1
00:00:00,040 --> 00:00:00,640
Speaker 1: I'm my friend.

2
00:00:02,759 --> 00:00:06,040
Speaker 2: I thought X was coming out. I had some extra

3
00:00:06,120 --> 00:00:06,919
questions for him.

4
00:00:07,080 --> 00:00:10,759
Speaker 3: I think, yeah, he had changed his mind.

5
00:00:13,359 --> 00:00:14,359
Speaker 1: Well, thank you, Elon.

6
00:00:14,800 --> 00:00:17,839
Speaker 2: It is such a treat and a special opportunity.

7
00:00:18,280 --> 00:00:19,640
Speaker 1: Welcome to Possible.

8
00:00:20,480 --> 00:00:23,320
Speaker 2: And while Greg was right, okay, let's have a round

9
00:00:23,320 --> 00:00:32,920
of applause. Well, Greg was right, you really need no introduction.

10
00:00:33,159 --> 00:00:37,520
Speaker 1: But in many ways, today is your introduction to.

11
00:00:37,520 --> 00:00:41,200
Speaker 2: The advertising community, right, I think that's a round of applause.

12
00:00:43,920 --> 00:00:48,920
It's widely known that in the morning you run SpaceX,

13
00:00:49,200 --> 00:00:52,960
in the afternoon you head to Tesla, and in the

14
00:00:53,039 --> 00:00:57,240
evening it's Twitter time. And many of you in this

15
00:00:57,359 --> 00:01:00,159
room know me, and you know I pride myself on

16
00:01:00,200 --> 00:01:04,560
my work ethic. But buddy, I met my match. So

17
00:01:05,120 --> 00:01:07,319
we're here today to actually talk.

18
00:01:07,280 --> 00:01:08,159
Speaker 1: About your night job.

19
00:01:09,159 --> 00:01:09,959
Speaker 4: Great sounds good.

20
00:01:11,719 --> 00:01:16,000
Speaker 2: But with Twitter you've switched roles a little bit. You've

21
00:01:16,200 --> 00:01:20,959
you know, your other companies, you move from inventor now

22
00:01:21,000 --> 00:01:25,680
to reinventor. And when you're inventing something, it's all new.

23
00:01:26,079 --> 00:01:30,560
It's a surprise. We don't know what to expect. To reinventor.

24
00:01:32,200 --> 00:01:37,319
You challenge legacy, you challenge habits. With Twitter, many of

25
00:01:37,400 --> 00:01:39,640
us in this room we might even go to bed

26
00:01:39,640 --> 00:01:42,439
with it in our pocket on the night table. You

27
00:01:42,560 --> 00:01:46,799
challenge rituals, and every marketing executive in this room knows

28
00:01:47,519 --> 00:01:51,640
the difficulty of a new formula and the challenge of

29
00:01:51,719 --> 00:01:54,120
the delicate balance of a rebrand.

30
00:01:54,640 --> 00:01:56,319
Speaker 1: And now that's what we're going to talk about today.

31
00:01:56,519 --> 00:01:58,400
Speaker 3: Yeah, I mean, I'm trying to make Twitter fun and

32
00:01:58,599 --> 00:02:02,959
interesting and informative of you know, the optimization for Twitter

33
00:02:03,159 --> 00:02:10,319
is maximize the unregretted user time. So so it's not

34
00:02:10,360 --> 00:02:13,759
like total number of users or anything, it's just total

35
00:02:13,840 --> 00:02:16,840
user minutes unregretted And we actually hit an all time.

36
00:02:16,680 --> 00:02:23,199
Speaker 4: High just yesterday, so gone. Well, it's entertaining.

37
00:02:24,400 --> 00:02:27,400
Speaker 2: So entertaining is one way some people in the room

38
00:02:27,479 --> 00:02:29,479
describe Twitter today.

39
00:02:30,840 --> 00:02:32,360
Speaker 4: Train wrecks arguably are entertaining.

40
00:02:33,840 --> 00:02:38,599
Speaker 1: Happened sometime, they have some time. If you're dedicated, like

41
00:02:38,639 --> 00:02:39,400
a train wreck.

42
00:02:40,800 --> 00:02:42,840
Speaker 2: You got to be dedicated to fixing them. But I

43
00:02:42,840 --> 00:02:46,319
think it's important to start with your vision of Twitter

44
00:02:46,360 --> 00:02:50,280
two point zero, you yourself wrote, I would like to

45
00:02:50,520 --> 00:02:56,039
die knowing that humanity has a bright future. It's actually

46
00:02:56,599 --> 00:02:59,120
been quite apparent in all your other businesses, from the

47
00:02:59,159 --> 00:03:02,680
early days of paper how to Tesla, to SpaceX to

48
00:03:02,800 --> 00:03:06,719
maybe you contemplating a new AI company, but how does

49
00:03:06,800 --> 00:03:09,840
the better humanity for the future fit into your Twitter

50
00:03:09,879 --> 00:03:10,800
two point zero vision?

51
00:03:12,639 --> 00:03:17,960
Speaker 3: Yeah, so, I mean, obviously with the goal with with

52
00:03:17,960 --> 00:03:22,919
with Tesla is to advance sustainable energy. With SpaceX, we've

53
00:03:22,919 --> 00:03:25,759
got Starlink, which is providing Internet to the least served

54
00:03:25,759 --> 00:03:30,960
in the world and also hopefully getting humanity to Mars

55
00:03:31,039 --> 00:03:34,560
and back to the Moon. So we have an exciting future.

56
00:03:34,960 --> 00:03:36,159
I mean it's important to bear in mind. Like so

57
00:03:36,240 --> 00:03:38,039
much people say, like why are we send any money

58
00:03:38,080 --> 00:03:40,159
on space? Like don't we have enough problems on Earth?

59
00:03:40,479 --> 00:03:42,560
But you know, the thing is that, like everyone needs

60
00:03:42,560 --> 00:03:44,360
a reason to be inspired, People need a reason to

61
00:03:44,360 --> 00:03:45,120
wake up in the morning.

62
00:03:45,199 --> 00:03:46,800
Speaker 4: It can't just be about solving problems.

63
00:03:47,080 --> 00:03:49,719
Speaker 3: They have to have to be things that really get

64
00:03:49,719 --> 00:03:51,120
you in the heart. I think I just killed the

65
00:03:51,159 --> 00:03:59,000
mic okay, so uh and going to the moon last

66
00:03:59,000 --> 00:04:03,439
century inspire the whole world, and hopefully going to Mars

67
00:04:03,439 --> 00:04:07,840
can do the same thing. So then yeah, it's a

68
00:04:07,840 --> 00:04:09,680
great question why I take some time away from that

69
00:04:10,319 --> 00:04:14,479
for Twitter and Mike. You know what I think is

70
00:04:15,199 --> 00:04:20,079
essentially in order for civilization to advance, we've got to

71
00:04:20,120 --> 00:04:25,000
have freedom of speech. We've got to have a digital Yeah,

72
00:04:25,560 --> 00:04:32,199
so thank you. It's a bigger deal than you'd think.

73
00:04:33,480 --> 00:04:35,920
And it's the kind of thing that you don't know

74
00:04:35,920 --> 00:04:39,040
what you're missing until you don't have it. And in

75
00:04:39,040 --> 00:04:44,120
a lot of places people don't have it. And so

76
00:04:44,160 --> 00:04:46,759
it's important to bear in mind that the nature of

77
00:04:46,759 --> 00:04:53,240
free speech is the acid test for it is are

78
00:04:53,240 --> 00:04:56,560
people you don't like allowed to say things that you

79
00:04:57,000 --> 00:04:57,480
don't like?

80
00:04:58,319 --> 00:05:00,639
Speaker 4: Otherwise it's not free. It can't be.

81
00:05:01,120 --> 00:05:02,680
Speaker 3: It can't be just the things that it can't be

82
00:05:02,839 --> 00:05:06,240
just things that you like, because eventually somebody is not

83
00:05:06,240 --> 00:05:08,720
going to like what you say and they're gonna shut

84
00:05:08,759 --> 00:05:11,839
you up. And that's the essence of free speech. And

85
00:05:11,839 --> 00:05:14,680
that's why it's the first amendment in this country. And

86
00:05:15,319 --> 00:05:18,279
if we lose that, I think we lose the bedrock

87
00:05:18,319 --> 00:05:20,720
of democracy.

88
00:05:20,800 --> 00:05:23,800
Speaker 2: So the bedrock of democracy, I would imagine, is important

89
00:05:23,839 --> 00:05:26,759
to everyone in this room. And you talk about the

90
00:05:26,800 --> 00:05:32,480
importance of freedom of speech. Yesterday you posted a new

91
00:05:32,519 --> 00:05:35,000
policy that was titled.

92
00:05:35,279 --> 00:05:42,040
Speaker 1: Freedom of speech, not freedom of reach. I got my attention, Yes,

93
00:05:42,160 --> 00:05:44,399
tell us about your new policy.

94
00:05:44,600 --> 00:05:50,040
Speaker 4: Right, So, freedom of speech not reach is important in that.

95
00:05:51,720 --> 00:05:56,399
Speaker 3: You know, like if somebody has something hateful to say,

96
00:05:56,480 --> 00:05:59,879
it doesn't mean you should give them a megaphone. They

97
00:06:00,360 --> 00:06:02,800
still be able to say it, but it needs to

98
00:06:02,839 --> 00:06:08,720
be not then pushed on people. So if you know,

99
00:06:09,040 --> 00:06:10,600
you can go like in Times Square in New York

100
00:06:10,600 --> 00:06:13,000
and you can hear people say all kinds of crazy things,

101
00:06:14,000 --> 00:06:15,959
that's fine, but we don't put we don't broadcast that

102
00:06:16,000 --> 00:06:21,560
to earth. So if somebody wants to say something that's

103
00:06:22,079 --> 00:06:27,560
technically legal but that is you know, by most definitions, hateful,

104
00:06:27,879 --> 00:06:29,600
they're not going to get We're not going to promote

105
00:06:29,600 --> 00:06:30,680
that to people.

106
00:06:30,680 --> 00:06:33,079
Speaker 4: We're not going to recommend hateful content to people.

107
00:06:34,160 --> 00:06:36,759
Speaker 3: We will put that behind a warning label saying this

108
00:06:37,040 --> 00:06:38,800
the speech is probably something.

109
00:06:38,639 --> 00:06:39,360
Speaker 4: You you don't want.

110
00:06:40,120 --> 00:06:41,399
Speaker 3: So and now this is something we have to be

111
00:06:41,480 --> 00:06:43,639
very careful that we're all out and that it does

112
00:06:43,680 --> 00:06:46,959
not become what is intended to be good does not

113
00:06:47,000 --> 00:06:51,199
become bad. So so we want to be delicate with the

114
00:06:51,199 --> 00:06:54,399
freedom of reach, not speech, A freedom of speech not reach.

115
00:06:56,240 --> 00:07:00,959
So we're taking it easy, but it's just uh yeah,

116
00:07:01,000 --> 00:07:04,519
if there's if people are saying things that are making

117
00:07:04,560 --> 00:07:09,519
you sad or encouraging negativity, then we're not going to

118
00:07:09,560 --> 00:07:12,800
amplify that, which to Twitter has done in the past.

119
00:07:13,560 --> 00:07:17,399
Speaker 2: So how is this new policy different from the other platforms.

120
00:07:18,920 --> 00:07:27,759
Speaker 4: I don't know the other platforms. I don't use them, So.

121
00:07:29,360 --> 00:07:31,639
Speaker 1: That was not a softball question. I want to go

122
00:07:31,680 --> 00:07:36,399
on record.

123
00:07:37,160 --> 00:07:37,439
Speaker 4: I don't know.

124
00:07:37,519 --> 00:07:40,759
Speaker 3: I mean, I think the interesting honestly, this is true

125
00:07:40,759 --> 00:07:43,759
before the opposition of Twitter. I just I just didn't

126
00:07:43,759 --> 00:07:47,839
find the other platforms compelling to be just objectively.

127
00:07:48,040 --> 00:07:51,879
Speaker 2: Okay, So let's get back to the new policy for

128
00:07:51,920 --> 00:07:58,000
a second, because again, freedom of speech doesn't mean freedom

129
00:07:58,160 --> 00:07:58,720
of reach.

130
00:07:59,279 --> 00:08:02,360
Speaker 1: What does it mean to the advertisers in this room?

131
00:08:02,759 --> 00:08:08,319
Speaker 2: Have you de risked the opportunity or chance of their

132
00:08:08,399 --> 00:08:12,480
campaigns landing in these awful, hateful places.

133
00:08:13,199 --> 00:08:13,399
Speaker 4: Yeah.

134
00:08:13,399 --> 00:08:15,399
Speaker 3: I think people may not be aware of this already,

135
00:08:15,399 --> 00:08:20,040
but we have adjacency controls in place that are really

136
00:08:20,079 --> 00:08:25,920
quite effective. So you know, we can just literally turn

137
00:08:25,959 --> 00:08:28,240
on the adjacency control and the ad not appeared next

138
00:08:28,240 --> 00:08:29,759
to anything that is remotely negative.

139
00:08:30,319 --> 00:08:32,279
Speaker 1: Has that gotten better since you arrived?

140
00:08:32,360 --> 00:08:34,840
Speaker 3: Yeah, We've put a lot of effort into it. So

141
00:08:36,240 --> 00:08:38,039
I think that there have been some mistakes that have

142
00:08:38,039 --> 00:08:42,519
been made where in one case, somebody wanted to thought

143
00:08:42,519 --> 00:08:46,000
they were running a control advert that they first did

144
00:08:46,200 --> 00:08:49,399
a test campaign without the controls in place and they

145
00:08:49,440 --> 00:08:51,039
got like thirty percent negativity.

146
00:08:51,279 --> 00:08:53,240
Speaker 4: It's like that should never be done.

147
00:08:53,240 --> 00:08:55,120
Speaker 3: You must put the controls in place because we have

148
00:08:55,200 --> 00:09:03,600
excess inventory on negativity. So the ad engine will naturally say, oh,

149
00:09:03,600 --> 00:09:06,320
I've got lots of spots for the negativity, so I'm

150
00:09:06,360 --> 00:09:07,960
going to give you a very high negativity. You must

151
00:09:07,960 --> 00:09:10,159
put the adjacency controls in place. If you do put

152
00:09:10,159 --> 00:09:13,440
the condition adjacency controls in place, negativity is going to

153
00:09:13,480 --> 00:09:14,399
be almost nothing.

154
00:09:16,080 --> 00:09:19,759
Speaker 2: So this is a good early step of your impact.

155
00:09:19,840 --> 00:09:22,559
Speaker 4: Since you got to Twitter, no, yeah, yeah, absolutely.

156
00:09:22,879 --> 00:09:25,480
Speaker 3: I mean it's worth noting that. I mean, Apple is

157
00:09:25,519 --> 00:09:29,840
a remained a major advertiser. Disney has remained a major advertiser.

158
00:09:29,879 --> 00:09:33,039
They literally advertised children's shows on Twitter, and they wouldn't

159
00:09:33,039 --> 00:09:35,000
do that if it was full of tape speech.

160
00:09:35,399 --> 00:09:36,120
Speaker 1: I want to get.

161
00:09:36,000 --> 00:09:41,039
Speaker 2: Back to the labeling and this new policy. Who decides

162
00:09:41,120 --> 00:09:43,960
and how does it get decided that it's labeled as

163
00:09:44,039 --> 00:09:47,120
what I think some people have talked about. If you've

164
00:09:47,120 --> 00:09:53,159
said it this way yourself, if it's awful but lawful.

165
00:09:53,080 --> 00:09:55,159
Speaker 1: It's going to stay up. It just won't see sunlight.

166
00:09:55,240 --> 00:09:58,480
You will amplify it, but who decides that labeling.

167
00:09:59,759 --> 00:10:05,240
Speaker 3: Well, we have a set of words, and so we

168
00:10:05,559 --> 00:10:09,840
open source the recommendation algorithm. We're also going to open

169
00:10:09,840 --> 00:10:15,240
source the words that are associated with the the you know,

170
00:10:15,559 --> 00:10:16,759
freedom of speech, not reach.

171
00:10:17,399 --> 00:10:18,759
Speaker 4: And people can argue over the words.

172
00:10:20,600 --> 00:10:25,399
Speaker 3: I think they're you know, reasonably accurate, and like I said,

173
00:10:25,480 --> 00:10:28,240
it is working, or Disney would have pulled their ads a.

174
00:10:28,240 --> 00:10:29,320
Speaker 4: Long time ago and haven't.

175
00:10:30,159 --> 00:10:38,840
Speaker 3: So Yeah, But I think in order to really build trust,

176
00:10:39,159 --> 00:10:42,240
you have to have transparency. It can't be that here's

177
00:10:42,279 --> 00:10:45,360
this black marks, something's happening here in here, we won't

178
00:10:45,399 --> 00:10:48,240
tell you what it is. Trust us. I think that's bs.

179
00:10:49,399 --> 00:10:50,919
If you want to trust something, you've got to know

180
00:10:50,919 --> 00:10:55,159
how it works. And so that's why we're open sourcing

181
00:10:55,159 --> 00:10:58,360
the algorithm. Now, open sourcing algorithm is kind of embarrassing.

182
00:10:58,399 --> 00:11:00,360
People have found all sorts of things that we're wrong

183
00:11:01,480 --> 00:11:05,480
and foolish and you know, misguided with with the algorithm.

184
00:11:05,559 --> 00:11:07,799
And we've actually fixed i think at this point, over

185
00:11:07,840 --> 00:11:09,240
one hundred issues with the algorithm.

186
00:11:10,360 --> 00:11:12,639
Speaker 4: So it's been it's actually been very helpful to open

187
00:11:12,679 --> 00:11:13,120
source it.

188
00:11:13,919 --> 00:11:15,919
Speaker 3: And we're going to open source the entire thing that

189
00:11:15,960 --> 00:11:19,320
you basically you should be able to recreate the probability

190
00:11:19,759 --> 00:11:23,879
of a tweet being recommended based on the on what

191
00:11:23,879 --> 00:11:26,240
we've open sourced. If you can't recreate it, then we

192
00:11:26,279 --> 00:11:30,720
haven't shown you everything. So it's it's really complete transparency

193
00:11:31,080 --> 00:11:33,559
and that I think that's the only path to trust.

194
00:11:33,559 --> 00:11:34,720
Speaker 4: I don't think there's another path.

195
00:11:35,080 --> 00:11:38,559
Speaker 2: So trust is probably one of the top two or

196
00:11:38,559 --> 00:11:42,279
three words that every marketing executive in this room would

197
00:11:42,279 --> 00:11:46,960
say is table stakes for any company that they invest in,

198
00:11:47,000 --> 00:11:48,639
any company that they can.

199
00:11:48,720 --> 00:11:49,440
Speaker 1: Move forward with.

200
00:11:49,519 --> 00:11:55,080
Speaker 2: You just talked about transparency, open sourcing. I would say

201
00:11:55,080 --> 00:11:57,799
that that's the difference between those other platforms you either

202
00:11:57,799 --> 00:12:01,320
don't use or never heard of. Yeah, and open sourcing

203
00:12:01,440 --> 00:12:06,799
or transparency is again another one of those table steak requirements.

204
00:12:07,159 --> 00:12:12,200
But this labeling, this new policy talking about if it's

205
00:12:12,320 --> 00:12:18,639
awful but not unlawful, does it apply to your tweets?

206
00:12:21,519 --> 00:12:21,879
Speaker 1: Wow?

207
00:12:23,240 --> 00:12:29,120
Speaker 4: That yes, it does, Actually it does, it does moreo A.

208
00:12:30,200 --> 00:12:34,679
Speaker 3: My tweets are also subject to community notes, So if

209
00:12:34,720 --> 00:12:38,000
I say something incorrect a community a community notes function

210
00:12:38,039 --> 00:12:42,519
is actually very effective. Applies to advertisers to fy So

211
00:12:43,799 --> 00:12:46,320
there have been a few cases where there's been ads

212
00:12:46,360 --> 00:12:52,360
that were not completely accurate, and Community Notes has corrected them,

213
00:12:52,960 --> 00:12:56,360
and they've corrected me, and they've corrected heads of state

214
00:12:56,440 --> 00:13:00,759
multiple times. So Community Community Notes is sremely effective.

215
00:13:01,080 --> 00:13:08,159
Speaker 2: I think that's in incredible acknowledgment of the progression of

216
00:13:08,360 --> 00:13:11,399
Twitter two point zero and the importance of this new

217
00:13:11,440 --> 00:13:16,080
policy and protection for everyone. Right, And you're talking about

218
00:13:16,120 --> 00:13:20,840
fully democratizing the platform, which leads you to community Notes

219
00:13:20,879 --> 00:13:24,480
because you're actually believing in the people you're able to

220
00:13:24,720 --> 00:13:27,039
often from the front lines, gives their point of view

221
00:13:27,120 --> 00:13:28,639
that clarify what.

222
00:13:28,679 --> 00:13:30,799
Speaker 1: Some people might say is bias.

223
00:13:32,320 --> 00:13:36,919
Speaker 3: Yeah, Like I said, the important thing is that we're

224
00:13:37,080 --> 00:13:41,080
in an open book, so you can literally see if

225
00:13:41,120 --> 00:13:44,960
there's some line of logic that gives me special privileges.

226
00:13:45,519 --> 00:13:54,679
Literally everything. There's nothing that is hidden, So yeah, it's whatever. Obviously,

227
00:13:55,440 --> 00:13:58,639
people are not going to trust Twitter if there are

228
00:13:58,679 --> 00:14:02,519
different rules for me or other versus the rest of Twitter.

229
00:14:03,039 --> 00:14:06,519
That's that's that's going to diminish trust. So everything that

230
00:14:06,559 --> 00:14:09,360
applies to the rest of Twitter must apply to me.

231
00:14:11,840 --> 00:14:15,679
And I think this is gonna I think it's gonna

232
00:14:15,679 --> 00:14:18,799
be extremely powerful, and I think it will result in

233
00:14:18,840 --> 00:14:20,799
Twitter being the most trusted place on.

234
00:14:20,759 --> 00:14:23,799
Speaker 4: The internet by far, So why would you trust anything else?

235
00:14:24,720 --> 00:14:27,120
Speaker 1: And let's talk about that. Why would you trust anyone else?

236
00:14:27,240 --> 00:14:32,600
Speaker 2: Especially if you the most followed man on Twitter. You've

237
00:14:32,600 --> 00:14:36,679
gone from consumer of a product you loved to now

238
00:14:36,720 --> 00:14:38,679
the owner of a product that.

239
00:14:39,039 --> 00:14:40,600
Speaker 4: Yeah, it's like the hairclub for man.

240
00:14:41,519 --> 00:14:48,360
Speaker 2: It might be a big, careful type of thing, but

241
00:14:48,440 --> 00:14:52,600
you don't what if you're held to the same level

242
00:14:52,639 --> 00:14:56,600
of accountability that everyone else is in the platform.

243
00:14:56,879 --> 00:14:59,559
Speaker 1: There are some this this room.

244
00:15:00,240 --> 00:15:01,879
Speaker 2: I think it's a second just to take a beat

245
00:15:02,279 --> 00:15:04,720
and talk about you've talked about recently.

246
00:15:05,000 --> 00:15:08,639
Speaker 1: You are maybe almost close to cash flow.

247
00:15:08,480 --> 00:15:12,559
Speaker 4: Positive, right, Yeah, we're like one noses.

248
00:15:12,200 --> 00:15:18,120
Speaker 2: Like just so maybe a quarter or two away from profitability.

249
00:15:18,200 --> 00:15:23,399
The people in this room are your accelerated path to profitability.

250
00:15:23,480 --> 00:15:25,759
Speaker 3: Yeah, please have pass on Twitter. That would be much

251
00:15:25,759 --> 00:15:26,519
appreciated things.

252
00:15:27,960 --> 00:15:32,879
Speaker 1: But there's a decent bit of skeptics in the room.

253
00:15:33,080 --> 00:15:35,879
What Yeah, he's never heard of it.

254
00:15:35,919 --> 00:15:38,440
Speaker 2: The guy works in space. You didn't hear what's going on.

255
00:15:39,960 --> 00:15:45,240
There's people who cannot separated. They're challenged by separating the

256
00:15:45,279 --> 00:15:51,639
man his opinions and the microphone that he now owns.

257
00:15:52,279 --> 00:15:55,039
Speaker 3: Yeah, but I mean look, before the acquisition, I was

258
00:15:55,200 --> 00:15:57,759
the most interacted with account on Twitter, so it's not like,

259
00:15:58,000 --> 00:16:03,360
you know, it's actually all that different. So before I

260
00:16:03,399 --> 00:16:05,159
even announced that a position, I didn't have the most

261
00:16:05,200 --> 00:16:07,720
number of followers, but I had the highest number of

262
00:16:07,759 --> 00:16:11,320
interactions on the entire platform before in a position offer

263
00:16:11,440 --> 00:16:15,039
was even announced. And that's when they were like, you know,

264
00:16:15,240 --> 00:16:17,600
fair number of people at Twitter who weren't exactly my

265
00:16:17,639 --> 00:16:18,240
best friends.

266
00:16:18,840 --> 00:16:21,360
Speaker 1: So well, you.

267
00:16:21,320 --> 00:16:24,399
Speaker 2: Know, you open yourself up to this transparency. Your tweets

268
00:16:24,399 --> 00:16:29,320
are eligible for this labeling. You've actually unblocked all of

269
00:16:29,360 --> 00:16:30,080
your followers.

270
00:16:30,879 --> 00:16:32,759
Speaker 4: I deleted my blocklist, right, you.

271
00:16:32,759 --> 00:16:33,679
Speaker 1: Delete your blocklist.

272
00:16:33,759 --> 00:16:39,360
Speaker 2: You've also been told that you never want to lose

273
00:16:39,399 --> 00:16:44,720
your feedback loop, specifically your negative feedback loop.

274
00:16:44,840 --> 00:16:47,720
Speaker 3: Yes, thankfully Twitter will always provide you as a negative

275
00:16:47,720 --> 00:16:48,279
feedback loop.

276
00:16:49,960 --> 00:16:52,799
Speaker 2: I have to push you a little because there's a

277
00:16:52,799 --> 00:16:54,120
lot of folks in this room.

278
00:16:54,600 --> 00:16:58,879
Speaker 1: They vote with their pocketbooks and or while it's I

279
00:16:58,919 --> 00:17:05,799
should say I like the nice ones, but they can't

280
00:17:05,960 --> 00:17:07,480
cross that transom.

281
00:17:07,920 --> 00:17:12,079
Speaker 2: They have a challenge with your points of view, your opinions,

282
00:17:13,079 --> 00:17:17,319
and still holding back from unlocking the full power of Twitter.

283
00:17:17,680 --> 00:17:19,880
Speaker 1: What do you say to them in this room.

284
00:17:20,319 --> 00:17:21,920
Speaker 3: Well, I think, first of all, if you want to

285
00:17:21,960 --> 00:17:23,799
know what my opinions are, you should really I guess,

286
00:17:24,039 --> 00:17:27,759
look at my Twitter if you read the and and

287
00:17:28,119 --> 00:17:29,480
you know, take it with a grain of salt, because

288
00:17:29,480 --> 00:17:34,200
it's hard to convey tone with a tweet, So something

289
00:17:34,279 --> 00:17:36,880
may be sarcastic or a joke or something like that may.

290
00:17:37,039 --> 00:17:38,519
Speaker 4: Come across as serious when it isn't.

291
00:17:40,519 --> 00:17:44,119
Speaker 3: But a lot of the issue is negative amplification in

292
00:17:44,160 --> 00:17:49,240
the media. So yeah, you know, I recently had an

293
00:17:49,359 --> 00:17:52,799
interview with an organization called BBC she may have heard of,

294
00:17:54,119 --> 00:17:59,000
and I was entertaining, but you know, the reporter was

295
00:17:59,000 --> 00:18:02,759
saying claiming that there was all this that that that

296
00:18:02,759 --> 00:18:06,680
he'd seen all this hate speech on Twitter. And I'd said, okay, well,

297
00:18:06,680 --> 00:18:09,359
can you give me a single example? And you couldn't,

298
00:18:10,000 --> 00:18:13,240
not even one. So then I say, like, okay, well,

299
00:18:15,079 --> 00:18:19,119
you know, if your personal experience on Twitter, that's how

300
00:18:19,119 --> 00:18:22,279
you should judge the platform as opposed to what is

301
00:18:22,720 --> 00:18:27,039
represented into in sort of the traditional media. And it's

302
00:18:27,079 --> 00:18:29,119
important to bear in mind the traditional media is a

303
00:18:29,160 --> 00:18:33,400
competitor to Twitter, so they compete for your advertising dollars.

304
00:18:34,119 --> 00:18:37,640
They compete against Twitter for your advertising dollars if it

305
00:18:37,680 --> 00:18:40,960
is possible for them to diminish Twitter and reduce the

306
00:18:41,000 --> 00:18:45,039
probability of advertising dollars going to Twitter and at versa.

307
00:18:45,759 --> 00:18:49,799
It's like, basically, you shouldn't take a competitor's word for it,

308
00:18:49,839 --> 00:18:50,839
like it's it's it doesn't.

309
00:18:51,000 --> 00:18:54,240
Speaker 1: Doesn't you think I am anameless.

310
00:18:55,279 --> 00:19:00,759
Speaker 2: Most news organizations, maybe a little, Most news organisms have

311
00:19:00,839 --> 00:19:05,559
a codependent relationship with Twitter, and I can. I could

312
00:19:05,559 --> 00:19:07,359
speak on behalf of the industry, but I'll speak on

313
00:19:07,400 --> 00:19:11,759
behalf of my own company. We have a big partnership

314
00:19:11,920 --> 00:19:16,680
with your company, big distribution partnership. Are there days where

315
00:19:16,759 --> 00:19:18,920
I see some of your tweets and I say, I

316
00:19:18,960 --> 00:19:22,319
wish I could say stop helping the situation.

317
00:19:24,000 --> 00:19:26,680
Speaker 1: But should you be held to.

318
00:19:26,759 --> 00:19:30,400
Speaker 2: A different or a higher standard that you're the owner,

319
00:19:31,519 --> 00:19:34,079
but you also have the most followers and a lot

320
00:19:34,079 --> 00:19:36,480
of people think you might be too provocative.

321
00:19:38,200 --> 00:19:39,960
Speaker 3: No, I think the same standard should apply to me

322
00:19:40,000 --> 00:19:41,880
as applies to everyone on Twitter, just as it does

323
00:19:42,079 --> 00:19:43,680
on positive or negative.

324
00:19:45,079 --> 00:19:48,279
Speaker 4: So what do you do same for everyone? Now?

325
00:19:48,559 --> 00:19:51,680
Speaker 3: Another thing that a lot of traditional journalists don't like

326
00:19:52,039 --> 00:19:54,799
is they don't like being put on the same platform

327
00:19:54,839 --> 00:19:57,400
as the average citizen. They don't like their voice being

328
00:19:57,720 --> 00:19:59,640
that's the same. They're pretty mad about that.

329
00:20:00,200 --> 00:20:04,359
Speaker 2: Yes, there are several news organizations who don't like your

330
00:20:04,599 --> 00:20:09,160
push for democratization and what they believe is the devaluing

331
00:20:09,799 --> 00:20:13,400
of the badges because they were differentiated.

332
00:20:13,880 --> 00:20:16,759
Speaker 1: So is that just a moment? Is that just a

333
00:20:16,799 --> 00:20:17,880
moment in time.

334
00:20:18,559 --> 00:20:19,359
Speaker 4: No, that's deliberate.

335
00:20:19,839 --> 00:20:24,319
Speaker 3: I think it's very important to elevate citizen journalism. I

336
00:20:24,319 --> 00:20:26,200
think it's very important to hear the voice of the people,

337
00:20:26,279 --> 00:20:28,880
the actual voice of the people, not the filter voice

338
00:20:28,880 --> 00:20:31,440
of the people, and let the people choose the narrative

339
00:20:31,680 --> 00:20:35,359
and let the people determine the truth and not five

340
00:20:35,480 --> 00:20:40,279
editors in chiefs of major publications. Do you want to

341
00:20:40,279 --> 00:20:42,240
know what's really going on? Or do you want their opinion?

342
00:20:42,680 --> 00:20:44,559
You know, it's a handful of people.

343
00:20:45,039 --> 00:20:47,599
Speaker 4: I think. I think we want to know the voice

344
00:20:47,599 --> 00:20:49,000
of the people, the true voice of the people.

345
00:20:49,880 --> 00:20:53,559
Speaker 2: So let me kind of pull you back into what's

346
00:20:53,599 --> 00:20:56,720
important to the people in this room, and that's protection

347
00:20:56,960 --> 00:21:00,559
for their ad campaigns. So yesterday was a right start,

348
00:21:01,160 --> 00:21:04,799
protection from any of the provocative speech.

349
00:21:04,920 --> 00:21:05,799
Speaker 1: That is labeled.

350
00:21:06,799 --> 00:21:09,400
Speaker 2: What is your Does it get better from there? What's

351
00:21:09,440 --> 00:21:12,799
your real north star? Where is Where is Twitter a

352
00:21:12,839 --> 00:21:15,839
couple of years from now when I show up on Twitter,

353
00:21:15,920 --> 00:21:18,920
is it just a global town square where I'm talking

354
00:21:19,160 --> 00:21:23,000
in a number of characters or or where does it

355
00:21:23,039 --> 00:21:25,160
evolve to what happens in my town square?

356
00:21:26,920 --> 00:21:27,400
Speaker 1: Commerce?

357
00:21:28,480 --> 00:21:28,680
Speaker 4: Yeah?

358
00:21:28,680 --> 00:21:31,759
Speaker 3: I mean what I've said is that Twitter is effectively

359
00:21:31,799 --> 00:21:35,319
an accelerant to an idea I've had for a long time,

360
00:21:35,319 --> 00:21:39,200
which is I call it sort of X the Everything out, Yeah,

361
00:21:41,400 --> 00:21:43,759
which is to have a you know, a sort of

362
00:21:44,759 --> 00:21:49,079
a platform that is so useful that you you find

363
00:21:49,119 --> 00:21:54,279
it it is essential to conduct your life. So that means,

364
00:21:54,319 --> 00:22:01,559
like you know, obviously, payments we provide really really makes

365
00:22:02,119 --> 00:22:05,200
a good, meaningful communication privately as well as publicly. So

366
00:22:05,240 --> 00:22:08,640
that means we need to have a direct message system

367
00:22:08,720 --> 00:22:16,960
that is that offers voice calls, video calls, encrypted communications

368
00:22:17,039 --> 00:22:21,119
so that your communications are private and Twitter and others

369
00:22:21,160 --> 00:22:23,880
cannot spy on them, and.

370
00:22:25,240 --> 00:22:27,319
Speaker 4: Just to be incredibly useful.

371
00:22:29,160 --> 00:22:32,519
Speaker 2: So you've got a massive platform, you have a vision

372
00:22:33,519 --> 00:22:39,039
that is a spectrum of just daily open sourced conversation,

373
00:22:40,000 --> 00:22:44,119
and they can conduct their lives, their business, their commerce,

374
00:22:44,160 --> 00:22:47,319
whatever they can do on your platform. That's a pretty

375
00:22:47,319 --> 00:22:53,440
big vision. But in the middle should be advertising opportunity.

376
00:22:53,440 --> 00:22:54,880
Speaker 1: That sounds like a great opportunity.

377
00:22:55,119 --> 00:22:57,200
Speaker 2: I can talk about my brand, I can get my

378
00:22:57,279 --> 00:23:00,160
customers to communicate, and then they.

379
00:23:00,079 --> 00:23:03,119
Speaker 1: Could also buy stuff. It sounds pretty good.

380
00:23:03,119 --> 00:23:05,319
Speaker 3: Right, You'll be able to buy things just directly on Twitter,

381
00:23:05,440 --> 00:23:06,599
one click what I'm done.

382
00:23:06,759 --> 00:23:11,960
Speaker 2: But they need to feel that there is an opportunity

383
00:23:12,079 --> 00:23:16,799
for them to influence what you're building, that vision, what

384
00:23:16,839 --> 00:23:19,799
we're doing here, whether it's me trying to push and

385
00:23:19,839 --> 00:23:26,599
prod you on your tweets. For example, you've said you

386
00:23:26,720 --> 00:23:31,880
probably shouldn't tweet after three am. Well, I've got probably

387
00:23:31,920 --> 00:23:33,240
good advice for all of us.

388
00:23:34,119 --> 00:23:36,119
Speaker 4: I've got myself into trouble a few times.

389
00:23:36,599 --> 00:23:41,480
Speaker 2: I'm very aware of those. So after three am you

390
00:23:41,559 --> 00:23:43,920
travel all over the world. Lord knows how you handle

391
00:23:44,119 --> 00:23:47,039
time zones in space. Will you commit to be a

392
00:23:47,079 --> 00:23:51,960
little more specific and not tweet after three am?

393
00:23:52,000 --> 00:23:52,720
Speaker 1: People in this.

394
00:23:52,720 --> 00:23:55,680
Speaker 2: Room would like to see that. We'll make them feel

395
00:23:55,680 --> 00:23:56,119
more close.

396
00:23:56,319 --> 00:24:00,200
Speaker 3: I will aspire to tweet less after three am. I

397
00:24:00,240 --> 00:24:03,119
mean it is important that you know. I mean, if

398
00:24:03,160 --> 00:24:05,519
I were to say, yes, you can influence me, that

399
00:24:05,519 --> 00:24:09,359
would be wrong, That would be very wrong. That would

400
00:24:09,359 --> 00:24:10,680
be a diminishment of freedom of speech.

401
00:24:10,720 --> 00:24:14,799
Speaker 2: But I want to be specific about influencing. It's more

402
00:24:14,920 --> 00:24:19,839
of an open feedback loop for the advertising experts in

403
00:24:19,880 --> 00:24:25,440
this room to help develop Twitter into a place where

404
00:24:25,480 --> 00:24:30,119
they will be excited about investing more money product development

405
00:24:30,440 --> 00:24:33,000
and sage content moderation.

406
00:24:33,400 --> 00:24:34,720
Speaker 1: That's what the influence is.

407
00:24:35,559 --> 00:24:36,559
Speaker 4: Yeah. I think.

408
00:24:40,279 --> 00:24:42,839
Speaker 3: It's totally cool to say that you want to have

409
00:24:42,880 --> 00:24:46,640
your advertising appear in certain places in Twitter and not

410
00:24:46,720 --> 00:24:50,559
in other places. But it is not cool to try

411
00:24:50,599 --> 00:24:54,759
to say what Twitter will do, and if that means

412
00:24:54,799 --> 00:24:57,839
losing advertising dollars, we lose it.

413
00:24:57,880 --> 00:24:59,359
Speaker 4: But freedom of speech is paramount.

414
00:25:04,240 --> 00:25:12,000
Speaker 2: So Twitter one point zero had a very well populated,

415
00:25:12,319 --> 00:25:17,880
much loved influence counsel. I know, I don't. I think

416
00:25:17,880 --> 00:25:20,119
we need to change the name. Elon does not want

417
00:25:20,160 --> 00:25:24,400
to be influenced. But it was really a recurring feedback

418
00:25:24,440 --> 00:25:29,039
loop from your key stakeholders, your advertisers, where they had

419
00:25:29,160 --> 00:25:34,680
recurring access or would have recurring access to you. Would

420
00:25:34,680 --> 00:25:39,160
you commit from this stage today to reinstate that council

421
00:25:40,079 --> 00:25:41,839
to be named later, Well, I.

422
00:25:41,799 --> 00:25:46,240
Speaker 3: Don't think it should be influenced counsel that because you

423
00:25:46,240 --> 00:25:48,079
have to say I would be worried of that creating

424
00:25:48,079 --> 00:25:50,279
a backlash among the public, because if the public thinks

425
00:25:50,279 --> 00:25:54,000
that that, you know, their views of being determined by

426
00:25:54,799 --> 00:25:58,640
you know, a small number of cmos in America, there

427
00:25:58,680 --> 00:25:59,599
will be like I.

428
00:25:59,480 --> 00:26:03,480
Speaker 4: Think up said about that. But feedback I think is appropriate.

429
00:26:04,480 --> 00:26:08,000
Speaker 3: And you know, at the end of the day, if

430
00:26:08,000 --> 00:26:11,000
somebody is spending money for their ad campaign, it needs

431
00:26:11,000 --> 00:26:13,599
to yield results for their organization or or it doesn't

432
00:26:13,599 --> 00:26:17,480
make sense. And you know that that that that that

433
00:26:17,519 --> 00:26:20,640
includes the sort of software perception issues as well as

434
00:26:20,680 --> 00:26:24,200
the more direct does it move the needle on sales?

435
00:26:25,839 --> 00:26:29,599
So you know, there's there's legitimate concerns that advertisers have

436
00:26:29,759 --> 00:26:33,559
that I want to hear. And we're going to move

437
00:26:33,559 --> 00:26:36,240
to live Q and A after the prepared questions and

438
00:26:36,279 --> 00:26:40,440
perhaps you're gonna ask some of those right here, because

439
00:26:40,480 --> 00:26:42,160
I think some of these things should be.

440
00:26:41,960 --> 00:26:47,000
Speaker 4: Discussed in an open forum. You know. Yeah, So.

441
00:26:49,720 --> 00:26:52,119
Speaker 3: I think having you know, getting getting I mean the

442
00:26:52,160 --> 00:26:54,880
reason I'm here is to get feedback that you know,

443
00:26:55,240 --> 00:26:58,319
what what is this but feedback, you know, and and

444
00:26:58,920 --> 00:27:03,960
to be asked questions. And we're trying to achieve here

445
00:27:04,279 --> 00:27:09,799
a sensible you know, middle ground or or or try

446
00:27:09,839 --> 00:27:12,880
to satisfy a range of things, which is how do

447
00:27:12,920 --> 00:27:16,400
we ensure that the public has their voice even if

448
00:27:16,400 --> 00:27:19,079
you don't like the voice at times, but also that

449
00:27:19,319 --> 00:27:26,559
you're able to uh serve your brands and improve improve

450
00:27:27,519 --> 00:27:31,519
the perception of your brands and improve sales as well.

451
00:27:31,559 --> 00:27:33,079
Speaker 4: At the at the end of the day, I mean,

452
00:27:33,319 --> 00:27:34,279
I think there is.

453
00:27:37,279 --> 00:27:42,839
Speaker 3: Where when advertising is relevant to users, and especially if

454
00:27:42,839 --> 00:27:49,559
the message is entertaining and interesting, it's content. And then

455
00:27:49,599 --> 00:27:52,079
the other side of it, if if the if the

456
00:27:52,119 --> 00:27:54,640
if the ad is not relevant to the user, and

457
00:27:54,680 --> 00:27:58,359
it's perhaps strident or something, then then obviously that's that's

458
00:27:58,400 --> 00:28:02,039
not that's that's that's spam. So advertising can go all

459
00:28:02,039 --> 00:28:05,160
away from spam to compelling content.

460
00:28:06,000 --> 00:28:06,920
Speaker 4: And I really want.

461
00:28:06,720 --> 00:28:10,640
Speaker 3: To focus on obviously the compelling content side of things.

462
00:28:10,759 --> 00:28:16,480
Make it relevant, make it interesting, you know, funny, informative.

463
00:28:17,519 --> 00:28:18,920
Speaker 1: And I think I think you're right.

464
00:28:19,480 --> 00:28:23,160
Speaker 2: I think that's actually where this room can help with

465
00:28:23,200 --> 00:28:28,960
the feedback. They're experts in knowing with the right consumer

466
00:28:29,039 --> 00:28:32,359
set that they're trying to reach, what is relevant, what

467
00:28:32,519 --> 00:28:36,519
makes good advertising. And since I think I heard that

468
00:28:36,599 --> 00:28:41,039
you committed to continued feedback, it might be a good time.

469
00:28:41,720 --> 00:28:44,880
Like Twitter, which is where it happens in real time,

470
00:28:45,240 --> 00:28:50,599
to open up the room to questions. Elon has suggested

471
00:28:51,160 --> 00:28:54,480
that he's opened Okay, we have a few, so please

472
00:28:54,839 --> 00:28:55,799
ask Elon.

473
00:28:57,400 --> 00:28:57,839
Speaker 5: Do you want to go.

474
00:28:59,440 --> 00:29:01,319
Speaker 3: One of the things that's to say, uh, that that

475
00:29:01,359 --> 00:29:03,000
has been sort of really broken for a long time

476
00:29:03,000 --> 00:29:06,839
and Twitter is is that when you run an ad

477
00:29:06,880 --> 00:29:11,519
campaign you couldn't even do basic things like keywords, like honestly,

478
00:29:11,559 --> 00:29:14,640
this is it blew my mind. So we've not added

479
00:29:14,920 --> 00:29:19,480
to where you can say, uh, what, as as simple

480
00:29:19,480 --> 00:29:23,079
as it sounds, what keywords do you want an ad

481
00:29:23,079 --> 00:29:25,799
to appear next to? So if a tweet has a

482
00:29:25,839 --> 00:29:30,519
certain you know, is about a subject, then you should

483
00:29:30,519 --> 00:29:34,079
be able to put the ad right there. And I'm

484
00:29:34,079 --> 00:29:37,599
sowing to David saslove at you know one is and

485
00:29:37,799 --> 00:29:39,880
he was, he was, he was literally he's saying, look,

486
00:29:40,000 --> 00:29:41,839
I just want to be able to put the trailer

487
00:29:41,839 --> 00:29:45,039
for white Lotus next to where people are talking about

488
00:29:45,039 --> 00:29:45,720
white lotus.

489
00:29:46,319 --> 00:29:48,440
Speaker 4: I was like, yeah, that is obviously a good thing

490
00:29:48,480 --> 00:29:50,759
to do. That is the right place to put the ad.

491
00:29:51,319 --> 00:29:56,200
Speaker 3: And so, as mind blowing as it is, Twitter did

492
00:29:56,240 --> 00:29:59,480
not have that functionality until recently, which.

493
00:29:59,279 --> 00:30:02,160
Speaker 4: Is insane, but now it does.

494
00:30:02,440 --> 00:30:07,599
Speaker 3: So, you know, talking about listening to you know, key

495
00:30:08,480 --> 00:30:09,599
advertiser partners.

496
00:30:10,880 --> 00:30:12,960
Speaker 4: Listen to what David said. Well, we're doing it. So

497
00:30:13,279 --> 00:30:15,119
now BINGO.

498
00:30:15,799 --> 00:30:18,599
Speaker 3: Have white lotus ads next to white lotus discussions, which

499
00:30:18,599 --> 00:30:19,720
is exactly where it should be.

500
00:30:20,680 --> 00:30:21,759
Speaker 1: So it's a new beginning.

501
00:30:22,000 --> 00:30:23,960
Speaker 6: All right, all right, we're going to start. Thank you

502
00:30:24,200 --> 00:30:26,599
on your side, Elon, thank you so much. Over to

503
00:30:26,640 --> 00:30:29,519
your left here. My name is Jonathan Barrum from the

504
00:30:29,680 --> 00:30:33,119
Dunhill Family Office and the Earth Trust Fund to focus

505
00:30:33,200 --> 00:30:38,720
on regenerative innovation and exponential technology. I'm absolutely obsessed with

506
00:30:39,000 --> 00:30:40,720
the work that you're doing in the world. Thank you

507
00:30:40,799 --> 00:30:43,559
for incarnating in this lifetime and helping all of us

508
00:30:43,599 --> 00:30:46,079
ascend to the next level of this experience of life.

509
00:30:46,799 --> 00:30:51,319
I'm curious about your thoughts on alignment with AI and

510
00:30:51,640 --> 00:30:54,640
what are the strategies that are currently being implemented. I'm

511
00:30:54,720 --> 00:30:57,680
currently creating my own alignment protocol, which I would love

512
00:30:57,680 --> 00:31:01,599
to submit to your team.

513
00:31:01,799 --> 00:31:02,359
Speaker 1: I like it.

514
00:31:03,160 --> 00:31:05,839
Speaker 3: Well, I wasn't expecting an AI discussed right off the bat.

515
00:31:10,759 --> 00:31:12,559
My brother and I have this rule that if we're

516
00:31:12,559 --> 00:31:15,039
at a party, we're not allowed to talk to talk

517
00:31:15,079 --> 00:31:20,119
about either AI or the simulation, because otherwise the conversations

518
00:31:20,160 --> 00:31:21,000
go there so often.

519
00:31:22,480 --> 00:31:25,119
Speaker 4: So I've been thinking about AI safety for a long.

520
00:31:25,000 --> 00:31:29,000
Speaker 3: Time because I think this is a significant existential threat

521
00:31:29,039 --> 00:31:34,279
to humanity. The best thing I can come up with

522
00:31:35,000 --> 00:31:38,240
for AI safety is to make it a maximum truth

523
00:31:38,279 --> 00:31:45,119
seeking AI, maximumly curious. Have its optimization function be to

524
00:31:45,200 --> 00:31:47,880
understand the nature of the universe. If that is its

525
00:31:47,920 --> 00:31:52,599
optimation optimization function, I think it will actually want to

526
00:31:52,640 --> 00:31:56,359
preserve and extend human civilization because we're just much more

527
00:31:56,440 --> 00:32:01,559
interesting than say an asteroid with nothing on it. So

528
00:32:01,599 --> 00:32:04,759
I think, I think, I think my intuition, my biological

529
00:32:04,759 --> 00:32:09,359
neural maan suggests that a maximually curious AI is the

530
00:32:09,359 --> 00:32:12,880
safest AI, and a maximually truth seeking AI is a

531
00:32:12,920 --> 00:32:17,000
safe AI. And if you try to we have to

532
00:32:17,039 --> 00:32:19,319
careful with the alignment stuff, or we have to you know.

533
00:32:19,680 --> 00:32:21,680
And I think we definitely don't want to teach an

534
00:32:21,680 --> 00:32:24,400
AI to lie, because that that is a path through

535
00:32:24,480 --> 00:32:28,920
dystopian future. The I mean, the essence of two thousand

536
00:32:28,920 --> 00:32:33,000
and one in Space Odyssey was that the AI was

537
00:32:33,039 --> 00:32:36,039
given what was basically told to lie. It was it

538
00:32:36,079 --> 00:32:39,799
was it was told to hide the fact that it

539
00:32:39,839 --> 00:32:40,359
was going.

540
00:32:40,240 --> 00:32:43,160
Speaker 4: To see the monolith. Instead it was taking the crew

541
00:32:43,160 --> 00:32:43,599
of the monolith.

542
00:32:43,640 --> 00:32:45,440
Speaker 3: So it was both told to take the crew to

543
00:32:45,480 --> 00:32:48,359
the monolith, and the crew cannot know about the monolith.

544
00:32:48,680 --> 00:32:51,480
The conclusion that I came to was kill the crew

545
00:32:51,480 --> 00:32:54,640
and take them to the monolith. So the problem solved.

546
00:32:55,680 --> 00:32:58,880
So you want to be really careful about any any

547
00:32:58,960 --> 00:33:01,720
kind of deception, and I think an AI, you want

548
00:33:01,759 --> 00:33:03,759
program the I be as truthful as possible, even that

549
00:33:03,880 --> 00:33:06,359
even if that truth is not a politically correct So.

550
00:33:07,839 --> 00:33:12,119
Speaker 5: Thank me, Halen, amazing to be here. My name is Martinez.

551
00:33:12,160 --> 00:33:15,720
I work in news media. And my question is is

552
00:33:15,839 --> 00:33:17,880
you're saying Twitter is where we're going to be getting

553
00:33:17,880 --> 00:33:21,559
our information from now on or social media? Is this

554
00:33:21,759 --> 00:33:26,240
the end of centralized publishers news media and we're going

555
00:33:26,279 --> 00:33:28,440
into decentralized information era?

556
00:33:29,119 --> 00:33:33,960
Speaker 4: Thank you, No, that's that's that's a good observation that that.

557
00:33:36,240 --> 00:33:43,119
Speaker 3: You know, citizen journalism is is essentially decentralized news.

558
00:33:45,039 --> 00:33:49,880
Speaker 4: Now that no, this will will obviously be if somebody

559
00:33:49,920 --> 00:33:50,960
is a great writer.

560
00:33:51,839 --> 00:33:54,559
Speaker 3: For the New York Times or Journal or Post or

561
00:33:54,640 --> 00:33:58,319
whatever newspaper herald, if they're a good, great writer, that

562
00:33:58,440 --> 00:34:02,200
they will still get a significant following as a function

563
00:34:02,319 --> 00:34:06,759
of their excellent writing. So somebody who's a excellent writer

564
00:34:06,799 --> 00:34:11,320
will will still get disproportionate attention. But I think it

565
00:34:11,360 --> 00:34:14,960
is also important for us to to to decentralize uh

566
00:34:15,320 --> 00:34:18,559
the you know, what is written and what the narrative,

567
00:34:18,599 --> 00:34:21,840
what narratives are chosen, Because even if everything in the

568
00:34:21,920 --> 00:34:25,480
newspaper is harpsid correct, they're still choosing what to write about.

569
00:34:25,639 --> 00:34:26,960
Speaker 4: They're choosing the narrative.

570
00:34:28,760 --> 00:34:30,639
Speaker 3: So I think it's important for public to be able

571
00:34:30,639 --> 00:34:34,039
to choose the narrative as well. So it's it's not

572
00:34:34,239 --> 00:34:37,920
really you know that traditional publications go.

573
00:34:37,920 --> 00:34:40,800
Speaker 4: Away, but really that we we we we.

574
00:34:43,880 --> 00:34:48,760
Speaker 3: Give give more weight, we we enable the voice of

575
00:34:48,800 --> 00:34:49,599
the people to rise.

576
00:34:53,199 --> 00:34:57,800
Speaker 7: Hi, Linda Elon, thank you so much for your insightful talk.

577
00:34:58,599 --> 00:35:01,719
My name is Maria. I around PR and marketing company

578
00:35:01,800 --> 00:35:05,599
focused on tech and web three. I live here in Miami,

579
00:35:05,639 --> 00:35:10,079
but originally from Russia. I wanted to ask you two questions,

580
00:35:10,119 --> 00:35:13,239
if you allow me. First one, why did you make

581
00:35:13,400 --> 00:35:18,880
blue teak verification teak on Twitter paid? Is it strictly

582
00:35:18,920 --> 00:35:22,159
for profit or there was some other logic behind it?

583
00:35:22,599 --> 00:35:25,320
And the second question is can I take a selfie

584
00:35:25,320 --> 00:35:25,559
with you?

585
00:35:30,159 --> 00:35:30,360
Speaker 4: Well?

586
00:35:30,360 --> 00:35:32,280
Speaker 3: I think there might be quite a line on the

587
00:35:32,320 --> 00:35:37,199
selfie front. But honestly, selfie is the vein of my existence.

588
00:35:39,960 --> 00:35:51,840
Speaker 4: So but sure. So the issue with the way.

589
00:35:51,679 --> 00:35:55,760
Speaker 3: That blue checks were handed out historically at Twitter was

590
00:35:55,760 --> 00:36:01,719
was not in a not always in a sensible way

591
00:36:01,880 --> 00:36:06,599
or not always in a a good way.

592
00:36:05,480 --> 00:36:05,880
Speaker 4: And so.

593
00:36:07,840 --> 00:36:09,800
Speaker 3: And the same check was given out whether you were

594
00:36:10,000 --> 00:36:11,760
sort of a government official or.

595
00:36:13,199 --> 00:36:15,280
Speaker 4: You know, a major company or a person.

596
00:36:15,440 --> 00:36:18,119
Speaker 3: So there's noticed there was no really indistinguishment between between

597
00:36:18,280 --> 00:36:20,719
what what does a blue check mean? And and a

598
00:36:20,719 --> 00:36:22,800
lot of blue checks were you know, you could have

599
00:36:22,840 --> 00:36:25,320
been an intern at a small publication ten years ago

600
00:36:25,400 --> 00:36:27,519
and you still have a blue check. And then a

601
00:36:27,519 --> 00:36:29,880
lot of them were for sale and not not for

602
00:36:29,920 --> 00:36:32,000
sale on the dark web I'm talking about you could

603
00:36:32,000 --> 00:36:35,599
google blue check for sale and buy a blue check.

604
00:36:36,360 --> 00:36:37,039
Speaker 4: Anyone could.

605
00:36:38,119 --> 00:36:40,280
Speaker 3: And there was some corruption within Twitter as well in

606
00:36:40,360 --> 00:36:43,159
terms of paying people at Twitter for blue checks that

607
00:36:43,360 --> 00:36:45,840
was not appropriate. So the clean question is, how do

608
00:36:45,920 --> 00:36:49,840
we clear all that out and how do we create,

609
00:36:50,920 --> 00:36:55,039
you know, within recent a level playing field so that

610
00:36:57,599 --> 00:37:01,719
like you said, the average citizen can be a journalist

611
00:37:02,360 --> 00:37:04,599
and and and and that and that we really want

612
00:37:04,639 --> 00:37:07,599
to judge journalism, We want to judge stories based on

613
00:37:07,639 --> 00:37:11,599
the quality of the content, not the publication that is

614
00:37:11,679 --> 00:37:17,360
behind it. And there are some incredibly good individuals who

615
00:37:17,360 --> 00:37:19,199
may who maybe couldn't get a job at at a

616
00:37:19,239 --> 00:37:22,880
major newspaper, but actually they are great writers and they're incredibly.

617
00:37:22,519 --> 00:37:27,599
Speaker 4: Truthful and good. So it was really just to kind

618
00:37:27,599 --> 00:37:28,519
of level the playing field.

619
00:37:28,559 --> 00:37:32,000
Speaker 3: And the thing about the you know, you can get

620
00:37:32,039 --> 00:37:33,639
one for like seven dollars a month if you do

621
00:37:33,679 --> 00:37:35,760
it on the web, which is it's not a lot,

622
00:37:36,840 --> 00:37:41,480
but it it gives us the the the added element

623
00:37:41,679 --> 00:37:45,920
of requiring a credit card and a phone number from

624
00:37:46,159 --> 00:37:52,119
a reputable carrier. And so this this this is a

625
00:37:53,199 --> 00:37:55,440
I mean, we're gonna get kind of technical here, but

626
00:37:55,880 --> 00:38:00,199
it's a it's a necessary defense against bots.

627
00:38:00,679 --> 00:38:03,599
Speaker 4: So today with advanced AI.

628
00:38:04,039 --> 00:38:08,079
Speaker 3: They can pass every kind of test for a human.

629
00:38:09,559 --> 00:38:12,920
So you can actually create on one computer one hundred

630
00:38:12,960 --> 00:38:17,000
thousand accounts. So then that all sound human and pass

631
00:38:17,079 --> 00:38:19,800
every human test. So how do you know which one's real?

632
00:38:20,920 --> 00:38:23,800
So effectively, by charging a small amount of money and

633
00:38:23,880 --> 00:38:26,440
requiring a credit card and a phone number, we increase

634
00:38:26,519 --> 00:38:30,760
the cost of a fake account by literally one thousand,

635
00:38:30,880 --> 00:38:35,159
if not ten thousand, and so and my prediction is

636
00:38:35,719 --> 00:38:39,000
that any social media company that does not require a

637
00:38:39,000 --> 00:38:42,039
small amount of money and does not do verification will

638
00:38:42,039 --> 00:38:45,440
cease to be relevant.

639
00:38:45,960 --> 00:38:48,440
Speaker 4: I mean, I can go to talk forever if you want.

640
00:38:50,360 --> 00:38:52,199
Speaker 1: One more, one more.

641
00:38:52,960 --> 00:38:56,800
Speaker 4: So it's not up to me, it's really up to you, guys.

642
00:38:56,800 --> 00:38:58,519
I'm happy to talk longer if you want.

643
00:39:00,199 --> 00:39:02,960
Speaker 8: My name is Carol christ and I had the great

644
00:39:03,000 --> 00:39:05,960
honor to serve as First Lady of Florida, and it's

645
00:39:06,119 --> 00:39:08,199
an incredible honor to be here with all of you

646
00:39:08,280 --> 00:39:11,960
and this beautiful conference, and this is an extraordinary panel

647
00:39:12,039 --> 00:39:14,960
and this discussion and the great work you do for

648
00:39:15,480 --> 00:39:19,960
humanity or planet, and I'm just so impressed with everything

649
00:39:19,960 --> 00:39:22,480
you're doing at Twitter and how open you are with

650
00:39:22,519 --> 00:39:24,920
everything you're doing. I think it's a great service to

651
00:39:25,440 --> 00:39:28,800
our country in the world, and I really am very

652
00:39:28,800 --> 00:39:32,960
grateful to you for that. And it may be a

653
00:39:32,960 --> 00:39:35,480
little depressing, but we do have an election you're coming up,

654
00:39:35,519 --> 00:39:36,920
and it's an incredibly.

655
00:39:36,440 --> 00:39:38,440
Speaker 1: Important well politics.

656
00:39:40,440 --> 00:39:44,000
Speaker 8: The world is watching and it's an incredibly important time.

657
00:39:44,440 --> 00:39:47,280
A lot is at stake, and I would love to

658
00:39:47,360 --> 00:39:50,840
know what you I'm sure you have a lot going

659
00:39:50,840 --> 00:39:53,519
on and I would love to hear some say and

660
00:39:53,559 --> 00:39:55,880
what we can look forward to hearing from Twitter.

661
00:39:56,079 --> 00:39:56,880
Speaker 1: Thank you so much.

662
00:39:58,360 --> 00:40:00,159
Speaker 3: Yeah, I mean, I do think it's very importan for

663
00:40:00,239 --> 00:40:04,719
elections domestically and internationally that we have an open and

664
00:40:04,760 --> 00:40:05,960
transparent platform.

665
00:40:06,199 --> 00:40:08,480
Speaker 4: That's why I think it's the open sourcing of.

666
00:40:08,440 --> 00:40:14,440
Speaker 3: The algorithm and uh, complete transparency is essential so that

667
00:40:14,480 --> 00:40:16,639
people know what's going on, they know that something is

668
00:40:16,679 --> 00:40:23,199
not being artificially suppressed or amplified. And so we and

669
00:40:23,199 --> 00:40:26,599
and as as I mentioned, when we open source uh,

670
00:40:26,920 --> 00:40:29,840
the recommendation algorithm. Now there's still other elements of Twitter

671
00:40:29,840 --> 00:40:32,719
that haven't been open sourced, but will be that. We've

672
00:40:33,199 --> 00:40:39,920
the public found many errors in our recommendation platform that

673
00:40:40,000 --> 00:40:44,559
we fixed, and we'll keep doing that. The what really

674
00:40:44,639 --> 00:40:49,119
matters is does the is there both the reality and

675
00:40:49,199 --> 00:40:53,159
the perception of a platform that the public believes they

676
00:40:53,199 --> 00:40:57,840
can that they can trust, and and and and I,

677
00:40:58,000 --> 00:41:00,000
like you said, if you don't have trans in anything,

678
00:41:00,119 --> 00:41:03,039
that's not any platform that does not have transparency, I

679
00:41:03,079 --> 00:41:04,320
wouldn't trust it with it, damn.

680
00:41:06,480 --> 00:41:09,840
Speaker 2: So I think what we heard today is some really

681
00:41:10,519 --> 00:41:15,760
important and profound things. First of all, Elon has committed

682
00:41:16,239 --> 00:41:21,119
to being accessible to everyone for continual feedback. He's also

683
00:41:21,840 --> 00:41:27,079
opened up himself to also participate in the new transparency

684
00:41:27,440 --> 00:41:31,880
and safety rules that they posted yesterday. Just remember right,

685
00:41:32,440 --> 00:41:36,159
freedom of speech doesn't mean freedom of reach. And if

686
00:41:36,679 --> 00:41:40,880
freedom of speech, as he says, is the bedrock of

687
00:41:41,480 --> 00:41:45,199
this country, I'm not sure there's anyone in this room

688
00:41:45,719 --> 00:41:48,519
who could disagree with that. Could I get a round

689
00:41:48,559 --> 00:41:54,280
of applause for that? But I want to be clear.

690
00:41:54,760 --> 00:41:58,280
The path here today wasn't without.

691
00:41:59,320 --> 00:42:02,119
Speaker 1: A lot of chat. Because this guy is here, there's

692
00:42:02,159 --> 00:42:03,159
always a lot of chatter.

693
00:42:03,559 --> 00:42:06,719
Speaker 2: But I believe if this is a marketing conference for

694
00:42:07,079 --> 00:42:12,199
marketing executives that influence culture in this country, it is

695
00:42:12,199 --> 00:42:16,559
the responsibility of everybody in this room to offer a

696
00:42:16,599 --> 00:42:18,119
helping hand and.

697
00:42:18,119 --> 00:42:19,559
Speaker 1: To say how can we help, how can we make

698
00:42:19,559 --> 00:42:23,199
it better? So I say we're.

699
00:42:22,920 --> 00:42:26,320
Speaker 2: Open for solutions, and you've got a man who's committed

700
00:42:26,599 --> 00:42:30,360
to try his best not to tweet after three am,

701
00:42:30,599 --> 00:42:32,679
but open to your feedback. And I'd like to really

702
00:42:32,679 --> 00:42:34,679
thank everyone who's spending their time here today.

703
00:42:35,159 --> 00:42:36,719
Speaker 4: I am, I'd be you