1
00:00:04,440 --> 00:00:12,360
Speaker 1: Welcome to tech Stuff, a production from iHeartRadio. Hey there,

2
00:00:12,360 --> 00:00:15,840
Speaker 1: and welcome to tech Stuff. I'm your host, Jonathan Strickland.

3
00:00:15,880 --> 00:00:19,040
Speaker 1: I'm an executive producer with iHeart Podcasts and How the

4
00:00:19,120 --> 00:00:24,720
Speaker 1: Tech Are Yet Now. Normally I say tech news items

5
00:00:24,760 --> 00:00:28,960
Speaker 1: for Tuesdays and Thursdays, But over this past weekend a

6
00:00:29,080 --> 00:00:33,919
Speaker 1: pretty big sequence of events happened and it really merits

7
00:00:34,200 --> 00:00:37,080
Speaker 1: a deeper discussion. I'm sure most of you have at

8
00:00:37,159 --> 00:00:42,000
Speaker 1: least heard something about this story. The short version is

9
00:00:42,200 --> 00:00:46,839
Speaker 1: Sam Altman, the CEO of open Ai, received his walking

10
00:00:46,920 --> 00:00:51,360
Speaker 1: papers from the company's board of directors. Then the board

11
00:00:51,720 --> 00:00:54,560
Speaker 1: kind of flipped out and begged him to come back

12
00:00:54,600 --> 00:00:58,560
Speaker 1: to the company, and he ultimately decided now I'm good,

13
00:00:59,160 --> 00:01:02,320
Speaker 1: y'all can do this on your own. And last I heard,

14
00:01:02,600 --> 00:01:06,840
Speaker 1: he has now joined Microsoft's Advanced AI department. Which is

15
00:01:06,840 --> 00:01:10,200
Speaker 1: a heck of a weekend. So today I thought we

16
00:01:10,240 --> 00:01:12,240
Speaker 1: would talk a bit about Altman, We talk a bit

17
00:01:12,280 --> 00:01:16,000
Speaker 1: about open Ai, We chat about what went down behind

18
00:01:16,040 --> 00:01:20,360
Speaker 1: closed doors this past Friday, why the board of directors

19
00:01:20,600 --> 00:01:25,039
Speaker 1: fired Altman, why they then switched gears so quickly, and

20
00:01:25,080 --> 00:01:28,280
Speaker 1: what it all means going forward. Now. I did an

21
00:01:28,280 --> 00:01:31,960
Speaker 1: episode titled the Story of Open AI at the beginning

22
00:01:32,000 --> 00:01:35,040
Speaker 1: of this year, which published in January. I am going

23
00:01:35,080 --> 00:01:37,399
Speaker 1: to retread a lot of that same ground here in

24
00:01:37,480 --> 00:01:41,800
Speaker 1: a slightly different context, because it's necessary to really unravel

25
00:01:41,880 --> 00:01:45,040
Speaker 1: what was going on over the last weekend. So first up,

26
00:01:46,000 --> 00:01:50,120
Speaker 1: who is Sam Altman. Well, he grew up in Saint Louis, Missouri,

27
00:01:50,640 --> 00:01:54,240
Speaker 1: and as a kid he became really interested in programming.

28
00:01:54,360 --> 00:01:57,279
Speaker 1: According to The New Yorker, he was programming as early

29
00:01:57,400 --> 00:02:00,240
Speaker 1: as age eight, and that he went so far to

30
00:02:00,240 --> 00:02:02,840
Speaker 1: take a part a Macintosh computer in order to learn

31
00:02:02,840 --> 00:02:06,840
Speaker 1: how it worked. He also challenged social restrictions and taboos.

32
00:02:07,240 --> 00:02:10,640
Speaker 1: When a Christian group announced a boycott for an assembly

33
00:02:10,680 --> 00:02:14,000
Speaker 1: that was supposed to be focused on sexuality, Altman came

34
00:02:14,040 --> 00:02:17,280
Speaker 1: out to his community as gay, and he challenged them

35
00:02:17,280 --> 00:02:20,720
Speaker 1: to adopt an open attitude toward different ideas. And he

36
00:02:20,760 --> 00:02:23,880
Speaker 1: was just a teenager at the time, so very much

37
00:02:24,639 --> 00:02:32,400
Speaker 1: someone who is curious, motivated, and by most accounts I've read, fearless.

38
00:02:33,080 --> 00:02:36,640
Speaker 1: Altman attended Stanford, but he was only there for two years.

39
00:02:36,800 --> 00:02:39,280
Speaker 1: He studied computer science while he was there. In fact,

40
00:02:39,280 --> 00:02:43,160
Speaker 1: he studied artificial intelligence by some of the leading thinkers

41
00:02:43,360 --> 00:02:46,440
Speaker 1: in the discipline at Stanford, but he dropped out of

42
00:02:46,440 --> 00:02:48,560
Speaker 1: college in order to work on an app and a

43
00:02:48,600 --> 00:02:51,600
Speaker 1: business idea. So in many ways, it was the stereotypical

44
00:02:52,040 --> 00:02:55,520
Speaker 1: founder story of Silicon Valley. Right. You go to Stanford,

45
00:02:55,840 --> 00:02:58,760
Speaker 1: and when you're there, you're really there to make connections.

46
00:02:59,120 --> 00:03:02,280
Speaker 1: You don't bother completing your studies, You drop out of school,

47
00:03:02,760 --> 00:03:05,839
Speaker 1: you make a company, and then you get rich. It's

48
00:03:05,960 --> 00:03:08,119
Speaker 1: kind of like, you know, the whole idea of step one,

49
00:03:08,760 --> 00:03:12,360
Speaker 1: go to Stanford, step two, drop out, step three profit.

50
00:03:13,639 --> 00:03:15,720
Speaker 1: I guess you could argue that if you can make

51
00:03:15,720 --> 00:03:19,639
Speaker 1: a successful tech business, there's no real point to completing

52
00:03:19,639 --> 00:03:21,720
Speaker 1: your studies. I mean, if your studies are all about

53
00:03:22,360 --> 00:03:25,480
Speaker 1: learning the technology and it turns out you already have

54
00:03:25,520 --> 00:03:27,680
Speaker 1: a good mastery of that and you can make a

55
00:03:27,720 --> 00:03:31,480
Speaker 1: profitable business, why would you continue to spend money going

56
00:03:31,520 --> 00:03:35,400
Speaker 1: to school, unless, of course, maybe you would grow more

57
00:03:35,400 --> 00:03:38,880
Speaker 1: as a person and develop a deeper appreciation and understanding

58
00:03:38,920 --> 00:03:42,480
Speaker 1: of things that could perhaps help you when you make

59
00:03:42,880 --> 00:03:45,600
Speaker 1: decisions in the future. But don't listen to me. I

60
00:03:45,640 --> 00:03:48,240
Speaker 1: graduated with a degree in the humanities, so I have

61
00:03:48,320 --> 00:03:52,520
Speaker 1: these wacky ideas about how the experience of college is

62
00:03:52,840 --> 00:03:56,800
Speaker 1: about more than just learning a subject. But that's beside

63
00:03:56,800 --> 00:03:59,120
Speaker 1: the point. Let's get back to Sam Altman. So the

64
00:03:59,240 --> 00:04:02,240
Speaker 1: app he was working on with a few friends was

65
00:04:02,280 --> 00:04:07,440
Speaker 1: called Looped loopt, and it was meant to let users

66
00:04:07,480 --> 00:04:11,760
Speaker 1: to share their location data with selected other users, So

67
00:04:11,800 --> 00:04:14,160
Speaker 1: you can make friends with people on the app, and

68
00:04:14,200 --> 00:04:18,200
Speaker 1: then you could share your precise location with that person,

69
00:04:18,680 --> 00:04:21,039
Speaker 1: kind of a shorthand way of saying here I am.

70
00:04:21,400 --> 00:04:24,360
Speaker 1: And it could facilitate stuff like real world meetups. Like

71
00:04:24,480 --> 00:04:27,920
Speaker 1: imagine that you're heading to a concert venue and it's

72
00:04:27,920 --> 00:04:30,560
Speaker 1: a big venue. There's a lot of different interests and stuff.

73
00:04:30,640 --> 00:04:32,440
Speaker 1: You get there, you're going to meet up with your friends.

74
00:04:32,800 --> 00:04:35,720
Speaker 1: You use this app to say this is specifically where

75
00:04:35,720 --> 00:04:37,920
Speaker 1: I am, so that you can find each other. That's

76
00:04:38,000 --> 00:04:41,360
Speaker 1: kind of a use case. The Looped team applied to

77
00:04:41,760 --> 00:04:46,799
Speaker 1: the Why Combinator accelerator company to become part of their program.

78
00:04:47,160 --> 00:04:50,600
Speaker 1: So let's talk about Why Combinator for a moment. It is,

79
00:04:50,720 --> 00:04:55,039
Speaker 1: as I said, a startup accelerator organization. So Why Combinator

80
00:04:55,400 --> 00:04:59,680
Speaker 1: its purpose is to provide early funding in promising startup

81
00:04:59,680 --> 00:05:03,119
Speaker 1: ideas in order for them to start to get off

82
00:05:03,279 --> 00:05:07,640
Speaker 1: the ground. So it's an early investment part of a

83
00:05:07,720 --> 00:05:10,479
Speaker 1: startup so that they can at least get a chance

84
00:05:10,560 --> 00:05:14,839
Speaker 1: to mature into an actual business. Now, in return for

85
00:05:14,920 --> 00:05:19,760
Speaker 1: this early investment, why Combinator takes a small percentage of

86
00:05:19,800 --> 00:05:24,960
Speaker 1: ownership in the startup. So let's say that why Combinator

87
00:05:25,000 --> 00:05:28,799
Speaker 1: provides you know, a fairly modest sum in the early days,

88
00:05:28,839 --> 00:05:30,880
Speaker 1: Like it was a lot of money, don't get me wrong,

89
00:05:30,960 --> 00:05:33,200
Speaker 1: Like maybe like one hundred thousand dollars or maybe one

90
00:05:33,240 --> 00:05:35,719
Speaker 1: hundred and twenty thousand dollars. That's a lot of money,

91
00:05:35,960 --> 00:05:38,159
Speaker 1: but it's a tiny amount when you think about what

92
00:05:38,279 --> 00:05:40,920
Speaker 1: a company needs to actually run. So this is really

93
00:05:41,120 --> 00:05:44,440
Speaker 1: just to get a startup to go from idea to

94
00:05:44,560 --> 00:05:50,120
Speaker 1: something slightly more you know, coherent. But in return, why

95
00:05:50,160 --> 00:05:55,080
Speaker 1: Combinator gets like, you know, seven percent ownership of that startup. Now,

96
00:05:55,160 --> 00:05:58,560
Speaker 1: let's say that startup is something like Dropbox, and then

97
00:05:58,920 --> 00:06:00,960
Speaker 1: years down the road it's worth you know, more than

98
00:06:01,000 --> 00:06:03,839
Speaker 1: ten billion dollars. Well, that becomes a heck of a

99
00:06:03,880 --> 00:06:07,919
Speaker 1: return on investment. Right, you can have a huge profit

100
00:06:08,080 --> 00:06:11,320
Speaker 1: as this startup accelerator, even if just a few of

101
00:06:11,320 --> 00:06:14,880
Speaker 1: the startups really hit it big. Ideally, you want them

102
00:06:14,960 --> 00:06:17,479
Speaker 1: all to hit big and then you get huge payouts

103
00:06:17,520 --> 00:06:22,279
Speaker 1: down the line, and you become an important part of

104
00:06:22,360 --> 00:06:25,839
Speaker 1: the whole tech startup ecosystem, which is exactly what why

105
00:06:25,920 --> 00:06:28,919
Speaker 1: Combinator said up to do now why Combinator launched in

106
00:06:28,920 --> 00:06:30,760
Speaker 1: two thousand and five, that was the same year that

107
00:06:30,880 --> 00:06:36,760
Speaker 1: Looped would become part of its inaugural class of startups. Essentially, now,

108
00:06:36,800 --> 00:06:40,880
Speaker 1: not all startups make it. Obviously, Looped at least appeared

109
00:06:40,960 --> 00:06:44,400
Speaker 1: to do well initially, at least on paper. Altman and

110
00:06:44,440 --> 00:06:48,360
Speaker 1: his colleagues secured a couple of major rounds of investment funding,

111
00:06:48,440 --> 00:06:52,280
Speaker 1: Series A and Series B. The company's valuation hit more

112
00:06:52,320 --> 00:06:55,480
Speaker 1: than one hundred and seventy million dollars. But they were

113
00:06:55,520 --> 00:06:59,679
Speaker 1: running into a tiny little problem. They had developed this app,

114
00:07:00,200 --> 00:07:03,279
Speaker 1: and they couldn't convince people to use it. They were

115
00:07:03,880 --> 00:07:06,000
Speaker 1: all thinking that, you know, Looped was going to be

116
00:07:06,000 --> 00:07:09,000
Speaker 1: this really useful and popular tool. Everyone was going to

117
00:07:09,080 --> 00:07:12,600
Speaker 1: download it. But turns out the general public didn't seem

118
00:07:12,640 --> 00:07:17,120
Speaker 1: to agree with that, and so in twenty twelve, Looped

119
00:07:17,600 --> 00:07:21,240
Speaker 1: that team accepted an offer from the company green Dot,

120
00:07:21,760 --> 00:07:25,200
Speaker 1: and they sold Looped for around forty three million dollars.

121
00:07:25,760 --> 00:07:27,840
Speaker 1: That's also a lot of money, but it did not

122
00:07:28,040 --> 00:07:32,040
Speaker 1: cover the amount of money that venture capitalists had invested

123
00:07:32,240 --> 00:07:35,600
Speaker 1: into looped, so it was a negative return for investors.

124
00:07:36,200 --> 00:07:38,239
Speaker 1: You know, sometimes you bet on the ponies and you lose.

125
00:07:38,720 --> 00:07:41,880
Speaker 1: But Altman walked away with around five million bucks, so

126
00:07:42,160 --> 00:07:45,200
Speaker 1: it was a pretty decent return for him, even though

127
00:07:45,840 --> 00:07:48,960
Speaker 1: the app that he had worked on for several years

128
00:07:49,000 --> 00:07:52,560
Speaker 1: never really gained traction. Now, in the wake of this disappointment,

129
00:07:52,720 --> 00:07:56,280
Speaker 1: Altman founded a venture capital company of his own, and

130
00:07:56,320 --> 00:07:59,800
Speaker 1: it was called Hydrazine Capital. So he sunk most of

131
00:07:59,800 --> 00:08:02,400
Speaker 1: his personal wealth that he had earned from this sale

132
00:08:03,080 --> 00:08:06,440
Speaker 1: into this new venture capital company, and he also raised

133
00:08:06,480 --> 00:08:10,520
Speaker 1: millions more from other investors. He focused on investing in

134
00:08:10,560 --> 00:08:14,240
Speaker 1: companies that were in the y Combinator program, and he

135
00:08:14,360 --> 00:08:17,960
Speaker 1: was largely successful in this. He was picking some really

136
00:08:18,000 --> 00:08:20,920
Speaker 1: good startups, and he was backing ones that would become

137
00:08:20,960 --> 00:08:22,760
Speaker 1: a big deal in the tech space a few years

138
00:08:22,840 --> 00:08:25,520
Speaker 1: down the road, and so he was seeing big returns

139
00:08:25,520 --> 00:08:29,280
Speaker 1: on those investments. Within just a few years, his venture

140
00:08:29,320 --> 00:08:32,920
Speaker 1: capital firm increased in value by an order of magnitude.

141
00:08:33,679 --> 00:08:37,959
Speaker 1: But Altman wasn't super happy doing this work. He didn't

142
00:08:38,000 --> 00:08:41,960
Speaker 1: find it rewarding on a personal level. Financially sure, but

143
00:08:42,080 --> 00:08:44,680
Speaker 1: on a personal level, he didn't really like the work,

144
00:08:45,280 --> 00:08:49,520
Speaker 1: so he then extricated himself from the venture capital company

145
00:08:49,679 --> 00:08:53,440
Speaker 1: to try and do something else. Now, around this same time,

146
00:08:53,480 --> 00:08:56,680
Speaker 1: the folks who were behind why Combinator were looking to

147
00:08:56,840 --> 00:08:59,839
Speaker 1: hand off the whole accelerator program to someone else to

148
00:09:00,120 --> 00:09:03,640
Speaker 1: lead it, and that someone else ended up being Sam Altman.

149
00:09:04,440 --> 00:09:07,720
Speaker 1: The guy had gone through the process with looped and

150
00:09:07,760 --> 00:09:10,480
Speaker 1: now he would be running it, and he reportedly agreed

151
00:09:10,520 --> 00:09:13,120
Speaker 1: without hesitation. He was eager to do this job. It

152
00:09:13,160 --> 00:09:15,800
Speaker 1: was something that he didn't even necessarily know he wanted

153
00:09:15,800 --> 00:09:18,400
Speaker 1: to do, but once he was offered it, he was

154
00:09:18,600 --> 00:09:22,280
Speaker 1: really enthusiastic about doing it. So he really ramped up

155
00:09:22,400 --> 00:09:26,000
Speaker 1: y Combinator, and Altman began recruiting, you know, startups that

156
00:09:26,000 --> 00:09:29,680
Speaker 1: were focused on science and technology, so we're talking about

157
00:09:29,720 --> 00:09:34,240
Speaker 1: bleeding edge stuff like quantum computing or nuclear power or

158
00:09:34,600 --> 00:09:39,360
Speaker 1: AI that kind of stuff. Around this time, he started

159
00:09:39,440 --> 00:09:43,560
Speaker 1: to be part of a group that included Elon Musk,

160
00:09:44,280 --> 00:09:48,720
Speaker 1: and this group ostensibly wanted to develop artificial intelligence in

161
00:09:48,760 --> 00:09:53,240
Speaker 1: a responsible, accountable way while being extra careful not to

162
00:09:53,320 --> 00:09:58,440
Speaker 1: do anything foolish to make safe artificial intelligence. You know,

163
00:09:58,520 --> 00:10:01,080
Speaker 1: they didn't want to go down the wrong path and

164
00:10:01,120 --> 00:10:04,959
Speaker 1: do something like accidentally unleash Skynet and terminators all over

165
00:10:05,000 --> 00:10:10,640
Speaker 1: the place, so this group would become open ai. Now,

166
00:10:10,640 --> 00:10:15,240
Speaker 1: the original open ai was a non profit organization, and

167
00:10:15,280 --> 00:10:18,080
Speaker 1: the whole idea was to help in this effort to

168
00:10:18,280 --> 00:10:22,520
Speaker 1: foster the development of artificial intelligence in a responsible and

169
00:10:22,600 --> 00:10:26,600
Speaker 1: safe way. It wasn't some for profit company pushing a

170
00:10:26,720 --> 00:10:30,600
Speaker 1: generative AI chatbot at that time, and it was not

171
00:10:30,720 --> 00:10:36,160
Speaker 1: yet a partner to massive companies like Microsoft. So Altman

172
00:10:36,320 --> 00:10:40,000
Speaker 1: was running y Combinator, and he also began to work

173
00:10:40,040 --> 00:10:43,920
Speaker 1: with the open ai folks and tried to recruit various

174
00:10:44,080 --> 00:10:49,440
Speaker 1: leaders in artificial intelligence to join open ai. In twenty fifteen,

175
00:10:49,800 --> 00:10:53,199
Speaker 1: Sam Altman published a two part blog post about machine

176
00:10:53,200 --> 00:10:57,880
Speaker 1: intelligence and quote why you should fear it end quote,

177
00:10:58,480 --> 00:11:01,160
Speaker 1: And it starts off with the comfort sentence and I

178
00:11:01,320 --> 00:11:08,559
Speaker 1: quote development of superhuman machine intelligence SMI is probably the

179
00:11:08,600 --> 00:11:13,520
Speaker 1: greatest threat to the continued existence of humanity end quote.

180
00:11:13,720 --> 00:11:18,240
Speaker 1: Now Altman allows that other massive threats, like say, an

181
00:11:18,320 --> 00:11:23,280
Speaker 1: asteroid hitting the Earth, are possibly more likely to happen

182
00:11:23,640 --> 00:11:27,480
Speaker 1: than super human intelligent machines run amok, but he also

183
00:11:27,559 --> 00:11:30,520
Speaker 1: points out that a lot of the other threats that

184
00:11:30,600 --> 00:11:34,840
Speaker 1: we think of, like super volcanoes and climate crisis might

185
00:11:35,000 --> 00:11:38,160
Speaker 1: end up having a massive impact on the human population,

186
00:11:38,360 --> 00:11:43,079
Speaker 1: but probably wouldn't just totally wipe out humans in totality.

187
00:11:44,160 --> 00:11:51,080
Speaker 1: But superhuman machine intelligence, he argued, did have that potential,

188
00:11:51,960 --> 00:11:54,000
Speaker 1: and that this is why he thought of it as

189
00:11:54,040 --> 00:11:59,480
Speaker 1: being the most important or perhaps most dangerous threat. Almand

190
00:11:59,520 --> 00:12:01,839
Speaker 1: goes on in those blog posts to point out that

191
00:12:02,000 --> 00:12:04,840
Speaker 1: SMI doesn't have to be malevolent to be a threat.

192
00:12:05,640 --> 00:12:09,000
Speaker 1: Just setting an SMI to complete a task like trying

193
00:12:09,000 --> 00:12:12,720
Speaker 1: to manage resources could end up causing massive human harm.

194
00:12:13,520 --> 00:12:17,079
Speaker 1: The SMI might determine that the biggest cause of resource

195
00:12:17,080 --> 00:12:19,920
Speaker 1: depletion is the human race, So presto, you get rid

196
00:12:19,960 --> 00:12:21,959
Speaker 1: of the people, and now you don't have to worry

197
00:12:22,000 --> 00:12:25,760
Speaker 1: about these resources running out anymore. Now that's an oversimplification,

198
00:12:25,840 --> 00:12:28,800
Speaker 1: but you get the idea. Altman's point is if you

199
00:12:29,000 --> 00:12:33,840
Speaker 1: don't develop artificial intelligence in a way that is safe,

200
00:12:34,400 --> 00:12:37,640
Speaker 1: you can get terrible consequences, whether that was your goal

201
00:12:37,840 --> 00:12:41,439
Speaker 1: or otherwise. And of course there are malicious ways to

202
00:12:41,520 --> 00:12:44,800
Speaker 1: use AI. Right You could develop AI in an effort

203
00:12:44,880 --> 00:12:48,360
Speaker 1: to try and come up with new biological weapons. For example,

204
00:12:48,440 --> 00:12:53,439
Speaker 1: that's often a scenario that's cited by concerned critics of

205
00:12:53,520 --> 00:12:58,080
Speaker 1: artificial intelligence. That's certainly something that could potentially happen. So

206
00:12:58,120 --> 00:13:01,200
Speaker 1: again Altman saying, well, you need to have the right

207
00:13:01,280 --> 00:13:05,520
Speaker 1: team responsible to develop AI in a way that is

208
00:13:05,559 --> 00:13:09,320
Speaker 1: most likely to benefit people and to protect people from

209
00:13:09,880 --> 00:13:15,320
Speaker 1: malicious or badly designed AI. So Altman makes an argument

210
00:13:15,400 --> 00:13:19,840
Speaker 1: that machine intelligence could hit an inflection point once recursive

211
00:13:19,880 --> 00:13:25,200
Speaker 1: self improvement becomes a real possibility. That means, if we

212
00:13:25,240 --> 00:13:28,080
Speaker 1: get to the point where we can create machines that

213
00:13:28,160 --> 00:13:32,559
Speaker 1: are smart enough to reprogram themselves, and to reprogram themselves

214
00:13:32,600 --> 00:13:35,480
Speaker 1: in a way that is better than what humans could do,

215
00:13:36,160 --> 00:13:40,920
Speaker 1: so to program these machines at a higher than human

216
00:13:41,000 --> 00:13:45,200
Speaker 1: level of capability, a superhuman capability, if you will, then

217
00:13:45,360 --> 00:13:48,520
Speaker 1: machines suddenly engage in self improvement and can do so

218
00:13:48,600 --> 00:13:52,440
Speaker 1: at increasingly shorter intervals. They get better at doing the

219
00:13:52,480 --> 00:13:55,840
Speaker 1: thing that they're doing, so they get better at improving themselves,

220
00:13:56,080 --> 00:13:58,880
Speaker 1: and they improve themselves over and over, and this becomes

221
00:13:58,920 --> 00:14:02,920
Speaker 1: a version of the singularity, which is a moment where

222
00:14:03,080 --> 00:14:06,200
Speaker 1: change is so sudden and it's happening all the time

223
00:14:06,400 --> 00:14:09,800
Speaker 1: that effectively it becomes impossible to even describe the present,

224
00:14:10,600 --> 00:14:13,640
Speaker 1: everything will change and continue to change at a rate

225
00:14:13,720 --> 00:14:18,440
Speaker 1: that's beyond our ability to describe. Altman says, we might

226
00:14:18,440 --> 00:14:21,240
Speaker 1: be creeping toward that now, and maybe we're creeping toward

227
00:14:21,280 --> 00:14:23,080
Speaker 1: it at a rate that's just impossible for us to

228
00:14:23,120 --> 00:14:26,560
Speaker 1: notice because it's so gradual. That makes it really tricky,

229
00:14:26,680 --> 00:14:30,120
Speaker 1: because it could be that it goes from it's happening

230
00:14:30,160 --> 00:14:33,720
Speaker 1: so slowly that we can't notice it, to it's happening

231
00:14:33,800 --> 00:14:38,360
Speaker 1: so quickly that we are unable to describe it, and

232
00:14:38,400 --> 00:14:40,640
Speaker 1: that there's no point in the middle where we can say,

233
00:14:41,560 --> 00:14:45,080
Speaker 1: wait a second. So in part two of his blog posts,

234
00:14:45,120 --> 00:14:48,440
Speaker 1: Altman makes a clear argument. He says, quote, the US

235
00:14:48,560 --> 00:14:53,680
Speaker 1: government and all other governments should regulate the development of SMI.

236
00:14:54,080 --> 00:14:57,240
Speaker 1: In an ideal world, regulation would slow down the bad

237
00:14:57,280 --> 00:15:00,720
Speaker 1: guys and speed up the good guys. It seems like

238
00:15:00,760 --> 00:15:03,600
Speaker 1: what happens with the first SMI to be developed will

239
00:15:03,600 --> 00:15:08,360
Speaker 1: be very important end quote. Essentially, what Altman is arguing

240
00:15:08,360 --> 00:15:12,320
Speaker 1: here is that if ethical researchers develop a superhuman machine

241
00:15:12,360 --> 00:15:16,360
Speaker 1: intelligence first, they can employ that SMI to prevent the

242
00:15:16,400 --> 00:15:21,640
Speaker 1: development or deployment of malevolent or poorly built SMIs. So

243
00:15:21,680 --> 00:15:26,280
Speaker 1: we unleash our good guy Superman against their bad guy

244
00:15:26,440 --> 00:15:31,560
Speaker 1: General Zod or you know whatever whichever superhero supervillain pairing

245
00:15:31,600 --> 00:15:35,120
Speaker 1: you happen to like. Interestingly, this is going to come

246
00:15:35,160 --> 00:15:37,720
Speaker 1: back again when we talk about Altman and his appearances

247
00:15:37,760 --> 00:15:41,240
Speaker 1: around the world while talking about the potential for AI regulations.

248
00:15:42,160 --> 00:15:44,520
Speaker 1: Before we dive any further into this, let's take a

249
00:15:44,560 --> 00:15:47,320
Speaker 1: quick break to thank our sponsors and we'll be right back.

250
00:15:56,720 --> 00:15:59,360
Speaker 1: So Altman went so far in his blog posts to

251
00:15:59,400 --> 00:16:03,640
Speaker 1: say that he thinks, generally speaking, that tech is often overregulated,

252
00:16:04,040 --> 00:16:06,280
Speaker 1: but on the flip side, he doesn't want to live

253
00:16:06,320 --> 00:16:09,080
Speaker 1: in a world that has no regulation at all. In

254
00:16:09,120 --> 00:16:12,080
Speaker 1: some cases, you can see regulation as a necessary evil

255
00:16:12,600 --> 00:16:16,920
Speaker 1: that maybe it does slow down innovation or it has

256
00:16:17,160 --> 00:16:21,960
Speaker 1: unintended consequences, but in the absence of regulation you can

257
00:16:22,040 --> 00:16:25,280
Speaker 1: have some really poorly thought out deployments that can cause

258
00:16:25,320 --> 00:16:29,200
Speaker 1: a lot of harm. From twenty fifteen to twenty eighteen,

259
00:16:29,800 --> 00:16:35,680
Speaker 1: Open Ai operated as a nonprofit organization. The organization championed

260
00:16:35,760 --> 00:16:39,920
Speaker 1: the open part of its name, claiming that it would

261
00:16:39,960 --> 00:16:44,360
Speaker 1: freely share research and its patents with AI researchers all

262
00:16:44,400 --> 00:16:47,680
Speaker 1: around the world, all in an effort to ensure safety

263
00:16:47,720 --> 00:16:51,320
Speaker 1: in AI development. Greg Brockman, one of the co founders,

264
00:16:51,600 --> 00:16:55,600
Speaker 1: identified a short list of top AI researchers, and the

265
00:16:55,960 --> 00:16:58,560
Speaker 1: organization as a whole began to recruit several of them

266
00:16:58,560 --> 00:17:02,560
Speaker 1: to join open ai as the first employees of the organization.

267
00:17:03,040 --> 00:17:07,119
Speaker 1: The talent helped attract more talent. Some folks said they

268
00:17:07,160 --> 00:17:10,520
Speaker 1: actually joined open ai because it was where you could

269
00:17:10,520 --> 00:17:13,560
Speaker 1: work on really exciting research with the most brilliant and

270
00:17:13,760 --> 00:17:17,600
Speaker 1: talented people in the discipline, even though it would mean

271
00:17:17,640 --> 00:17:20,359
Speaker 1: you wouldn't be making as much money there as you

272
00:17:20,400 --> 00:17:23,960
Speaker 1: could somewhere else. Even high paid individuals at companies like

273
00:17:24,040 --> 00:17:27,160
Speaker 1: Google found themselves switching jobs for the chance to work

274
00:17:27,160 --> 00:17:31,239
Speaker 1: on something that they saw as important and challenging and

275
00:17:31,280 --> 00:17:35,680
Speaker 1: potentially critical to the survival of humanity. One of those

276
00:17:35,720 --> 00:17:38,840
Speaker 1: people who would also be listed as a co founder

277
00:17:38,920 --> 00:17:44,040
Speaker 1: of open ai was Ilia Sutzkiver, who would become chief

278
00:17:44,080 --> 00:17:47,959
Speaker 1: Scientist at open ai and would join the board of directors.

279
00:17:48,600 --> 00:17:52,840
Speaker 1: Musk reportedly played a critical role in recruiting Sutzkiver over

280
00:17:52,840 --> 00:17:55,120
Speaker 1: to open Ai. Like it went back and forth between

281
00:17:55,560 --> 00:17:58,080
Speaker 1: open Ai and Google, which really wanted to hold on

282
00:17:58,119 --> 00:18:01,320
Speaker 1: to him, and reportedly the Musk was a big reason

283
00:18:01,400 --> 00:18:05,440
Speaker 1: why Sutskever eventually moved over to open ai, and also

284
00:18:05,480 --> 00:18:08,639
Speaker 1: Ilia Sutzkever is one of the people who would ultimately

285
00:18:09,359 --> 00:18:12,320
Speaker 1: be part of the decision making group that fired Sam

286
00:18:12,359 --> 00:18:16,200
Speaker 1: Altman this past weekend. Anyway, we're up to twenty eighteen,

287
00:18:16,240 --> 00:18:19,280
Speaker 1: and behind the scenes there was drama a bruin, and

288
00:18:19,400 --> 00:18:23,240
Speaker 1: much of it was in the cauldron known as Elon Musk.

289
00:18:24,200 --> 00:18:26,960
Speaker 1: Not a big surprise there, right, because Elon Musk is

290
00:18:27,080 --> 00:18:30,040
Speaker 1: kind of a magnet for drama in the tech sphere

291
00:18:30,200 --> 00:18:33,440
Speaker 1: and the business sector. So Musk was on the board

292
00:18:33,440 --> 00:18:36,320
Speaker 1: of directors for open Ai, but in twenty eighteen he

293
00:18:36,600 --> 00:18:40,800
Speaker 1: left open ai entirely, and the official story was that

294
00:18:40,960 --> 00:18:44,000
Speaker 1: Musk chose to step down because of a potential conflict

295
00:18:44,040 --> 00:18:46,679
Speaker 1: of interest because there he was on the board of

296
00:18:46,680 --> 00:18:50,960
Speaker 1: directors for an organization working on artificial intelligence, but he

297
00:18:51,119 --> 00:18:55,119
Speaker 1: also was CEO of Tesla, a car company that was

298
00:18:55,119 --> 00:18:59,439
Speaker 1: pushing hard to develop and deploy autonomous driving capabilities to

299
00:18:59,480 --> 00:19:03,240
Speaker 1: the market, and autonomous driving is of course a subset

300
00:19:03,520 --> 00:19:07,560
Speaker 1: of artificial intelligence, so stepping down was the responsible thing

301
00:19:07,600 --> 00:19:10,680
Speaker 1: to do because of this potential conflict of interest between

302
00:19:10,680 --> 00:19:15,040
Speaker 1: the two companies. There was, however, more to his decision

303
00:19:15,119 --> 00:19:19,280
Speaker 1: than just that. So, according to Business Insider, Musk was

304
00:19:19,359 --> 00:19:23,760
Speaker 1: not happy with open Ay's progress. He compared it negatively

305
00:19:23,840 --> 00:19:27,000
Speaker 1: to Google. He was saying Google is spending huge amounts

306
00:19:27,000 --> 00:19:31,080
Speaker 1: of money and is getting ahead in artificial intelligence research,

307
00:19:31,520 --> 00:19:37,160
Speaker 1: and he argued Musk argued that Google, and he specifically

308
00:19:37,200 --> 00:19:42,120
Speaker 1: targeted Larry Page in this criticism, was not paying any

309
00:19:42,160 --> 00:19:45,439
Speaker 1: attention to safety, that safety was not a factor when

310
00:19:45,480 --> 00:19:49,479
Speaker 1: it came to Google's approach to artificial intelligence, and so

311
00:19:49,720 --> 00:19:52,479
Speaker 1: that was one of the things he said that it

312
00:19:52,560 --> 00:19:55,600
Speaker 1: was critical of open ai, saying you're not doing enough,

313
00:19:56,119 --> 00:19:59,240
Speaker 1: and he was kind of pointing at Sam Altman as

314
00:19:59,320 --> 00:20:01,919
Speaker 1: the reason for them, that Altman's leadership was the reason

315
00:20:02,000 --> 00:20:08,040
Speaker 1: why open ai was lagging behind. So Musk then reportedly

316
00:20:08,880 --> 00:20:14,600
Speaker 1: went to other co founders of open ai, including Sam Altman,

317
00:20:15,080 --> 00:20:18,000
Speaker 1: and essentially he said, I want to run open ai,

318
00:20:18,640 --> 00:20:21,040
Speaker 1: and he was told in no uncertain terms that this

319
00:20:21,160 --> 00:20:24,600
Speaker 1: would not happen, and so, again, according to Business Insider,

320
00:20:24,720 --> 00:20:27,960
Speaker 1: Musk decided to take his ball and leave. His ball

321
00:20:28,000 --> 00:20:31,640
Speaker 1: also included a sizable investment or donation to open ai,

322
00:20:32,680 --> 00:20:35,480
Speaker 1: so when he left, he left with a whole bunch

323
00:20:35,520 --> 00:20:38,119
Speaker 1: of money that otherwise was going to go to the

324
00:20:38,280 --> 00:20:42,520
Speaker 1: organization and didn't. Musk would later say he disagreed with

325
00:20:42,560 --> 00:20:45,720
Speaker 1: the direction of open ai and that the company wasn't

326
00:20:45,920 --> 00:20:50,560
Speaker 1: nearly as open as its name would suggest. That last

327
00:20:50,600 --> 00:20:54,080
Speaker 1: criticism happened after open Ai would create a for profit

328
00:20:54,119 --> 00:20:58,679
Speaker 1: company in twenty nineteen. Musk actually leveled the lack of

329
00:20:58,720 --> 00:21:03,040
Speaker 1: openness at open Ai that critique. Around twenty twenty, Musk

330
00:21:03,119 --> 00:21:06,960
Speaker 1: also was founding his own ai research organization and would

331
00:21:06,960 --> 00:21:10,879
Speaker 1: occasionally throw shade at open Ai and Sam Altman. And

332
00:21:11,160 --> 00:21:14,200
Speaker 1: I am not an Elon Musk fan. Most of y'all

333
00:21:14,200 --> 00:21:16,640
Speaker 1: know this. I'm not a huge fan of Elon Musk. However,

334
00:21:17,000 --> 00:21:19,360
Speaker 1: at least some of the criticisms he had toward open

335
00:21:19,400 --> 00:21:23,160
Speaker 1: Ai I actually agree with, or at least I think

336
00:21:23,320 --> 00:21:26,040
Speaker 1: they were true, like the fact that open ai was

337
00:21:26,080 --> 00:21:30,400
Speaker 1: becoming less open. I think that criticism has merit. Meanwhile,

338
00:21:30,440 --> 00:21:33,440
Speaker 1: open Ai was in a pretty tough position because, as

339
00:21:33,480 --> 00:21:39,199
Speaker 1: it turns out, artificial intelligence research is expensive, so you

340
00:21:39,280 --> 00:21:42,479
Speaker 1: need access to a whole lot of compute power and

341
00:21:42,560 --> 00:21:45,800
Speaker 1: that's not cheap, and then you also need to have

342
00:21:45,880 --> 00:21:49,280
Speaker 1: the money to attract the best talent, especially if your

343
00:21:49,320 --> 00:21:55,360
Speaker 1: goal is to be the first to develop superhuman machine

344
00:21:55,400 --> 00:22:00,440
Speaker 1: intelligence that is ethically sound. Like, if that's your goal

345
00:22:01,040 --> 00:22:04,600
Speaker 1: and you need to outpace everybody else who's also working

346
00:22:04,680 --> 00:22:08,960
Speaker 1: on developing superhuman machine intelligence, you got to spend the

347
00:22:08,960 --> 00:22:12,000
Speaker 1: big bucks to get the top of the class to

348
00:22:12,080 --> 00:22:16,399
Speaker 1: come over to your organization. And a nonprofit organization is

349
00:22:16,520 --> 00:22:18,680
Speaker 1: just not the fastest way to gather huge amounts of

350
00:22:18,720 --> 00:22:21,879
Speaker 1: money needed to fund research and operations. It would be

351
00:22:21,920 --> 00:22:25,240
Speaker 1: way easier if you could get investors to pour money in,

352
00:22:25,520 --> 00:22:29,080
Speaker 1: but investors want a return. Meanwhile, a nonprofit is a

353
00:22:29,080 --> 00:22:33,040
Speaker 1: place where you donate money. You're not expecting return on

354
00:22:33,080 --> 00:22:36,520
Speaker 1: your donation. It's no an investment. This is what led

355
00:22:36,560 --> 00:22:39,840
Speaker 1: to the decision to create a for profit arm of

356
00:22:39,880 --> 00:22:43,800
Speaker 1: open ai, which in turn would generate money that could

357
00:22:43,880 --> 00:22:47,960
Speaker 1: be theoretically at least used by the nonprofit part of

358
00:22:48,040 --> 00:22:52,320
Speaker 1: open ai to further the original organization's goals and mission.

359
00:22:52,920 --> 00:22:56,840
Speaker 1: So the result was open Ai LP, which open Ai

360
00:22:56,960 --> 00:23:00,800
Speaker 1: called a capped profit company. So what the heck is

361
00:23:00,840 --> 00:23:04,399
Speaker 1: a capped profit company? That's actually a really good question,

362
00:23:04,520 --> 00:23:10,320
Speaker 1: because I've found two somewhat conflicting answers from various sources.

363
00:23:10,400 --> 00:23:13,560
Speaker 1: Like they lay it out in two different ways that

364
00:23:13,640 --> 00:23:17,199
Speaker 1: are similar but distinct. So I'm going to give you

365
00:23:17,280 --> 00:23:19,840
Speaker 1: both of the ways that it has been explained in

366
00:23:19,960 --> 00:23:22,840
Speaker 1: various sources, because I'm gonna be honest with y'all. I'm

367
00:23:22,840 --> 00:23:25,040
Speaker 1: not a business person. Despite the fact that I have

368
00:23:25,200 --> 00:23:28,120
Speaker 1: hosted a business podcast of the Best, I'm not really

369
00:23:28,119 --> 00:23:30,760
Speaker 1: a business person, so I can't pretend like I have

370
00:23:31,240 --> 00:23:34,280
Speaker 1: a firm grip on this. And also, open ai was

371
00:23:34,359 --> 00:23:37,159
Speaker 1: kind of charting new territory while they were announcing this.

372
00:23:37,680 --> 00:23:42,120
Speaker 1: But here are the two ways that it is frequently described.

373
00:23:42,400 --> 00:23:46,840
Speaker 1: So version number one means that open ai would accept

374
00:23:46,880 --> 00:23:50,880
Speaker 1: investments from venture capitalists and that would pay out returns

375
00:23:50,880 --> 00:23:54,600
Speaker 1: on those investments from profits, but only up to a

376
00:23:54,600 --> 00:23:58,240
Speaker 1: certain amount. So in open AI's case, the early backers,

377
00:23:58,280 --> 00:24:00,640
Speaker 1: the people who first poured money in to open ai,

378
00:24:01,280 --> 00:24:04,800
Speaker 1: would have a cap of one hundred times their initial investment.

379
00:24:05,359 --> 00:24:08,880
Speaker 1: So let's take a very simple scenario. Let's say that

380
00:24:08,920 --> 00:24:11,200
Speaker 1: some kids in your neighborhood want to start a lemonade

381
00:24:11,240 --> 00:24:15,959
Speaker 1: stand and you invest one dollar into their lemonade stand. Now,

382
00:24:16,040 --> 00:24:18,320
Speaker 1: let's say the kids running the stand turn out to

383
00:24:18,320 --> 00:24:23,160
Speaker 1: be business geniuses and your dollar investment helps lead that

384
00:24:23,280 --> 00:24:29,600
Speaker 1: stand into making tens of thousands of dollars in profits, Like,

385
00:24:29,680 --> 00:24:32,560
Speaker 1: even after the expenses, these kids are raking in tens

386
00:24:32,560 --> 00:24:37,520
Speaker 1: of thousands of dollars. However, when you invested, you did

387
00:24:37,560 --> 00:24:41,359
Speaker 1: so knowing there was a one hundred time cap on returns,

388
00:24:41,760 --> 00:24:43,560
Speaker 1: So that means the most you're ever going to get

389
00:24:43,600 --> 00:24:46,440
Speaker 1: from the stand is one hundred dollars. It's a one

390
00:24:46,480 --> 00:24:50,240
Speaker 1: hundred times return on your investment. Meanwhile, those snot nosed

391
00:24:50,280 --> 00:24:52,560
Speaker 1: kids who never could have made the stand without your

392
00:24:52,640 --> 00:24:56,800
Speaker 1: dollar are pocketing thousands of bucks and they're franchising across

393
00:24:56,840 --> 00:25:01,000
Speaker 1: the town, those rotten kids. Anyway, that's one version of

394
00:25:01,040 --> 00:25:05,400
Speaker 1: how the capped profit structure work. It works, investors can

395
00:25:05,400 --> 00:25:07,680
Speaker 1: make a return, but only up to a certain amount,

396
00:25:08,359 --> 00:25:11,080
Speaker 1: the early backers being one hundred times whatever they put in,

397
00:25:11,480 --> 00:25:13,920
Speaker 1: and that means if they put in ten million dollars,

398
00:25:14,240 --> 00:25:16,920
Speaker 1: they could potentially make as much as a billion dollars

399
00:25:16,960 --> 00:25:20,919
Speaker 1: in returns if open ai profited that much. So you

400
00:25:20,960 --> 00:25:25,600
Speaker 1: know it does add up. However, there is a second

401
00:25:25,680 --> 00:25:29,840
Speaker 1: explanation for capped profit that, like I said, is slightly different.

402
00:25:30,200 --> 00:25:33,760
Speaker 1: So in this version, investors would pour money into open

403
00:25:33,800 --> 00:25:37,960
Speaker 1: ai and open ai would hold back on distributing any

404
00:25:38,040 --> 00:25:42,880
Speaker 1: returns on profits until those profits reached at least one

405
00:25:42,960 --> 00:25:47,120
Speaker 1: hundred times the investments that had been made. So using

406
00:25:47,119 --> 00:25:50,840
Speaker 1: our limonade stand example, you've donated one dollar to the

407
00:25:50,840 --> 00:25:54,199
Speaker 1: limonade stand business, you would not see a return on

408
00:25:54,240 --> 00:25:58,280
Speaker 1: that investment until the liminade stand made at least one

409
00:25:58,320 --> 00:26:02,440
Speaker 1: hundred dollars in profit. At that point you could start

410
00:26:02,440 --> 00:26:06,240
Speaker 1: to receive returns. And a few explanations kind of combine

411
00:26:06,880 --> 00:26:10,680
Speaker 1: the first version I mentioned in this version, and frankly,

412
00:26:10,800 --> 00:26:13,480
Speaker 1: just to be transparent, this kind of confuses me. So

413
00:26:13,800 --> 00:26:18,639
Speaker 1: for example, Time at time dot Com uses the second

414
00:26:18,720 --> 00:26:22,520
Speaker 1: explanation right that you don't get any returns until the

415
00:26:22,520 --> 00:26:26,040
Speaker 1: profits reach one hundred times whatever your investment level was,

416
00:26:26,600 --> 00:26:30,840
Speaker 1: but then includes the phrase quote anything above that being

417
00:26:30,880 --> 00:26:33,960
Speaker 1: the one hundred times profit would be donated back to

418
00:26:34,040 --> 00:26:37,320
Speaker 1: the nonprofit. So if that's the case, it means you

419
00:26:37,320 --> 00:26:39,840
Speaker 1: wouldn't get a return until the profits hit that one

420
00:26:39,960 --> 00:26:44,080
Speaker 1: hundred times your investment, and then anything over one hundred

421
00:26:44,080 --> 00:26:50,240
Speaker 1: times your investment would be going toward the investment into

422
00:26:50,280 --> 00:26:53,640
Speaker 1: the nonprofit or a donation into the nonprofit, which means

423
00:26:53,640 --> 00:26:56,800
Speaker 1: I guess you would be limited to one hundred times.

424
00:26:57,160 --> 00:26:59,600
Speaker 1: I don't know, like maybe it's a combination of these two,

425
00:26:59,680 --> 00:27:03,919
Speaker 1: but it's just been poorly reported in various places. It

426
00:27:04,000 --> 00:27:06,399
Speaker 1: just it seems a little confusing to me, and it

427
00:27:06,440 --> 00:27:09,480
Speaker 1: also seems like it'd be confusing from an investor standpoint

428
00:27:09,560 --> 00:27:12,159
Speaker 1: of whether or not it would even make sense to

429
00:27:12,440 --> 00:27:15,960
Speaker 1: pour money into this. I think a lot of reporting

430
00:27:16,000 --> 00:27:19,480
Speaker 1: around the cap profit nature is just incomplete, and that's

431
00:27:19,520 --> 00:27:22,560
Speaker 1: the problem that there were just there's just a lack

432
00:27:22,640 --> 00:27:26,520
Speaker 1: of good explanations of this. And also, I mean, I'm dense,

433
00:27:26,680 --> 00:27:29,520
Speaker 1: so that's the other part of the problem. But anyway,

434
00:27:30,800 --> 00:27:34,880
Speaker 1: however you frame the context of a capped profit company,

435
00:27:35,520 --> 00:27:38,440
Speaker 1: the structure would give open ai the chance to court

436
00:27:38,520 --> 00:27:42,399
Speaker 1: investors and to hold a whole bunch of money that

437
00:27:42,440 --> 00:27:46,040
Speaker 1: they could then pour into research and recruiting. A one

438
00:27:46,080 --> 00:27:50,480
Speaker 1: hundred times factor is pretty darn big, And arguably you

439
00:27:50,520 --> 00:27:53,640
Speaker 1: could say this was necessary because while open ai had

440
00:27:53,760 --> 00:27:58,080
Speaker 1: this noble mission, the truth is you still had massive

441
00:27:58,080 --> 00:28:02,959
Speaker 1: companies like Amazon and Google and Meta. These companies have

442
00:28:03,040 --> 00:28:07,600
Speaker 1: really deep pockets and a desire to invest in AI research,

443
00:28:07,960 --> 00:28:11,440
Speaker 1: and if you didn't do something, there was just no way,

444
00:28:11,520 --> 00:28:14,119
Speaker 1: no matter how noble the cause you were going to

445
00:28:14,200 --> 00:28:17,760
Speaker 1: keep up with these companies. So that was kind of

446
00:28:17,840 --> 00:28:21,520
Speaker 1: the decision making factors that drove open eye to launch

447
00:28:21,840 --> 00:28:28,560
Speaker 1: this for profit arm of the organization, and that didn't

448
00:28:28,600 --> 00:28:32,840
Speaker 1: make everybody happy. In fact, it was controversial, to put

449
00:28:32,840 --> 00:28:35,720
Speaker 1: it lightly, There were critics who were asking if it

450
00:28:35,720 --> 00:28:38,400
Speaker 1: would even be possible for open ai to continue to

451
00:28:38,440 --> 00:28:43,760
Speaker 1: pursue its mission of ethical AI development while also operating

452
00:28:43,840 --> 00:28:49,760
Speaker 1: a commercial business that was profiting off of artificial intelligence development.

453
00:28:50,280 --> 00:28:54,080
Speaker 1: That these two things could not be in alignment and

454
00:28:54,160 --> 00:28:57,240
Speaker 1: would mean that ultimately open ai would not be able

455
00:28:57,280 --> 00:29:01,120
Speaker 1: to achieve its mission. Complicating matters was that open ai

456
00:29:01,240 --> 00:29:04,360
Speaker 1: began to back away from that whole open part of

457
00:29:04,400 --> 00:29:08,240
Speaker 1: the philosophy, which again Elon Musk would criticize. In twenty twenty.

458
00:29:08,840 --> 00:29:12,840
Speaker 1: Open a Eye sighted a concern that malicious developers might

459
00:29:12,920 --> 00:29:17,240
Speaker 1: take the information the research being shared from open ai

460
00:29:17,960 --> 00:29:22,040
Speaker 1: and use that information to develop nasty and harmful applications,

461
00:29:22,240 --> 00:29:24,800
Speaker 1: or at least poorly designed ones. So now they're saying,

462
00:29:25,680 --> 00:29:29,000
Speaker 1: you know, our knowledge is dangerous. So I know, we

463
00:29:29,080 --> 00:29:30,840
Speaker 1: said we were going to share it so that we

464
00:29:30,840 --> 00:29:33,760
Speaker 1: could benefit humanity, but now we're scared that if we

465
00:29:33,840 --> 00:29:36,680
Speaker 1: share it people will misuse it, so we're not going

466
00:29:36,760 --> 00:29:39,600
Speaker 1: to do that anymore. So again, like Musk was saying,

467
00:29:40,160 --> 00:29:43,080
Speaker 1: it was no longer being as open as the name

468
00:29:43,240 --> 00:29:46,960
Speaker 1: had implied. So yeah, his criticisms had weight. Open ai

469
00:29:47,120 --> 00:29:51,880
Speaker 1: really was moving away from an unassailable nonprofit status and

470
00:29:52,000 --> 00:29:55,480
Speaker 1: also was getting less open in the process. Sure, the

471
00:29:55,480 --> 00:29:59,000
Speaker 1: folks that open ai had explanations for why they were

472
00:29:59,000 --> 00:30:01,440
Speaker 1: doing this, but it didn't change the fact that the

473
00:30:01,480 --> 00:30:06,160
Speaker 1: open Ai of twenty nineteen was fundamentally different than the

474
00:30:06,280 --> 00:30:10,560
Speaker 1: organization that had started in twenty fifteen. All Right, we're

475
00:30:10,600 --> 00:30:12,960
Speaker 1: going to take another break. When we come back, we'll

476
00:30:13,000 --> 00:30:16,600
Speaker 1: talk more about what happened in the following years that

477
00:30:16,680 --> 00:30:22,280
Speaker 1: then led to the situation that we saw unfold this

478
00:30:22,400 --> 00:30:36,600
Speaker 1: past weekend. But first, let's thank our sponsors. Okay, So

479
00:30:37,240 --> 00:30:39,800
Speaker 1: we left off in around twenty nineteen. We're gonna skip

480
00:30:39,840 --> 00:30:43,280
Speaker 1: ahead a few years. So the battle for AI talent

481
00:30:43,520 --> 00:30:46,680
Speaker 1: was a constant one in the tech space, but the

482
00:30:46,720 --> 00:30:50,000
Speaker 1: world at large remained pretty much oblivious to open ai

483
00:30:50,360 --> 00:30:54,000
Speaker 1: and the folks who were involved in that company. Open

484
00:30:54,040 --> 00:30:57,920
Speaker 1: ai just wasn't a name that your average person was

485
00:30:57,960 --> 00:31:01,480
Speaker 1: aware of. But that would change in November twenty twenty two.

486
00:31:02,520 --> 00:31:08,120
Speaker 1: That's when OpenAI introduced the chatbot called chat GPT. This

487
00:31:08,240 --> 00:31:11,920
Speaker 1: chat bot drew on a large language model, the GPT model,

488
00:31:11,920 --> 00:31:15,520
Speaker 1: which had gone through a couple of iterations, and it

489
00:31:15,560 --> 00:31:20,240
Speaker 1: would use that language model to generate responses to queries

490
00:31:20,320 --> 00:31:25,280
Speaker 1: and input. The responses often seemed like a human being

491
00:31:25,280 --> 00:31:27,640
Speaker 1: had actually written it. It didn't come across as your

492
00:31:27,680 --> 00:31:32,400
Speaker 1: typical AI generated text. It seemed more natural than that.

493
00:31:32,840 --> 00:31:35,040
Speaker 1: It also seemed like it was a really smart person

494
00:31:35,360 --> 00:31:37,560
Speaker 1: who wrote the response, someone who appeared to be an

495
00:31:37,600 --> 00:31:41,720
Speaker 1: authority on whatever the subject was. Like any subject you

496
00:31:41,760 --> 00:31:44,000
Speaker 1: could think of, you could put into this thing, and

497
00:31:44,120 --> 00:31:46,840
Speaker 1: chat gpt would generate a response that seemed to be

498
00:31:46,920 --> 00:31:52,840
Speaker 1: pretty definitive. And there were limitations on chat GPT's expertise.

499
00:31:53,280 --> 00:31:56,360
Speaker 1: The open Ai announced that chat gpt really only had

500
00:31:56,400 --> 00:31:59,720
Speaker 1: access to information leading up to September twenty twenty one,

501
00:32:00,400 --> 00:32:03,120
Speaker 1: and if you asked it to explain anything that happened

502
00:32:03,200 --> 00:32:06,240
Speaker 1: after September twenty twenty one, you'd be out of luck

503
00:32:06,440 --> 00:32:09,880
Speaker 1: because chat GPT wouldn't have access to that information. But

504
00:32:10,000 --> 00:32:14,280
Speaker 1: right out of the gate, chat gpt seemed incredible. Now

505
00:32:14,320 --> 00:32:18,400
Speaker 1: over the following weeks after its introduction, we would start

506
00:32:18,440 --> 00:32:24,040
Speaker 1: to see various critics and skeptics raise concerns about generative

507
00:32:24,080 --> 00:32:28,440
Speaker 1: AI in general, really and chat GPT in particular. Now,

508
00:32:28,520 --> 00:32:31,720
Speaker 1: some of these conversations had already started because there were

509
00:32:31,800 --> 00:32:35,880
Speaker 1: already text to image generative AI tools out there that

510
00:32:36,000 --> 00:32:41,200
Speaker 1: had prompted some concern. Chat GPT created new discussions about

511
00:32:41,240 --> 00:32:47,800
Speaker 1: how generative AI could make misinformation. They could engage in plagiarism,

512
00:32:48,200 --> 00:32:51,640
Speaker 1: it could slander someone, or it could just produce the

513
00:32:51,680 --> 00:32:56,080
Speaker 1: wrong response due to something that the AI field calls hallucinations.

514
00:32:56,720 --> 00:33:00,920
Speaker 1: Sometimes they call it confabulations instead. So this is when

515
00:33:00,920 --> 00:33:05,160
Speaker 1: an AI model fabricates an answer for whatever reason. Like

516
00:33:05,240 --> 00:33:08,200
Speaker 1: one reason that AI might just make something up is

517
00:33:08,200 --> 00:33:12,240
Speaker 1: that it doesn't have access to relevant information that relates

518
00:33:12,240 --> 00:33:16,840
Speaker 1: to the query. So instead the chatbot produces an answer that,

519
00:33:16,920 --> 00:33:22,040
Speaker 1: from a linguistic perspective, is statistically relevant. In other words,

520
00:33:22,320 --> 00:33:28,160
Speaker 1: it's creating sentences that are linguistically correct but factually incorrect

521
00:33:28,440 --> 00:33:31,800
Speaker 1: because it doesn't know the difference, and it's just trying

522
00:33:31,840 --> 00:33:34,280
Speaker 1: to provide a response to the question that was asked

523
00:33:34,320 --> 00:33:36,920
Speaker 1: of it. Now, this meant that sometimes you could ask

524
00:33:37,000 --> 00:33:39,360
Speaker 1: chat GBT to solve a problem for you, and the

525
00:33:39,400 --> 00:33:43,800
Speaker 1: response you would get would sound authoritative and sound like

526
00:33:43,880 --> 00:33:47,920
Speaker 1: it's correct, but in fact it was entirely wrong. And

527
00:33:48,320 --> 00:33:52,640
Speaker 1: following these criticisms came concerns from lawmakers who began to

528
00:33:52,720 --> 00:33:56,400
Speaker 1: ask the very same questions that open AI was intended

529
00:33:56,480 --> 00:33:59,560
Speaker 1: to address when folks first got together back in twenty

530
00:33:59,680 --> 00:34:02,240
Speaker 1: fifteen to create it in the first place. Now, as

531
00:34:02,280 --> 00:34:06,520
Speaker 1: we all know, the law trails behind technological development, sometimes

532
00:34:06,960 --> 00:34:10,880
Speaker 1: by years. It takes time to make laws, and then

533
00:34:10,920 --> 00:34:13,600
Speaker 1: it takes time to approve them and to pass them

534
00:34:13,640 --> 00:34:17,840
Speaker 1: into law. If you rush, you're likely to make problems worse,

535
00:34:18,440 --> 00:34:21,200
Speaker 1: or at the very least, you're likely to complicate matters

536
00:34:21,320 --> 00:34:23,880
Speaker 1: so that it becomes very difficult to comply with the

537
00:34:23,960 --> 00:34:28,200
Speaker 1: laws you've written. So Altman, who had already made his

538
00:34:28,239 --> 00:34:31,880
Speaker 1: philosophy around regulation known in that blog post from twenty fifteen,

539
00:34:32,640 --> 00:34:36,799
Speaker 1: began to meet with various officials all around the world. Now,

540
00:34:36,880 --> 00:34:41,520
Speaker 1: the idea was that sam Altman would help legislators understand

541
00:34:41,560 --> 00:34:46,719
Speaker 1: the potential risks of artificial intelligence and presumably create the

542
00:34:46,760 --> 00:34:52,160
Speaker 1: most responsible approach to regulations to ensure safety. But skeptics

543
00:34:52,200 --> 00:34:55,960
Speaker 1: were worried that what sam Altman was actually doing was

544
00:34:56,040 --> 00:34:59,680
Speaker 1: just stacking the deck to favor Open a Eye over

545
00:35:00,080 --> 00:35:04,200
Speaker 1: other AI companies. You see, Altman had long held this

546
00:35:04,280 --> 00:35:07,480
Speaker 1: position that it's really important for an ethical group of

547
00:35:07,520 --> 00:35:11,240
Speaker 1: researchers to beat everybody else to the punch to develop

548
00:35:11,280 --> 00:35:15,480
Speaker 1: that superhuman machine intelligence in order to prevent catastrophe. That

549
00:35:15,520 --> 00:35:20,240
Speaker 1: if you don't do that, you're essentially sealing your own doom.

550
00:35:20,360 --> 00:35:25,120
Speaker 1: And of course Altman viewed open AI as that ethical group.

551
00:35:25,480 --> 00:35:30,640
Speaker 1: It is the group that's dedicated to creating ethical, safe AI,

552
00:35:31,120 --> 00:35:34,759
Speaker 1: and Altmann felt that regulations could help mitigate the risk

553
00:35:34,840 --> 00:35:40,600
Speaker 1: of bad actors or inept creators from making dangerous machine intelligence.

554
00:35:41,000 --> 00:35:45,239
Speaker 1: And so these skeptics were arguing Altman's position was that

555
00:35:45,320 --> 00:35:50,200
Speaker 1: regulations should really hold back everybody else and then favor

556
00:35:50,320 --> 00:35:52,840
Speaker 1: open AI and allow it to move forward towards this

557
00:35:52,960 --> 00:35:58,879
Speaker 1: goal of creating benign, protective machine intelligence. So in other words, yeah,

558
00:35:58,920 --> 00:36:02,239
Speaker 1: the AI field needs rights, but more importantly, those regulations

559
00:36:02,280 --> 00:36:05,480
Speaker 1: need to stop every one other than open AI. That

560
00:36:05,600 --> 00:36:08,920
Speaker 1: was how the skeptics saw Sam Altman's position as he

561
00:36:09,000 --> 00:36:12,400
Speaker 1: met with all these different leaders, and in fact we

562
00:36:12,480 --> 00:36:16,560
Speaker 1: saw proposals for putting AI research on hold for half

563
00:36:16,600 --> 00:36:20,480
Speaker 1: a year. In fact, Elon Musk argued for this. Now,

564
00:36:20,560 --> 00:36:24,600
Speaker 1: the skeptics came out again and said, well, I see

565
00:36:24,640 --> 00:36:28,040
Speaker 1: the need to kind of pump the brakes so that

566
00:36:28,120 --> 00:36:31,720
Speaker 1: we make sure that all this work in AI isn't

567
00:36:31,800 --> 00:36:35,520
Speaker 1: going to cause enormous harm in the future. But of

568
00:36:35,560 --> 00:36:38,200
Speaker 1: course Musk is arguing for this because he wants to

569
00:36:38,239 --> 00:36:41,920
Speaker 1: create his own AI research division. And if you force

570
00:36:42,080 --> 00:36:44,880
Speaker 1: the industry as a whole to hold off for six

571
00:36:45,000 --> 00:36:49,000
Speaker 1: months on moving forward, it would give Musk the opportunity

572
00:36:49,080 --> 00:36:52,279
Speaker 1: to start to build out the foundations for his own

573
00:36:52,320 --> 00:36:57,160
Speaker 1: AI division while not losing more ground to competitors like

574
00:36:57,239 --> 00:37:02,200
Speaker 1: open Ai. It's a cutthroat world out in that AI field, y'all. Now.

575
00:37:02,200 --> 00:37:06,360
Speaker 1: In January of this year, in twenty twenty three, news

576
00:37:06,400 --> 00:37:10,120
Speaker 1: broke that Microsoft was investing around ten billion with a

577
00:37:10,200 --> 00:37:15,239
Speaker 1: B dollars in open Ai. Microsoft had already invested billions

578
00:37:15,280 --> 00:37:18,120
Speaker 1: in open Ai over the previous years, in twenty nineteen

579
00:37:18,120 --> 00:37:21,359
Speaker 1: and twenty twenty one, specifically, so this was seen as

580
00:37:21,400 --> 00:37:25,800
Speaker 1: Microsoft's effort to catch up to rivals like Google and Amazon,

581
00:37:25,920 --> 00:37:29,800
Speaker 1: which had already been spending their own billions in AI research.

582
00:37:30,520 --> 00:37:33,520
Speaker 1: The relationship between Microsoft and open Ai would manifest in

583
00:37:33,600 --> 00:37:38,120
Speaker 1: lots of different ways, including in Microsoft incorporating chat GPT

584
00:37:38,440 --> 00:37:42,640
Speaker 1: into its search feature in bing. For a long time,

585
00:37:42,680 --> 00:37:47,719
Speaker 1: Microsoft has been pushing being an edge toward becoming more

586
00:37:47,760 --> 00:37:52,000
Speaker 1: important in the market, but as come up time and

587
00:37:52,000 --> 00:37:56,120
Speaker 1: again up against the brick wall that is Google. In August,

588
00:37:56,480 --> 00:37:59,760
Speaker 1: open Ai announced in an enterprise version of chat gpt,

589
00:38:00,320 --> 00:38:04,520
Speaker 1: and then in September, open Ai allowed the chatbot to

590
00:38:04,680 --> 00:38:07,279
Speaker 1: access information on the Internet for the first time. So

591
00:38:07,760 --> 00:38:11,799
Speaker 1: now that restriction where it could only access information up

592
00:38:11,880 --> 00:38:15,480
Speaker 1: to September twenty twenty one, had been lifted. Now it

593
00:38:15,520 --> 00:38:19,400
Speaker 1: could access real time information around the world. On the

594
00:38:19,440 --> 00:38:22,959
Speaker 1: political side, lawmakers around the world, particularly in the United

595
00:38:23,000 --> 00:38:26,160
Speaker 1: States and in the European Union, began to grow more

596
00:38:26,200 --> 00:38:30,040
Speaker 1: concerned about AI and its possible uses and the risks

597
00:38:30,080 --> 00:38:33,839
Speaker 1: associated with it. So more pressure was building on the

598
00:38:34,040 --> 00:38:37,760
Speaker 1: artificial intelligence discipline in general and open ai in particular,

599
00:38:37,760 --> 00:38:40,720
Speaker 1: because open ai was seen as sort of the leading

600
00:38:40,800 --> 00:38:46,600
Speaker 1: authority in artificial intelligence. Chad GPT had really captured a

601
00:38:46,640 --> 00:38:49,640
Speaker 1: lot of interest around the world, and at the same

602
00:38:49,680 --> 00:38:53,719
Speaker 1: time you had some people within open ai who were

603
00:38:53,760 --> 00:38:57,719
Speaker 1: really clinging onto the ideals of the original nonprofit organization

604
00:38:57,920 --> 00:39:02,160
Speaker 1: and who had growing concerns about where open ai was

605
00:39:02,239 --> 00:39:05,520
Speaker 1: headed because of the for profit arm of the company.

606
00:39:05,960 --> 00:39:09,080
Speaker 1: So similar to Elon Musk, there were people, high level

607
00:39:09,120 --> 00:39:13,480
Speaker 1: people in open ai who were starting to feel uncomfortable

608
00:39:13,800 --> 00:39:16,600
Speaker 1: with where the organization was going. Now, just a couple

609
00:39:16,640 --> 00:39:19,800
Speaker 1: of weeks ago, open ai held its first developer conference.

610
00:39:20,200 --> 00:39:23,920
Speaker 1: Sam Altman took the stage on November sixth, so not

611
00:39:24,080 --> 00:39:27,400
Speaker 1: long ago, and he took the stage as CEO of

612
00:39:27,400 --> 00:39:30,560
Speaker 1: open Ai, and he listened off some pretty incredible statistics

613
00:39:30,880 --> 00:39:33,399
Speaker 1: like the fact that open ai can count more than

614
00:39:33,560 --> 00:39:40,640
Speaker 1: ninety percent of Fortune five hundred companies as customers. That's incredible.

615
00:39:40,680 --> 00:39:44,880
Speaker 1: It shows how influential open ai is in this field.

616
00:39:45,400 --> 00:39:49,240
Speaker 1: Open AI's partnership with Microsoft also played a huge part

617
00:39:49,680 --> 00:39:54,319
Speaker 1: in Sam Altman's presentation. But again, behind the scenes, things

618
00:39:54,360 --> 00:39:58,200
Speaker 1: were far from honky dory. You had open Ai doing

619
00:39:58,280 --> 00:40:01,680
Speaker 1: gangbusters on a business level, but again some of the

620
00:40:01,719 --> 00:40:04,440
Speaker 1: scientists who were part of the board of directors were

621
00:40:04,440 --> 00:40:07,920
Speaker 1: growing increasingly concerned that the company was guilty of the

622
00:40:08,080 --> 00:40:11,520
Speaker 1: very behaviors that open Ai was meant to head off.

623
00:40:12,120 --> 00:40:15,840
Speaker 1: That open ai was developing and deploying tools without putting

624
00:40:15,840 --> 00:40:20,560
Speaker 1: in appropriate safeguards or considering the consequences of unleashing these

625
00:40:20,600 --> 00:40:25,680
Speaker 1: tools that open ai had become more about monetizing technologies

626
00:40:25,719 --> 00:40:30,600
Speaker 1: and innovation and less about ethical development of artificial intelligence.

627
00:40:31,320 --> 00:40:33,960
Speaker 1: There were also some personal tensions that were growing between

628
00:40:34,000 --> 00:40:39,239
Speaker 1: Altman and other members of the board, such as Ilia Sutzkever. So,

629
00:40:39,360 --> 00:40:43,960
Speaker 1: according to Time, Altman reduced Ilia's role in the company,

630
00:40:44,440 --> 00:40:48,719
Speaker 1: while Ilia worried that Altman was launching side projects that

631
00:40:49,040 --> 00:40:52,719
Speaker 1: would benefit from open AI's work but also not be

632
00:40:52,880 --> 00:40:55,920
Speaker 1: accountable to open ai. So, in other words, ILIO was

633
00:40:55,920 --> 00:40:59,600
Speaker 1: worried about this sort of conflict of interest that Altman

634
00:40:59,680 --> 00:41:03,280
Speaker 1: was going to end up pursuing some developments of artificial

635
00:41:03,280 --> 00:41:07,000
Speaker 1: intelligence that were not governed by the board of open

636
00:41:07,040 --> 00:41:12,719
Speaker 1: ai and thus not constrained by these ethical concerns. So

637
00:41:13,239 --> 00:41:16,960
Speaker 1: this really boiled over. During the developer conference, Altman made

638
00:41:16,960 --> 00:41:21,280
Speaker 1: several announcements that Ilya reportedly objected to, including the unveiling

639
00:41:21,320 --> 00:41:25,040
Speaker 1: of a customizable version of chat GPT that, in theory,

640
00:41:25,520 --> 00:41:28,600
Speaker 1: could run autonomously once it was told what tasks it

641
00:41:28,680 --> 00:41:31,480
Speaker 1: was supposed to handle. So the critics on the board

642
00:41:32,080 --> 00:41:36,200
Speaker 1: outnumbered Altman's supporters. You had essentially two camps. You had

643
00:41:36,200 --> 00:41:38,600
Speaker 1: the people who thought Altman was in the right, including

644
00:41:38,600 --> 00:41:42,239
Speaker 1: Altman and Brockman, and then you had other members who

645
00:41:42,280 --> 00:41:46,799
Speaker 1: were more concerned. And last Friday, that's when the board

646
00:41:46,880 --> 00:41:50,000
Speaker 1: decided it was time to fire Sam Altman. They saw

647
00:41:50,040 --> 00:41:54,640
Speaker 1: Altman as being too reckless, not nearly cautious enough, and

648
00:41:54,680 --> 00:41:58,200
Speaker 1: despite open AI's market performance, which was incredible, they felt

649
00:41:58,200 --> 00:42:00,400
Speaker 1: the company was moving in the wrong direction and that

650
00:42:00,480 --> 00:42:04,000
Speaker 1: it needed new leadership as a result, so they decided

651
00:42:04,000 --> 00:42:07,520
Speaker 1: they had to fire Sam Altman as CEO. Now. Reportedly,

652
00:42:07,960 --> 00:42:11,200
Speaker 1: Altman learned of this fate in a Zoom meeting. Ilia

653
00:42:11,320 --> 00:42:12,520
Speaker 1: was the one who told him that he had to

654
00:42:12,520 --> 00:42:15,279
Speaker 1: go to the Zoom meeting, and it happened shortly before

655
00:42:15,320 --> 00:42:19,880
Speaker 1: the board announced their decision Publicly. Greg Brockman, co founder

656
00:42:20,080 --> 00:42:23,720
Speaker 1: president of open Ai, he was not told of this meeting,

657
00:42:24,000 --> 00:42:26,560
Speaker 1: and in fact, he found out about Sam Altman getting

658
00:42:26,600 --> 00:42:30,840
Speaker 1: fired just shortly before open Ai released the news to

659
00:42:30,880 --> 00:42:36,320
Speaker 1: the public. Similarly, Satya Nadella, the CEO of Microsoft, also

660
00:42:36,719 --> 00:42:40,080
Speaker 1: found out essentially when the news got released to the public,

661
00:42:40,840 --> 00:42:46,120
Speaker 1: and this set off a metaphorical explosion in the tech world. First,

662
00:42:46,480 --> 00:42:50,320
Speaker 1: open AI's board didn't exactly have a real good transition

663
00:42:50,520 --> 00:42:54,439
Speaker 1: plan in place, to handle this, nor did it seem

664
00:42:54,480 --> 00:42:58,000
Speaker 1: to really comprehend the extent of the fallout this decision

665
00:42:58,040 --> 00:43:01,279
Speaker 1: would have, particularly in the way they did it, where

666
00:43:01,320 --> 00:43:05,640
Speaker 1: they did not consult with Microsoft, a partner that was

667
00:43:05,719 --> 00:43:09,520
Speaker 1: going to invest ten billion dollars into the company, or

668
00:43:09,960 --> 00:43:14,240
Speaker 1: talk it over with the other executives before making the decision.

669
00:43:14,760 --> 00:43:18,360
Speaker 1: So even the folks who agreed that Altman was perhaps

670
00:43:18,440 --> 00:43:22,239
Speaker 1: not being cautious enough felt that the board's move was

671
00:43:22,320 --> 00:43:26,160
Speaker 1: poorly thought out and even more poorly executed. It's really

672
00:43:26,160 --> 00:43:29,279
Speaker 1: hard to argue against that. Like, even if you feel

673
00:43:29,320 --> 00:43:32,440
Speaker 1: that Sam Altman was absolutely leading open Ai in the

674
00:43:32,480 --> 00:43:35,960
Speaker 1: wrong way the wrong direction, you can also say that

675
00:43:36,040 --> 00:43:42,319
Speaker 1: the way the board handled this ultimately was disastrous. So

676
00:43:42,480 --> 00:43:46,720
Speaker 1: Greg Brockman announced that he was leaving open Ai once

677
00:43:47,040 --> 00:43:49,640
Speaker 1: the news went public that Sam Altman had been fired,

678
00:43:50,360 --> 00:43:53,759
Speaker 1: So open Ai would see both its CEO and its

679
00:43:53,800 --> 00:43:58,400
Speaker 1: president leave the company in one day. Some members of

680
00:43:58,440 --> 00:44:01,960
Speaker 1: the board, like Ilia, expressed regret for having supported the

681
00:44:01,960 --> 00:44:05,680
Speaker 1: measure to fire Altman, like they said later like I

682
00:44:05,760 --> 00:44:09,759
Speaker 1: kind of wish we hadn't done that. Iliosutskiv would go

683
00:44:09,840 --> 00:44:12,960
Speaker 1: on to sign an open letter saying as much, and

684
00:44:13,040 --> 00:44:15,600
Speaker 1: even threatened to leave Open a Eye along with more

685
00:44:15,600 --> 00:44:19,759
Speaker 1: than five hundred other staff members over this decision. Just

686
00:44:19,760 --> 00:44:22,920
Speaker 1: a big old whoopsie, right, so the board found itself

687
00:44:23,040 --> 00:44:27,600
Speaker 1: in extremely hot water. They had done the classy thing

688
00:44:27,680 --> 00:44:31,320
Speaker 1: of waiting until a Friday to announce a massive decision.

689
00:44:32,200 --> 00:44:34,480
Speaker 1: My guess is this was probably in an effort to

690
00:44:34,560 --> 00:44:36,719
Speaker 1: take at least some of the sting out of the

691
00:44:36,760 --> 00:44:39,600
Speaker 1: news cycle, the idea of being like, well, if it's

692
00:44:39,600 --> 00:44:42,000
Speaker 1: on a Friday afternoon, no one's going to pay attention

693
00:44:42,080 --> 00:44:44,239
Speaker 1: because we're going into the weekend. By the time it

694
00:44:44,280 --> 00:44:46,640
Speaker 1: comes around to Monday, things are going to cool off

695
00:44:46,640 --> 00:44:49,120
Speaker 1: a bit. Plus here in the United States, we're going

696
00:44:49,160 --> 00:44:52,840
Speaker 1: into a holiday week with Thanksgiving, so there won't be

697
00:44:52,880 --> 00:44:54,839
Speaker 1: a whole lot of opportunity to bring a whole lot

698
00:44:54,880 --> 00:44:56,680
Speaker 1: of attention to this, and we'll be able to get

699
00:44:56,680 --> 00:45:01,880
Speaker 1: away relatively unscathed. That is not how it turned out, however. Instead,

700
00:45:02,520 --> 00:45:07,359
Speaker 1: the news media went bonkers with this decision, and how

701
00:45:07,400 --> 00:45:11,160
Speaker 1: could you not chat GPT and open AI. They had

702
00:45:11,200 --> 00:45:13,799
Speaker 1: been the center of so many headlines throughout the whole year.

703
00:45:13,880 --> 00:45:16,640
Speaker 1: Of course, this was going to get a lot of attention,

704
00:45:17,440 --> 00:45:21,000
Speaker 1: and so while they were hoping that they could get

705
00:45:21,000 --> 00:45:25,520
Speaker 1: away with this without it being too painful. Immediately, investors

706
00:45:25,560 --> 00:45:28,840
Speaker 1: started to freak out about this change. A bunch of

707
00:45:28,840 --> 00:45:31,120
Speaker 1: them essentially indicated that they would pull out of open

708
00:45:31,200 --> 00:45:34,479
Speaker 1: ai and they would back whatever Altman chose to do next.

709
00:45:34,480 --> 00:45:38,200
Speaker 1: So if Altman launched his own competing artificial intelligence company,

710
00:45:38,719 --> 00:45:41,640
Speaker 1: they were going to back Altman, not open Ai. There

711
00:45:41,640 --> 00:45:44,760
Speaker 1: were hints that Microsoft could potentially even do the same,

712
00:45:44,840 --> 00:45:49,760
Speaker 1: and that's ten billion dollars plus. You had the general

713
00:45:49,840 --> 00:45:53,279
Speaker 1: staff who felt that this was the wrong move, and

714
00:45:53,320 --> 00:45:56,200
Speaker 1: they felt that this was a terrible mistake, and they

715
00:45:56,200 --> 00:46:00,799
Speaker 1: were threatening a mass walkout of the company. It was

716
00:46:00,840 --> 00:46:04,640
Speaker 1: pretty much the worst reaction you could expect from a

717
00:46:04,640 --> 00:46:07,640
Speaker 1: big announcement. So it did not take very long for

718
00:46:08,040 --> 00:46:10,799
Speaker 1: news to break that the board was trying hard to

719
00:46:10,960 --> 00:46:13,960
Speaker 1: take back what it had done and to try and

720
00:46:14,000 --> 00:46:18,440
Speaker 1: convince Altman and Brockman to return to the company, but

721
00:46:18,600 --> 00:46:22,200
Speaker 1: by then the damage had been done. Altman was not

722
00:46:22,360 --> 00:46:26,040
Speaker 1: interested in coming back. Specifically, he said unless the board

723
00:46:26,160 --> 00:46:29,719
Speaker 1: stepped down, he would not come back, and that would

724
00:46:29,760 --> 00:46:34,160
Speaker 1: become a real sticking point. Mira Murrati, the chief technology

725
00:46:34,200 --> 00:46:37,960
Speaker 1: officer for open Ai, would serve as interim CEO for

726
00:46:38,000 --> 00:46:41,560
Speaker 1: about two whole days. Murati reportedly was the person who

727
00:46:41,560 --> 00:46:43,719
Speaker 1: actually reached out to Altman to try and convince him

728
00:46:43,719 --> 00:46:48,160
Speaker 1: to come back to open AIE. But while Altman did

729
00:46:48,200 --> 00:46:51,759
Speaker 1: return to open AI's headquarters to negotiate, and he said

730
00:46:51,760 --> 00:46:53,480
Speaker 1: it was the first and only time he would ever

731
00:46:53,520 --> 00:46:58,640
Speaker 1: have a guest badge to open AIE, those negotiations didn't

732
00:46:58,920 --> 00:47:02,799
Speaker 1: really go very far, so the board decided on a

733
00:47:02,840 --> 00:47:06,200
Speaker 1: new interim CEO, perhaps because of a perception that Murrati

734
00:47:06,360 --> 00:47:09,120
Speaker 1: was maybe a bit too pro Altman and they needed

735
00:47:09,120 --> 00:47:11,480
Speaker 1: to get someone who would be more in their pocket.

736
00:47:12,440 --> 00:47:16,879
Speaker 1: They chose the former CEO of Twitch, Emmitt Sheer, who

737
00:47:16,880 --> 00:47:21,400
Speaker 1: doesn't have any experience with artificial intelligence. By this time,

738
00:47:21,719 --> 00:47:25,080
Speaker 1: the board directors consisted of just four people, people who

739
00:47:25,080 --> 00:47:28,600
Speaker 1: have been pressured to step down but refused to do so.

740
00:47:29,000 --> 00:47:31,840
Speaker 1: Thus Altman did not come back to the company. Altman

741
00:47:31,920 --> 00:47:35,000
Speaker 1: and Brockman, meanwhile, weren't exactly on the job market for

742
00:47:35,120 --> 00:47:39,160
Speaker 1: very long, because Microsoft swiftly hired both of them to

743
00:47:39,239 --> 00:47:43,279
Speaker 1: head up a new advanced AI research team within Microsoft.

744
00:47:43,680 --> 00:47:47,440
Speaker 1: Nadella also said Microsoft remains committed to supporting open AI,

745
00:47:48,000 --> 00:47:51,359
Speaker 1: hinting that those ten billion dollars, most of which has

746
00:47:51,480 --> 00:47:54,399
Speaker 1: not made its way to open Ai yet. Open Ai

747
00:47:54,560 --> 00:47:58,320
Speaker 1: has received just a fraction of that ten billion dollars,

748
00:47:58,760 --> 00:48:01,799
Speaker 1: but the hint is that that money will continue to

749
00:48:01,920 --> 00:48:05,080
Speaker 1: go to OpenAI, that Microsoft is not backing out of

750
00:48:05,120 --> 00:48:08,240
Speaker 1: that agreement, but Altman is going to end up working

751
00:48:08,280 --> 00:48:11,560
Speaker 1: directly with Microsoft and will have the title of CEO

752
00:48:11,760 --> 00:48:16,400
Speaker 1: for whatever this Advanced AI part of Microsoft ends up

753
00:48:16,400 --> 00:48:21,120
Speaker 1: being called. A few other prominent executives and scientists from

754
00:48:21,200 --> 00:48:24,960
Speaker 1: open Ai are apparently moving over to Microsoft as well,

755
00:48:25,080 --> 00:48:31,680
Speaker 1: so there are already other defections from open Ai to Microsoft. Meanwhile,

756
00:48:31,719 --> 00:48:34,520
Speaker 1: back at open Ai, a lot of folks who work

757
00:48:34,600 --> 00:48:37,720
Speaker 1: within the company have been posting their support for Altman

758
00:48:38,080 --> 00:48:41,640
Speaker 1: on platforms like x so there's a concern that there's

759
00:48:41,680 --> 00:48:45,160
Speaker 1: going to be a mass walkout and resignation following this move.

760
00:48:45,440 --> 00:48:49,359
Speaker 1: Certainly other companies like Microsoft, Google, Amazon, and Meta would

761
00:48:49,360 --> 00:48:51,440
Speaker 1: be eager to get hold of some of that talent,

762
00:48:52,080 --> 00:48:54,920
Speaker 1: and it's entirely possible that the board of open Ai,

763
00:48:55,000 --> 00:48:57,520
Speaker 1: in a move made out of concern for the company's

764
00:48:57,520 --> 00:49:02,600
Speaker 1: safety and humanity safety, may have actually doomed the organization entirely.

765
00:49:02,800 --> 00:49:05,279
Speaker 1: I'm not sure hiring a former CEO of Twitch is

766
00:49:05,320 --> 00:49:08,600
Speaker 1: going to be enough to prevent disaster. Now, all that

767
00:49:08,640 --> 00:49:12,080
Speaker 1: being said, open ai is in an incredible position. A

768
00:49:12,120 --> 00:49:16,680
Speaker 1: recent evaluation placed the company at around eighty six billion dollars.

769
00:49:17,280 --> 00:49:20,960
Speaker 1: Microsoft says it is committed to this ongoing relationship with

770
00:49:21,040 --> 00:49:25,600
Speaker 1: open Ai. Chad GPT is still an incredibly important tool

771
00:49:25,719 --> 00:49:28,680
Speaker 1: in the tech space, particularly with the introduction of the

772
00:49:28,840 --> 00:49:33,080
Speaker 1: enterprise product of chat GPT. So could open ai just

773
00:49:33,120 --> 00:49:37,920
Speaker 1: be too big to fail? Maybe? I think this monumental

774
00:49:38,640 --> 00:49:43,080
Speaker 1: misstep will test that hypothesis. I do not know how

775
00:49:43,080 --> 00:49:45,800
Speaker 1: it's all going to shake out. From a business perspective,

776
00:49:46,000 --> 00:49:48,640
Speaker 1: I would say open ai is in a really strong position.

777
00:49:49,160 --> 00:49:52,080
Speaker 1: But then if the organization suddenly sees a mass defection

778
00:49:52,280 --> 00:49:55,280
Speaker 1: from its researchers and staff, that could very well change.

779
00:49:55,560 --> 00:49:57,279
Speaker 1: So we'll have to see. And of course, now we're

780
00:49:57,280 --> 00:49:59,759
Speaker 1: on a holiday week, so it might be another week

781
00:49:59,760 --> 00:50:03,040
Speaker 1: but before we start getting answers. But yeah, that's kind

782
00:50:03,080 --> 00:50:06,160
Speaker 1: of an update on what went down this past weekend

783
00:50:06,160 --> 00:50:09,279
Speaker 1: and why it happened, Like all those different factors that

784
00:50:09,400 --> 00:50:14,040
Speaker 1: built up to this big explosion of activity. Now you

785
00:50:14,160 --> 00:50:16,960
Speaker 1: have a bit more background as to what was going on.

786
00:50:17,760 --> 00:50:20,239
Speaker 1: As to whose side I'm on, I don't know. I

787
00:50:20,280 --> 00:50:26,200
Speaker 1: do think that Altman's leadership was not always the best

788
00:50:26,239 --> 00:50:29,800
Speaker 1: as far as trying to achieve the goal of ethical AI.

789
00:50:30,440 --> 00:50:33,880
Speaker 1: I do not think that it was almost like engaging

790
00:50:33,880 --> 00:50:36,200
Speaker 1: in a necessary evil kind of thing. But I'm not

791
00:50:36,200 --> 00:50:40,239
Speaker 1: sure that the evil is really necessary. But what do

792
00:50:40,320 --> 00:50:43,560
Speaker 1: I know. I know that AI is very hard and

793
00:50:43,680 --> 00:50:45,799
Speaker 1: very expensive, and I don't know how you get the

794
00:50:45,840 --> 00:50:48,680
Speaker 1: money to do it the right way and still beat

795
00:50:48,680 --> 00:50:51,759
Speaker 1: out all the companies that don't have those restrictions on them.

796
00:50:52,200 --> 00:50:56,120
Speaker 1: So I don't know. I just know that it's a mess.

797
00:50:56,680 --> 00:50:58,920
Speaker 1: But now it's a mess we can put behind us

798
00:50:59,000 --> 00:51:02,160
Speaker 1: until we see what happens next. I hope you are

799
00:51:02,200 --> 00:51:06,040
Speaker 1: all well, and I'll talk to you again really soon.

800
00:51:12,400 --> 00:51:17,040
Speaker 1: Tech Stuff is an iHeartRadio production. For more podcasts from iHeartRadio,

801
00:51:17,360 --> 00:51:21,040
Speaker 1: visit the iHeartRadio app, Apple Podcasts, or wherever you listen

802
00:51:21,120 --> 00:51:22,200
Speaker 1: to your favorite shows.