1
00:00:04,600 --> 00:00:08,320
Speaker 1: Sleepwalkers is a production of I Heart Radio and Unusual Productions.

2
00:00:15,240 --> 00:00:18,320
Speaker 1: I'm Lain and I'm Kara Price. Welcome to a special

3
00:00:18,360 --> 00:00:34,880
Speaker 1: bonus episode of Sleepwalkers from the Consumer Electronic show, So Kara,

4
00:00:35,000 --> 00:00:37,920
Speaker 1: I'd never been to Las Vegas before, which is the

5
00:00:38,000 --> 00:00:41,360
Speaker 1: difference between us. I've been to Vegas too many times. Well,

6
00:00:41,479 --> 00:00:43,720
Speaker 1: I could tell, and it did feel good to be

7
00:00:43,800 --> 00:00:45,960
Speaker 1: in good hands with an old Vegas hand like you.

8
00:00:46,360 --> 00:00:48,880
Speaker 1: One of the new things though for me was slats,

9
00:00:48,880 --> 00:00:52,000
Speaker 1: which I don't normally play. I think so consciously. I

10
00:00:52,040 --> 00:00:54,400
Speaker 1: was thinking about what Tristan Harris talked about in the

11
00:00:54,400 --> 00:00:57,000
Speaker 1: first season Sleepwalkers. You know he was that former googler

12
00:00:57,240 --> 00:00:59,640
Speaker 1: who told us that Instagram is actually supposed to feel

13
00:00:59,640 --> 00:01:02,360
Speaker 1: a lot like slot machines. Well, that's right. Tristan studied

14
00:01:02,400 --> 00:01:05,360
Speaker 1: at the Stanford Persuasion Lab and told us about how

15
00:01:05,720 --> 00:01:10,320
Speaker 1: casino architecture has influenced the development of highly addictive type

16
00:01:10,360 --> 00:01:13,440
Speaker 1: products like Instagram. So it's interesting for me to actually

17
00:01:13,440 --> 00:01:16,720
Speaker 1: see Vegas and the bright lights and the impossibility of

18
00:01:16,880 --> 00:01:20,600
Speaker 1: escape firsthand, not to mention the replicas of the Empire

19
00:01:20,640 --> 00:01:23,520
Speaker 1: State Building, the Canals of Venice, the Colisseum of Rome,

20
00:01:23,680 --> 00:01:25,520
Speaker 1: and you know, I was lucky enough to see the

21
00:01:25,520 --> 00:01:27,640
Speaker 1: Seattle Space Needle for the first time. I just didn't

22
00:01:27,640 --> 00:01:30,240
Speaker 1: know that it was in Las Vegas. But that's not

23
00:01:30,280 --> 00:01:33,120
Speaker 1: why we were there. We were there for CEES, the

24
00:01:33,160 --> 00:01:37,040
Speaker 1: Consumer Electronics Show, and this episode we're actually going to

25
00:01:37,160 --> 00:01:39,480
Speaker 1: talk about some of the coolest things we saw there,

26
00:01:39,520 --> 00:01:42,039
Speaker 1: but we're going to focus more on the innovations that

27
00:01:42,040 --> 00:01:44,959
Speaker 1: are at the intersection of technology and humanity rather than

28
00:01:45,000 --> 00:01:48,240
Speaker 1: talk about you know, infamous toilet paper dispensers. One of

29
00:01:48,240 --> 00:01:50,680
Speaker 1: the big reasons we went is because we were invited

30
00:01:50,720 --> 00:01:54,480
Speaker 1: by WaveMaker, which is an agency part of w p P,

31
00:01:55,160 --> 00:01:58,280
Speaker 1: to do an interview on stage, a live podcast, so

32
00:01:58,320 --> 00:02:00,960
Speaker 1: to speak with Matt and on a hand, who is

33
00:02:01,000 --> 00:02:04,440
Speaker 1: head of product at ARC Publishing and ARC Publishing is

34
00:02:04,520 --> 00:02:07,280
Speaker 1: part of the Washington Post. Yeah, and ARC is also

35
00:02:07,400 --> 00:02:10,880
Speaker 1: an interesting case of AI and action because they're forward

36
00:02:10,880 --> 00:02:13,880
Speaker 1: thinking in terms of increasing the visibility of content through

37
00:02:13,880 --> 00:02:18,560
Speaker 1: personalization and optimizing everything from headlines to photo selection, all

38
00:02:18,680 --> 00:02:21,560
Speaker 1: using machine learning, and those are things that really matter

39
00:02:21,680 --> 00:02:24,400
Speaker 1: for journalists and readers. Yeah, and this use of AI

40
00:02:24,480 --> 00:02:27,040
Speaker 1: stands out to me because it provides a solution to

41
00:02:27,160 --> 00:02:30,040
Speaker 1: real problem. How do you get eyeballs on the right

42
00:02:30,080 --> 00:02:33,080
Speaker 1: content when there's just so much. That said, the issue

43
00:02:33,080 --> 00:02:36,639
Speaker 1: of personalization does also raise questions about what happens when

44
00:02:36,680 --> 00:02:39,200
Speaker 1: machines start to know us better than we know ourselves.

45
00:02:39,600 --> 00:02:42,079
Speaker 1: Not to mention, what are the appropriate limits of how

46
00:02:42,120 --> 00:02:45,960
Speaker 1: companies use AI and dature about us. Yeah, AI can

47
00:02:46,000 --> 00:02:51,040
Speaker 1: definitely streamline processes by detecting patterns that you know, human

48
00:02:51,080 --> 00:02:54,440
Speaker 1: beings cannot see, or it can allow you to do

49
00:02:54,480 --> 00:02:57,000
Speaker 1: things at the scale like tag hundreds of thousands of

50
00:02:57,080 --> 00:03:01,560
Speaker 1: articles that again, human beings just cannot do. So greater

51
00:03:01,600 --> 00:03:04,680
Speaker 1: efficiency is on one side of the spectrum and extremely

52
00:03:04,720 --> 00:03:06,960
Speaker 1: attractive to people, but on the other side you have

53
00:03:07,080 --> 00:03:10,040
Speaker 1: issues of taking humans out of the loop, like the

54
00:03:10,080 --> 00:03:13,960
Speaker 1: black box problem and authenticity in a world of deep fakes.

55
00:03:14,000 --> 00:03:17,359
Speaker 1: So a question for businesses and users of technology is

56
00:03:17,400 --> 00:03:21,040
Speaker 1: sort of when does AI add to our experience and

57
00:03:21,080 --> 00:03:24,000
Speaker 1: when does it maybe hold us back or take advantage

58
00:03:24,000 --> 00:03:27,040
Speaker 1: of us. You know, for example, from seeing news stories

59
00:03:27,040 --> 00:03:29,680
Speaker 1: that we should see, but maybe the algorithm doesn't think

60
00:03:29,720 --> 00:03:32,080
Speaker 1: we want to see it or that we won't click

61
00:03:32,120 --> 00:03:34,720
Speaker 1: on it. Right in the old days when everyone received

62
00:03:34,760 --> 00:03:37,400
Speaker 1: a print use of paper on their doorstep, everyone has

63
00:03:37,440 --> 00:03:40,640
Speaker 1: the same front page and the same headlines. Nowadays, when

64
00:03:40,640 --> 00:03:43,080
Speaker 1: you log onto a news website or onto social media,

65
00:03:43,520 --> 00:03:47,160
Speaker 1: everybody has a different version of the world, and that

66
00:03:47,320 --> 00:03:50,280
Speaker 1: is obviously positive for driving engagement, but may not be

67
00:03:50,320 --> 00:03:53,520
Speaker 1: so positive in terms of having conversations with the same

68
00:03:53,520 --> 00:03:56,600
Speaker 1: facts about the same stories. Equally, we have to ask

69
00:03:56,800 --> 00:03:59,200
Speaker 1: do we want articles where the headline has been written

70
00:03:59,200 --> 00:04:02,160
Speaker 1: by an algorithm or do we prefer headlines written by

71
00:04:02,200 --> 00:04:04,800
Speaker 1: a person. And that's something we talked about with Matt

72
00:04:05,160 --> 00:04:08,520
Speaker 1: because ARC actually tested the headline writing technology. Let's talk

73
00:04:08,560 --> 00:04:13,800
Speaker 1: to Matt seriously. Let's cut to the chase. Are really

74
00:04:13,920 --> 00:04:18,320
Speaker 1: came out of a collaboration trying to better understand what

75
00:04:18,760 --> 00:04:22,040
Speaker 1: actual journalists needed. Can you talk a little bit more. Yeah,

76
00:04:22,120 --> 00:04:24,240
Speaker 1: at the very beginning, you know, we were just trying

77
00:04:24,240 --> 00:04:27,000
Speaker 1: to solve problems for ourselves seven or eight years ago.

78
00:04:27,400 --> 00:04:28,520
Speaker 1: You know, we knew and he had to make some

79
00:04:28,520 --> 00:04:32,080
Speaker 1: pretty fundamental transformation to the post and to really prepare

80
00:04:32,080 --> 00:04:34,760
Speaker 1: ourselves for the digital future. We didn't have the right

81
00:04:34,800 --> 00:04:36,760
Speaker 1: tools to do it, and we couldn't really find the

82
00:04:36,839 --> 00:04:38,920
Speaker 1: right tools on the market either. What we did was

83
00:04:39,400 --> 00:04:40,800
Speaker 1: spent a lot of time with the journalists and the

84
00:04:40,920 --> 00:04:42,840
Speaker 1: editors trying to figure out what it was that make

85
00:04:42,880 --> 00:04:45,359
Speaker 1: their lives easier. It's trying to figure out how do

86
00:04:45,400 --> 00:04:47,440
Speaker 1: you make journalists work better, how can they publish faster?

87
00:04:47,880 --> 00:04:49,320
Speaker 1: What are the little things you can do inside of

88
00:04:49,320 --> 00:04:51,320
Speaker 1: a product to make it easier for them to write

89
00:04:51,360 --> 00:04:54,600
Speaker 1: stories or publish From there? About four years ago is

90
00:04:54,640 --> 00:04:57,520
Speaker 1: when we started evolving it into a commercial offering. Today

91
00:04:57,600 --> 00:05:00,360
Speaker 1: we're running hundreds of websites around the world. We're in

92
00:05:00,400 --> 00:05:04,360
Speaker 1: about twenty different countries. We're running companies like BP their

93
00:05:04,400 --> 00:05:07,040
Speaker 1: internal communications as well as some of their marketing. We're

94
00:05:07,120 --> 00:05:09,640
Speaker 1: running large broadcasters and all their live video and body

95
00:05:09,680 --> 00:05:11,440
Speaker 1: and of course we're still running a lot of newspapers

96
00:05:11,440 --> 00:05:14,080
Speaker 1: and news publishers like The Post and many others around

97
00:05:14,080 --> 00:05:17,160
Speaker 1: the world, Lucky and publishing. You know that AI and

98
00:05:17,240 --> 00:05:21,919
Speaker 1: artificial intelligence are made in headlines, and there was a

99
00:05:22,040 --> 00:05:25,039
Speaker 1: story in the Financial Times last year we said fort

100
00:05:25,880 --> 00:05:31,520
Speaker 1: of AI startups use no AI whatsoever. So I bet

101
00:05:31,600 --> 00:05:35,279
Speaker 1: it's probably higher. But when we talk about using AI,

102
00:05:35,400 --> 00:05:37,400
Speaker 1: or when you talk about using AI, what do we

103
00:05:37,400 --> 00:05:40,479
Speaker 1: actually mean? So it can span the range of technologies

104
00:05:40,480 --> 00:05:43,600
Speaker 1: from something like machine learning, which is basically a way

105
00:05:43,600 --> 00:05:45,760
Speaker 1: to use algorithms, to take large sets of data and

106
00:05:45,800 --> 00:05:48,520
Speaker 1: either uncovered patterns in it or try to model a

107
00:05:48,520 --> 00:05:52,040
Speaker 1: way to predict a certain outcome to technologies like computer vision,

108
00:05:52,040 --> 00:05:53,960
Speaker 1: which you can use to look at images or video

109
00:05:54,320 --> 00:05:56,839
Speaker 1: and extract information about them by recognizing patterns and trying

110
00:05:56,839 --> 00:05:59,920
Speaker 1: to identify objects inside of them, and so a lot

111
00:05:59,920 --> 00:06:01,720
Speaker 1: of those technologies, then when you put them together, you

112
00:06:01,760 --> 00:06:04,400
Speaker 1: can form some really interesting workflows that you know in

113
00:06:04,400 --> 00:06:06,080
Speaker 1: the past you might have had to use humans to

114
00:06:06,080 --> 00:06:09,040
Speaker 1: do that, you can actually do much more simple automatically.

115
00:06:09,520 --> 00:06:12,640
Speaker 1: Was there a particular business challenge or challenge at the

116
00:06:12,640 --> 00:06:15,560
Speaker 1: Washington Post that you couldn't have solved if you hadn't

117
00:06:15,600 --> 00:06:19,880
Speaker 1: been using AI. Any story that we write on Washington Post,

118
00:06:19,920 --> 00:06:22,880
Speaker 1: we're mapping to a set of two or three hundred topics.

119
00:06:22,880 --> 00:06:24,800
Speaker 1: Maybe an example of one of those might be like

120
00:06:25,000 --> 00:06:28,680
Speaker 1: congressional policy or narcotics crime. What you're trying to do

121
00:06:28,720 --> 00:06:30,800
Speaker 1: is say, if I look at all this content, I'm

122
00:06:30,839 --> 00:06:33,080
Speaker 1: not just pulling specific words out of it. I'm actually

123
00:06:33,080 --> 00:06:35,520
Speaker 1: trying to figure out what is this content about, what

124
00:06:35,640 --> 00:06:38,120
Speaker 1: is the fundamental concept of this content. So you pick

125
00:06:38,160 --> 00:06:40,680
Speaker 1: a set of articles, let's say a hundred thousand news

126
00:06:40,760 --> 00:06:43,040
Speaker 1: articles in the case of this example for the post,

127
00:06:43,960 --> 00:06:47,000
Speaker 1: and at first you use humans it's called micro labor

128
00:06:47,120 --> 00:06:49,160
Speaker 1: to do this training set. And the goal is you're

129
00:06:49,160 --> 00:06:52,320
Speaker 1: building an algorithm based on a set of real data.

130
00:06:52,480 --> 00:06:54,760
Speaker 1: And so the humans are going there and saying this article, yeah,

131
00:06:54,760 --> 00:06:57,200
Speaker 1: this is about congressional policy. Why because I know it is,

132
00:06:57,240 --> 00:06:59,360
Speaker 1: I read it, that's what it's about. This one's about

133
00:06:59,440 --> 00:07:02,960
Speaker 1: narcotics time, and this one's about soccer. And so you

134
00:07:03,000 --> 00:07:06,680
Speaker 1: train all these articles against that algorithm until finally the

135
00:07:06,720 --> 00:07:10,640
Speaker 1: algorithm is basically sufficiently advanced to predict a new article

136
00:07:10,680 --> 00:07:13,360
Speaker 1: that you put into it and determine an outcome with

137
00:07:13,400 --> 00:07:16,320
Speaker 1: the same high probability of success that you were able

138
00:07:16,360 --> 00:07:19,600
Speaker 1: to with humans training it. Now, every time a journalist

139
00:07:19,920 --> 00:07:22,840
Speaker 1: saves or publishes a story, we're able to parse over

140
00:07:22,880 --> 00:07:25,600
Speaker 1: all the content inside that story. Then we can predict

141
00:07:25,680 --> 00:07:28,200
Speaker 1: the strength at which it's likely to belong to that topic.

142
00:07:28,840 --> 00:07:32,920
Speaker 1: How do you create a better user experience, in your case,

143
00:07:32,960 --> 00:07:36,480
Speaker 1: news experience for an individual or consumer With that metadat

144
00:07:36,520 --> 00:07:38,080
Speaker 1: you can do a lot of interesting things. We can

145
00:07:38,120 --> 00:07:40,680
Speaker 1: figure out that hey, this is something that they're interested

146
00:07:40,720 --> 00:07:42,200
Speaker 1: in reading, perhaps they'd like to read more, and it

147
00:07:42,240 --> 00:07:45,920
Speaker 1: actually serves the signal into our recommendation algorithms. From your perspective,

148
00:07:46,000 --> 00:07:50,840
Speaker 1: where can businesses sort of harness the power of machine

149
00:07:50,920 --> 00:07:55,320
Speaker 1: learning to really hone in on who their customer is

150
00:07:55,400 --> 00:07:59,200
Speaker 1: and what that customer wants. We want to deliver more

151
00:07:59,240 --> 00:08:01,080
Speaker 1: content to our readers who want to help them find

152
00:08:01,080 --> 00:08:04,760
Speaker 1: more content that we've created. We have about nine journalists

153
00:08:04,760 --> 00:08:07,360
Speaker 1: at the Washington Post. We write something like, you know,

154
00:08:07,360 --> 00:08:10,040
Speaker 1: three or four hundred original stories a day, so there's

155
00:08:10,040 --> 00:08:12,360
Speaker 1: a lot of content there. To get readers to all

156
00:08:12,360 --> 00:08:14,840
Speaker 1: that different content and have them continue moving through your

157
00:08:14,960 --> 00:08:17,360
Speaker 1: content you spend a lot of money to produce is

158
00:08:17,360 --> 00:08:20,200
Speaker 1: really challenging, and so that's a great use case for personalization.

159
00:08:20,840 --> 00:08:22,680
Speaker 1: But where you can make it really come alive is

160
00:08:23,040 --> 00:08:26,880
Speaker 1: by having more sophisticated metadata, more sophisticated information about that content.

161
00:08:26,960 --> 00:08:29,080
Speaker 1: It's more likely to bring readers to it, and so

162
00:08:29,160 --> 00:08:31,400
Speaker 1: that's where these machine learning models really come in handy.

163
00:08:32,160 --> 00:08:34,400
Speaker 1: I think part of what's fun about this conversation is

164
00:08:34,440 --> 00:08:37,520
Speaker 1: there's a lot of cases out there where average users,

165
00:08:38,200 --> 00:08:39,920
Speaker 1: you know, they imagine they see something like that. You

166
00:08:39,960 --> 00:08:42,160
Speaker 1: see the boots on Instagram and you think, oh my god,

167
00:08:42,240 --> 00:08:47,120
Speaker 1: like these companies must crazy, like you know, indiscernible for magic, right, like,

168
00:08:47,200 --> 00:08:49,160
Speaker 1: there must be some crazy model out there doing this,

169
00:08:49,200 --> 00:08:51,960
Speaker 1: and perhaps there is. But in a lot of ways,

170
00:08:52,640 --> 00:08:57,640
Speaker 1: you know, your users aren't necessarily as aware of the

171
00:08:57,679 --> 00:09:01,400
Speaker 1: advertising ecosystem, the data ecosystem, and how these things tied

172
00:09:01,440 --> 00:09:04,240
Speaker 1: together between platforms and sites, and I think, as like

173
00:09:04,280 --> 00:09:07,360
Speaker 1: industry professionals, we always kind of underestimate that fact. And

174
00:09:07,400 --> 00:09:10,040
Speaker 1: so the net effect is that users are completely surprised

175
00:09:10,040 --> 00:09:11,720
Speaker 1: by this. They think you must be doing something completely

176
00:09:11,800 --> 00:09:13,880
Speaker 1: unheard of to achieve it, when in fact, you know

177
00:09:14,080 --> 00:09:16,599
Speaker 1: it could be really simple data sharing. And so the

178
00:09:16,640 --> 00:09:18,160
Speaker 1: reason I think that's important is then when you do

179
00:09:18,200 --> 00:09:21,080
Speaker 1: build technologies that actually utilize some of these more sophisticated

180
00:09:21,120 --> 00:09:24,600
Speaker 1: methods to build data sets, you have to be aware

181
00:09:24,640 --> 00:09:27,000
Speaker 1: that your users. You know, first of all, your users

182
00:09:27,000 --> 00:09:29,520
Speaker 1: aren't gonna necessarily anticipate the outcomes that you can create,

183
00:09:30,120 --> 00:09:31,400
Speaker 1: and if you don't do a good job on the

184
00:09:31,400 --> 00:09:33,559
Speaker 1: product side of making sure that you really think through

185
00:09:33,600 --> 00:09:36,000
Speaker 1: the use case and how you're leveraging technology to solve it,

186
00:09:36,400 --> 00:09:39,360
Speaker 1: you can generate unexpected outcomes. You know, there was the

187
00:09:39,400 --> 00:09:43,600
Speaker 1: example of a retailer who produced advertising flyers that were

188
00:09:43,600 --> 00:09:46,520
Speaker 1: able to predict folks who are pregnant, right, even if

189
00:09:47,000 --> 00:09:49,120
Speaker 1: some of those folks didn't necessarily know that themselves yet

190
00:09:49,200 --> 00:09:51,800
Speaker 1: or hadn't shared it with with their family or their spouses.

191
00:09:52,280 --> 00:09:56,480
Speaker 1: And so that was a case really of both the

192
00:09:56,559 --> 00:10:00,680
Speaker 1: company and the consumer being shocked by outcomes we're generating

193
00:10:00,679 --> 00:10:03,520
Speaker 1: exactly right. I mean, the you know, the algorithm doesn't

194
00:10:03,520 --> 00:10:06,360
Speaker 1: do anything magic, but that's a case of you know,

195
00:10:06,400 --> 00:10:08,520
Speaker 1: putting together in that case, like a marketing program, where

196
00:10:08,559 --> 00:10:10,920
Speaker 1: you don't really think through what's the possible data that

197
00:10:10,920 --> 00:10:13,280
Speaker 1: this could produce and what are my users? What do

198
00:10:13,320 --> 00:10:15,280
Speaker 1: they already know about this data? You know, you need

199
00:10:15,360 --> 00:10:17,320
Speaker 1: to think really hard about your users and what they

200
00:10:17,360 --> 00:10:18,880
Speaker 1: want and what they're trying to achieve, and what the

201
00:10:18,960 --> 00:10:21,959
Speaker 1: dangers are and leveraging this technology. It's no different than

202
00:10:22,200 --> 00:10:24,679
Speaker 1: in that way than any previous technology solutions you might

203
00:10:24,679 --> 00:10:26,520
Speaker 1: have used to build a great product for people, and

204
00:10:26,520 --> 00:10:28,800
Speaker 1: it can be misused just as easily. Funny enough, the

205
00:10:28,840 --> 00:10:32,360
Speaker 1: first episode of Sleepwalkers season one open with a story

206
00:10:32,480 --> 00:10:37,840
Speaker 1: of Washington Post employee Gillian Brockell, who, to your point

207
00:10:37,840 --> 00:10:41,520
Speaker 1: about pregnancy and data stuffered a miscarriage but continue to

208
00:10:41,559 --> 00:10:46,320
Speaker 1: receive targeted ads for pregnancy goods after a miscarriage, and

209
00:10:46,360 --> 00:10:49,640
Speaker 1: she wrote this openless is the technology companies saying please

210
00:10:49,640 --> 00:10:52,200
Speaker 1: stop targeting me. But that raised a big question for us,

211
00:10:52,240 --> 00:10:55,040
Speaker 1: which is what happens when the algorithms go wrong? Yeah,

212
00:10:55,200 --> 00:10:57,440
Speaker 1: I'd almost be more specific with the way that you

213
00:10:57,520 --> 00:11:00,200
Speaker 1: say that, and like the algorithm didn't go wrong, right,

214
00:11:00,240 --> 00:11:02,559
Speaker 1: but like the implementation of it and the product that

215
00:11:02,600 --> 00:11:05,320
Speaker 1: they built around it did because it wasn't really correctly conceived.

216
00:11:05,360 --> 00:11:07,160
Speaker 1: And we have to make sure that like what you're

217
00:11:07,200 --> 00:11:09,160
Speaker 1: trying to do automatically fits really well with what your

218
00:11:09,240 --> 00:11:11,559
Speaker 1: users are trying to accomplish, doesn't happen in a way

219
00:11:11,559 --> 00:11:14,600
Speaker 1: that's not expected. Is a well designed product, you know,

220
00:11:14,640 --> 00:11:17,079
Speaker 1: So in that specific case, yeah, I mean it always

221
00:11:17,120 --> 00:11:19,240
Speaker 1: starts and ends with kind of good product design. If

222
00:11:19,520 --> 00:11:21,760
Speaker 1: you're not doing that, just like any other tool, you

223
00:11:21,800 --> 00:11:23,920
Speaker 1: can misuse it. One of the other things we did

224
00:11:23,960 --> 00:11:27,640
Speaker 1: on the show was we used a language generator to

225
00:11:27,720 --> 00:11:31,360
Speaker 1: can't with pickup lines based on a data set of

226
00:11:31,400 --> 00:11:36,480
Speaker 1: all none of them were actually I don't have kind

227
00:11:36,480 --> 00:11:39,560
Speaker 1: of things like you are a thing and I love you,

228
00:11:39,559 --> 00:11:41,199
Speaker 1: you know, which is now the name of the book

229
00:11:41,360 --> 00:11:44,320
Speaker 1: by the woman who. Yeah, woman Jell Shane her wrote

230
00:11:44,320 --> 00:11:46,079
Speaker 1: a book about it, and then she also did these

231
00:11:46,080 --> 00:11:48,800
Speaker 1: things like AI recipes, like one was for chocolate chocolate

232
00:11:48,880 --> 00:11:52,400
Speaker 1: chocolate chicken cake. So there is funny things and Shakespeare

233
00:11:52,559 --> 00:11:55,960
Speaker 1: on it, and I didn't revealed two things. One is

234
00:11:56,040 --> 00:12:00,199
Speaker 1: when you turn these deep learning algorithms onto big data sets,

235
00:12:00,520 --> 00:12:02,959
Speaker 1: they reveal passions you might not necessarily be aware of,

236
00:12:03,000 --> 00:12:05,000
Speaker 1: like with a lot of chicken and quite a lot

237
00:12:05,040 --> 00:12:07,880
Speaker 1: of chocolates. On the other hand, like these were clearly

238
00:12:08,360 --> 00:12:10,840
Speaker 1: not something human would ever make. So how do you

239
00:12:10,880 --> 00:12:13,800
Speaker 1: think about the line between doing fun things in AI

240
00:12:13,960 --> 00:12:16,680
Speaker 1: and doing stuff which is valuable for business and also

241
00:12:16,800 --> 00:12:19,040
Speaker 1: not getting lost in the uncanny valley. So a good

242
00:12:19,080 --> 00:12:21,559
Speaker 1: example of this for instances. We spent some time at

243
00:12:21,559 --> 00:12:24,520
Speaker 1: the Post trying to build a headline generation algorithm we

244
00:12:24,520 --> 00:12:28,640
Speaker 1: could automatically create headlines for stories. And you know, the

245
00:12:28,679 --> 00:12:31,240
Speaker 1: idea I think at the beginning wasn't necessarily that, you know,

246
00:12:31,280 --> 00:12:33,760
Speaker 1: journalists are never read headlines again, but we'd be able

247
00:12:33,760 --> 00:12:35,800
Speaker 1: to create some alternative headlines in different ways to think

248
00:12:35,840 --> 00:12:38,280
Speaker 1: about a story. Our intention was, let's see if we

249
00:12:38,320 --> 00:12:39,800
Speaker 1: can come up with something so that we can create

250
00:12:39,840 --> 00:12:42,360
Speaker 1: several different variants of a headline. Part of our software

251
00:12:42,400 --> 00:12:45,440
Speaker 1: platform we include content testing framework. So one of the

252
00:12:45,480 --> 00:12:48,040
Speaker 1: things that we can do is say, for a given story,

253
00:12:48,120 --> 00:12:50,480
Speaker 1: let's have three different headlines for it, Let's run a

254
00:12:50,520 --> 00:12:53,280
Speaker 1: test as soon as it publishest of the audience is

255
00:12:53,320 --> 00:12:55,640
Speaker 1: going to get each variant, and then as people start

256
00:12:55,720 --> 00:12:57,120
Speaker 1: to click one more than the other, we're gonna shift

257
00:12:57,120 --> 00:12:59,600
Speaker 1: the burden of traffic to the most successful variant. And that,

258
00:12:59,640 --> 00:13:02,240
Speaker 1: I would them, by itself works really well. If you know,

259
00:13:02,280 --> 00:13:04,120
Speaker 1: folks in the audience here were to look at the

260
00:13:04,160 --> 00:13:06,520
Speaker 1: homepage of our site right now, there's probably two or

261
00:13:06,559 --> 00:13:08,480
Speaker 1: three stories that are running those types of tests where

262
00:13:08,480 --> 00:13:11,040
Speaker 1: different people would se different headlines or different images, or

263
00:13:11,240 --> 00:13:14,680
Speaker 1: in fact maybe actually just complutely different stories, and those

264
00:13:14,720 --> 00:13:17,320
Speaker 1: tests will resolve in like fifteen or twenty minutes. So

265
00:13:17,400 --> 00:13:19,280
Speaker 1: that works well enough by itself. But then we realized, well,

266
00:13:19,280 --> 00:13:20,960
Speaker 1: we could probably create more of these tests if only

267
00:13:20,960 --> 00:13:23,439
Speaker 1: we could automatically create headlines for them. We could just

268
00:13:23,480 --> 00:13:25,600
Speaker 1: be running these tests all the time for every single story.

269
00:13:26,080 --> 00:13:28,840
Speaker 1: But what we found was, you know, not exactly so

270
00:13:29,320 --> 00:13:31,760
Speaker 1: if the idea was to save journalists time and doing

271
00:13:31,800 --> 00:13:33,679
Speaker 1: that in the end, I mean, you'd have to come

272
00:13:33,760 --> 00:13:36,240
Speaker 1: up with something that's fairly solid and ready to publish out.

273
00:13:36,600 --> 00:13:38,720
Speaker 1: We were able to create something that allowed you know,

274
00:13:38,800 --> 00:13:41,200
Speaker 1: journalists basically have different formulations that they could play with

275
00:13:41,240 --> 00:13:43,000
Speaker 1: and maybe gave them some ideas of what to create,

276
00:13:43,240 --> 00:13:44,839
Speaker 1: but it still require people to look at it. In

277
00:13:44,880 --> 00:13:48,720
Speaker 1: the end, how can businesses work better with their engineers,

278
00:13:48,840 --> 00:13:53,920
Speaker 1: with their tech teams to sort of create and not

279
00:13:54,280 --> 00:13:57,240
Speaker 1: stay siloed in a way that like somebody who works

280
00:13:57,240 --> 00:13:59,840
Speaker 1: in marketing feels like, well, you know, there's actually this

281
00:14:00,040 --> 00:14:01,520
Speaker 1: need that I have, but I don't know who to

282
00:14:01,559 --> 00:14:03,600
Speaker 1: talk to about it, and that don't really know what

283
00:14:03,640 --> 00:14:05,800
Speaker 1: to do. It's an awesome question to me, Like one

284
00:14:05,800 --> 00:14:06,920
Speaker 1: of the best things that you can do as a

285
00:14:06,960 --> 00:14:10,080
Speaker 1: business is to put those people together, sometimes even physically.

286
00:14:10,200 --> 00:14:12,920
Speaker 1: So when we started this project, you know, we literally

287
00:14:12,960 --> 00:14:17,560
Speaker 1: co located engineers, product people directly inside the news room

288
00:14:17,600 --> 00:14:19,760
Speaker 1: to sit with the folks who are doing this work. Now,

289
00:14:19,760 --> 00:14:22,480
Speaker 1: when it comes to a I M L, you remember,

290
00:14:22,480 --> 00:14:24,640
Speaker 1: these are just tools. These are tools to make work easier.

291
00:14:24,680 --> 00:14:27,160
Speaker 1: Their tools in a lot of cases for automation and efficiency.

292
00:14:27,560 --> 00:14:29,920
Speaker 1: There's some problems that can't be solved without it. In

293
00:14:30,000 --> 00:14:32,000
Speaker 1: the end, though, you know, you're still trying to solve

294
00:14:32,040 --> 00:14:34,040
Speaker 1: some business problem, and most of those involved some sort

295
00:14:34,040 --> 00:14:35,560
Speaker 1: of users that you need to get to know. So

296
00:14:36,200 --> 00:14:38,440
Speaker 1: you know, even at the post we had data sciencests

297
00:14:38,440 --> 00:14:40,640
Speaker 1: who were on those teams embedded in the news room

298
00:14:40,680 --> 00:14:42,680
Speaker 1: as well. You know, they weren't kind of seated somewhere

299
00:14:42,720 --> 00:14:44,960
Speaker 1: else thinking of problems on their own. There's a time

300
00:14:45,000 --> 00:14:47,440
Speaker 1: and a place for creating room for prototyping sometimes that

301
00:14:47,480 --> 00:14:50,440
Speaker 1: has to happen to especially with really advanced technologies. But

302
00:14:50,480 --> 00:14:54,000
Speaker 1: beyond prototyping, putting those teams together is super crucial. So

303
00:14:54,040 --> 00:14:57,160
Speaker 1: how do you make sure, speaking metaphorically, you write a

304
00:14:57,200 --> 00:15:00,160
Speaker 1: good brief to your AI team, well, your engineering team.

305
00:15:00,200 --> 00:15:03,200
Speaker 1: I still think, you know, start with the problem that

306
00:15:03,200 --> 00:15:05,120
Speaker 1: you're trying to solve. Like, if you're going in thinking

307
00:15:05,200 --> 00:15:08,120
Speaker 1: let's use AI to solve something, I think you're probably

308
00:15:08,720 --> 00:15:10,960
Speaker 1: starting the problem the wrong way, and start by framing

309
00:15:11,000 --> 00:15:13,240
Speaker 1: up the problem in business terms. For people, you'd be surprised,

310
00:15:13,280 --> 00:15:15,800
Speaker 1: I think how much you know engineers and product folks

311
00:15:15,800 --> 00:15:17,880
Speaker 1: really actually prefer to get that first before they start

312
00:15:17,920 --> 00:15:19,880
Speaker 1: diving into what's the technology that I'm getting used to

313
00:15:19,880 --> 00:15:23,160
Speaker 1: solve this problem? With a buzz around AI, especially right now.

314
00:15:23,920 --> 00:15:25,480
Speaker 1: You know, people tend to go into it. I think

315
00:15:25,480 --> 00:15:28,080
Speaker 1: thinking this is kind of something that's pretty close to magic.

316
00:15:28,560 --> 00:15:30,440
Speaker 1: We just use AI. It's going to solve these problems

317
00:15:30,440 --> 00:15:32,320
Speaker 1: that we haven't been able to solve some other way.

318
00:15:32,480 --> 00:15:34,280
Speaker 1: And that's not really the right way to approach it.

319
00:15:34,560 --> 00:15:37,119
Speaker 1: But a I will be transformative, I think for organizations

320
00:15:37,120 --> 00:15:39,640
Speaker 1: that apply it the right way, with the product mindset,

321
00:15:39,720 --> 00:15:41,320
Speaker 1: with a good knowledge of the problem that they're trying

322
00:15:41,320 --> 00:15:44,600
Speaker 1: to solve, with empathy for their users. When we're doing

323
00:15:44,680 --> 00:15:48,239
Speaker 1: our research for this panel, as an article in Bloomberg

324
00:15:48,240 --> 00:15:53,520
Speaker 1: News saying that Jeff Bezos is pestimally very invested in

325
00:15:53,520 --> 00:15:56,480
Speaker 1: the product, and then you called him Jeff in conversation,

326
00:15:56,720 --> 00:16:01,160
Speaker 1: I found very impress So we will unders people sending

327
00:16:01,160 --> 00:16:04,400
Speaker 1: me an email already without obviously telling us the contented

328
00:16:04,440 --> 00:16:07,640
Speaker 1: your meetings. How does his vision imbue what you do?

329
00:16:08,160 --> 00:16:11,440
Speaker 1: So certainly for us it's boon to have him owning

330
00:16:11,440 --> 00:16:14,000
Speaker 1: the company. I think that's one of the greatest things

331
00:16:14,040 --> 00:16:16,360
Speaker 1: is you know, obviously at the Washington Post were known

332
00:16:16,400 --> 00:16:19,040
Speaker 1: for an amazing newsroom, but we've also spent a lot

333
00:16:19,040 --> 00:16:21,480
Speaker 1: of time investing in our engineering team that started to

334
00:16:21,520 --> 00:16:23,720
Speaker 1: some extent before you know, we were purchased by Jeff,

335
00:16:23,760 --> 00:16:26,000
Speaker 1: but certainly after we purchased us. It opened a lot

336
00:16:26,040 --> 00:16:28,000
Speaker 1: of new doors for us, and it gets people excited

337
00:16:28,040 --> 00:16:29,120
Speaker 1: to come and work for us and some of the

338
00:16:29,120 --> 00:16:32,080
Speaker 1: problems that we're trying to solve. It really inspires people

339
00:16:32,160 --> 00:16:34,160
Speaker 1: to be able to build like a platform like we built,

340
00:16:34,760 --> 00:16:36,920
Speaker 1: you know, within a newspaper company. I think would have

341
00:16:36,960 --> 00:16:39,280
Speaker 1: been hired to Fathom probably ten years ago, but I

342
00:16:39,280 --> 00:16:41,040
Speaker 1: mean today we really can say that, you know, we're

343
00:16:41,040 --> 00:16:44,320
Speaker 1: a content company and we're a technology company. And I

344
00:16:44,360 --> 00:16:46,680
Speaker 1: think part of that starts with him in the leadership

345
00:16:46,720 --> 00:17:04,920
Speaker 1: that he provides. More Sleepwalkers after the break, So, Kara,

346
00:17:05,080 --> 00:17:08,720
Speaker 1: that was our conversation on stage with Matt monahantan CS

347
00:17:08,800 --> 00:17:11,960
Speaker 1: in early January. It was interesting because we hear so

348
00:17:12,040 --> 00:17:15,800
Speaker 1: much about tech companies becoming publishers, whether it's Facebook, YouTube,

349
00:17:15,880 --> 00:17:19,600
Speaker 1: or Twitter, but we hear less about publishers becoming tech companies.

350
00:17:20,080 --> 00:17:22,280
Speaker 1: I guess that's where Jeff Basos as an owner is

351
00:17:22,320 --> 00:17:25,120
Speaker 1: what we might call a differentiator. So I was personally

352
00:17:25,160 --> 00:17:28,480
Speaker 1: struck with Matt's experiments with the headline generator. You know,

353
00:17:28,520 --> 00:17:30,960
Speaker 1: for the time being, it doesn't work well enough to

354
00:17:31,000 --> 00:17:34,760
Speaker 1: be a commercial product, but I think it will soon.

355
00:17:34,840 --> 00:17:37,280
Speaker 1: You know, look at autocomplete when you send a Gmail

356
00:17:37,359 --> 00:17:41,200
Speaker 1: like Sincerely Comma. You know I get those all the time,

357
00:17:41,640 --> 00:17:44,960
Speaker 1: and it works. You know, in an apocalyptic reading, that

358
00:17:45,040 --> 00:17:47,520
Speaker 1: means that machines will take over our lives and there

359
00:17:47,520 --> 00:17:49,919
Speaker 1: will be no work left for humans. We won't have

360
00:17:49,960 --> 00:17:52,119
Speaker 1: to come up with smart headlines. But I think in

361
00:17:52,160 --> 00:17:56,320
Speaker 1: a more optimistic reading, using algorithms to generate writing suggestions

362
00:17:56,320 --> 00:18:00,840
Speaker 1: could actually enable originality. Reminds me of that Chinese science

363
00:18:00,840 --> 00:18:03,200
Speaker 1: fiction writer who you and I have talked about named

364
00:18:03,280 --> 00:18:08,120
Speaker 1: Chen show Fund, who actually used an algorithm to create

365
00:18:08,280 --> 00:18:11,800
Speaker 1: ideas for his own work, and he used it when

366
00:18:11,840 --> 00:18:14,200
Speaker 1: he had writer's block. He wasn't using it to replace

367
00:18:14,680 --> 00:18:17,400
Speaker 1: his creative skill. He was using it as an enhancement tool.

368
00:18:17,440 --> 00:18:19,639
Speaker 1: And I think that's really interesting. Yeah, And in season

369
00:18:19,640 --> 00:18:21,920
Speaker 1: one of sleep Walkers, we spoke to a filmmaker called

370
00:18:21,960 --> 00:18:24,760
Speaker 1: Oscar Shop who actually shot a whole film written by

371
00:18:24,800 --> 00:18:28,880
Speaker 1: Ai called Sunspring. Oscar and Chen turned the technology into

372
00:18:28,920 --> 00:18:31,359
Speaker 1: a tool that actually serves their purposes. You know, you

373
00:18:31,400 --> 00:18:34,359
Speaker 1: can develop all kinds of technology in a vacuum. About

374
00:18:34,359 --> 00:18:37,040
Speaker 1: the technology that really serves people and fills a need.

375
00:18:37,400 --> 00:18:39,760
Speaker 1: Is the technology that sticks around, you know, speaking of

376
00:18:39,760 --> 00:18:42,800
Speaker 1: technology that really sticks around, and then some technology that

377
00:18:43,160 --> 00:18:46,399
Speaker 1: might not stick around. There's so much stuff on the

378
00:18:46,440 --> 00:18:49,879
Speaker 1: floor of CS you and I have never been to

379
00:18:49,880 --> 00:18:52,159
Speaker 1: see us before. I think we were very overwhelmed by

380
00:18:52,160 --> 00:18:54,880
Speaker 1: what we saw and excited. It was kind of inspected

381
00:18:54,920 --> 00:18:58,679
Speaker 1: gudgets paradise, and obviously, as someone who's obsessed with technology

382
00:18:58,880 --> 00:19:01,639
Speaker 1: and consumer technology, I would have bought every single thing

383
00:19:01,640 --> 00:19:04,320
Speaker 1: I thought you tried to buy. I did try to

384
00:19:04,359 --> 00:19:07,080
Speaker 1: buy that. I mean that keyboard with the mouse burnt

385
00:19:07,119 --> 00:19:11,400
Speaker 1: into the keyboard. How much did this cost? I almost

386
00:19:11,400 --> 00:19:17,040
Speaker 1: bought a laser cool laser patch for my back, which

387
00:19:17,119 --> 00:19:21,560
Speaker 1: placebo or not made me look very hot. But no,

388
00:19:21,760 --> 00:19:24,520
Speaker 1: in all seriousness, you know, there are things that were

389
00:19:24,560 --> 00:19:27,320
Speaker 1: on the floor that are kind of amazing when you

390
00:19:27,359 --> 00:19:30,320
Speaker 1: think about it, Like from this company called Pillow Health.

391
00:19:30,680 --> 00:19:33,359
Speaker 1: They've developed this device called Priya that looks like a

392
00:19:33,359 --> 00:19:35,520
Speaker 1: little face, a cute little face, as they always do,

393
00:19:36,040 --> 00:19:39,840
Speaker 1: and it's basically a pill dispenser that is voice and

394
00:19:39,880 --> 00:19:43,080
Speaker 1: face activated, so anybody could have one. I could have one,

395
00:19:43,080 --> 00:19:45,919
Speaker 1: You could have one. But I think they've developed it

396
00:19:46,040 --> 00:19:50,080
Speaker 1: mostly for elderly people who have many pills that they

397
00:19:50,080 --> 00:19:54,080
Speaker 1: have to take throughout the day, and who's children or

398
00:19:54,640 --> 00:19:57,800
Speaker 1: health aids want to be able to control when their

399
00:19:57,840 --> 00:20:01,360
Speaker 1: medicine is dispensed. And I think for someone who might

400
00:20:01,480 --> 00:20:07,040
Speaker 1: have memory impairment, physical impairment, the idea that someone who

401
00:20:07,160 --> 00:20:10,920
Speaker 1: isn't in the room with that person could control when

402
00:20:10,960 --> 00:20:15,440
Speaker 1: they're getting you know, vital medicine is really amazing. And

403
00:20:15,760 --> 00:20:18,359
Speaker 1: you know, you say what you will about privacy. I

404
00:20:18,400 --> 00:20:22,800
Speaker 1: think being able to do something like take care of

405
00:20:22,880 --> 00:20:26,359
Speaker 1: your elderly parent with a device is you know, the

406
00:20:26,480 --> 00:20:30,359
Speaker 1: perfect intersection of technology and humanity, right. I spoke to

407
00:20:30,359 --> 00:20:32,760
Speaker 1: the founder about exactly that. And you know, we have

408
00:20:32,800 --> 00:20:35,440
Speaker 1: a lot of concerns about facial recognition that we've discussed

409
00:20:35,480 --> 00:20:38,840
Speaker 1: at length on Sleepwalkers and will continue to discuss in

410
00:20:38,920 --> 00:20:41,600
Speaker 1: season two. But in a narrow use case like this,

411
00:20:41,680 --> 00:20:44,479
Speaker 1: in a voluntary use case where it can help somebody

412
00:20:44,480 --> 00:20:47,639
Speaker 1: out to remember something very important, like what pills are taken,

413
00:20:47,720 --> 00:20:50,280
Speaker 1: when it may well be that that's a sacrifice which

414
00:20:50,320 --> 00:20:53,840
Speaker 1: is very much worth taking. There was another startup on

415
00:20:53,840 --> 00:20:56,600
Speaker 1: the floor that really caught my eye, which was called

416
00:20:57,119 --> 00:20:59,399
Speaker 1: in New Pathy, and according to the card which have

417
00:20:59,440 --> 00:21:01,439
Speaker 1: in front of me. It's the first device in the

418
00:21:01,480 --> 00:21:04,760
Speaker 1: world which is equipped with technology to visualize your dog's

419
00:21:04,840 --> 00:21:09,080
Speaker 1: status from his or her heart rate information. And this

420
00:21:09,200 --> 00:21:11,399
Speaker 1: is basically a harness that you put on your dog

421
00:21:11,640 --> 00:21:14,359
Speaker 1: and it recalls your dog's heart rate and in particular

422
00:21:14,680 --> 00:21:17,240
Speaker 1: the variability in your dog's heart rate to tell you

423
00:21:17,280 --> 00:21:19,720
Speaker 1: if your dog is happy or sad, or anxious or

424
00:21:19,760 --> 00:21:23,480
Speaker 1: excited or curious. And you know, people struggle to know

425
00:21:23,520 --> 00:21:25,679
Speaker 1: what that dogs are thinking. And if you can use

426
00:21:25,800 --> 00:21:30,040
Speaker 1: data from historical doggie feelings to model what a current

427
00:21:30,080 --> 00:21:32,400
Speaker 1: dog is feeling and use that to have better interaction

428
00:21:32,440 --> 00:21:34,800
Speaker 1: with your dog, more power to you. I think it's cool.

429
00:21:34,920 --> 00:21:36,840
Speaker 1: There was this other piece of technology from a company

430
00:21:36,840 --> 00:21:40,560
Speaker 1: called we Labs. They were Japanese, right, and it kind

431
00:21:40,560 --> 00:21:43,000
Speaker 1: of blew my mind in the same way that like

432
00:21:43,119 --> 00:21:45,919
Speaker 1: thinking about automation of drive through blew my mind. You know.

433
00:21:46,640 --> 00:21:50,000
Speaker 1: It was this like would beam that looked like a

434
00:21:50,040 --> 00:21:53,639
Speaker 1: beam in a house, and it had a computer that

435
00:21:53,760 --> 00:21:57,960
Speaker 1: was inside of it. And the woman who was showcasing

436
00:21:57,960 --> 00:22:01,159
Speaker 1: it basically asked ours to stay end up against it

437
00:22:01,240 --> 00:22:04,479
Speaker 1: like you would when you're charting at child's height, and

438
00:22:04,520 --> 00:22:08,120
Speaker 1: she took a pen or stylus and marked Oz's height,

439
00:22:08,920 --> 00:22:12,760
Speaker 1: and then immediately that marking was uploaded into the cloud

440
00:22:13,240 --> 00:22:17,520
Speaker 1: and displayed on a device next to this would beam

441
00:22:17,560 --> 00:22:20,600
Speaker 1: And it just made me think, like this thing that

442
00:22:20,960 --> 00:22:24,240
Speaker 1: millions of families do as their children are growing up

443
00:22:24,920 --> 00:22:27,960
Speaker 1: is now being digitized, and again like going back to

444
00:22:28,000 --> 00:22:33,040
Speaker 1: the intersection of technology and human behavior, like imagine if

445
00:22:33,080 --> 00:22:36,840
Speaker 1: someone moves from the house, those height markings that were

446
00:22:36,920 --> 00:22:40,240
Speaker 1: such a part of your child's growing up can be

447
00:22:40,320 --> 00:22:42,880
Speaker 1: taken with you in the cloud. It's just I mean,

448
00:22:42,920 --> 00:22:44,600
Speaker 1: that's the kind of stuff where I'm like, do I

449
00:22:44,640 --> 00:22:46,960
Speaker 1: need it? Does someone need it? Who cares? But the

450
00:22:47,000 --> 00:22:54,560
Speaker 1: idea that it's like replicating this very very personal feeling

451
00:22:54,720 --> 00:22:57,679
Speaker 1: and you know activity that we do in our childhood

452
00:22:57,880 --> 00:22:59,720
Speaker 1: is I don't know, it kind of blew my mind.

453
00:23:00,040 --> 00:23:01,879
Speaker 1: All three of the things we ended up talking about,

454
00:23:01,880 --> 00:23:04,800
Speaker 1: you know, pillow Health, the doggy heart rate monitor, and

455
00:23:04,840 --> 00:23:08,000
Speaker 1: this Japanese WOULD device. You know, they go back to

456
00:23:08,080 --> 00:23:11,840
Speaker 1: the most human things are our parents, okay, is our

457
00:23:11,880 --> 00:23:14,879
Speaker 1: dog okay? Our children growing up? What does it make

458
00:23:15,000 --> 00:23:17,680
Speaker 1: us feel as they grow up? And so technology that

459
00:23:17,720 --> 00:23:21,760
Speaker 1: addresses those questions in a sensitive and humanistic way will

460
00:23:21,800 --> 00:23:24,520
Speaker 1: always be interesting to us because it really allows us

461
00:23:24,520 --> 00:23:27,480
Speaker 1: to think about and tell stories about ourselves. The oldest

462
00:23:27,480 --> 00:23:29,960
Speaker 1: stories we tell, the stories that are parts of novels

463
00:23:29,960 --> 00:23:32,680
Speaker 1: and films and all other kinds of art. So that's

464
00:23:32,760 --> 00:23:34,920
Speaker 1: to me where technology is most interesting and the types

465
00:23:34,960 --> 00:23:37,480
Speaker 1: of stories that will continue to tell on Sleepwalkers. So

466
00:23:37,560 --> 00:23:40,800
Speaker 1: everything we just talked about is consumer focus and very interesting,

467
00:23:41,160 --> 00:23:43,680
Speaker 1: but a I can also help address problems at scale,

468
00:23:44,040 --> 00:23:47,119
Speaker 1: you know, issues ranging from climate change to pain management,

469
00:23:47,160 --> 00:23:49,640
Speaker 1: and those are all things that we're going to talk

470
00:23:49,640 --> 00:23:53,719
Speaker 1: about in our very exciting season two. Thank you for listening,

471
00:23:53,760 --> 00:23:55,959
Speaker 1: and we're looking forward to seeing you for Season two

472
00:23:55,960 --> 00:24:07,840
Speaker 1: of Sleepwalkers very soon. Und the Roots under the Roots

473
00:24:07,880 --> 00:24:08,680
Speaker 1: for the f