How To Re-think News in the AI Era

WEBVTT

1
00:00:00.040 --> 00:00:01.919
Welcome to this enter life Alex.
You know, great to have you on

2
00:00:02.200 --> 00:00:06.519
the pod today, and you know, the visionary founder. I think of

3
00:00:06.519 --> 00:00:09.880
other web you know and the like
really ground blacking and cool platform that's helping

4
00:00:09.960 --> 00:00:14.519
us rethink what news can be and
what what information can be online, and

5
00:00:14.599 --> 00:00:17.920
so I was excited to get in
touch with you to think about this and

6
00:00:18.320 --> 00:00:24.199
basically how we can rethink how we
consume information online and what is our responsibility

7
00:00:24.199 --> 00:00:27.440
both as creators and as consumers.
So, first off, welcome to the

8
00:00:27.480 --> 00:00:29.760
program, and great to have you. Thank you so much. It's great

9
00:00:29.760 --> 00:00:32.000
to be on this kind of open
up with this idea, why start caring

10
00:00:32.039 --> 00:00:34.119
about digital joke? You know,
when you said it and I heard about

11
00:00:34.159 --> 00:00:35.520
it, I was like, okay, this is this is the thing that

12
00:00:35.520 --> 00:00:38.520
matters. You know. How did
you find yourself asking these questions around joke?

13
00:00:38.560 --> 00:00:41.560
Well, I think I was primed
to care about it from my childhood

14
00:00:41.600 --> 00:00:45.039
a little bit because I was born
on the Soviet Union and I kind of

15
00:00:45.439 --> 00:00:50.520
saw what a society looks like if
nobody has any idea what's going on interesting,

16
00:00:50.640 --> 00:00:54.679
and everybody just makes the wrong assumptions
or they're pretending to know something,

17
00:00:54.759 --> 00:00:58.960
but they're kind of ninety percent sure
it's not true. And I have these

18
00:00:59.039 --> 00:01:02.359
vivid memories of my and slocking themselves
in the closet or in the bathroom or

19
00:01:02.399 --> 00:01:04.640
someplace where the neighbors won't hear,
and listening to the radio to hear a

20
00:01:04.719 --> 00:01:07.799
voice of America at four am.
Wow. Interesting to get at least some

21
00:01:08.079 --> 00:01:11.879
idea of something that might be true, even though it's foreign propaganda. Right,

22
00:01:12.159 --> 00:01:15.239
But it was foreign propaganda in the
country that had its own propaganda that

23
00:01:15.400 --> 00:01:19.879
was worse. So I kind of
thought that knowing stuff is important and it

24
00:01:19.959 --> 00:01:25.000
probably leads to better actions. And
I thought that my entire life. And

25
00:01:25.079 --> 00:01:29.680
so when at some point I had
this crisis of conscience a few years ago

26
00:01:29.680 --> 00:01:32.760
and thought about what I want to
care about next, what I want to

27
00:01:32.760 --> 00:01:37.040
work on next, that seemed like
a good idea. Right, the stuff

28
00:01:37.079 --> 00:01:41.159
we're all consuming is becoming worse.
Yeah, nobody seems to be talking about

29
00:01:41.159 --> 00:01:44.400
it, or if they're talking about
it, they're all trying to come up

30
00:01:44.400 --> 00:01:48.719
with some villain to blame for this, which is typically not why bad stuff

31
00:01:48.760 --> 00:01:52.719
happens in the entire world all at
once. It's usually not a villain,

32
00:01:53.079 --> 00:01:57.079
it's something and the evolution of content
has gone wrong essentially. Yeah, that's

33
00:01:57.079 --> 00:02:00.519
that's actually really interesting, and so
I guess, yeah, the one head,

34
00:02:00.680 --> 00:02:02.439
it's like our brains wanted to jump
to and say, oh, it's

35
00:02:02.480 --> 00:02:06.359
it's this person or this country's or
this idea's fault. But but you're kind

36
00:02:06.400 --> 00:02:09.240
of saying that we're seeing at the
same time just the rise and I'm like

37
00:02:09.319 --> 00:02:13.039
visualizing in my head, right,
just the writhing piles of paper or trash,

38
00:02:13.120 --> 00:02:15.719
right, that just kind of like
piles around us. But it's funny

39
00:02:15.759 --> 00:02:19.039
that digitally we we yes, like
people may not think about that idea,

40
00:02:19.159 --> 00:02:21.879
right, like that there's so much
that's kind of around us all the time.

41
00:02:21.879 --> 00:02:23.960
But you know, I think people
might we definitely recognize, like do

42
00:02:24.039 --> 00:02:25.960
we all feel busier all the time? And I think part of it too,

43
00:02:27.039 --> 00:02:30.280
is like our attention is always grabbed, being pulled towards different little you

44
00:02:30.319 --> 00:02:32.240
know, whatever blips and blaps of
our phone or emails. But most of

45
00:02:32.280 --> 00:02:35.199
the time, I think, to
your point, it's a lot of junk,

46
00:02:35.280 --> 00:02:38.120
right, like the reason we have
a spammer junk filter in our email.

47
00:02:38.159 --> 00:02:39.439
But beyond that too, like how
do we know what we're consuming?

48
00:02:39.479 --> 00:02:43.120
So uh, I'm curious to you
have to kind of get get your perspective

49
00:02:43.159 --> 00:02:46.039
on this too, because like,
as we've gone in like deeper deeper into

50
00:02:46.039 --> 00:02:47.599
the digital realm, and we're not
going to come out of that like back

51
00:02:47.639 --> 00:02:52.759
to back to paper or something else
that's more analog entirely. You know,

52
00:02:52.919 --> 00:02:55.120
has as we see have you seen
like people's kind of perception about this like

53
00:02:55.319 --> 00:02:59.199
shift in terms of like do they
feel like they're getting more junk now than

54
00:02:59.240 --> 00:03:02.280
we had before? Like is it
something qualitatively different? I guess about like

55
00:03:02.319 --> 00:03:06.840
the kind of junk that we're seeing
than twenty years ago. I mean,

56
00:03:07.039 --> 00:03:09.039
you can both quantify it and look
at it qualitatively, right, So we

57
00:03:09.080 --> 00:03:14.680
tried to quantify it recently by hiring
the Harris Polling Company to do a poll.

58
00:03:14.800 --> 00:03:16.960
And one of the questions they asked
is are you spending more time than

59
00:03:17.000 --> 00:03:21.719
you did five years ago reading the
news? And more than eighty percent of

60
00:03:21.759 --> 00:03:24.000
people said yes. Interesting, And
they asked, okay, do you have

61
00:03:24.039 --> 00:03:28.000
more confidence than you did five years
ago that you know what is true?

62
00:03:28.919 --> 00:03:31.520
And most people said no. Then, so people are spending more time and

63
00:03:31.560 --> 00:03:36.280
they know less and they are certain
of less, and they find it harder

64
00:03:36.319 --> 00:03:39.120
to figure out what is true.
So quantitatively, that's happening all around us

65
00:03:39.280 --> 00:03:44.199
now. Qualitatively we are noticing this. Right when you see CNN publish an

66
00:03:44.280 --> 00:03:46.919
article with a title like stop what
you're doing and watch your slefant play with

67
00:03:46.960 --> 00:03:52.560
bubbles, that's an actual see in
an article, right you realize in your

68
00:03:52.639 --> 00:03:59.520
childhood this wasn't the thing, and
see that stuff pass editorial review in the

69
00:04:00.080 --> 00:04:03.000
you know, Cayble News Network in
the nineties. So something is new,

70
00:04:03.080 --> 00:04:08.759
something is happening here. It's happening
gradually, but it's now reached this kind

71
00:04:08.800 --> 00:04:12.080
of monstrous proportion. Where by the
way, you mentioned spam filter. One

72
00:04:12.120 --> 00:04:15.879
of the questions we asked was do
you think there's a need for a spam

73
00:04:15.920 --> 00:04:20.040
filter on the endternet? And eighty
nine percent said yes, I would agree

74
00:04:20.040 --> 00:04:23.680
with that. Yeah, that's an
interesting question. So it's like part of

75
00:04:23.680 --> 00:04:26.639
it is we might have a filter
problem. But then I think you know,

76
00:04:26.639 --> 00:04:30.560
as you know there too, like
what used to not pass editorial content

77
00:04:31.439 --> 00:04:35.000
walls now seems to do that,
right? Is that part because we've seen

78
00:04:35.040 --> 00:04:38.879
like I don't know, business model
shift, you know, from from tech

79
00:04:39.079 --> 00:04:43.279
advertising and things like that in terms
of like what or like why we're getting

80
00:04:43.279 --> 00:04:46.279
so much content, Like like why
in your experience have we seen that shift

81
00:04:46.279 --> 00:04:48.319
happening, like in terms of we're
getting more stuff that wouldn't pass, you

82
00:04:48.360 --> 00:04:51.240
know, twenty years ago. So
there's a few reasons, but I think

83
00:04:51.279 --> 00:04:56.480
the big one it's not that the
business model changed. It's that we actually

84
00:04:56.480 --> 00:05:00.600
became better at implementing the business model
we already had. So starting from let's

85
00:05:00.600 --> 00:05:04.120
say two thousand and three or so, we became really really good at tracking

86
00:05:04.160 --> 00:05:09.519
how ads perform. We can now
track every impression, every click, every

87
00:05:09.519 --> 00:05:15.000
single thing. And there's no longer
this consumption of a packaged good called and

88
00:05:15.319 --> 00:05:18.079
daily issue of something where you have
a whole bunch of articles at once,

89
00:05:18.360 --> 00:05:23.600
every article fence for itself, and
it's all judged by this perfect tracking on

90
00:05:24.000 --> 00:05:28.800
how ads on that article perform.
And once you have that combination, now

91
00:05:28.959 --> 00:05:31.839
suddenly there is just no incentive to
produce anything other than clicks and views.

92
00:05:32.199 --> 00:05:35.319
So you have a single selective pressure. You're an evolution guy, so you

93
00:05:35.399 --> 00:05:40.839
understand what that leads to, right. It all leads towards all content evolving

94
00:05:40.879 --> 00:05:45.319
over time towards that thing, regardless
of what the intentions of the people creating

95
00:05:45.360 --> 00:05:48.120
it are so if the intentions are
good that it doesn't matter. The ones

96
00:05:48.120 --> 00:05:51.399
who have bad intentions probably outperformed the
ones who have good intentions in the long

97
00:05:51.480 --> 00:05:55.959
run. For as to say,
for a better offit worse, we're probably

98
00:05:55.959 --> 00:05:59.040
just for worse, right, Yep, that's but yes, I think I

99
00:05:59.079 --> 00:06:01.240
think that's a great point too,
in a great view of like the incentive

100
00:06:01.240 --> 00:06:05.120
structure of just the business model getting
better, we're going to see than the

101
00:06:05.160 --> 00:06:09.759
pressure kind of push us in one
direction. So as you think about that,

102
00:06:09.800 --> 00:06:12.600
like, I think there's I mean, I think a ton of directions

103
00:06:12.600 --> 00:06:15.879
that I want to explore here,
but like to keep us like moving in

104
00:06:16.079 --> 00:06:18.199
this one for now, like thinking
about this idea of like, as we're

105
00:06:18.199 --> 00:06:21.399
seeing this explosion of this kind of
content and we have this pressure that's basically

106
00:06:21.399 --> 00:06:27.759
just trying to feed this one algorithm, Like how can we I guess in

107
00:06:27.839 --> 00:06:30.800
one hand, like think about sort
of shifting that model, right, you

108
00:06:30.800 --> 00:06:32.519
know, and this could be kind
of work you're doing with other web but

109
00:06:32.600 --> 00:06:36.920
just also thinking that just like how
do we I guess? Are like in

110
00:06:36.959 --> 00:06:40.279
addition to yourself, are we seeing
other people kind of ask this question and

111
00:06:40.319 --> 00:06:42.399
say, hey, hold on,
like, there's just a lot coming at

112
00:06:42.439 --> 00:06:45.399
me all the time. We should
ask about how we think about sources and

113
00:06:45.439 --> 00:06:48.600
like what's what's actually moving in this
way, like what it looks like to

114
00:06:48.639 --> 00:06:51.000
you. So there's a lot of
people ask him the same question, and

115
00:06:51.040 --> 00:06:55.519
they are all coming up with different
solutions. So I have some reasons why

116
00:06:55.519 --> 00:06:59.000
I think mine is best, right, But in reality, you have a

117
00:06:59.000 --> 00:07:01.600
lot of companies like use Guard that
are basically saying which sources are good,

118
00:07:01.839 --> 00:07:05.519
and all they do is kind of
the JD Power approach which works in automotive.

119
00:07:05.639 --> 00:07:09.600
They apply it here. Just the
rank sources, give them some number

120
00:07:09.600 --> 00:07:13.519
of stars, publish annual rankings,
things like that. The problem I see

121
00:07:13.560 --> 00:07:16.720
with that approach as it's basically spreadsheet
level filtering. Right, So first of

122
00:07:16.720 --> 00:07:20.000
all, you're assuming that all sources
are consistent. But what is substack?

123
00:07:20.079 --> 00:07:25.040
Is substack good or bad? Question? It's both. There's a huge variance

124
00:07:25.040 --> 00:07:27.879
there, right, But Okay,
what is the Washington Examiner? Is it

125
00:07:27.879 --> 00:07:30.759
good or bad? Seems to me
like the news articles are good, the

126
00:07:30.839 --> 00:07:33.800
op eds kind of suck, right, So there's a lot of variance even

127
00:07:33.839 --> 00:07:39.199
within that in itself. So that's
one approach and I have my columns with

128
00:07:39.240 --> 00:07:42.040
it, but it's good that it
exists that people are working on it.

129
00:07:42.199 --> 00:07:46.560
Then there's the Tristan Harris approach of
just going to Facebook and saying please be

130
00:07:46.639 --> 00:07:50.519
nice. And I'm sure that there's
some value in that. It's creating this

131
00:07:50.639 --> 00:07:56.560
sort of public pressure to not be
evil, to use a Google slogan for

132
00:07:56.720 --> 00:08:01.240
years past, right, But I
don't know that that works when it collides

133
00:08:01.319 --> 00:08:05.439
with the c Corps fiduciary duty to
maximize the older value. It seems like

134
00:08:05.560 --> 00:08:11.079
being nice if you're a company that
makes money from ads is kind of it

135
00:08:11.120 --> 00:08:16.560
goes against what makes you money and
what you're supposed to do as an executive.

136
00:08:16.600 --> 00:08:18.759
So I'm not sure that's going to
work in the long run. So

137
00:08:18.839 --> 00:08:22.560
our approach has been for as long
as this is the ecosystem we are in.

138
00:08:24.319 --> 00:08:28.319
We want to create some sort of
penalty for bad content, so we

139
00:08:28.360 --> 00:08:31.720
want to help people select good content. And when people select good content,

140
00:08:33.039 --> 00:08:37.360
the secondary consequence is that they're not
selecting bad content, and therefore people creating

141
00:08:37.360 --> 00:08:43.159
bad content get penalized less people see
them. Now, this will only become

142
00:08:43.639 --> 00:08:48.919
really meaningful when we become really large, or if one hundred more other webs

143
00:08:48.960 --> 00:08:52.399
come up and do the same thing, and then all the people who consume

144
00:08:52.879 --> 00:08:56.440
content through something like this will basically
filter the bad stuff out. Bad stuff

145
00:08:56.480 --> 00:09:03.000
gets pinalized, air go less incentive
to produce it. Now it could be

146
00:09:03.120 --> 00:09:07.200
less binary. So I see a
future in which it's not just filtered out

147
00:09:07.279 --> 00:09:11.080
or not filtered out, but in
which we can go to ad networks or

148
00:09:11.080 --> 00:09:16.320
advertisers and say, here's how you
evaluate how each article would score. Give

149
00:09:16.519 --> 00:09:22.320
that to advertisers, or if you're
an advertiser, yourself scale how much you're

150
00:09:22.360 --> 00:09:24.159
willing to pay for an ad or
for a click or for a view on

151
00:09:24.559 --> 00:09:30.279
this content, right, and that
would create a real kind of sliding scale

152
00:09:30.440 --> 00:09:33.960
incentive for people to produce better content
than they otherwise would have. It's like

153
00:09:33.960 --> 00:09:37.080
like provide And that's the interesting idea
though, because it's like providing a scale

154
00:09:37.120 --> 00:09:41.879
of like what would be worth for
them monetary value of based on like how

155
00:09:41.960 --> 00:09:46.279
how I guess? I mean.
The question that naturally comes to mind then

156
00:09:46.360 --> 00:09:48.639
is like, how are we thinking
about what makes good and bad content?

157
00:09:48.080 --> 00:09:52.639
Right? Like if it's a one
handed filter? But then also like how

158
00:09:52.639 --> 00:09:58.159
can we determine these ideas like beyond
a qualitative opinion, like a clickbait comes

159
00:09:58.200 --> 00:10:00.759
to mind, is like a topic
that doesn't actually tell you anything article itself,

160
00:10:00.799 --> 00:10:03.000
you know, But like what makes
something good or bad in this context?

161
00:10:03.320 --> 00:10:05.960
It's a tough question. I've talked
to a bunch of people with degrees

162
00:10:05.960 --> 00:10:11.960
in philosophy and they don't agree with
each other, right, But yeah,

163
00:10:11.000 --> 00:10:16.000
in general, I would say that
good content is content that serves its stated

164
00:10:16.039 --> 00:10:18.519
purpose. To the degree that something
does, it's good, right, And

165
00:10:18.559 --> 00:10:22.919
you can basically rank content along how
well does it serve the purpose. So

166
00:10:22.960 --> 00:10:26.639
when I look at news, it's
supposed to inform you about what happened,

167
00:10:28.200 --> 00:10:31.440
right, So to the degree that
it informs you, it's good. To

168
00:10:31.519 --> 00:10:35.120
the degree that it tries to persuade
you without actually giving you new information,

169
00:10:35.240 --> 00:10:39.080
it's probably not so good. Right. So, if you look at how

170
00:10:39.080 --> 00:10:43.519
we rank content, it tends to
track pretty well with your impression of how

171
00:10:43.559 --> 00:10:46.159
informative versus opinionated this thing is interesting, okay, Right, But then if

172
00:10:46.159 --> 00:10:50.200
you're looking at opinion content, obviously
it's not supposed to be purely informative.

173
00:10:50.240 --> 00:10:54.799
It's supposed to be opinionated and controversial
in all those things. So You're probably

174
00:10:54.840 --> 00:11:00.720
going to grade it based on how
interesting it is or how much it stimulates

175
00:11:00.720 --> 00:11:05.240
your gray matter, that kind of
thing. So every type of content is

176
00:11:05.240 --> 00:11:09.120
going to have its own criteria.
Now how do you rank it? That

177
00:11:09.200 --> 00:11:13.279
in itself is a hard question.
Even once you know what the criteria are,

178
00:11:13.000 --> 00:11:16.919
how exactly do you rank content?
Our approach to it has been,

179
00:11:16.120 --> 00:11:20.000
at least in the news department,
we basically say, what are all the

180
00:11:20.039 --> 00:11:24.559
bad things we know how to detect? Your score is going to be one

181
00:11:24.600 --> 00:11:28.919
hundred minus adding up all the bad
things. A clickbait is an example you

182
00:11:28.960 --> 00:11:33.440
mentioned. That's one of the things
that we detect, and if the headline

183
00:11:33.480 --> 00:11:37.279
has clickbait, you lose something like
five points, and then over time we

184
00:11:37.399 --> 00:11:41.279
just add up all the penalties.
Cool is that? Is that that list

185
00:11:41.279 --> 00:11:43.559
of bad things? Is this kind
of an open source list or is this

186
00:11:43.600 --> 00:11:48.480
kind of a trade trade? Each
one of those models is source available,

187
00:11:48.559 --> 00:11:50.200
so which basically means you can look
at it, you can test that,

188
00:11:50.200 --> 00:11:56.759
you can't reuse it at open source
and in general, you can see a

189
00:11:56.799 --> 00:12:01.440
nutrition label next to each article in
our platform, which includes most of these

190
00:12:01.480 --> 00:12:03.440
outputs. Some of them are not
their row outputs, but with some additional

191
00:12:03.480 --> 00:12:07.879
processing. There's also source available,
right, But that was the general idea

192
00:12:07.919 --> 00:12:11.679
that you can see kind of what
we ran. In fact, you can

193
00:12:11.720 --> 00:12:15.960
see some things there that don't actually
end up in the big score. So

194
00:12:16.559 --> 00:12:18.679
yeah, if the article has affiliate
links, I don't know whether I should

195
00:12:18.720 --> 00:12:22.960
penalize it or not, but I
think you want to know about it,

196
00:12:22.120 --> 00:12:26.679
yeah, because if you're using it
basically as a product endorsement, you should

197
00:12:26.679 --> 00:12:28.559
probably take it with a grain of
salt. If you're just looking at it

198
00:12:28.600 --> 00:12:33.759
to figure out which new laptops came
out this year, then you don't care

199
00:12:33.840 --> 00:12:35.960
that it has affiliate lengths. It's
still good for you, right, So

200
00:12:35.000 --> 00:12:37.759
it depends on your purpose. It
seems like that's the kind of part of

201
00:12:37.799 --> 00:12:41.519
the work of kind of retraining or
thinking around around getting rid of Junker.

202
00:12:41.559 --> 00:12:46.360
Being smarter or mindful about how you
consume. Is understanding kind of user intent

203
00:12:46.360 --> 00:12:48.840
also, or like you're helping users
recognize what their intent is. We're looking

204
00:12:48.919 --> 00:12:50.879
for for some content, right,
Because if I you're right, if I'm

205
00:12:50.879 --> 00:12:54.720
looking for the best laptops I came
out this year, I'm actually happy to

206
00:12:54.720 --> 00:12:56.279
have links, right because I can
go there and find where to buy them

207
00:12:56.360 --> 00:13:01.559
versus you know, if it doesn't
show me so is that kind of part

208
00:13:01.600 --> 00:13:03.159
of the model. But if you're
trying to select which one is the best

209
00:13:03.159 --> 00:13:05.679
one, then that's probably a bad
article for you, right because all I

210
00:13:05.759 --> 00:13:09.399
did was copypaste the marketing materials from
the manufacturers. That's a good point too.

211
00:13:09.480 --> 00:13:13.120
Yeah, So I guess do you
do you write that kind of or

212
00:13:13.159 --> 00:13:15.240
do you kind of check for that
that type of thing, or you can

213
00:13:15.240 --> 00:13:20.399
see the articles actually just doing that, like pulling press releases. Well,

214
00:13:20.480 --> 00:13:24.679
so we definitely see that. I
wouldn't even think that we need to penalize

215
00:13:24.679 --> 00:13:26.840
for that because at that point we
don't know the users intent. So what

216
00:13:26.879 --> 00:13:31.480
we can do is we can help
users kind of structure their feed the way

217
00:13:31.480 --> 00:13:35.080
they want. And here I don't
know that I can define how to even

218
00:13:35.120 --> 00:13:39.000
ask the user a direct question what
do you want and then adjust their feed

219
00:13:39.080 --> 00:13:41.960
based on that. So I just
try to give the user as many knobs

220
00:13:41.960 --> 00:13:46.639
as possible. So actually, if
you go to our advanced customizations, you're

221
00:13:46.639 --> 00:13:50.240
going to see a lot of things
that kind of look bizarre. We even

222
00:13:50.279 --> 00:13:54.440
classify every article based on which emotion
it's most likely to evoken the reader.

223
00:13:54.879 --> 00:13:58.279
And then allow you to customize how
much of each emotion you'd like to get

224
00:13:58.320 --> 00:14:01.080
from your feet interesting, and we
try to detect that automatically and adjusted,

225
00:14:01.120 --> 00:14:05.919
but then you can go and override
everything we detected automatically because we don't believe

226
00:14:05.960 --> 00:14:09.759
in black boxes. Yeah, I
think like that's the question of transparency.

227
00:14:09.799 --> 00:14:11.759
I think is has become so important
right as we to go back to your

228
00:14:11.759 --> 00:14:16.559
hair's poll point above, like that
we read more, but we like trust

229
00:14:16.600 --> 00:14:18.840
less right that the more kind of
we consume, and some part of that

230
00:14:18.879 --> 00:14:22.320
too is recognizing what's coming at us. But then also to this point of

231
00:14:22.360 --> 00:14:26.039
like can we see how and why
it's coming at us right now? Can

232
00:14:26.080 --> 00:14:28.120
I what role of control do I
have? And that I think it's really

233
00:14:28.159 --> 00:14:31.360
interesting point. I'm curious like as
you think about like both data, like

234
00:14:31.399 --> 00:14:35.879
do you see like patterns of behavior
emerge that are interesting, Like in terms

235
00:14:35.919 --> 00:14:39.759
of you see a lot of people
pre selecting for happiness or like they're they're

236
00:14:39.799 --> 00:14:41.600
like are they are? They are
they not selecting for emotion? But then

237
00:14:41.639 --> 00:14:46.320
it turns out we see like a
you know, a likelihood or a preponderance

238
00:14:46.360 --> 00:14:52.360
of of like I don't know slightly
enraging articles pop up as people click through

239
00:14:52.399 --> 00:14:54.799
or something. You know. Yeah, I think the sad thing that I'm

240
00:14:54.840 --> 00:14:56.840
seeing for the most part is the
most most people, even if they find

241
00:14:56.840 --> 00:15:01.159
the configurations, don't use them.
They just use it default. So,

242
00:15:01.279 --> 00:15:03.440
you know, I wish I could
change that, but it's really hard to

243
00:15:03.440 --> 00:15:07.559
get users to change the default.
So we have this parallel track in our

244
00:15:07.600 --> 00:15:13.080
mind where we want to give users
all the controls that they might ever want

245
00:15:13.120 --> 00:15:16.320
to have, but at the same
time, we have to spend most of

246
00:15:16.320 --> 00:15:20.799
our efforts tuning the defaults because that's
what most users are going to end up

247
00:15:20.039 --> 00:15:24.519
using. Now, I wanted to
mention one other thing based on what you

248
00:15:24.559 --> 00:15:28.759
said earlier, which is that there
is probably a nefarious element here of what

249
00:15:28.840 --> 00:15:33.759
you call the single algorithm. Right, The vast majority of social media and

250
00:15:33.840 --> 00:15:39.000
other media companies out there right now
are essentially tuning everything for maximum engagement.

251
00:15:39.360 --> 00:15:43.360
Right. They want to get you
more engaged, they want you to spend

252
00:15:43.399 --> 00:15:46.039
more time on site, they want
to share more, to reply more,

253
00:15:46.080 --> 00:15:50.639
etc. Now, engagement is something
they get out of you, it's not

254
00:15:50.799 --> 00:15:54.399
value that they give to you,
right, And so if users are feeling

255
00:15:54.440 --> 00:15:58.200
like they're spending more time and they're
getting less in return. It's not accidental.

256
00:16:00.279 --> 00:16:03.559
That's exactly what the algorithm has been
tuned for to get as much as

257
00:16:03.639 --> 00:16:06.720
possible out of you, and how
much it gives you in return. This

258
00:16:07.039 --> 00:16:11.039
kind of beside the point, right, and therefore it kind of gets kewed

259
00:16:11.039 --> 00:16:14.039
in one direction. So I think
as a user, you always have to

260
00:16:14.039 --> 00:16:18.120
be cognizant that this thing wants something
from you, but you're not here to

261
00:16:18.200 --> 00:16:22.000
give the thing what it wants.
You're here to get value, right,

262
00:16:22.039 --> 00:16:26.200
and you're optimizing for something that is
the opposite of the algorithm. From that

263
00:16:26.279 --> 00:16:30.639
perspective, any algorithm that is a
black box is already bad for you,

264
00:16:30.159 --> 00:16:34.440
right. You want an algorithm that's
maximally tunable because then you at least have

265
00:16:34.480 --> 00:16:37.240
a chance to get some value out
of it. If it's not tunable at

266
00:16:37.240 --> 00:16:41.480
all, then you've lost the battle
already. That's a super important point.

267
00:16:41.879 --> 00:16:44.879
And it is always like one of
those back and forth challenges like working in

268
00:16:44.919 --> 00:16:48.360
user experience. Right, in an
experience design that you want to give the

269
00:16:48.440 --> 00:16:51.360
user a sense of both control in
terms of like what they what parameters they

270
00:16:51.360 --> 00:16:52.919
may want to use. But then
you know this is a good example where

271
00:16:53.519 --> 00:16:57.000
they either only use so much ago
for the default, but then also but

272
00:16:57.159 --> 00:17:02.879
this fundamental point is great where it's
like, engagement is not about me as

273
00:17:02.919 --> 00:17:04.920
the user, right, it's about
me serving an organization which feels weird.

274
00:17:04.920 --> 00:17:07.119
And I mean that's a great way
to like define this idea. I was

275
00:17:07.160 --> 00:17:12.200
talking with a documentarian David Donnelly a
few episodes ago about kind of the price

276
00:17:12.240 --> 00:17:15.799
of like convenience rather the cost of
convenience for this kind of like algorithmic software,

277
00:17:15.799 --> 00:17:18.039
and this is a it's a good
one here where it's like there's the

278
00:17:18.039 --> 00:17:22.599
the emotional social media challenge point,
but then there's other side here that that

279
00:17:22.640 --> 00:17:26.119
you're saying to which is really important
to recognize that like what I'm getting out

280
00:17:26.160 --> 00:17:30.359
of my typical news media consumption is
incidental to what they're actually doing getting out

281
00:17:30.359 --> 00:17:32.799
of me, which is you know, my engagement, my clicks looking through

282
00:17:32.839 --> 00:17:36.680
ads and things like that. That's
that's a that's a great point too to

283
00:17:36.680 --> 00:17:40.640
think about. And so as as
you approach that is this like when we're

284
00:17:41.119 --> 00:17:42.920
kind of giving users between those two, its like of like you know,

285
00:17:44.160 --> 00:17:48.480
as many options as they might ever
need versus like tuning that that default you

286
00:17:48.519 --> 00:17:52.799
know, I've got thinking about this, I guess in in the of the

287
00:17:52.920 --> 00:17:56.119
broader question of like how do we
move towards what we feel is kind of

288
00:17:56.160 --> 00:18:00.200
like ethical technology right now? Theical
technology development, you know, because it

289
00:18:00.200 --> 00:18:03.920
sounds on one hand like there's the
Tristan Harris model and all tech is human

290
00:18:03.000 --> 00:18:07.079
kind of approach, which sounds nice
and humanistic because on one hand it is

291
00:18:07.160 --> 00:18:11.039
right. But then there's this other
side here of like, okay, but

292
00:18:11.720 --> 00:18:15.480
how are we actually making and deploying
ethical technology or ethics into our technology itself

293
00:18:17.359 --> 00:18:18.319
and for people to access? And
so I'm curious, like how you think

294
00:18:18.319 --> 00:18:22.200
about this idea of like and I
regize I'm putting the word ethics into this

295
00:18:22.240 --> 00:18:25.759
conversation, but just like, I'm
curious how you approach that idea of developing

296
00:18:26.000 --> 00:18:27.519
technology for people. I mean,
if you see the wall behind me and

297
00:18:27.559 --> 00:18:30.240
the bookshelf behind me, you see
the word ethics appear many times in many

298
00:18:30.240 --> 00:18:34.000
places. I care about the subject
deeply. I don't know that you can

299
00:18:34.160 --> 00:18:37.640
put that as a parallel track.
I mean, we tried in some sounds,

300
00:18:37.680 --> 00:18:41.559
we registered ourselves as a public benefit
corporation to have this parallel track of

301
00:18:41.960 --> 00:18:45.920
we have a mission that we're trying
to maximize and we're trying to make money,

302
00:18:45.240 --> 00:18:48.200
and we are not forced to choose
just one and not the other.

303
00:18:48.400 --> 00:18:52.680
Right, we can maximize both,
But most companies aren't like that. So

304
00:18:52.759 --> 00:18:56.079
I guess my first answer would be, let's try to maximize the number of

305
00:18:56.119 --> 00:19:00.799
public benefit corporations in the US instead
of just sea corps, because a sea

306
00:19:00.839 --> 00:19:04.559
corp is a very limited structure.
If even if an executive really wants to

307
00:19:04.599 --> 00:19:10.440
be ethical, sometimes it goes against
their fiduciary duty and they might get sued

308
00:19:10.440 --> 00:19:14.519
by doing the ethical thing. So
that's one answer. The other one,

309
00:19:14.599 --> 00:19:19.079
which I guess is more higher level
abstract, is that incentives drive everything,

310
00:19:19.680 --> 00:19:26.319
and every time you try to regulate
something against the incentives of the population,

311
00:19:26.000 --> 00:19:30.119
regulation fails. In some sense,
it becomes a corrupt system because now all

312
00:19:30.160 --> 00:19:34.119
you need to do is maximize your
incentives, and they then buy the regulators

313
00:19:34.119 --> 00:19:37.160
with the money you just made.
So that's what we see in a lot

314
00:19:37.160 --> 00:19:41.720
of different industries, and I think
the solution to that is to just think

315
00:19:41.759 --> 00:19:45.920
through the incentives, think through the
selective pressures that we are creating for this

316
00:19:45.079 --> 00:19:51.799
population of people and population of information. Meames for lack of a better term,

317
00:19:52.039 --> 00:19:56.279
and try to optimize that because ultimately
that's what's going to happen, whether

318
00:19:56.319 --> 00:20:00.240
you'll legislate against it or not.
Yeah, that's true. I guess here,

319
00:20:00.440 --> 00:20:03.960
like the social often will structure,
but then follow the incentives, right

320
00:20:03.039 --> 00:20:06.759
and then and then whether or not
you like set up rules in that place

321
00:20:07.240 --> 00:20:08.720
in front of those, it's going
to then either find a way around.

322
00:20:08.799 --> 00:20:14.759
And people are ultimately really good at
convincing themselves that they're doing the right thing

323
00:20:15.119 --> 00:20:18.759
when they're doing what benefits them,
right, And so nobody just or at

324
00:20:18.839 --> 00:20:22.119
least very few people set out with
I'm going to do the bad thing and

325
00:20:22.160 --> 00:20:25.200
make a lot a lot of money, right. No, yeah, hopefully

326
00:20:25.279 --> 00:20:30.359
that they first they're following the path
that seems to make a lot of money

327
00:20:30.400 --> 00:20:33.000
for them, right, and then
time up they come up with the reason

328
00:20:33.039 --> 00:20:37.960
of why this is amazing and it's
actually the good thing and the ethical thing,

329
00:20:37.079 --> 00:20:41.720
right, don't do evil connecting the
world. We see that with every

330
00:20:41.759 --> 00:20:45.680
company, right, they all have
good missions and then they just make a

331
00:20:45.680 --> 00:20:48.079
bunch of money and yeah, but
they still believe in the mission. It's

332
00:20:48.119 --> 00:20:52.200
not like they just scratched that and
throw it out. It's just somehow the

333
00:20:52.240 --> 00:20:56.079
mission morphs into whatever happens to make
the most money. Yeah, that's interesting,

334
00:20:56.079 --> 00:20:57.079
I mean, and so how are
you thinking about that? I guess

335
00:20:57.160 --> 00:21:00.319
like as other web is growing,
I think you have over four million users

336
00:21:00.359 --> 00:21:03.960
now, right, and like how
are you getting close to ten? See

337
00:21:04.000 --> 00:21:07.400
my name is that to day?
So there you go. So that's amazing,

338
00:21:07.480 --> 00:21:07.960
you know. And so it's like, how do you think about that

339
00:21:08.079 --> 00:21:12.240
as you scale, like both having
the mission in play and then and then

340
00:21:12.359 --> 00:21:18.440
kind of keeping that So one way
as again the public benefit corporation, right,

341
00:21:18.519 --> 00:21:21.799
the other one is the source available
right. Both of these are essentially

342
00:21:21.880 --> 00:21:26.240
ways to bind our future selves and
two so that if in the future we

343
00:21:26.319 --> 00:21:32.160
have the incentive to more for mission
into something slightly less beneficial, it's going

344
00:21:32.240 --> 00:21:36.480
to be harder for us. Right. So that's I guess the best I

345
00:21:36.480 --> 00:21:40.279
can do at the company level.
Now. One other thing to kind of

346
00:21:40.279 --> 00:21:45.599
consider here is that everything that we
discussed so far, and I'm sorry if

347
00:21:45.599 --> 00:21:48.559
this is going to explode the discussion, right, but everything we discussed so

348
00:21:48.640 --> 00:21:56.000
far maybe temporary because we're getting really
close to this inflection point where AI models

349
00:21:56.799 --> 00:22:00.799
are as good at writing as the
best humans. Yeah. Right, it's

350
00:22:00.839 --> 00:22:04.519
basically like chess circa nineteen ninety seven. Right, Deep Blue is just about

351
00:22:04.519 --> 00:22:10.359
to beat casparav And the moment that
happens, all the incentives that we just

352
00:22:10.359 --> 00:22:14.839
discussed they changed a lot, and
it's probably going to be pretty binary.

353
00:22:14.839 --> 00:22:18.160
It's not going to be gradual,
right, there's just a magical moment in

354
00:22:18.240 --> 00:22:22.039
time at which the work of sitting
down and writing an article becomes worth zero

355
00:22:22.079 --> 00:22:26.960
dollars. A scary prospect, I
think, but that's interesting, Yeah,

356
00:22:26.160 --> 00:22:29.880
it is. Right now, it's
still really hard to find facts to write

357
00:22:29.920 --> 00:22:32.960
about, and it's still really hard
to figure out what will people care about

358
00:22:33.039 --> 00:22:36.799
that I should actually write about.
Right, both of these decisions are probably

359
00:22:36.839 --> 00:22:40.839
still human for a while, right, AI models can't go to a garage

360
00:22:40.839 --> 00:22:42.160
and talk to Deep Throat and figure
out what our gate, Right, it

361
00:22:42.240 --> 00:22:47.119
has to be humans. But once
you have a set of fact, it

362
00:22:47.160 --> 00:22:49.680
doesn't need to be written down by
humans anymore. It can be just documented

363
00:22:49.880 --> 00:22:53.319
in shorthand or filled out in a
spreadsheet somewhere, right. Yeah, And

364
00:22:53.359 --> 00:22:57.920
then after that, the act of
writing itself seems like a very well defined,

365
00:22:59.079 --> 00:23:02.480
well bound task. So once that
happens, a lot of what we

366
00:23:02.640 --> 00:23:07.079
just discussed might change a lot,
because do you really need this many people

367
00:23:07.160 --> 00:23:12.640
writing stuff? Do you really need
publications that package many written articles together and

368
00:23:12.720 --> 00:23:18.839
call it a daily issue? That
already seems questionable even since the Internet started,

369
00:23:18.880 --> 00:23:22.880
because nobody's reading daily issues anymore.
People are consuming one article at a

370
00:23:22.960 --> 00:23:26.759
time, so the whole idea of
a publication is kind of pass a already.

371
00:23:26.000 --> 00:23:30.039
People are just picking specific articles from
Twitter. Right, But do you

372
00:23:30.079 --> 00:23:36.319
need Twitter in this world? Right? In a world where writing is free,

373
00:23:36.720 --> 00:23:40.920
Why can't something just be written for
you on demand with a target audience

374
00:23:40.960 --> 00:23:45.960
of n equals one? Why do
you need a bunch of written things that

375
00:23:47.240 --> 00:23:51.640
have each been read by eighty eighty
thousand different people, right, and you

376
00:23:51.720 --> 00:23:55.160
just need an algorithm that selects the
ones that you might like. Why not

377
00:23:55.240 --> 00:23:56.839
just have it written for you?
It costs zero dollars to do that?

378
00:23:57.400 --> 00:24:02.160
Right? And so the this is
why you asked me before we started about

379
00:24:02.200 --> 00:24:06.279
our new news concierge. That's the
first kind of prototype towards doing something like

380
00:24:06.319 --> 00:24:10.559
this, writing for you on demand
whilstill you know, being trustful about the

381
00:24:10.559 --> 00:24:14.880
facts and not just inventing stuff you
might like like some other companies might do.

382
00:24:15.200 --> 00:24:18.240
Yeah, but that's the general direction. And I don't know how this

383
00:24:18.279 --> 00:24:21.640
affects the incentives of everybody here.
I can only guess. Yeah, that's

384
00:24:21.640 --> 00:24:26.559
a great question too. Yeah,
And it's it's like we are at that

385
00:24:26.680 --> 00:24:30.799
interesting inflection point right where we're we've
seen like now we've been you know,

386
00:24:30.960 --> 00:24:33.720
a year and a half with like
CHATGBT in consumer mindset, you know,

387
00:24:33.920 --> 00:24:37.319
as a as a consumer facing product, And it's like, it's to your

388
00:24:37.359 --> 00:24:41.640
point because we have not seen the
hallucination question go away at all, right,

389
00:24:41.839 --> 00:24:45.599
in terms of like it still tends
to make up answers we've seen,

390
00:24:45.720 --> 00:24:49.839
like you know, there are products
trying to rectify some of that like perplexity

391
00:24:49.920 --> 00:24:52.319
right to like add in like their
sources as part of the search engine.

392
00:24:52.319 --> 00:24:56.880
But you make a good point too
in terms of like as a generative product,

393
00:24:56.640 --> 00:25:00.640
you know, how do we both
think about what content is created news

394
00:25:00.640 --> 00:25:03.880
already? Then? Also yet,
like when we get to a quality level

395
00:25:03.880 --> 00:25:04.960
of production, what might that look
like? So this is interesting. I

396
00:25:04.960 --> 00:25:08.160
mean, so in this case are
you talking about like, on the one

397
00:25:08.160 --> 00:25:11.759
hand, like there's the we can
talk about like the capacity to be able

398
00:25:11.839 --> 00:25:12.880
to chat with the news, right, is kind of the news concierge.

399
00:25:12.880 --> 00:25:15.240
But then there's the other side of
that, there's a content creation side to

400
00:25:15.359 --> 00:25:18.440
that that journalists are like, I
mean, could I guess questions, could

401
00:25:18.440 --> 00:25:23.160
anybody generate quote unquote the news from
other sources? Like when you think about

402
00:25:23.200 --> 00:25:26.799
that, the idea of making something
or talking with it. In some sense,

403
00:25:26.839 --> 00:25:30.480
I think journalists are serving two different
functions, right, They're trying to

404
00:25:30.519 --> 00:25:36.000
find things worth writing about and trying
to figure out which things are true and

405
00:25:36.079 --> 00:25:40.079
not true and which things should be
packaged together in a single story or not

406
00:25:40.599 --> 00:25:44.279
right, And then they are writing, And we are selecting as our journalists

407
00:25:44.319 --> 00:25:48.640
people who are in the intersection of
the Van diagram of these two skills that

408
00:25:48.799 --> 00:25:52.480
are completely different. Right, one
is basically a PI and the other one

409
00:25:52.720 --> 00:25:56.720
is a fiction writer. But we
are demanding that their journalists be both.

410
00:25:56.240 --> 00:26:00.839
Interuestion, should we or can we
make this job much easier by just hiring

411
00:26:00.880 --> 00:26:04.559
pis for the job of a journalist, right and saying, yeah, but

412
00:26:06.200 --> 00:26:08.599
your writing can be completely incoherent,
it can be dyslexic, it doesn't matter.

413
00:26:08.640 --> 00:26:11.359
As long as you can find the
facts, you're going to be a

414
00:26:11.359 --> 00:26:15.480
great journalist. And then there's this
magical black box which, when you feed

415
00:26:15.480 --> 00:26:19.160
the facts into it, is going
to write it really well and not hallucinate.

416
00:26:19.440 --> 00:26:23.279
Now, hallucination is an interesting topic
because it happens on two levels.

417
00:26:23.359 --> 00:26:29.400
Right. There's the level of the
larger language model itself, which basically,

418
00:26:29.400 --> 00:26:34.000
we train the system that looks at
sentences with a missing word and guesses the

419
00:26:34.039 --> 00:26:38.119
missing word. Right, That's what
we did, right, And then once

420
00:26:38.119 --> 00:26:41.519
you ask it to guess one thousand
words in a row and produce a thousand

421
00:26:41.519 --> 00:26:47.119
word article for you. The problem
is that the error compounds, and basically

422
00:26:47.160 --> 00:26:51.920
the level of error it makes in
the one thousands word is much larger because

423
00:26:52.400 --> 00:26:55.680
it's just assuming that all the previous
nine hundred ninety nine are correct, but

424
00:26:56.039 --> 00:26:59.599
each one of them had some sort
of error chance, right, So it's

425
00:26:59.599 --> 00:27:03.240
built in them allel itself. Now. The other problem is we then train

426
00:27:03.359 --> 00:27:06.079
this system on the entire Internet,
and most of the Internet is junk.

427
00:27:06.119 --> 00:27:08.319
We kind of start also fair,
right, So it learned a bunch of

428
00:27:08.319 --> 00:27:11.839
things that are just not true.
And I've caught Chadgipt giving me answers that

429
00:27:11.880 --> 00:27:17.559
are basically a part of somebody is
April's fool's joke right, which, Okay,

430
00:27:17.759 --> 00:27:19.839
one person said it, zero person
said the opposite. Therefore it's true

431
00:27:21.519 --> 00:27:25.440
that's how LOLM has learned. But
obviously that's not what you want from your

432
00:27:25.440 --> 00:27:30.000
news articles. So all of these
problems exist. But then the question is

433
00:27:30.079 --> 00:27:33.880
what if we start stop thinking about
it as generation, and we just started

434
00:27:33.920 --> 00:27:38.000
thinking about it as synthesis. Right, here's the data, now write it

435
00:27:38.079 --> 00:27:42.799
up in the best way without adding
anything of your own. It seems like

436
00:27:42.839 --> 00:27:45.480
we already have the tech to do
this. In fact, we already have

437
00:27:45.519 --> 00:27:52.079
the tech to do this. It's
just not launched yet, and so maybe

438
00:27:52.079 --> 00:27:56.200
it's still not as good as the
best journalists at writing, but give it

439
00:27:56.240 --> 00:27:57.359
a year. The scary but also
helpful part, right, is that these

440
00:27:57.400 --> 00:28:02.400
things learn fast, right, The
technology adopts very quickly to that point.

441
00:28:02.599 --> 00:28:07.240
That's interesting too, like just being
so, is it kind of like getting

442
00:28:07.480 --> 00:28:14.000
a well put together like you know, retrieval augmented generation search tool like that

443
00:28:14.119 --> 00:28:15.480
just says like, just stick with
this data only, do not let go.

444
00:28:15.839 --> 00:28:19.240
That's what concierge is. Yeah,
there you go. It is a

445
00:28:19.319 --> 00:28:22.839
RAG. There's a few more steps
in there beyond the RAG because one of

446
00:28:22.880 --> 00:28:29.000
the problems with just a basic rag
for the news is that it wouldn't know

447
00:28:29.039 --> 00:28:32.480
what's important. Yeah, right,
So if you just ask it to summarize

448
00:28:32.480 --> 00:28:36.720
the news and AI research over the
past two months, it doesn't know which

449
00:28:36.799 --> 00:28:40.079
of the news items or the topics
are important. So you need a few

450
00:28:40.079 --> 00:28:44.960
additional steps to figure out this was
actually more important than all the stuff that

451
00:28:45.000 --> 00:28:48.000
came after it. So don't sort
by recency, sort by something else,

452
00:28:48.359 --> 00:28:52.559
right, But other than that,
Yes, that is the basics, right

453
00:28:52.720 --> 00:28:56.960
of that last step, the personalization. Now the first step I'm talking about

454
00:28:56.160 --> 00:29:00.440
is that's a bigger step, right, Yeah. Should AI be createyating most

455
00:29:00.480 --> 00:29:03.559
of the content to begin with from
facts that were curated by humans in some

456
00:29:03.599 --> 00:29:08.039
way and we're working on something new
that hasn't been released yet. There cool?

457
00:29:08.480 --> 00:29:11.640
I don't know if, yeah,
but in a broad sense then,

458
00:29:11.680 --> 00:29:15.240
but it sounds like that is a
direction that that you feel shows promise rights

459
00:29:15.279 --> 00:29:19.640
as a thing to think about as
it's inevitable. Also it could be inevitable.

460
00:29:19.680 --> 00:29:25.839
Is it just seems like one of
those things that maybe the industry is

461
00:29:25.880 --> 00:29:30.640
still undnial about it, right,
but it is completely inevitable. I just

462
00:29:30.720 --> 00:29:33.359
don't see another way. It's just
the question is who does it? Does

463
00:29:33.359 --> 00:29:38.559
it asks or somebody else? And
whoever does it actually? Like are they

464
00:29:38.599 --> 00:29:41.400
going to do what I said,
which is make it actually write things that

465
00:29:41.440 --> 00:29:47.160
are true? Just have AI do
the writing, or will it have AI

466
00:29:47.720 --> 00:29:52.079
write whatever you want to hear,
which is probably going to result in more

467
00:29:52.079 --> 00:29:56.920
clicks but might not be very good
for the world. Yeah, it's an

468
00:29:56.920 --> 00:29:59.279
interesting question too in terms of like
if I pick it abat, if I

469
00:29:59.440 --> 00:30:00.799
if I'm a reporter, I mean
organ as anthropologist, right, I mean

470
00:30:00.960 --> 00:30:04.880
I do interviews with folks and do
you know, surveys and if I'm collecting

471
00:30:04.920 --> 00:30:08.079
a bunch of data observational out in
the field and blah blah blah, like

472
00:30:08.400 --> 00:30:12.240
come back and I dumped my notebook
into into a tool and say help me

473
00:30:12.279 --> 00:30:15.119
summarize and like bring out you know, the top four themes or whatever,

474
00:30:15.279 --> 00:30:18.400
like you know, say help me
write a story about something that we saw

475
00:30:18.400 --> 00:30:22.200
in the data. Like there's also
the interesting question too on the on the

476
00:30:22.319 --> 00:30:26.960
editorial side, in terms of like
fact checking and making sure that it's doing

477
00:30:26.440 --> 00:30:29.920
it's not to your point, making
anything up. And then on top of

478
00:30:29.920 --> 00:30:33.440
that, like is it telling a
story that is like helpful to humans?

479
00:30:33.519 --> 00:30:37.799
Right? To think about that.
That's an interesting question. Like I'm curious

480
00:30:37.799 --> 00:30:40.359
with on that too, because it's
like there's a couple levels where human in

481
00:30:40.359 --> 00:30:41.319
the loop would would be necessary.
I mean, I can see how this

482
00:30:41.359 --> 00:30:45.920
can get mitigator automated at some time
over time, but like, how do

483
00:30:45.920 --> 00:30:51.480
we think about that too in terms
of like who who might or when?

484
00:30:51.559 --> 00:30:55.400
In what scenarios does do something?
Check what's get written before gets published?

485
00:30:55.480 --> 00:30:59.640
You know, well, I think
in general, if we just use another

486
00:30:59.680 --> 00:31:03.400
industry as kind of an analogy of
how this might go. Right, think

487
00:31:03.400 --> 00:31:07.839
about the fact that initially computer was
a profession. A human was computing.

488
00:31:07.039 --> 00:31:11.920
It was using some sort of a
calculator as an aid. Right. Well,

489
00:31:11.960 --> 00:31:15.839
before that there was no calculator.
Then the calculation part was automated with

490
00:31:15.880 --> 00:31:19.079
a calculator. The computer was still
doing the rest. Right. Then we

491
00:31:19.119 --> 00:31:23.480
started creating digital computers that replaced those
humans, right, but they still had

492
00:31:23.519 --> 00:31:27.319
programmers who would program everything, single
thing, one command at a time.

493
00:31:27.799 --> 00:31:32.960
Right, and fast forward to today. You might be clicking some button in

494
00:31:33.000 --> 00:31:37.240
a big piece of software. Somewhere
underneath there are multiple abstraction layers. You

495
00:31:37.319 --> 00:31:41.160
are not expected to know what commands
run on what core on your computer,

496
00:31:41.599 --> 00:31:44.519
right, So generally speaking, the
human is always in the loop. The

497
00:31:44.599 --> 00:31:48.920
question is how many layers underneath are
obstructed the way, And so we're just

498
00:31:48.279 --> 00:31:52.799
getting to the point where the writing
level is going to be abstracted. The

499
00:31:52.920 --> 00:31:56.119
editorial level is probably still not going
to be for a while, but then

500
00:31:56.200 --> 00:32:00.920
eventually that might go and we'll see
how far this gets at some point.

501
00:32:01.839 --> 00:32:07.039
I mean, not all technologies scale
exponentially all the time, right, Some

502
00:32:07.119 --> 00:32:09.759
things reach a plateau and get stuck
there. I mean, we haven't had

503
00:32:09.759 --> 00:32:16.599
a faster airplane since nineteen sixty seven, so you can't expect all the condor

504
00:32:19.480 --> 00:32:22.920
I'm talking about commercial air flights,
like, I'm not talking about military.

505
00:32:22.960 --> 00:32:25.799
Yes, there's some advances there,
but in general, yeah, not all

506
00:32:25.799 --> 00:32:30.480
technologies progress all the time. At
some point, all the smartest humans just

507
00:32:30.480 --> 00:32:32.880
say, Okay, progress here is
really difficult, let's switch to this field,

508
00:32:32.960 --> 00:32:36.799
right, and they start working on
something else. So I don't know

509
00:32:36.920 --> 00:32:43.440
at which point this generative AI trend
might just reach its apex, and unless

510
00:32:43.440 --> 00:32:46.599
there's a big kind of change coming
afterwards, we're just going to be stuck

511
00:32:46.640 --> 00:32:51.519
there for a while, But then
breakthrough might happen. Again, I mentioned

512
00:32:51.599 --> 00:32:52.680
just before. Sorry if I'm going
on tangents, right, that's good.

513
00:32:52.839 --> 00:32:58.880
Nineteen ninety seven the Blue Beats cast
part of Right, the next big progress

514
00:32:58.880 --> 00:33:01.759
in chess is two thousand thirteen.
If you look at the strength of chess

515
00:33:01.759 --> 00:33:07.119
programs between nineteen ninety seven and twenty
thirteen, it's basically flat, even though

516
00:33:07.160 --> 00:33:10.599
it was progressing the entire time until
nineteen ninety seven. Right, that's for

517
00:33:10.680 --> 00:33:15.799
sixteen years, and then from twenty
thirteen onwards. Humans are nowhere in the

518
00:33:15.839 --> 00:33:19.880
picture. Right, They're not even
necessary in the loop playing in tandem with

519
00:33:19.960 --> 00:33:22.359
the computer. They're just too slow
to make a difference. Yeah, which

520
00:33:22.400 --> 00:33:24.559
is that's why we just play in
the park, you know, after work,

521
00:33:24.640 --> 00:33:28.880
because otherwise it makes it sad that
a computer can beat us so quickly.

522
00:33:29.799 --> 00:33:32.200
Yeah. I mean the reality is
if you look at high level chess

523
00:33:32.200 --> 00:33:37.000
players as well, Right, they
themselves play in their own kind of park.

524
00:33:37.119 --> 00:33:38.799
Right. They're not even allowed to
wear a watch, because you know

525
00:33:39.680 --> 00:33:45.039
they might be cheating using any device. That's a good point. Yeah,

526
00:33:45.160 --> 00:33:47.599
I guess it's again there's there's always
ways to evolve. We how we do

527
00:33:47.640 --> 00:33:52.079
it, you know, I'm curious
you're thinking about this as as we're like,

528
00:33:52.799 --> 00:33:54.039
you know, society we're wrestling with
with a lot of these kind of

529
00:33:54.119 --> 00:33:58.799
changes in terms of, you know, how we use technology, what role

530
00:33:58.839 --> 00:34:01.079
technology plays in different parts of our
work lives, you know, and writing

531
00:34:01.160 --> 00:34:05.200
is an interesting area. And then
obviously around journalism too, it's like do

532
00:34:05.279 --> 00:34:09.079
you I'm curious, like how have
you spoken with or getting like what are

533
00:34:09.119 --> 00:34:14.239
reactions from journalists or like news corps
about these ideas? Right? You know,

534
00:34:14.360 --> 00:34:15.639
in terms of that if you kind
of chat chat with them about like

535
00:34:16.159 --> 00:34:21.280
developing this kind of tech that can
help in the in the creation process of

536
00:34:22.280 --> 00:34:28.639
not at scale, not yet so
and even if you go beyond the idea

537
00:34:28.679 --> 00:34:30.800
of using technology to do that.
Right now, there's a pretty big fight

538
00:34:30.880 --> 00:34:36.639
going on between most news corporations and
journalists themselves, right, because news corporations

539
00:34:36.639 --> 00:34:42.519
are trying to basically use humans to
reuse what other humans did. You're seeing

540
00:34:42.639 --> 00:34:45.400
companies like old like Gonnet kind of
buying up small newspapers, right, and

541
00:34:45.440 --> 00:34:50.199
then some of those newspapers they just
get rid of all the journalists and borrow

542
00:34:50.320 --> 00:34:55.000
content from other newspapers, right,
yep, Because again the incentives are to

543
00:34:55.039 --> 00:35:00.400
make that happen, but journalists typically
unionize, go and strike, fight against

544
00:35:00.440 --> 00:35:04.639
that, et cetera. Again,
understandably, so so when we're talking about

545
00:35:04.800 --> 00:35:08.880
and now it's not even going to
be that all the efficiency comes from reducing

546
00:35:08.920 --> 00:35:15.079
redundancy between humans. Now just this
entire group of humans becomes redundant because AI

547
00:35:15.159 --> 00:35:17.840
can do their work right, and
you only need the actual investigative guys.

548
00:35:19.199 --> 00:35:22.880
There's going to be a lot of
resistance towards that, right. I'm not

549
00:35:22.960 --> 00:35:25.920
even sure that we would get good
press if I just kept talking about that

550
00:35:27.000 --> 00:35:30.000
part over and over. You asked
about it, so I talked about it,

551
00:35:30.039 --> 00:35:32.719
but most of the time I talk
about reducing the junk, and I

552
00:35:32.719 --> 00:35:37.039
try to avoid the part about yeh, and I'm not sure how your profession

553
00:35:37.239 --> 00:35:40.840
is going to still be there,
of course. Yeah, if it does,

554
00:35:40.880 --> 00:35:45.079
then it's probably going to be about
ten percent of the size that is

555
00:35:45.119 --> 00:35:47.920
now in terms of number of people
engaged in it. Again, I love

556
00:35:49.000 --> 00:35:52.039
historic parallels, as you notice,
right, agriculture In the nineteen twenties and

557
00:35:52.320 --> 00:35:58.239
nineteen twenty it's thirty percent of people
in the US working in agriculture, and

558
00:35:58.320 --> 00:36:00.679
in nineteen thirty it's three percent.
Yeah, it totally changed, right,

559
00:36:01.119 --> 00:36:06.639
I think we're in a similar kind
of decade for anything that involves words as

560
00:36:06.679 --> 00:36:08.559
output. Yeah, I mean that's
interesting. Yeah, yeah, it's it's

561
00:36:08.599 --> 00:36:10.400
and it's a fair question. I
mean the part of it too is like

562
00:36:10.400 --> 00:36:13.599
we can't shy away from the challenge
of the question, right, And so

563
00:36:13.639 --> 00:36:15.320
that's like people that put their head
in the sand are going to get buried

564
00:36:15.400 --> 00:36:19.199
first, you know, and not
that and to be negative, but it's

565
00:36:19.239 --> 00:36:22.119
just like it's it's we have to
be willing to ask the questions, right

566
00:36:22.159 --> 00:36:25.199
of like how how might our work
change in our outputs? Yeah, it's

567
00:36:25.199 --> 00:36:30.280
like and we can imagine a future
that the creation of words, like you

568
00:36:30.280 --> 00:36:32.559
can do it for fun, but
like as a as a job or something

569
00:36:32.559 --> 00:36:37.159
that it's it's automated you know,
in many ways like that. And then

570
00:36:37.199 --> 00:36:39.440
it's one of those two like you
know, farming is automated. It did,

571
00:36:39.480 --> 00:36:42.639
it did change a lot of like
like cultural landscapes. But we also

572
00:36:42.639 --> 00:36:45.840
still need food, right, like
how that happens. So it's like in

573
00:36:45.880 --> 00:36:47.119
the same kind of way, like
I mean, even they have digital junk

574
00:36:47.119 --> 00:36:51.159
also, right, it's like we
need we need i mean more than everything

575
00:36:51.159 --> 00:36:52.679
good information, right, we need
information that we can trust and feel and

576
00:36:52.679 --> 00:36:57.639
feel safe with, and so like
whether written by a person or a computer.

577
00:36:57.760 --> 00:37:00.639
Like on the one hand, like
and that level that question, it's

578
00:37:00.679 --> 00:37:01.199
like, either way, how do
we get to the answer of like,

579
00:37:01.239 --> 00:37:07.639
what's what's trustworthy information that can help, you know, disseminate knowledge in a

580
00:37:07.679 --> 00:37:09.639
way that we can we can work
with. Where there's risk, there's also

581
00:37:09.679 --> 00:37:14.559
opportunity. Right, So if the
model that is writing it is actually taking

582
00:37:14.679 --> 00:37:17.800
facts from a source you can trust, right, And if the model of

583
00:37:19.880 --> 00:37:22.360
or the model itself and the data
sets it was trained on is at least

584
00:37:22.480 --> 00:37:27.239
partially open to you so you can
trust that it's doing it well, that

585
00:37:27.360 --> 00:37:30.239
is much more trustworthy than anything we
have in the ecosystem right now. So

586
00:37:30.280 --> 00:37:35.159
we could lament that the current ecosystem, you know, good people are losing

587
00:37:35.159 --> 00:37:39.360
their jobs, is very lamentable,
right, But that ecosystem is what's producing

588
00:37:39.360 --> 00:37:43.960
the junk right now, and so
if we can replace it with something that

589
00:37:44.320 --> 00:37:47.880
produces a higher signal to noise ratio, that's ultimately good for us. Sort

590
00:37:47.920 --> 00:37:52.840
of in the same way as the
mechanization of agriculture resulted more food being produced,

591
00:37:52.440 --> 00:37:59.480
right, we got more food,
but obviously an entire region got ninety

592
00:37:59.480 --> 00:38:02.519
percent on the pointment and the big
out migration from it and to other regions.

593
00:38:02.559 --> 00:38:06.800
Right, So I don't know how
that works for white collar workers,

594
00:38:06.800 --> 00:38:08.400
but I guess we'll see. Yeah, I mean knowledge worker is it's an

595
00:38:08.400 --> 00:38:12.159
interesting it's an interesting point of what
that's going to be, right, And

596
00:38:12.159 --> 00:38:15.039
and it is in you're right,
like, and we're talking about writing a

597
00:38:15.039 --> 00:38:17.000
lot here too, obviously because we're
approaching this from from news media, but

598
00:38:17.079 --> 00:38:21.599
like creative industries in general, right, because both visual and video creation like

599
00:38:21.760 --> 00:38:23.840
you know, are also in this
space as well. Like there's a lot

600
00:38:23.840 --> 00:38:29.079
of you know, generative image that's
a little more difficult, right, So,

601
00:38:29.079 --> 00:38:31.159
so it is progressing, but it's
a few years behind in a sense.

602
00:38:31.320 --> 00:38:35.800
Yeah, essentially if you use the
same method that I described as alarms

603
00:38:35.840 --> 00:38:38.400
our train, just a missing thing
that doesn't work in video. So video

604
00:38:38.440 --> 00:38:43.360
requires a little bit extra work to
try to generate something useful. But yeah,

605
00:38:43.440 --> 00:38:45.760
yeah, it's only a few years
behind probably. Yeah. I think

606
00:38:45.760 --> 00:38:50.559
you're right there too. It's in
like but even to that point too,

607
00:38:50.599 --> 00:38:53.920
like we're seeing the early implementations,
like in in mainstream software like Adobe's putting

608
00:38:53.920 --> 00:38:58.920
it in through Firefly and and other
you know, there's what's it called SOA

609
00:38:59.000 --> 00:39:01.199
or sona from open air eye,
you know. Yeah, Like I think

610
00:39:01.280 --> 00:39:05.280
there's some interesting, interesting things that
are that are at play there. So

611
00:39:05.360 --> 00:39:10.159
I'm curious too, Like as we
think about the h I guess we talked

612
00:39:10.159 --> 00:39:13.239
to I guess a lot of but
don't think about like the kind of tech

613
00:39:13.280 --> 00:39:15.960
challenges that we're facing in terms of
as we see these these shifts happen,

614
00:39:16.079 --> 00:39:22.119
like how do you approach trying to
you know, build something new or break

615
00:39:22.239 --> 00:39:24.280
you know, kind of new kind
of breakthrough technologies in the space. It's

616
00:39:24.280 --> 00:39:29.000
like super rapidly advancing. Uh.
And there's again a lot of like you

617
00:39:29.039 --> 00:39:31.880
know, social buzz because it's like
what it is that we're potentially making or

618
00:39:31.880 --> 00:39:36.000
how we're changing conversations around digital clutter, Like how do you think about the

619
00:39:36.639 --> 00:39:40.199
technical challenges that you're facing as you're
putting the other platform syndicating content and then

620
00:39:40.199 --> 00:39:44.840
potentially like putting the concierge in place
and like having people chat Like what's I

621
00:39:44.840 --> 00:39:46.519
guess, what's been the toughest thing
to kind of help or not to crack

622
00:39:46.679 --> 00:39:50.599
as you're trying to put the tech
stack together to make this thing work.

623
00:39:51.199 --> 00:39:55.480
Yeah, I think generally speaking,
both I and all the people that's gravitated

624
00:39:55.519 --> 00:40:01.000
towards me in this project are better
at building stuff than we are getting eyeballs

625
00:40:01.000 --> 00:40:05.760
to stuff. So generally speaking,
if you ask me what are our biggest

626
00:40:05.840 --> 00:40:07.440
challenges, I will list ten things, and none of them will be a

627
00:40:07.480 --> 00:40:14.280
tech challenge. Organ Yeah, the
tech challenges are the easiest thing for us,

628
00:40:14.320 --> 00:40:17.559
again just because of the kind of
people that the project attracted early on,

629
00:40:17.679 --> 00:40:22.400
right, we tend to be tech
is the biggest challenges. How do

630
00:40:22.440 --> 00:40:24.920
you build a community of people that
actually talk to each other. That's something

631
00:40:25.000 --> 00:40:29.920
where I'm incompetent and I can't even
stay in touch with my best friends,

632
00:40:30.079 --> 00:40:35.280
right, and so getting me to
build a community that's completely foreign to me.

633
00:40:35.679 --> 00:40:40.119
So it kind of comes out of
the personality of me and the rest

634
00:40:40.159 --> 00:40:45.000
of the team. But in general, yeah, that tech landscape is evolving

635
00:40:45.039 --> 00:40:49.960
a lot, and that means that
I always have to watch what happens,

636
00:40:49.960 --> 00:40:52.360
and everybody else has to watch what
happens, and we have to constantly try

637
00:40:52.400 --> 00:40:59.159
to adjust. And it's a little
difficult for some team members because you know,

638
00:40:59.199 --> 00:41:05.199
they were something for six months and
then that's something that was supposed to

639
00:41:05.239 --> 00:41:07.920
be on the main tab and the
app has been moved to the fourth tab

640
00:41:07.960 --> 00:41:10.440
because three more more important things came
out. Yeah, and they're upset,

641
00:41:10.639 --> 00:41:14.519
right, and I understand them.
Right. They were working on something that

642
00:41:14.599 --> 00:41:16.960
was supposed to be a part of
the flagship feature, the main flow that

643
00:41:17.000 --> 00:41:20.960
the all users go through, and
if it's in the fourth tab, then

644
00:41:21.559 --> 00:41:24.760
maybe only fifteen percent of users even
get there and know what's there. But

645
00:41:25.599 --> 00:41:30.519
there's three cooler things that happened just
because now the technology allows them. Yeah,

646
00:41:30.519 --> 00:41:32.559
it's a good point too, right. It seems like it's the especially

647
00:41:32.559 --> 00:41:36.199
like if you're working in a modular
development environment, you know, it's like

648
00:41:36.199 --> 00:41:37.960
you can't we can't always know as
I build this thing when it'll get down

649
00:41:38.000 --> 00:41:40.239
downstream and they get out and by
the time it gets out, like what

650
00:41:40.320 --> 00:41:45.719
new has popped up elsewhere? That
would change. That's interesting too, that

651
00:41:45.760 --> 00:41:50.519
I mean, like, how do
we think about like the as an online

652
00:41:50.519 --> 00:41:53.679
platform too, you know, like
we've got ways of being able to select

653
00:41:53.719 --> 00:41:57.840
topics and emotions even if if I
want, like what I'm looking for in

654
00:41:57.920 --> 00:42:00.320
stories, Like how do you think
about competing with like a CNN or BBC.

655
00:42:01.360 --> 00:42:05.000
Obviously those are individual publishers, right, and so part of it might

656
00:42:05.039 --> 00:42:07.000
be the syndication question, but like, how do you how do you kind

657
00:42:07.000 --> 00:42:10.719
of try to position other web as
a way of like being in competition but

658
00:42:10.800 --> 00:42:15.199
conversation with these others lublishers. Yeah, I'm not trying to compete with them,

659
00:42:15.320 --> 00:42:19.880
right. Essentially, I'm trying to
stay within the guidlands of fair use.

660
00:42:20.440 --> 00:42:23.000
I try to add something they don't
have, which has summaries and nutrition

661
00:42:23.119 --> 00:42:27.760
labels and better ways to select an
order content. Right, and then if

662
00:42:27.760 --> 00:42:30.679
somebody wants the original content that they
created, I have a link back to

663
00:42:30.719 --> 00:42:37.360
CNN or BBC. Yeah. So
I'm basically strictly in the aggregator space of

664
00:42:37.719 --> 00:42:39.800
let's say, the Google News and
Apple News and Flipboard of the world,

665
00:42:39.840 --> 00:42:44.880
not the CNNs of the world.
Right. But again this is up until

666
00:42:45.840 --> 00:42:51.360
March. Yeah, and now I'm
gradually getting into a completely different space,

667
00:42:51.400 --> 00:42:54.159
which we'll see how that works.
Right. Where Again, I still don't

668
00:42:54.199 --> 00:42:58.880
want to compete with CNN or BBC, right, especially not with BBC.

669
00:42:59.199 --> 00:43:06.400
They have endless funding that we will
never have right in government, but even

670
00:43:06.440 --> 00:43:10.760
within organizations of that caliber, still
we want to always partner up with them.

671
00:43:10.920 --> 00:43:15.199
We want to even when we have
a model that writes really well,

672
00:43:15.960 --> 00:43:19.199
we're going to go to Gannett and
Alden and McClatchy and all these guys and

673
00:43:19.239 --> 00:43:22.000
say, hey, here's this school
thing that can save you a lot of

674
00:43:22.039 --> 00:43:25.880
time, and it proved the quality
of what you guys are producing. It

675
00:43:27.039 --> 00:43:31.519
probably doesn't make sense to compete with
the guys who are the most trusted names

676
00:43:31.559 --> 00:43:36.920
in news. We should try to
improve them because that's good for everybody,

677
00:43:37.199 --> 00:43:39.880
or just help them and whatever way
we can. And it's worth noting that

678
00:43:39.920 --> 00:43:45.280
all these companies are losing money crisis
the industry. Yeah yeah, yeah,

679
00:43:45.320 --> 00:43:50.920
I mean Gannett is probably they're the
biggest, the biggest company in the print

680
00:43:50.960 --> 00:43:55.760
industry essentially. Right. Their revenue
went down from three point four billion dollars

681
00:43:55.760 --> 00:44:01.760
four years ago to two point six
last year, right, And they're cutting

682
00:44:02.559 --> 00:44:07.199
employees, cutting essentially their expenses every
single year. But the cuts are not

683
00:44:07.320 --> 00:44:10.559
going as fast as the revenue declines, right. And if you look at

684
00:44:10.760 --> 00:44:15.880
their market gap right now, it's
something like three hundred million on almost three

685
00:44:15.960 --> 00:44:19.559
billion of revenue. Right, So
basically the market is betting that they go

686
00:44:19.639 --> 00:44:22.000
chapter eleven. They say in every
quarterly call. They won't, right,

687
00:44:22.599 --> 00:44:27.840
But yeah, so they're in a
tough situation. But that's three hundred different

688
00:44:27.840 --> 00:44:31.239
newspapers, including USA Today, and
a lot of people read them and trust

689
00:44:31.280 --> 00:44:36.760
them. They don't want to go
in and compete with them. I want

690
00:44:36.800 --> 00:44:39.519
to go in and try to help
them. They need the help, yes,

691
00:44:39.719 --> 00:44:44.480
and the readers need good information that's
refreshing to hear then too, because

692
00:44:44.519 --> 00:44:46.920
yeah, it's not a like zero
sum game of information, because that's its

693
00:44:46.920 --> 00:44:51.840
own challenge, their own problem potentially
too. Right of being a singular source

694
00:44:51.880 --> 00:44:53.960
of truth, I think is part
of the issue that we're discussing around in

695
00:44:54.159 --> 00:44:59.199
a philosophical sense. Yeah, yeah, and that in itself I think is

696
00:44:59.239 --> 00:45:04.159
a part of Maybe it's the ethos
of the profession that generates that to some

697
00:45:04.320 --> 00:45:08.440
extent, right, But I haven't
seen that many journalists who would like openly

698
00:45:08.519 --> 00:45:13.440
say I'm the equivalent of the line
cock at McDonald's. Right. All they

699
00:45:13.480 --> 00:45:16.519
do is just do the same thing
that my neighbors do in the most standardized

700
00:45:16.559 --> 00:45:22.639
way possible. No, every single
journalist, thanks their Emichelin Starshift, right,

701
00:45:22.800 --> 00:45:25.480
they are creating something unique and if
you ask them, is your version

702
00:45:25.480 --> 00:45:29.840
of the story better than every other
version of the same story out there.

703
00:45:30.119 --> 00:45:32.280
They're obviously going to say yes.
Now, the problem is there's more than

704
00:45:32.280 --> 00:45:36.559
fifty versions of every single story out
there, even within the same set of

705
00:45:36.639 --> 00:45:40.000
fact published in the same hour,
and so they can't all be right.

706
00:45:40.599 --> 00:45:44.960
But it's a part of the ethos
of the profession, and I understand why

707
00:45:44.960 --> 00:45:49.239
it exists because then they take pride
in their work. They're actually trying to

708
00:45:49.239 --> 00:45:52.760
produce something really good. Right.
But when I look at this as an

709
00:45:52.880 --> 00:45:57.960
entire ecosystem, obviously that's a part
of the reason why the industry is losing

710
00:45:57.960 --> 00:46:02.079
money all the time, right,
because you can't have every single restaurant in

711
00:46:02.119 --> 00:46:07.039
the world operating as if at some
Michelin Star restaurant. They have to kind

712
00:46:07.079 --> 00:46:10.679
of pick their lane. And right
now, everybody's trying to be the best,

713
00:46:12.599 --> 00:46:15.440
and in part because that's that's to
our other part before of like following

714
00:46:15.480 --> 00:46:20.800
them, the incentive structure of like
you capture the most eyeballs slash advertising dollars

715
00:46:21.920 --> 00:46:23.480
and then you do I guess,
is there like I mean, or do

716
00:46:23.519 --> 00:46:30.599
you envision in terms of like the
future of the industry, like both either

717
00:46:30.639 --> 00:46:34.559
ways or models you've seen of different
kinds of incentives, like and like could

718
00:46:35.199 --> 00:46:37.239
services platforms like other web in terms
of whether it's producing content or like,

719
00:46:37.280 --> 00:46:40.119
how we can engage with it and
get, you know, a better sense

720
00:46:40.119 --> 00:46:44.440
of why I am consuming, presuming
in what it means like to shift some

721
00:46:44.480 --> 00:46:47.599
of those models, either away from
like the chokehold of advertising or something else

722
00:46:47.679 --> 00:46:52.079
like you know, do do do
we see anything else emerging as potential contenders

723
00:46:52.119 --> 00:46:57.000
for that? In theory, yes, but then practice trying to innovate and

724
00:46:57.079 --> 00:47:01.679
business models as a very dangerous proposition. And so do I want the industry

725
00:47:01.679 --> 00:47:07.559
to shift away from advertising? Yes? Do I want to be one of

726
00:47:07.559 --> 00:47:10.440
the companies that tries? Not really, because that usually ends badly, right,

727
00:47:10.519 --> 00:47:15.920
And so we're trying to innovate and
the tech side trying to produce better

728
00:47:15.159 --> 00:47:21.559
experiences for our readers. Right,
We're not trying to innovate in the business

729
00:47:21.599 --> 00:47:28.760
model side, but somebody should.
Yeah. Yeah, I just don't think

730
00:47:28.800 --> 00:47:32.199
that we want to be it.
And the other reality is that not all

731
00:47:32.239 --> 00:47:37.960
experiments in business models have been good
for the ecosystem at all as well.

732
00:47:37.480 --> 00:47:44.519
Right, you can argue that the
iTunes experiment basically kill the ability of most

733
00:47:44.599 --> 00:47:49.159
musicians to make money, and so
I think generally Near did the study on

734
00:47:49.199 --> 00:47:52.039
that, the father of the art
right, but he's also a musician,

735
00:47:52.039 --> 00:47:55.679
and so he did a study on
this and basically showing that the curve of

736
00:47:55.760 --> 00:48:00.119
how much money musicians make has become
much much steeper. Interesting we've been able

737
00:48:00.159 --> 00:48:06.000
to buy songs for one cent each
separately instead of buying entire albums, and

738
00:48:06.039 --> 00:48:08.360
the result is that the number of
musicians that can make a living for music

739
00:48:08.559 --> 00:48:13.719
is much much smaller than it was
even twenty years ago. And so that's

740
00:48:13.719 --> 00:48:16.360
a business model innovation. It looked
great when it happened because it was a

741
00:48:16.400 --> 00:48:22.440
solution to the problem of piracy in
some sense, but in reality it actually

742
00:48:22.480 --> 00:48:28.880
made everybody's life in that industry worse
probably, And so again I believe that

743
00:48:28.920 --> 00:48:31.159
we need to try to innovate this. And it would be great if everybody

744
00:48:31.159 --> 00:48:36.639
paid for the content they want to
pay for, as opposed to being monetized

745
00:48:36.639 --> 00:48:42.440
in this backhanded way by somebody trying
to extract their attention. But I'm not

746
00:48:42.519 --> 00:48:45.519
equipped to innovate in this field.
Neither is anybody else in the company,

747
00:48:45.519 --> 00:48:50.480
I think, so they'll leave that
to others. We'll just focus on good

748
00:48:50.559 --> 00:48:52.800
content. Yeah. Cool, No, that makes sense. Yeah too.

749
00:48:52.880 --> 00:48:53.719
And I appreciate the honest answer too, because it's like, you know,

750
00:48:53.880 --> 00:48:55.800
you talk to people like, oh
yeah, we could do something else.

751
00:48:55.800 --> 00:48:59.079
It's like okay, yeah, I
mean, like what could it be?

752
00:48:59.119 --> 00:49:02.440
Is the question? You know,
that's not but that's really helpful to to

753
00:49:02.519 --> 00:49:05.880
kind of think about too, because
it's like and part of it as folks

754
00:49:05.880 --> 00:49:08.480
think about, like as we're building
what the future is from content creation,

755
00:49:08.599 --> 00:49:12.960
production, consumption, you know,
we have to think about to your point,

756
00:49:13.000 --> 00:49:15.039
like what can you be good at
as a business, as a as

757
00:49:15.039 --> 00:49:17.320
a as a researcher, as a
writer, as a as a creator,

758
00:49:17.360 --> 00:49:22.000
as as a developer, you know, and then create in that area,

759
00:49:22.119 --> 00:49:23.800
right, and like and not trying
to necessarily change everything at the same time,

760
00:49:23.840 --> 00:49:25.960
because it's like it feels nice to
be able to do that, but

761
00:49:27.000 --> 00:49:29.880
like also that that's a hard call
to be able to do in the first

762
00:49:29.880 --> 00:49:32.880
place. I think I would say, just as a kind of wrap up

763
00:49:32.920 --> 00:49:37.400
question, if folks are thinking about
the wanting to be more mindful of how

764
00:49:37.400 --> 00:49:42.000
they consume content, you know,
like whether it's like practices that you have

765
00:49:42.039 --> 00:49:44.719
in other way or other things that
you've seen in terms of helping us have

766
00:49:44.800 --> 00:49:49.360
a better diet, media diet and
not consume so much junk, Like what

767
00:49:49.400 --> 00:49:52.519
are some strategies that you you share
like or you thought about you're built into

768
00:49:52.559 --> 00:49:55.480
the platform for like how to be
more effective at like being smarter about my

769
00:49:55.519 --> 00:50:00.360
consumption of media, or you know, making sure that I'm knowing what it

770
00:50:00.400 --> 00:50:05.039
is that I'm consuming. And why
let me start at a higher level,

771
00:50:05.119 --> 00:50:07.880
because the platform only covers a part
of what you could do. But if

772
00:50:07.920 --> 00:50:12.360
you go to a higher level,
if you want to consume good stuff most

773
00:50:12.360 --> 00:50:15.679
of the time, then you should
actually plan what you're going to consume.

774
00:50:15.639 --> 00:50:19.599
I think most people don't actually if
you ask them, even how much time

775
00:50:19.599 --> 00:50:22.360
you're spending right now on Instagram versus
reading books, most people can't answer that

776
00:50:22.599 --> 00:50:27.000
question. But I think you should
have a plan of how many hours per

777
00:50:27.000 --> 00:50:29.960
week are you going to spend on
Instagram? How many hours per week are

778
00:50:29.960 --> 00:50:32.079
you going to spend reading books?
The minute you come up with that plan,

779
00:50:32.239 --> 00:50:35.639
even if you can't stick to it
most of the time, it's still

780
00:50:35.679 --> 00:50:37.199
going to be better than if you
had never had the plan. We see

781
00:50:37.239 --> 00:50:42.119
the same thing with regular dieting as
well. Right the moment people start taking

782
00:50:42.119 --> 00:50:45.559
photos of their food, they start
eating better. They don't even need to

783
00:50:45.559 --> 00:50:49.360
start planning their diet. Just taking
photos is typically enough, but if they

784
00:50:49.360 --> 00:50:52.599
also plan their diet, it becomes
even better. So that's my first piece

785
00:50:52.639 --> 00:50:55.880
of advice to everybody. Just try
to plan this thing. And then once

786
00:50:55.960 --> 00:51:00.800
you notice how much time you're spending
on recent versus that are trying to extract

787
00:51:00.800 --> 00:51:06.199
your attention, chances are you're going
to want to change that. Now,

788
00:51:06.400 --> 00:51:10.519
how much up to you? Right? Some people like their Instagram because they're

789
00:51:10.519 --> 00:51:15.079
already hooked. It's okay, But
the question is how much right? Even

790
00:51:15.639 --> 00:51:19.800
drugs are you know, some of
them are okay some of the time they're

791
00:51:19.880 --> 00:51:22.960
not okay, multiple times per day
for very long time, being strung out

792
00:51:22.039 --> 00:51:27.039
right, Yes, sorry, I'm
not at what they're advocating for drugs anyone.

793
00:51:27.079 --> 00:51:30.480
But let's say alcohol, Okay,
one drink per night seems to actually

794
00:51:30.519 --> 00:51:34.719
extend your life even though it's clearly
unhealthy. Why because I guess moderations okay,

795
00:51:35.000 --> 00:51:37.079
right, but seven drinks three times
a day probably not even that's great?

796
00:51:37.119 --> 00:51:43.280
Yeah, So basically, try to
plan what you think is the right

797
00:51:43.360 --> 00:51:47.840
level of consumption for you for every
particular drug, and information that produces dopamine

798
00:51:47.960 --> 00:51:52.480
is a drug, just like chemicals
that produce dopamine. That's a great points

799
00:51:52.639 --> 00:51:58.079
and something that I'm curious to hear
what listener's viewers think about there. It's

800
00:51:58.079 --> 00:52:00.840
like, like, have you ever
planned your media consumption? Because I have

801
00:52:00.920 --> 00:52:02.679
not been, Like, let me, let me make sure I only consume

802
00:52:02.719 --> 00:52:06.639
fifteen minutes of media news today or
something whatever it is, like, but

803
00:52:06.639 --> 00:52:09.039
that's that's a really great point actually
to be aware of, like what it

804
00:52:09.079 --> 00:52:14.800
is that you're choosing to expose yourself
to. And I think the most ironic

805
00:52:14.840 --> 00:52:17.280
thing that happened to me when when
I got small kids, right, is

806
00:52:17.440 --> 00:52:22.440
I start planning like, Okay,
she's only allowed to watch cartoons for fifteen

807
00:52:22.440 --> 00:52:25.360
minutes in the morning and fifteen minutes
in the evening. And then I noticed,

808
00:52:25.519 --> 00:52:29.239
okay, but my wife is looking
at Instagram for the past hour.

809
00:52:29.760 --> 00:52:31.639
You're like, wait, yeah,
I'm not supposed to say anything. Yeah,

810
00:52:32.079 --> 00:52:37.239
yeah, exactly. And I'm thinking
with different standards. Somebody is planning

811
00:52:37.280 --> 00:52:38.800
it for the kids, but for
us adults, nobody is planning it.

812
00:52:38.840 --> 00:52:43.480
We're kind of waning it, right, And so how about we start planning

813
00:52:43.599 --> 00:52:45.280
for ourselves too. Yeah, that's
a great idea, right, we need

814
00:52:45.320 --> 00:52:49.639
a media planner to help us out
here. You do awesome? Cool,

815
00:52:49.679 --> 00:52:51.800
Well, Alex Like this has been
great to chat with you. I appreciate

816
00:52:51.800 --> 00:52:54.119
you taking the time out and excited
you know about the work, about the

817
00:52:54.119 --> 00:52:59.760
work that you're doing and raising I
think a ton of really important questions for

818
00:52:59.840 --> 00:53:01.880
media consumption and how we can be
smarter about that. How do we like

819
00:53:02.039 --> 00:53:07.159
tack all the challenges facing journalism and
media creation too? You know, I

820
00:53:07.159 --> 00:53:09.639
think it's a it's an interesting challenge
space that you've doven into and so I'm

821
00:53:09.639 --> 00:53:14.039
excited for folks to check out the
site that they haven't already and to kind

822
00:53:14.039 --> 00:53:16.280
of share some thoughts on this idea
of how can we be smarter about media

823
00:53:16.320 --> 00:53:21.159
consumption? Sounds good. It's been
great, It's awesome, Thank you much.

824
00:53:21.239 --> 00:53:21.880
And it was easy.