,600 --> 00:20:24,959
Speaker 4: you can do that, just to give you an intuition,

401
00:20:25,080 --> 00:20:27,399
Speaker 4: you could just have an essay that you write with

402
00:20:27,480 --> 00:20:30,840
Speaker 4: pencil and paper. At the beginning, you get a baseline

403
00:20:30,840 --> 00:20:33,520
Speaker 4: of what you're writing is like. And then later when

404
00:20:33,560 --> 00:20:37,440
Speaker 4: you you know generated, there will be obvious differences around

405
00:20:37,480 --> 00:20:39,000
Speaker 4: what kind of writing has been generated.

406
00:20:39,040 --> 00:20:42,360
Speaker 2: On the other way, but you've turned it's everything you're

407
00:20:42,400 --> 00:20:46,280
Speaker 2: describing makes sense. Put it greatly in this In this respect,

408
00:20:46,320 --> 00:20:49,000
Speaker 2: at least it seems to greatly come complicate the life

409
00:20:49,000 --> 00:20:51,920
Speaker 2: of the teacher, whereas the other two use cases seem

410
00:20:52,000 --> 00:20:56,879
Speaker 2: to kind of clarify and simplify the role. Right suddenly,

411
00:20:57,200 --> 00:21:00,959
Speaker 2: you know, reaching student perspective students, sounds like I can

412
00:21:01,000 --> 00:21:03,080
Speaker 2: do that much more kind of efficienty a lot. Yeah,

413
00:21:03,119 --> 00:21:05,720
Speaker 2: I can bring out administration costs, but the teaching thing

414
00:21:05,800 --> 00:21:06,760
Speaker 2: is tricky.

415
00:21:07,880 --> 00:21:11,639
Speaker 4: Well, until we developed the new norms, right, I mean again,

416
00:21:11,800 --> 00:21:14,120
Speaker 4: I mean, I know it's not abuse analogy, but calculators

417
00:21:14,119 --> 00:21:17,680
Speaker 4: we deal. We've done with that too, right, And it says, well, calculator,

418
00:21:17,680 --> 00:21:18,680
Speaker 4: what is the purpose of math?

419
00:21:18,720 --> 00:21:19,520
Speaker 3: How are we going to do this?

420
00:21:19,600 --> 00:21:22,840
Speaker 2: And so can I tell you my dad's calculator story?

421
00:21:22,960 --> 00:21:23,520
Speaker 3: Yes? Please.

422
00:21:23,800 --> 00:21:27,760
Speaker 2: My father was a mathematician, taught mathematics at University of Waterloo, Agada,

423
00:21:28,480 --> 00:21:31,800
Speaker 2: and in the seventies when people started to get pocket calculators,

424
00:21:32,240 --> 00:21:34,840
Speaker 2: his students demanded that they'd be able to use them,

425
00:21:35,000 --> 00:21:36,760
Speaker 2: and he said no, and they took him to the

426
00:21:36,800 --> 00:21:42,600
Speaker 2: administration and he lost. So he then changed completely throughout

427
00:21:42,640 --> 00:21:45,679
Speaker 2: all of his old exams and introduced new exams where

428
00:21:46,080 --> 00:21:50,320
Speaker 2: there was no calculation. It was all like deep think,

429
00:21:50,440 --> 00:21:52,760
Speaker 2: you know, figure out the problem on a conceptual level

430
00:21:52,760 --> 00:21:56,040
Speaker 2: and describe it to me. And they were all students

431
00:21:56,080 --> 00:22:00,520
Speaker 2: deeply unhappy that he'd made their lives for complicated. But

432
00:21:59,760 --> 00:22:03,800
Speaker 2: to your point, I mean he probably the result was

433
00:22:03,840 --> 00:22:07,960
Speaker 2: probably a better education. He just removed the element that

434
00:22:08,000 --> 00:22:10,919
Speaker 2: they could game with their pocket calculators. I suppose it's

435
00:22:10,920 --> 00:22:11,560
Speaker 2: a version.

436
00:22:11,359 --> 00:22:13,119
Speaker 4: Of I think it's a version of that, And so

437
00:22:13,200 --> 00:22:14,919
Speaker 4: I think they will develop the equivalent of what your

438
00:22:14,920 --> 00:22:16,520
Speaker 4: father did. And I think people say, you know what

439
00:22:16,600 --> 00:22:18,920
Speaker 4: if like these kinds of things, everybody's doing it generically

440
00:22:19,000 --> 00:22:21,200
Speaker 4: and none of us have any meaning because all you're

441
00:22:21,240 --> 00:22:23,240
Speaker 4: doing is pressing buttons, and like the intent of this

442
00:22:23,400 --> 00:22:24,960
Speaker 4: was something which was to teach you how to write

443
00:22:25,160 --> 00:22:26,119
Speaker 4: or to think or something.

444
00:22:26,400 --> 00:22:28,160
Speaker 5: There may be a variant of how we do all

445
00:22:28,160 --> 00:22:29,520
Speaker 5: of this. I mean, obviously some.

446
00:22:29,560 --> 00:22:31,720
Speaker 4: Version of that that has happened is like Okay, we're

447
00:22:31,720 --> 00:22:33,240
Speaker 4: all going to sit down and doing with pencil and

448
00:22:33,280 --> 00:22:35,800
Speaker 4: paper and computers in their classroom. But there'll be other

449
00:22:35,880 --> 00:22:38,680
Speaker 4: variants of creativity that people will put forth to say,

450
00:22:38,680 --> 00:22:40,760
Speaker 4: you know what, you know, that's a way to solve

451
00:22:40,760 --> 00:22:41,400
Speaker 4: that problem too.

452
00:22:41,480 --> 00:22:44,520
Speaker 2: But this is interesting because to stay on this analogy,

453
00:22:45,160 --> 00:22:50,119
Speaker 2: we're really talking about a profound rethinking just using a

454
00:22:50,200 --> 00:22:54,080
Speaker 2: college as an example, a real profound rethinking of the

455
00:22:54,119 --> 00:22:57,000
Speaker 2: way there's no part of this college it's unaffected by

456
00:22:57,240 --> 00:23:02,280
Speaker 2: aia B. In one case, I've made everyone's job easier.

457
00:23:02,320 --> 00:23:05,280
Speaker 2: In one case, I've made I'm asking us to really

458
00:23:05,320 --> 00:23:09,560
Speaker 2: rethink from the ground up what teaching means. In another case,

459
00:23:09,560 --> 00:23:11,600
Speaker 2: I've automated systems that I didn't think of it. I mean,

460
00:23:11,640 --> 00:23:14,680
Speaker 2: it's like, that's right, that's it's not that's a lot

461
00:23:14,720 --> 00:23:17,399
Speaker 2: to ask someone who got a pH d in medieval

462
00:23:17,480 --> 00:23:19,800
Speaker 2: language literature, you know, forty years ago.

463
00:23:20,440 --> 00:23:22,640
Speaker 4: Yeah, but you know, I'll tell you a positive sort

464
00:23:22,640 --> 00:23:25,320
Speaker 4: of development that I'm seeing the sciences around this, which

465
00:23:25,400 --> 00:23:28,760
Speaker 4: is you're seen as you see more and more examples

466
00:23:29,080 --> 00:23:33,320
Speaker 4: of applying AI technology within the context of like historians

467
00:23:33,320 --> 00:23:36,800
Speaker 4: to as an example, Right, you have archival and you know,

468
00:23:36,800 --> 00:23:38,439
Speaker 4: and you have all these books and being able to

469
00:23:38,520 --> 00:23:41,600
Speaker 4: actually help you as an assistant right around that, but

470
00:23:41,640 --> 00:23:44,880
Speaker 4: not only with text now, but with diagrams, right. And

471
00:23:45,240 --> 00:23:48,960
Speaker 4: I've seen it in anthropology too, write and archaeology with

472
00:23:49,119 --> 00:23:52,359
Speaker 4: examples of engravings and translations and things that can happen.

473
00:23:52,720 --> 00:23:56,280
Speaker 4: So so as you see in diverse fields people applying

474
00:23:56,320 --> 00:23:58,800
Speaker 4: these techniques to advance and how to do physics or

475
00:23:58,840 --> 00:24:02,160
Speaker 4: how to do chemistry. They inspire each other, right, and

476
00:24:02,160 --> 00:24:04,000
Speaker 4: this said, you know, how does it apply actually to

477
00:24:04,040 --> 00:24:07,680
Speaker 4: my area? So once as that happens, it becomes less

478
00:24:07,680 --> 00:24:09,239
Speaker 4: of a chore of like, my god, you know, how

479
00:24:09,240 --> 00:24:11,359
Speaker 4: do I have to deal with this? But actually it's

480
00:24:11,359 --> 00:24:15,040
Speaker 4: triggered by Curiosity's triggered by you know, there'll be like,

481
00:24:15,200 --> 00:24:16,840
Speaker 4: you know, faculty that will be like, you know what,

482
00:24:16,960 --> 00:24:19,240
Speaker 4: you know, let me explore what this means for my area,

483
00:24:19,560 --> 00:24:21,879
Speaker 4: and they will adapt it to the local context, to

484
00:24:21,920 --> 00:24:25,480
Speaker 4: the local you know, language, and the professional itself. So

485
00:24:25,760 --> 00:24:28,560
Speaker 4: I see that as a positive vector that is not

486
00:24:28,600 --> 00:24:29,880
Speaker 4: all going to feel like homework.

487
00:24:30,119 --> 00:24:31,760
Speaker 5: You know, it's not going to feel like, oh my god,

488
00:24:31,760 --> 00:24:33,560
Speaker 5: this is so overwhelming.

489
00:24:33,200 --> 00:24:35,520
Speaker 4: But rather to be very practical to see what works,

490
00:24:35,680 --> 00:24:37,800
Speaker 4: What have I seen others to do that is inspiring?

491
00:24:38,000 --> 00:24:40,000
Speaker 3: And what am I inspired to do? You know what?

492
00:24:40,000 --> 00:24:42,000
Speaker 3: What is how is this going to help my career?

493
00:24:42,280 --> 00:24:44,040
Speaker 4: I think that that's going to be an interesting question

494
00:24:44,160 --> 00:24:47,119
Speaker 4: for you know, those faculty members, for the students.

495
00:24:46,720 --> 00:24:49,720
Speaker 2: The professionals. Sorry, I'm gonna stick with this example along

496
00:24:49,760 --> 00:24:52,439
Speaker 2: because it's really interesting. I'm curious following up on what

497
00:24:52,520 --> 00:24:55,919
Speaker 2: you just said, that one of the most persistent critiques

498
00:24:55,960 --> 00:25:00,320
Speaker 2: of academic but also of many of many corporate institutions

499
00:25:01,320 --> 00:25:05,280
Speaker 2: in recent UITs, has been siloing. Right, different parts of

500
00:25:05,320 --> 00:25:08,480
Speaker 2: the of the organization are going off on their own

501
00:25:08,480 --> 00:25:11,280
Speaker 2: and not speaking to each other. Is a potential is

502
00:25:12,359 --> 00:25:17,520
Speaker 2: a real potential benefit to AI the kind of breaking

503
00:25:17,560 --> 00:25:21,440
Speaker 2: down a simple tool for breaking down those kinds of barriers,

504
00:25:21,440 --> 00:25:23,320
Speaker 2: is that a very is that an elegant way of

505
00:25:23,359 --> 00:25:25,200
Speaker 2: sort of saying what.

506
00:25:24,640 --> 00:25:25,159
Speaker 3: I really think?

507
00:25:25,200 --> 00:25:28,119
Speaker 4: And I was actually just having a conversation with Provos

508
00:25:28,119 --> 00:25:30,840
Speaker 4: stuff and very much on this topic, very recently exactly

509
00:25:30,840 --> 00:25:34,000
Speaker 4: on that, which is all these this you know, this

510
00:25:34,119 --> 00:25:37,040
Speaker 4: appetite right to collaborate across disciplines. There's a lot of

511
00:25:37,520 --> 00:25:41,960
Speaker 4: attempts stores a goal, right, creating interdisciplinary centers, creating dual

512
00:25:42,000 --> 00:25:46,240
Speaker 4: degree programs or dual appointment programs. But actually in a

513
00:25:46,320 --> 00:25:50,600
Speaker 4: lot of progress in academia happens by methodology too. Write

514
00:25:50,640 --> 00:25:53,680
Speaker 4: like a new you know, when when some methodology gets adopted,

515
00:25:53,880 --> 00:25:56,040
Speaker 4: I mean the most famous example of that is a

516
00:25:56,080 --> 00:25:58,679
Speaker 4: scientific method as an example of that. But when you

517
00:25:58,760 --> 00:26:01,840
Speaker 4: have a methodology that gets adopted, it also provides a

518
00:26:01,920 --> 00:26:05,879
Speaker 4: way to speak to your colleagues across different disciplines. And

519
00:26:05,920 --> 00:26:08,879
Speaker 4: I think what's happened in AI is linked to that

520
00:26:08,880 --> 00:26:12,120
Speaker 4: that within the context of the scientific method as an example,

521
00:26:12,680 --> 00:26:17,359
Speaker 4: the methodology about we about what we do discovery, the

522
00:26:17,480 --> 00:26:20,119
Speaker 4: role of data, the role of these neural networks, of

523
00:26:20,200 --> 00:26:23,040
Speaker 4: how we actually find proximity to concepts to one another

524
00:26:23,480 --> 00:26:28,240
Speaker 4: is actually fundamentally different than how we've traditionally applied it.

525
00:26:28,640 --> 00:26:31,960
Speaker 4: So as we see across more professions, people applying this

526
00:26:32,080 --> 00:26:35,439
Speaker 4: methodology is also going to give some element of common

527
00:26:35,520 --> 00:26:38,959
Speaker 4: language to each other. Right And in fact, you know,

528
00:26:39,119 --> 00:26:42,399
Speaker 4: in this very high dimensional representation of information that is

529
00:26:42,400 --> 00:26:46,439
Speaker 4: pressent to neural networks, we may find amazing adjacencies or

530
00:26:46,440 --> 00:26:50,240
Speaker 4: connections of themes and topics in ways that the individual

531
00:26:50,280 --> 00:26:53,800
Speaker 4: practitioners cannot describe, but yet will be latent.

532
00:26:53,720 --> 00:26:55,240
Speaker 3: In these large calural networks.

533
00:26:56,080 --> 00:26:58,480
Speaker 4: We are going to suffer a little bit from causality,

534
00:26:58,520 --> 00:27:00,640
Speaker 4: from the problem of like, hey, what's the root cause

535
00:27:00,720 --> 00:27:04,600
Speaker 4: of that? Because I think one of the unsatisfying aspects

536
00:27:04,640 --> 00:27:07,560
Speaker 4: that this methodology will provide is they may give you

537
00:27:07,640 --> 00:27:10,480
Speaker 4: answers for which they don't give you good reasons for

538
00:27:10,600 --> 00:27:13,840
Speaker 4: where the answers came from, and then there will be

539
00:27:13,880 --> 00:27:16,639
Speaker 4: the traditional process of discovery of saying, if that is

540
00:27:16,680 --> 00:27:19,440
Speaker 4: the answer, what are the reasons? So we're going to

541
00:27:19,560 --> 00:27:22,720
Speaker 4: have to do this sort of hybrid way of understanding

542
00:27:22,760 --> 00:27:25,320
Speaker 4: the world. But I do think that common layer of

543
00:27:25,359 --> 00:27:26,960
Speaker 4: AI is a powerful new thing.

544
00:27:27,280 --> 00:27:30,560
Speaker 2: Yeah, well, a couple of random questions. I couldn't manage

545
00:27:30,600 --> 00:27:33,560
Speaker 2: you talk in the In the Writer's strike that just

546
00:27:33,680 --> 00:27:36,680
Speaker 2: ended in Hollywood, one of the sticking points was how

547
00:27:36,720 --> 00:27:40,920
Speaker 2: the studios and writers would treat AI generated content. Good

548
00:27:40,960 --> 00:27:44,880
Speaker 2: writers get credit if their material with somehow the source

549
00:27:44,960 --> 00:27:49,679
Speaker 2: for A but more broadly, did the writers need protections

550
00:27:49,720 --> 00:27:51,720
Speaker 2: against the use of I could go on? You know what,

551
00:27:51,920 --> 00:27:54,400
Speaker 2: I'll be worth familiar with all of this. Had you been,

552
00:27:54,440 --> 00:27:57,240
Speaker 2: I don't know whether you were, but had either side

553
00:27:57,280 --> 00:28:01,159
Speaker 2: called you in for advice? During that the writers, the

554
00:28:01,200 --> 00:28:03,920
Speaker 2: writers called you and said, Daria, what should we do

555
00:28:04,119 --> 00:28:06,879
Speaker 2: about AI? And how should we that should be reflect

556
00:28:06,880 --> 00:28:10,080
Speaker 2: should that be reflected in our contract negotiations? What would

557
00:28:10,080 --> 00:28:12,200
Speaker 2: you have told them?

558
00:28:12,880 --> 00:28:14,919
Speaker 5: The way I think about that is that I divided.

559
00:28:15,000 --> 00:28:16,440
Speaker 3: I would divide it into two pieces.

560
00:28:16,600 --> 00:28:20,840
Speaker 4: First, is what's technically possible right and anticipate scenarios like

561
00:28:21,680 --> 00:28:24,200
Speaker 4: you know, what can you do with voice cloning for example?

562
00:28:24,520 --> 00:28:27,240
Speaker 4: You know, now, for example, it is possible this being

563
00:28:28,480 --> 00:28:31,280
Speaker 4: dubbing right legis take that topic right around the world.

564
00:28:31,280 --> 00:28:34,159
Speaker 4: There was all these folks that would dub people in

565
00:28:34,240 --> 00:28:38,080
Speaker 4: other languages. Well, now you can do these incredible renderings.

566
00:28:38,160 --> 00:28:40,200
Speaker 4: I mean, I know if you've seen them, where you

567
00:28:40,240 --> 00:28:42,800
Speaker 4: know you match the lips is your original voice, but

568
00:28:42,880 --> 00:28:45,400
Speaker 4: speaking any language that you want. As an example, so

569
00:28:45,480 --> 00:28:47,640
Speaker 4: busy that has a set of implications around that. I mean,

570
00:28:47,720 --> 00:28:49,320
Speaker 4: just to give an example, So I would say, create

571
00:28:49,360 --> 00:28:53,560
Speaker 4: a taxonomy that describes technical capabilities that we know of

572
00:28:53,640 --> 00:28:58,440
Speaker 4: today and applications to the industry and to examples of

573
00:28:58,520 --> 00:29:00,000
Speaker 4: like hey, you know I could film you for five

574
00:29:00,080 --> 00:29:02,080
Speaker 4: minutes and I could generate two hours of content of

575
00:29:02,120 --> 00:29:04,000
Speaker 4: you and I don't have to you know, then if

576
00:29:04,040 --> 00:29:06,120
Speaker 4: you'll get paid by the hour, obviously I'm not paying

577
00:29:06,120 --> 00:29:08,320
Speaker 4: you for that other thing. So I would say technological

578
00:29:08,320 --> 00:29:12,280
Speaker 4: capability and then map with their expertise consequences of how

579
00:29:12,320 --> 00:29:14,560
Speaker 4: it changes the way they work or the way they

580
00:29:14,600 --> 00:29:16,840
Speaker 4: interact or the way they negotiate and so on.

581
00:29:16,880 --> 00:29:19,120
Speaker 3: So that would be one element of it, and then the.

582
00:29:19,080 --> 00:29:22,000
Speaker 4: Other one is like a non technology related matter, which

583
00:29:22,040 --> 00:29:25,000
Speaker 4: is an element of almost distributed justices like who deserves

584
00:29:25,000 --> 00:29:27,320
Speaker 4: what right and who has the power to get what?

585
00:29:28,160 --> 00:29:31,920
Speaker 4: And then that's a completely different discussion. That is to say, well,

586
00:29:31,960 --> 00:29:34,760
Speaker 4: if this is the scenario of what's possible, you know,

587
00:29:34,840 --> 00:29:37,560
Speaker 4: what do we want and what are we able to get?

588
00:29:38,000 --> 00:29:40,200
Speaker 4: And I think that that's a different discussion, which.

589
00:29:40,040 --> 00:29:42,200
Speaker 2: Is which all that life? Which when do you do?

590
00:29:42,320 --> 00:29:42,600
Speaker 3: First?

591
00:29:43,840 --> 00:29:47,600
Speaker 4: I think it's very helpful to have an understanding of

592
00:29:47,640 --> 00:29:50,800
Speaker 4: what's possible and how it changes a landscape.

593
00:29:51,200 --> 00:29:53,400
Speaker 3: As part of a broader.

594
00:29:53,560 --> 00:29:58,239
Speaker 4: Discussion, right, and a broad negotiation, because you also have

595
00:29:58,280 --> 00:30:00,280
Speaker 4: to see the opportunities because there will be a lot

596
00:30:00,360 --> 00:30:03,240
Speaker 4: of ground to say, actually, you know, if we can

597
00:30:03,280 --> 00:30:05,760
Speaker 4: do it in this way and we can all be

598
00:30:05,840 --> 00:30:08,480
Speaker 4: that much more efficient in getting these piece work done

599
00:30:08,600 --> 00:30:12,120
Speaker 4: or these filming done, but we have a reasonable agreement

600
00:30:12,280 --> 00:30:16,360
Speaker 4: about how we both sides benefit from it, right, then

601
00:30:16,480 --> 00:30:16,800
Speaker 4: that's a.

602
00:30:16,760 --> 00:30:17,760
Speaker 3: Win win for everybody.

603
00:30:17,840 --> 00:30:18,040
Speaker 2: Yeah.

604
00:30:18,160 --> 00:30:21,160
Speaker 5: Right, so that's I think that would be a golden triangle.

605
00:30:21,240 --> 00:30:21,360
Speaker 3: Right.

606
00:30:21,440 --> 00:30:24,040
Speaker 2: Here's my reading, and I would like you to correct

607
00:30:24,040 --> 00:30:25,800
Speaker 2: me if I'm wrong, and I'm likely to be wrong.

608
00:30:26,920 --> 00:30:29,000
Speaker 2: When I looked at that strike, I said if they're

609
00:30:29,040 --> 00:30:33,040
Speaker 2: worried about AI. The writers are worried about AI. That

610
00:30:33,160 --> 00:30:35,720
Speaker 2: seems silly. It should be the studios who are worried

611
00:30:35,760 --> 00:30:38,200
Speaker 2: about the economic impact of AI. Does it? In the

612
00:30:38,200 --> 00:30:41,280
Speaker 2: long run AI put the studios out of business long

613
00:30:41,320 --> 00:30:43,240
Speaker 2: before it puts the writers out of business. I only

614
00:30:43,240 --> 00:30:47,239
Speaker 2: need the studio because the costs of production are as

615
00:30:47,320 --> 00:30:49,920
Speaker 2: high as the sky, and the cost of production are overwhelming.

616
00:30:49,960 --> 00:30:53,200
Speaker 2: And whereas if I don't, if I have a tool

617
00:30:53,240 --> 00:30:58,600
Speaker 2: which brings introduces massive technological efficiencies to the production of movies,

618
00:30:59,040 --> 00:31:02,360
Speaker 2: then why don't studio? Why would they the scared ones?

619
00:31:02,520 --> 00:31:04,560
Speaker 4: Or maybe maybe you need like a different kind of

620
00:31:04,560 --> 00:31:06,520
Speaker 4: studio or a different kind of different kind of studies.

621
00:31:06,560 --> 00:31:09,680
Speaker 2: But I mean the in the but in the in

622
00:31:09,720 --> 00:31:12,840
Speaker 2: this strike the fright the frightened ones with the writers

623
00:31:12,880 --> 00:31:17,080
Speaker 2: and the you know, with the studios. Wasn't that backwards?

624
00:31:18,520 --> 00:31:19,480
Speaker 3: I haven't thought about it.

625
00:31:19,640 --> 00:31:21,719
Speaker 5: Uh, it can be about the implications of it.

626
00:31:21,720 --> 00:31:23,960
Speaker 4: It goes back to we're talking before the implications because

627
00:31:23,960 --> 00:31:26,600
Speaker 4: are so horizontal, it is right to think about it,

628
00:31:26,640 --> 00:31:27,200
Speaker 4: like what does.

629
00:31:27,080 --> 00:31:28,560
Speaker 3: It do to the studios as well? Right?

630
00:31:28,720 --> 00:31:29,920
Speaker 2: Yeah?

631
00:31:29,960 --> 00:31:32,560
Speaker 4: But then you know the reason why that happens is

632
00:31:32,600 --> 00:31:36,640
Speaker 4: that it's the order of either negotiations or or who

633
00:31:36,720 --> 00:31:40,760
Speaker 4: first got concerned about it and did something about it, right,

634
00:31:40,800 --> 00:31:43,560
Speaker 4: which is in the context of the strike. You know,

635
00:31:43,680 --> 00:31:46,080
Speaker 4: I don't know what the equivalent conversations are going inside

636
00:31:46,080 --> 00:31:47,960
Speaker 4: the studio and whether they have a war room saying

637
00:31:47,960 --> 00:31:49,719
Speaker 4: what this is going to mean to us? Right, but

638
00:31:50,000 --> 00:31:52,600
Speaker 4: it doesn't get exercise through a strike, but maybe through

639
00:31:52,600 --> 00:31:54,960
Speaker 4: a task force inside you know, the companies about what

640
00:31:55,000 --> 00:31:55,560
Speaker 4: are they going to do?

641
00:31:55,640 --> 00:31:55,800
Speaker 3: Right?

642
00:31:56,040 --> 00:31:58,280
Speaker 2: Well, and to go back to your thing you said,

643
00:31:58,280 --> 00:31:59,560
Speaker 2: the first thing you do is you make a list

644
00:31:59,600 --> 00:32:04,320
Speaker 2: of what techological capabilities are. But don't technological capabilities change

645
00:32:04,320 --> 00:32:07,880
Speaker 2: every I mean they do. You're racing ahead so fast,

646
00:32:08,080 --> 00:32:11,040
Speaker 2: so you can't Can you have a contract? I'm sorry

647
00:32:11,040 --> 00:32:13,000
Speaker 2: for begetting in the little weeds here, but this is interesting.

648
00:32:13,440 --> 00:32:16,239
Speaker 2: Can you You can't have a five year contract if

649
00:32:16,280 --> 00:32:20,000
Speaker 2: the contract is based on an assessment of technological capabilities

650
00:32:20,000 --> 00:32:22,240
Speaker 2: in twenty twenty three, because by the time we get

651
00:32:22,240 --> 00:32:28,080
Speaker 2: to twenty eight twenty three eight, it's totally different. Right.

652
00:32:28,200 --> 00:32:30,280
Speaker 4: Yeah, But like you know, I mean where I was

653
00:32:30,320 --> 00:32:34,200
Speaker 4: going is like there are some abstractions around that is like,

654
00:32:35,240 --> 00:32:37,680
Speaker 4: you know, whate can we do with my image? Right, Like,

655
00:32:37,720 --> 00:32:40,239
Speaker 4: if I generally get the category that my image can

656
00:32:40,280 --> 00:32:43,360
Speaker 4: be reproduced, generated contents and so on, it's like, let's

657
00:32:43,360 --> 00:32:45,760
Speaker 4: talk about the abstract notion about who has rights to

658
00:32:45,840 --> 00:32:48,520
Speaker 4: that or do we both get to benefit from that?

659
00:32:48,720 --> 00:32:51,520
Speaker 4: If you get that straight, Yes, the nature of how

660
00:32:51,520 --> 00:32:54,880
Speaker 4: the image gets alter created as something will change underneath,

661
00:32:55,080 --> 00:32:57,640
Speaker 4: but the concept will stay the same. And so I

662
00:32:57,680 --> 00:32:59,880
Speaker 4: think what's important is to get the categories right.

663
00:33:00,200 --> 00:33:04,160
Speaker 2: Yeah. Yeah, if you had to you to think about

664
00:33:04,200 --> 00:33:10,480
Speaker 2: the biggest technological revolutions of the post war era last

665
00:33:10,600 --> 00:33:14,040
Speaker 2: seventy five years, you can all come up with a list. Actually,

666
00:33:14,040 --> 00:33:15,479
Speaker 2: it's really fun to come up with a list. I

667
00:33:15,520 --> 00:33:18,920
Speaker 2: was thinking about this when we were you know, containerized

668
00:33:18,920 --> 00:33:25,719
Speaker 2: shipping is my favorite, the green revolution, the internet is

669
00:33:25,960 --> 00:33:27,320
Speaker 2: Where is the I in that list?

670
00:33:29,960 --> 00:33:32,760
Speaker 4: So I would put it first in that context that

671
00:33:32,800 --> 00:33:36,920
Speaker 4: you put forth over since World War Two, undoubtedly, like

672
00:33:37,080 --> 00:33:41,320
Speaker 4: computing as a category is one of those trajectories that

673
00:33:41,440 --> 00:33:43,040
Speaker 4: has reshaped.

674
00:33:42,560 --> 00:33:43,320
Speaker 3: Right or world.

675
00:33:43,760 --> 00:33:47,600
Speaker 4: And I think we think computing, I would say the

676
00:33:47,720 --> 00:33:52,160
Speaker 4: role that semiconductors have had has been increwdly defining. I

677
00:33:52,160 --> 00:33:56,160
Speaker 4: would say AI is the second example of that as

678
00:33:56,200 --> 00:33:59,560
Speaker 4: a core architecture that is going to have an equivalent

679
00:33:59,640 --> 00:34:02,400
Speaker 4: level of impact. And then the third leg I would

680
00:34:02,400 --> 00:34:03,520
Speaker 4: put to that equation.

681
00:34:03,200 --> 00:34:04,880
Speaker 3: Will be quantum and quantum information.

682
00:34:05,400 --> 00:34:07,400
Speaker 4: And that's sort of like I like to summarize that

683
00:34:07,440 --> 00:34:09,960
Speaker 4: the future of computing its bits, neurons, and cubits, and

684
00:34:10,040 --> 00:34:13,319
Speaker 4: it is that idea of high precision computation, the world

685
00:34:13,360 --> 00:34:15,960
Speaker 4: of neural networks and artificial intelligence, and the world of

686
00:34:16,040 --> 00:34:19,400
Speaker 4: quantum and the combination of those things is going to

687
00:34:19,440 --> 00:34:21,880
Speaker 4: be the defining force of the next hundred years in

688
00:34:21,920 --> 00:34:23,160
Speaker 4: that category of computing.

689
00:34:23,200 --> 00:34:24,640
Speaker 3: But it makes the list for sure.

690
00:34:24,880 --> 00:34:27,239
Speaker 2: If it's that high up on the list. This is

691
00:34:27,280 --> 00:34:31,080
Speaker 2: a total hypothetical. Would you if you were starting over

692
00:34:31,560 --> 00:34:35,080
Speaker 2: if you're starting IBM right now, would you say, oh,

693
00:34:35,360 --> 00:34:38,839
Speaker 2: our AI operations actually should be way bigger, Like how

694
00:34:38,840 --> 00:34:40,880
Speaker 2: many how many thousands of people working for you?

695
00:34:41,760 --> 00:34:45,080
Speaker 4: So within the research division it's about like three thousand,

696
00:34:45,120 --> 00:34:46,040
Speaker 4: five hundred scientists.

697
00:34:46,040 --> 00:34:48,200
Speaker 2: So in a perfect world, would you if it's that big,

698
00:34:48,280 --> 00:34:51,280
Speaker 2: isn't that too small as a group?

699
00:34:51,719 --> 00:34:53,839
Speaker 4: Yeah, Well that's like in the ricer division. I mean

700
00:34:53,880 --> 00:34:57,879
Speaker 4: IBM overall, but I mean, like.

701
00:34:58,000 --> 00:35:01,080
Speaker 2: So starting from first, so you have a we've got

702
00:35:01,120 --> 00:35:05,880
Speaker 2: a technology that you're ranking with compute and you know,

703
00:35:06,160 --> 00:35:08,800
Speaker 2: up there with as in terms of a world changer,

704
00:35:11,000 --> 00:35:13,359
Speaker 2: are we So what I'm basically asking is are we

705
00:35:13,480 --> 00:35:16,440
Speaker 2: underinvested in this huge you.

706
00:35:16,360 --> 00:35:18,919
Speaker 4: Know, but so so yeah, it's a it's a good question.

707
00:35:19,000 --> 00:35:20,799
Speaker 4: So like what I would say is that I think

708
00:35:20,800 --> 00:35:23,839
Speaker 4: we should segment how many people do you need on

709
00:35:24,000 --> 00:35:27,680
Speaker 4: the creation of the technology itself, and what is the

710
00:35:27,760 --> 00:35:30,480
Speaker 4: right size of research and engineers and compute to do that,

711
00:35:31,080 --> 00:35:33,359
Speaker 4: And how many people do you need in the sort

712
00:35:33,400 --> 00:35:38,239
Speaker 4: of application of the technology to create better products, to

713
00:35:38,320 --> 00:35:41,960
Speaker 4: deliver services and consulting and then ultimately to diffuse it

714
00:35:42,000 --> 00:35:44,719
Speaker 4: through you know, sort of all spheres of society. And

715
00:35:44,760 --> 00:35:47,240
Speaker 4: the numbers are very different, and that is not different

716
00:35:47,239 --> 00:35:49,279
Speaker 4: than anywhere else. I mean, I mean, if you give

717
00:35:49,320 --> 00:35:52,000
Speaker 4: examples of since you were talking about in context of

718
00:35:52,000 --> 00:35:54,080
Speaker 4: World War two, how many people does it take to

719
00:35:54,120 --> 00:35:57,440
Speaker 4: create you know, an atomic weapon as an example, it's

720
00:35:57,480 --> 00:35:58,200
Speaker 4: a large number.

721
00:35:58,280 --> 00:35:59,760
Speaker 3: I mean, it wasn't just loss animals.

722
00:35:59,800 --> 00:36:01,600
Speaker 4: There was a lot of people in Okay, it's a

723
00:36:01,640 --> 00:36:04,920
Speaker 4: large number, but it wasn't a million people, right, So

724
00:36:05,440 --> 00:36:09,360
Speaker 4: you could have highly concentrated teams of people that, with

725
00:36:09,560 --> 00:36:13,880
Speaker 4: enough resources, can do extraordinary scientific and technological achievements, and

726
00:36:13,920 --> 00:36:16,160
Speaker 4: that's always by definition, is going to be a fraction

727
00:36:16,320 --> 00:36:19,040
Speaker 4: of like one percent compared to the total volume that

728
00:36:19,120 --> 00:36:20,240
Speaker 4: is going to require to then.

729
00:36:20,120 --> 00:36:20,560
Speaker 3: Deal with it.

730
00:36:21,200 --> 00:36:24,080
Speaker 2: But the application side is infinite almost.

731
00:36:23,760 --> 00:36:24,520
Speaker 3: That's exactly.

732
00:36:24,600 --> 00:36:27,040
Speaker 5: So that is where like in the end, the bottleneck

733
00:36:27,080 --> 00:36:27,560
Speaker 5: really is.

734
00:36:28,040 --> 00:36:31,680
Speaker 4: So with you know, thousands of scientists and engineers, you

735
00:36:31,719 --> 00:36:35,480
Speaker 4: can create world class AI, right, And so no, you

736
00:36:35,480 --> 00:36:37,719
Speaker 4: don't need ten thousand to be able to create the

737
00:36:37,800 --> 00:36:39,799
Speaker 4: large language model and the generatic model and some but

738
00:36:39,840 --> 00:36:43,480
Speaker 4: you need thousands, and you need you know, very significant

739
00:36:43,480 --> 00:36:44,479
Speaker 4: amount of computer and data.

740
00:36:44,520 --> 00:36:44,960
Speaker 3: You need that.

741
00:36:45,640 --> 00:36:49,640
Speaker 4: The rest is okay, I build software, I build databases,

742
00:36:49,760 --> 00:36:52,440
Speaker 4: or I build a software product that allows you to

743
00:36:52,480 --> 00:36:55,320
Speaker 4: do inventory management, or I build you know, a photo

744
00:36:55,440 --> 00:37:00,840
Speaker 4: editor and so on. Now that product incorporating the AI, modifying,

745
00:37:00,960 --> 00:37:03,880
Speaker 4: expanding it and so on. Well, now you're talking about

746
00:37:03,920 --> 00:37:06,800
Speaker 4: the entire software industries. So now you're talking about millions

747
00:37:06,840 --> 00:37:09,480
Speaker 4: of people right who are neted, you know, who are

748
00:37:09,560 --> 00:37:12,680
Speaker 4: required to bring AI into their product. Then you go

749
00:37:12,760 --> 00:37:15,759
Speaker 4: on a step beyond the technology creators in terms of

750
00:37:15,800 --> 00:37:18,600
Speaker 4: software and you say, well, okay, now what the skills

751
00:37:18,640 --> 00:37:22,880
Speaker 4: to help organizations go undeployed in the department of you know,

752
00:37:22,960 --> 00:37:25,279
Speaker 4: the interior, right, And then I said, okay, well, now

753
00:37:25,320 --> 00:37:28,759
Speaker 4: you need like consultants and experts and people to work

754
00:37:28,800 --> 00:37:31,160
Speaker 4: they are to integer into the workflow. So now you're

755
00:37:31,200 --> 00:37:33,960
Speaker 4: talking into the many tens of millions of people around that.

756
00:37:34,239 --> 00:37:36,600
Speaker 4: So I see it as these concentric circles of it.

757
00:37:37,080 --> 00:37:40,040
Speaker 4: But to some degree in many of these core technology areas,

758
00:37:40,080 --> 00:37:41,600
Speaker 4: just saying like, well, I need a team of like

759
00:37:41,600 --> 00:37:43,960
Speaker 4: one hundred thousand people to create like AI or a

760
00:37:44,280 --> 00:37:46,799
Speaker 4: or a new transistor or a new quantum computer. It's

761
00:37:46,840 --> 00:37:49,040
Speaker 4: actually a diminished in return right in the end, like

762
00:37:49,120 --> 00:37:50,920
Speaker 4: too many people connecting with each other.

763
00:37:50,800 --> 00:37:53,480
Speaker 2: Is very difficult. But on the application side of just

764
00:37:53,880 --> 00:37:58,600
Speaker 2: to go back to our our example of that college,

765
00:37:59,160 --> 00:38:03,200
Speaker 2: just a task of sitting down with a faculty and

766
00:38:03,320 --> 00:38:07,640
Speaker 2: working with them to reimagine what they do with these

767
00:38:07,680 --> 00:38:10,319
Speaker 2: new set of tools in mind, with the understanding that

768
00:38:10,360 --> 00:38:12,200
Speaker 2: the students coming in are probably going to know more

769
00:38:12,200 --> 00:38:14,839
Speaker 2: about it than they do that a lot. I mean,

770
00:38:14,880 --> 00:38:18,680
Speaker 2: that's that is a hurriculeion people problem.

771
00:38:18,880 --> 00:38:19,720
Speaker 3: It's a people problem.

772
00:38:19,800 --> 00:38:21,760
Speaker 4: Yeah, that's why I started in terms of the barriers

773
00:38:21,760 --> 00:38:23,800
Speaker 4: of adoption of that, I mean the context of IBM,

774
00:38:23,840 --> 00:38:28,680
Speaker 4: an example That's why we have a consulting organization, Ivan Consulting,

775
00:38:28,719 --> 00:38:32,799
Speaker 4: that complements ib and technology, and the Ivan Consulting Organization

776
00:38:32,880 --> 00:38:35,840
Speaker 4: has over one hundred and fifty thousand employees because of

777
00:38:35,880 --> 00:38:38,279
Speaker 4: this question, right, because you have to sit down and

778
00:38:38,320 --> 00:38:40,920
Speaker 4: you say, Okay, what problem are you trying to solve?

779
00:38:41,239 --> 00:38:43,160
Speaker 4: What is a methodology we're going to do, and here's

780
00:38:43,160 --> 00:38:45,200
Speaker 4: the technology options that we have to be able to

781
00:38:45,200 --> 00:38:49,279
Speaker 4: bring into the table. In the end, the adoption across

782
00:38:49,920 --> 00:38:54,239
Speaker 4: or society will be limited by this part. The technology

783
00:38:54,280 --> 00:38:56,640
Speaker 4: is going to make it easier, more cost effective to

784
00:38:56,719 --> 00:39:01,319
Speaker 4: implement those solutions. You first have to think about what

785
00:39:01,360 --> 00:39:03,239
Speaker 4: you want to do, how you're going to do it,

786
00:39:03,360 --> 00:39:04,719
Speaker 4: and how are you going to bring it into a

787
00:39:04,760 --> 00:39:08,120
Speaker 4: life of this in this context faculty member or you know,

788
00:39:08,200 --> 00:39:10,200
Speaker 4: the administrator and so on in the college.

789
00:39:10,480 --> 00:39:13,279
Speaker 2: With that Hollywood that that notion I thought, which was

790
00:39:13,320 --> 00:39:18,600
Speaker 2: absolutely I thought really interesting that in a Hollywood strike

791
00:39:18,680 --> 00:39:22,160
Speaker 2: you have to have this conversation about a distributive justice,

792
00:39:22,360 --> 00:39:25,200
Speaker 2: conversation about how do we That's it's a really hard

793
00:39:25,239 --> 00:39:28,560
Speaker 2: conversation right to have and a So this brings me

794
00:39:28,560 --> 00:39:30,080
Speaker 2: to my netflie, which is that you we were talking

795
00:39:30,080 --> 00:39:34,360
Speaker 2: a backstage you have. You have two daughters, one in college,

796
00:39:34,400 --> 00:39:37,239
Speaker 2: one about to go to college. That's right, so they're

797
00:39:37,280 --> 00:39:41,719
Speaker 2: both scients minded. So tell me about the conversations you

798
00:39:41,719 --> 00:39:44,560
Speaker 2: you have with your daughter. You have a unique conversation

799
00:39:44,680 --> 00:39:47,680
Speaker 2: with your daughters because your conversation, your advice to them

800
00:39:47,880 --> 00:39:51,280
Speaker 2: is is influenced by what you do for a living.

801
00:39:51,560 --> 00:39:52,400
Speaker 3: Yes, it's true.

802
00:39:52,560 --> 00:39:56,880
Speaker 2: So did you warn your daughters away from certain fields?

803
00:39:56,920 --> 00:40:02,440
Speaker 2: Did you say, whatever you do, don't be No, no.

804
00:40:01,440 --> 00:40:04,040
Speaker 4: No, that's not my style. I mean, for me, no,

805
00:40:04,320 --> 00:40:06,439
Speaker 4: I try not to be like you know, preachy about that.

806
00:40:07,400 --> 00:40:09,760
Speaker 4: So for me, it was just about showing by example

807
00:40:09,840 --> 00:40:12,520
Speaker 4: all things I love, right and things I care about,

808
00:40:12,920 --> 00:40:14,920
Speaker 4: and then you know, bringing them to the lab and

809
00:40:14,960 --> 00:40:18,440
Speaker 4: seeing things, and then the natural conversations of things working

810
00:40:18,480 --> 00:40:21,800
Speaker 4: on or interesting people I meet, so to the extent

811
00:40:21,840 --> 00:40:23,759
Speaker 4: that they have chosen that and obviously this has an

812
00:40:23,800 --> 00:40:27,920
Speaker 4: influence on them. It has been through seeing it, you know,

813
00:40:28,000 --> 00:40:29,400
Speaker 4: perhaps through my eyes, right.

814
00:40:29,280 --> 00:40:30,560
Speaker 3: And what do you see me do? And that I

815
00:40:30,640 --> 00:40:31,600
Speaker 3: like my profession? Right.

816
00:40:31,680 --> 00:40:34,600
Speaker 2: But one of your daughters, you said, is thinking that

817
00:40:34,640 --> 00:40:37,920
Speaker 2: she wants to be a doctor. But being a doctor

818
00:40:38,000 --> 00:40:40,520
Speaker 2: in a post AI world is surely a very different

819
00:40:40,520 --> 00:40:43,160
Speaker 2: proposition than being a doctor in a pre AI world.

820
00:40:43,600 --> 00:40:46,279
Speaker 2: Do you think you have you tried to prepare her

821
00:40:46,320 --> 00:40:49,120
Speaker 2: for that difference. Have you explained to her what you

822
00:40:49,200 --> 00:40:51,319
Speaker 2: think will happen to this profession she might enter.

823
00:40:51,840 --> 00:40:54,760
Speaker 4: Yeah, I mean not in like, you know, incredible amount

824
00:40:54,800 --> 00:40:57,640
Speaker 4: of detail, but but but yes, at the level of

825
00:40:57,760 --> 00:41:02,000
Speaker 4: understanding what is changing, like lens of the information, lens

826
00:41:02,040 --> 00:41:03,920
Speaker 4: with which you can look at the world and what

827
00:41:04,080 --> 00:41:07,600
Speaker 4: is possible and what it can do, Like what is

828
00:41:07,600 --> 00:41:09,960
Speaker 4: our role and what is the role of the technology

829
00:41:09,960 --> 00:41:12,839
Speaker 4: and how that shapes At that level of abstraction, for sure,

830
00:41:13,160 --> 00:41:15,760
Speaker 4: but not at the level of like, don't be a radiologist,

831
00:41:15,880 --> 00:41:17,319
Speaker 4: you know, because this is what we.

832
00:41:17,239 --> 00:41:19,320
Speaker 2: Want for you. I was gonna say, if you'ren't happy

833
00:41:19,320 --> 00:41:21,359
Speaker 2: with your current job, you could do a podcast called

834
00:41:21,360 --> 00:41:25,240
Speaker 2: Parenting Tips with Dario, which is just an AI person

835
00:41:25,840 --> 00:41:28,080
Speaker 2: gives you advice on what your kids should do based

836
00:41:28,120 --> 00:41:30,720
Speaker 2: on exactly this, Like should I be a radiologist? Dario

837
00:41:30,920 --> 00:41:35,240
Speaker 2: tell me, Like it seems to be a really important question. Yeah,

838
00:41:35,560 --> 00:41:37,360
Speaker 2: let me ask this question in a more I'm joking,

839
00:41:37,400 --> 00:41:41,640
Speaker 2: but in a more serious way. Surely it would If

840
00:41:41,719 --> 00:41:43,320
Speaker 2: I don't mean to use your daughter as an example.

841
00:41:43,360 --> 00:41:45,560
Speaker 2: But let's imagine we're giving advice to someone who wants

842
00:41:45,600 --> 00:41:49,920
Speaker 2: to enter medicine. A really useful conversation to have is

843
00:41:50,440 --> 00:41:54,120
Speaker 2: what are the skills that are will be most prized

844
00:41:55,040 --> 00:41:58,239
Speaker 2: in that profession fifteen years from now, and are they

845
00:41:58,280 --> 00:42:00,799
Speaker 2: different from the skills that are prized now. How would

846
00:42:00,840 --> 00:42:01,880
Speaker 2: you answer that question?

847
00:42:02,719 --> 00:42:03,000
Speaker 3: Yeah?

848
00:42:03,320 --> 00:42:07,160
Speaker 4: I think for example, this goes back to how is

849
00:42:07,160 --> 00:42:10,080
Speaker 4: the scientific method in this context, like the practice of

850
00:42:10,120 --> 00:42:12,839
Speaker 4: medicine going to change. I think we will see more

851
00:42:12,920 --> 00:42:15,319
Speaker 4: changes on how we practice a scientific method and so

852
00:42:15,360 --> 00:42:19,360
Speaker 4: on as a consequence of what is happening with the

853
00:42:19,360 --> 00:42:22,600
Speaker 4: world of computing and information. How we represent information, how

854
00:42:22,600 --> 00:42:26,120
Speaker 4: we represent knowledge, how we extract meaning from knowledge as

855
00:42:26,120 --> 00:42:29,120
Speaker 4: a method than we have seen in the last two

856
00:42:29,160 --> 00:42:32,880
Speaker 4: hundred years. So therefore, what I would like strongly encourage

857
00:42:32,960 --> 00:42:35,080
Speaker 4: is not about like, hey, use this tool for doing

858
00:42:35,120 --> 00:42:38,080
Speaker 4: this or doing that, but in the curriculum itself, in

859
00:42:38,320 --> 00:42:41,920
Speaker 4: understanding how we do problems solving in the age of

860
00:42:42,040 --> 00:42:45,120
Speaker 4: like data and data representation and so on, that needs

861
00:42:45,120 --> 00:42:48,400
Speaker 4: to be embedded in the curriculum of everybody you know.

862
00:42:48,440 --> 00:42:51,240
Speaker 3: That is I would say, actually quite horizontally.

863
00:42:50,640 --> 00:42:53,360
Speaker 4: But certainly in the context of medicine and scientists and

864
00:42:53,360 --> 00:42:56,600
Speaker 4: so on for sure, And to the extent that that

865
00:42:56,800 --> 00:42:59,400
Speaker 4: gets ingrained, that will give us a lens that no

866
00:42:59,480 --> 00:43:03,480
Speaker 4: matter what what specialty they go within medicine, they will say, Actually,

867
00:43:03,880 --> 00:43:06,440
Speaker 4: the way I want to be able to tackle improving

868
00:43:06,440 --> 00:43:08,799
Speaker 4: the quality of care, the way to do that is

869
00:43:08,920 --> 00:43:11,320
Speaker 4: in addition to all the elements that we have practiced

870
00:43:11,719 --> 00:43:14,120
Speaker 4: in the field of medicine, is this new lens?

871
00:43:14,360 --> 00:43:14,920
Speaker 3: And are we.

872
00:43:14,920 --> 00:43:17,160
Speaker 4: Representing the data the right way? Do we have the

873
00:43:17,239 --> 00:43:20,080
Speaker 4: right tools to be able to represent that knowledge? Am

874
00:43:20,080 --> 00:43:23,160
Speaker 4: I incorporating that in my own so with my own

875
00:43:23,239 --> 00:43:25,320
Speaker 4: knowledge in a way that gives me better outcomes?

876
00:43:25,400 --> 00:43:25,640
Speaker 3: Right?

877
00:43:25,760 --> 00:43:29,080
Speaker 4: Do I have the rigor of benchmarking too? And quality

878
00:43:29,400 --> 00:43:31,839
Speaker 4: of the results? So that is what needs to be incorporated.

879
00:43:32,000 --> 00:43:37,759
Speaker 2: How in a perfect world, if I asked you your

880
00:43:37,800 --> 00:43:42,320
Speaker 2: team to rewrite curriculum for American medical schools, how dramatic

881
00:43:42,440 --> 00:43:45,640
Speaker 2: a revision is that? Are we tinkering with ten percent

882
00:43:45,680 --> 00:43:47,880
Speaker 2: of the curriculum or we tinkering with fifty percent of it?

883
00:43:50,080 --> 00:43:54,279
Speaker 4: I think they would be a subset of classes that

884
00:43:54,400 --> 00:43:56,800
Speaker 4: is about the method the methodology, what has changed? Like

885
00:43:57,280 --> 00:44:01,000
Speaker 4: have these lens of it to understand? And then within

886
00:44:01,239 --> 00:44:05,799
Speaker 4: each class that methodology will represent something that is embedded

887
00:44:05,880 --> 00:44:10,799
Speaker 4: in it, right, so it will be substantive but not

888
00:44:11,000 --> 00:44:14,320
Speaker 4: But but doesn't mean replacing the specialization and the context

889
00:44:14,320 --> 00:44:16,759
Speaker 4: and the knowledge of each domain. But I do think

890
00:44:16,840 --> 00:44:20,839
Speaker 4: everybody should have sort of a basic knowledge of the horizontal, right,

891
00:44:20,880 --> 00:44:23,480
Speaker 4: what is it, how does it work, what tools you have,

892
00:44:23,840 --> 00:44:25,839
Speaker 4: what is the technology, and like you know what are.

893
00:44:25,800 --> 00:44:27,160
Speaker 3: The dos and don'ts around that?

894
00:44:27,640 --> 00:44:29,799
Speaker 4: And then every area you say, and you know that

895
00:44:29,880 --> 00:44:32,920
Speaker 4: thing that you learn, this is how it applies to anatomy,

896
00:44:33,160 --> 00:44:34,880
Speaker 4: and this is how you know how it applies to

897
00:44:35,080 --> 00:44:37,600
Speaker 4: you know, radiology if you study that, or or this

898
00:44:37,640 --> 00:44:39,960
Speaker 4: is how you apply you know, in the context of discovery,

899
00:44:40,040 --> 00:44:42,000
Speaker 4: right of self structure and this is how we can

900
00:44:42,080 --> 00:44:44,480
Speaker 4: use it. Or protein folding and this is how it

901
00:44:44,800 --> 00:44:47,960
Speaker 4: does so that way you'll see a connecting tissue through

902
00:44:48,360 --> 00:44:49,160
Speaker 4: throughout the whole thing.

903
00:44:49,440 --> 00:44:52,359
Speaker 2: Yeah, I mean I would add to that because I

904
00:44:52,400 --> 00:44:57,360
Speaker 2: was thinking of the that it's also this incredible opportunity

905
00:44:57,400 --> 00:45:00,239
Speaker 2: to do what doctors are supposed to do but don't

906
00:45:00,280 --> 00:45:03,160
Speaker 2: have time to do now, which is they're so consumed

907
00:45:03,200 --> 00:45:07,799
Speaker 2: with figuring out what's wrong with you that they have

908
00:45:07,880 --> 00:45:11,600
Speaker 2: little time to talk about the implications of the diagnosisness

909
00:45:11,800 --> 00:45:14,720
Speaker 2: and what we really want to if we can freedom

910
00:45:14,719 --> 00:45:17,600
Speaker 2: of some of the burden of what is actually quite

911
00:45:17,600 --> 00:45:20,120
Speaker 2: a prosaic question of what's wrong with you and leave

912
00:45:20,160 --> 00:45:23,160
Speaker 2: the hard human thing of let make should you be

913
00:45:23,320 --> 00:45:26,880
Speaker 2: scared or hopeful? Should you you know? What do you

914
00:45:27,000 --> 00:45:28,719
Speaker 2: need to do? Or what? Let me put this in

915
00:45:28,760 --> 00:45:31,360
Speaker 2: the context of all the patients I've seen. That conversation,

916
00:45:31,400 --> 00:45:33,040
Speaker 2: which is the most important one, is the one that

917
00:45:33,760 --> 00:45:36,000
Speaker 2: seems to me. So like, if I had to, I

918
00:45:36,040 --> 00:45:39,880
Speaker 2: would add, if we're reimagining the curriculum of med school,

919
00:45:40,239 --> 00:45:43,879
Speaker 2: I'd like with whatever, by the way, very little time.

920
00:45:43,880 --> 00:45:46,880
Speaker 2: Maybe we have to add two more years to med school,

921
00:45:47,000 --> 00:45:52,120
Speaker 2: but like a whole but the whole thing about bringing

922
00:45:52,200 --> 00:45:56,840
Speaker 2: back the human side of you know, now, if I

923
00:45:56,840 --> 00:45:59,440
Speaker 2: can give you ten more minutes, how do you use

924
00:45:59,480 --> 00:46:00,520
Speaker 2: that ten more men?

925
00:46:00,640 --> 00:46:03,759
Speaker 4: But in that, in that reconceptualization that you just did,

926
00:46:04,120 --> 00:46:06,080
Speaker 4: is what we should be doing around that, because I

927
00:46:06,080 --> 00:46:08,920
Speaker 4: think the debate as to like, well I'm not going

928
00:46:09,000 --> 00:46:10,719
Speaker 4: to need doctors or not, it's actually a not very

929
00:46:10,760 --> 00:46:13,960
Speaker 4: useful debate. But rather this other question is how is

930
00:46:14,000 --> 00:46:16,720
Speaker 4: your time being spent? What problems are you getting stuck.

931
00:46:17,040 --> 00:46:19,880
Speaker 4: I mean I generalize this by like the obvious observation

932
00:46:20,040 --> 00:46:22,160
Speaker 4: that if you look around in your professions, in our

933
00:46:22,239 --> 00:46:24,839
Speaker 4: daily lives, we have not run out of problems to solve.

934
00:46:25,120 --> 00:46:27,200
Speaker 4: So as an example of that is, hey, if I'm

935
00:46:27,200 --> 00:46:29,319
Speaker 4: spending all my time trying to do diagnosis, and I

936
00:46:29,320 --> 00:46:31,600
Speaker 4: could do that ten times faster, and it allows me

937
00:46:31,680 --> 00:46:34,560
Speaker 4: actually to go, you know, and take care of the

938
00:46:34,600 --> 00:46:36,560
Speaker 4: patients and all the next steps of what we have

939
00:46:36,600 --> 00:46:38,719
Speaker 4: to do about it. That's probably a trade off that

940
00:46:38,760 --> 00:46:42,160
Speaker 4: a lot of doctors would take, right, And then you say, well,

941
00:46:42,200 --> 00:46:43,880
Speaker 4: you know, to what degree does it allow me to

942
00:46:43,920 --> 00:46:45,839
Speaker 4: do that? And I can do these other things and

943
00:46:45,880 --> 00:46:49,200
Speaker 4: these other things are critically important for my profession around that.

944
00:46:49,640 --> 00:46:52,920
Speaker 4: So when you actually become less abstract and like we

945
00:46:53,080 --> 00:46:56,160
Speaker 4: get past the futile conversation of like, oh, there's no

946
00:46:56,200 --> 00:46:57,920
Speaker 4: more jobs and I's going to take it all of it,

947
00:46:57,920 --> 00:47:01,040
Speaker 4: which is kind of nonsense, is you go back to say,

948
00:47:01,080 --> 00:47:05,120
Speaker 4: in practice in your context, right, for you, what does

949
00:47:05,160 --> 00:47:06,680
Speaker 4: it mean, how do you work?

950
00:47:06,840 --> 00:47:08,480
Speaker 3: What can you do differently around that?

951
00:47:08,600 --> 00:47:11,040
Speaker 4: Actually that's a much richer conversation, and very often we

952
00:47:11,080 --> 00:47:13,279
Speaker 4: would find ourselves that there's a portion of the work

953
00:47:13,320 --> 00:47:15,439
Speaker 4: we do that we say I would rather do less

954
00:47:15,440 --> 00:47:17,880
Speaker 4: of that. This is this other part I like a lot,

955
00:47:18,200 --> 00:47:21,040
Speaker 4: And if it is possible that technology could help us

956
00:47:21,040 --> 00:47:22,360
Speaker 4: make that trade off, I'll.

957
00:47:22,160 --> 00:47:23,080
Speaker 3: Take it in a heartbeat.

958
00:47:23,800 --> 00:47:27,479
Speaker 4: Now, poorly implemented technology can also create another problem.

959
00:47:27,640 --> 00:47:29,480
Speaker 3: You say, Hey, this was supposed to solve.

960
00:47:29,200 --> 00:47:32,399
Speaker 4: Me things, but the way it's being implemented is not

961
00:47:32,440 --> 00:47:35,200
Speaker 4: helping me, right, it's making my life more more miserable,

962
00:47:35,320 --> 00:47:37,920
Speaker 4: or so on, or I've lost connection in how I

963
00:47:38,000 --> 00:47:41,400
Speaker 4: used to work, et cetera. So that is why design

964
00:47:42,080 --> 00:47:45,200
Speaker 4: is so important. That is why I also workflow is

965
00:47:45,239 --> 00:47:48,000
Speaker 4: so important in being able to solve these problems. But

966
00:47:48,840 --> 00:47:52,360
Speaker 4: it begins by, you know, going from the intergalactic to

967
00:47:52,440 --> 00:47:54,840
Speaker 4: the reality of it, of that faculty member in the

968
00:47:54,840 --> 00:47:57,759
Speaker 4: liberal Arts college or you know, or a you know,

969
00:47:57,920 --> 00:48:00,560
Speaker 4: a practitioner in medicine in a hospital and what it

970
00:48:00,680 --> 00:48:01,399
Speaker 4: means for them.

971
00:48:01,520 --> 00:48:01,680
Speaker 3: Right.

972
00:48:02,400 --> 00:48:06,640
Speaker 2: Yeah. What struck me Daria throughout our conversation is how

973
00:48:06,719 --> 00:48:11,880
Speaker 2: much of this revolution is non technical, as to say,

974
00:48:12,160 --> 00:48:14,120
Speaker 2: you guys are doing the technical thing here, but the

975
00:48:14,160 --> 00:48:17,080
Speaker 2: real the revolution is going to require a whole range

976
00:48:17,120 --> 00:48:20,840
Speaker 2: of people doing things that have nothing to do with software,

977
00:48:21,080 --> 00:48:24,520
Speaker 2: that have to do with working out new new human arrangements.

978
00:48:24,960 --> 00:48:27,600
Speaker 2: Talking about that, I mean, does keep coming back to

979
00:48:27,640 --> 00:48:30,400
Speaker 2: the Hollywood strike thing that you have to have a

980
00:48:30,400 --> 00:48:37,120
Speaker 2: conversation about our values is creators of of of of movies?

981
00:48:37,160 --> 00:48:39,920
Speaker 2: How are we going to divide up the exactly credit

982
00:48:39,960 --> 00:48:44,319
Speaker 2: and the like. That's a that's a conversation about philosophy,

983
00:48:44,400 --> 00:48:46,319
Speaker 2: and you know, you know it's it is.

984
00:48:46,280 --> 00:48:49,600
Speaker 5: And is it's in the grand tradition of why you know,

985
00:48:51,080 --> 00:48:52,160
Speaker 5: a liberal.

986
00:48:51,960 --> 00:48:55,359
Speaker 4: Education is so important in the broadest possible sense. Right,

987
00:48:55,760 --> 00:48:59,480
Speaker 4: there's no common conception of the good right that is

988
00:48:59,520 --> 00:49:03,799
Speaker 4: always tested a dialogue that happens within our society, and

989
00:49:03,920 --> 00:49:06,239
Speaker 4: technology is going to fit in that context too, right.

990
00:49:06,280 --> 00:49:08,520
Speaker 4: So that's why I personally, as a philosophy I'm not

991
00:49:08,560 --> 00:49:12,280
Speaker 4: a technological determinists, right, And I don't like when colleagues

992
00:49:12,280 --> 00:49:14,960
Speaker 4: in my profession right starts saying like, well, this is

993
00:49:15,000 --> 00:49:17,560
Speaker 4: the way the technology is going to be, and by consequence,

994
00:49:17,760 --> 00:49:19,719
Speaker 4: this is how society is going to be. I'm like,

995
00:49:19,800 --> 00:49:22,920
Speaker 4: that's a highly contested goal. And if you want to

996
00:49:23,000 --> 00:49:25,560
Speaker 4: enter into realm of politics or the real other ones,

997
00:49:25,560 --> 00:49:28,120
Speaker 4: go and stand up on a stool and discuss whether

998
00:49:28,200 --> 00:49:30,279
Speaker 4: that's what society wants. You will find that it's a

999
00:49:30,400 --> 00:49:34,560
Speaker 4: huge diversity of opinions and perspective and that's what makes

1000
00:49:34,600 --> 00:49:37,200
Speaker 4: you know, you know, in a democracy, the richness of

1001
00:49:37,239 --> 00:49:39,440
Speaker 4: our society, and in the end that is going to

1002
00:49:39,440 --> 00:49:42,399
Speaker 4: be the centerpiece of the conversation what do we want?

1003
00:49:43,320 --> 00:49:45,879
Speaker 4: You know, who gets what? And so on? And that

1004
00:49:46,200 --> 00:49:48,480
Speaker 4: is actually, I don't think it's anything negative. That's acid

1005
00:49:48,520 --> 00:49:51,400
Speaker 4: should be because in the end is anchored of who

1006
00:49:51,480 --> 00:49:55,000
Speaker 4: we want as humans, you know, you know, as friends, family, citizens,

1007
00:49:55,160 --> 00:49:57,960
Speaker 4: and we have many overlapping sets of responsibilities, right and

1008
00:49:58,000 --> 00:50:01,120
Speaker 4: as a technology creator my only version ponsibilities not just

1009
00:50:01,160 --> 00:50:02,960
Speaker 4: as a scientist and a technology creator.

1010
00:50:03,320 --> 00:50:04,400
Speaker 3: I'm also a member.

1011
00:50:04,239 --> 00:50:06,000
Speaker 4: Of a family, I'm a citizen, and I'm many other

1012
00:50:06,080 --> 00:50:08,160
Speaker 4: things that I care about. And I think that that

1013
00:50:08,400 --> 00:50:13,560
Speaker 4: sometimes in the debate of the technological determinists, they start

1014
00:50:13,640 --> 00:50:18,200
Speaker 4: now budding into what is the realm of you know,

1015
00:50:18,600 --> 00:50:22,319
Speaker 4: justice and you know, in society and philosophy and democracy,

1016
00:50:22,680 --> 00:50:25,279
Speaker 4: And that's where they get the most uncomfortable because it's

1017
00:50:25,280 --> 00:50:28,120
Speaker 4: like I'm just telling you, like, you know, what's possible,

1018
00:50:28,400 --> 00:50:32,239
Speaker 4: and when there's pushback, it's like, yeah, but now we're

1019
00:50:32,280 --> 00:50:35,160
Speaker 4: talking about how we live and how we work and

1020
00:50:36,280 --> 00:50:36,839
Speaker 4: how much I.

1021
00:50:36,800 --> 00:50:38,040
Speaker 3: Get paid or not paid.

1022
00:50:38,360 --> 00:50:42,920
Speaker 4: So that technology is important. Technology shapes that conversation, but

1023
00:50:43,000 --> 00:50:45,759
Speaker 4: we're going to have the conversation with a different language,

1024
00:50:46,160 --> 00:50:49,360
Speaker 4: as it should be, and technologists need to get accustomed

1025
00:50:49,400 --> 00:50:51,319
Speaker 4: to if they want to participate in that world with

1026
00:50:51,400 --> 00:50:54,360
Speaker 4: the broad consequences. Hey, get a custom to deal with

1027
00:50:54,400 --> 00:50:59,160
Speaker 4: the complexity of that world of politics, society, institutions, unions,

1028
00:50:59,200 --> 00:51:01,279
Speaker 4: all that stuff. And you know, you can be like

1029
00:51:01,360 --> 00:51:03,840
Speaker 4: whiny about it. It's like they're not adopting my technology.

1030
00:51:04,000 --> 00:51:06,280
Speaker 4: That's what it takes to bring technology into the world.

1031
00:51:07,200 --> 00:51:13,759
Speaker 2: Yeah, well, said thank you Dario for this wonderful conversation.

1032
00:51:13,920 --> 00:51:17,600
Speaker 2: Thank you to all of you for coming and listening,

1033
00:51:17,800 --> 00:51:19,239
Speaker 2: and thank you.

1034
00:51:19,520 --> 00:51:21,360
Speaker 1: Thank you.

1035
00:51:23,360 --> 00:51:26,480
Speaker 2: Dario gild transformed how I think about the future of AI.

1036
00:51:27,200 --> 00:51:29,480
Speaker 2: He explained to me how huge of a leap it

1037
00:51:29,640 --> 00:51:33,000
Speaker 2: was when we went from chess playing models to language

1038
00:51:33,080 --> 00:51:36,279
Speaker 2: learning models, and he talked about how we still have

1039
00:51:36,360 --> 00:51:39,000
Speaker 2: a lot of room to grow. That's why it's important

1040
00:51:39,360 --> 00:51:42,640
Speaker 2: that we get things right. The future of AI is

1041
00:51:42,760 --> 00:51:47,040
Speaker 2: impossible to predict, but the technology has so much potential

1042
00:51:47,120 --> 00:51:50,880
Speaker 2: in every industry. Zooming into an academic or medical setting,

1043
00:51:51,080 --> 00:51:54,200
Speaker 2: showed just how close we are to the widespread adoption

1044
00:51:54,560 --> 00:51:58,200
Speaker 2: of AI. Even Hollywood is being forced to figure this out.

1045
00:51:58,840 --> 00:52:01,560
Speaker 2: Institutions have all Soortz will have to be at the

1046
00:52:01,600 --> 00:52:05,280
Speaker 2: forefront of integration in order to unlock the full power

1047
00:52:05,280 --> 00:52:10,239
Speaker 2: of AI thoughtfully and responsibly. Humans have the power and

1048
00:52:10,280 --> 00:52:14,279
Speaker 2: the responsibility to shape the tech for our world. I

1049
00:52:14,520 --> 00:52:17,399
Speaker 2: for one, I'm excited to see how things play out.

1050
00:52:18,719 --> 00:52:22,879
Speaker 2: Smart Talks with IBM is produced by Matt Romano, Joey Fishground,

1051
00:52:23,120 --> 00:52:27,520
Speaker 2: David jaw and Jacob Goldstein. We're edited by Lydia Jane Kott.

1052
00:52:27,840 --> 00:52:32,320
Speaker 2: Our engineers are Jason Gambrel, Sarah Bruguier, and Ben Holliday.

1053
00:52:32,960 --> 00:52:38,200
Speaker 2: Theme song by Gramoscope. Special thanks to Andy Kelly, Kathy Callahan,

1054
00:52:38,560 --> 00:52:41,399
Speaker 2: and the eight Bar and IBM teams, as well as

1055
00:52:41,400 --> 00:52:45,160
Speaker 2: the Pushkin marketing team. Smart Talks with IBM is a

1056
00:52:45,160 --> 00:52:49,759
Speaker 2: production of Pushkin Industries and Ruby Studio at iHeartMedia. To

1057
00:52:49,800 --> 00:52:54,880
Speaker 2: find more Pushkin podcasts, listen on the iHeartRadio app, Apple Podcasts,

1058
00:52:55,000 --> 00:53:00,279
Speaker 2: or wherever you listen to podcasts. I'm Malcolm Gladwell. This

1059
00:53:00,360 --> 00:53:07,239
Speaker 2: is a paid advertisement from IBM.