WEBVTT - The Fallout from OpenAI Firing CEO Sam Altman 0:00:04.440 --> 0:00:12.360 Welcome to tech Stuff, a production from iHeartRadio. Hey there, 0:00:12.360 --> 0:00:15.840 and welcome to tech Stuff. I'm your host, Jonathan Strickland. 0:00:15.880 --> 0:00:19.040 I'm an executive producer with iHeart Podcasts and How the 0:00:19.120 --> 0:00:24.720 Tech Are Yet Now. Normally I say tech news items 0:00:24.760 --> 0:00:28.960 for Tuesdays and Thursdays, But over this past weekend a 0:00:29.080 --> 0:00:33.919 pretty big sequence of events happened and it really merits 0:00:34.200 --> 0:00:37.080 a deeper discussion. I'm sure most of you have at 0:00:37.159 --> 0:00:42.000 least heard something about this story. The short version is 0:00:42.200 --> 0:00:46.839 Sam Altman, the CEO of open Ai, received his walking 0:00:46.920 --> 0:00:51.360 papers from the company's board of directors. Then the board 0:00:51.720 --> 0:00:54.560 kind of flipped out and begged him to come back 0:00:54.600 --> 0:00:58.560 to the company, and he ultimately decided now I'm good, 0:00:59.160 --> 0:01:02.320 y'all can do this on your own. And last I heard, 0:01:02.600 --> 0:01:06.840 he has now joined Microsoft's Advanced AI department. Which is 0:01:06.840 --> 0:01:10.200 a heck of a weekend. So today I thought we 0:01:10.240 --> 0:01:12.240 would talk a bit about Altman, We talk a bit 0:01:12.280 --> 0:01:16.000 about open Ai, We chat about what went down behind 0:01:16.040 --> 0:01:20.360 closed doors this past Friday, why the board of directors 0:01:20.600 --> 0:01:25.039 fired Altman, why they then switched gears so quickly, and 0:01:25.080 --> 0:01:28.280 what it all means going forward. Now. I did an 0:01:28.280 --> 0:01:31.960 episode titled the Story of Open AI at the beginning 0:01:32.000 --> 0:01:35.040 of this year, which published in January. I am going 0:01:35.080 --> 0:01:37.399 to retread a lot of that same ground here in 0:01:37.480 --> 0:01:41.800 a slightly different context, because it's necessary to really unravel 0:01:41.880 --> 0:01:45.040 what was going on over the last weekend. So first up, 0:01:46.000 --> 0:01:50.120 who is Sam Altman. Well, he grew up in Saint Louis, Missouri, 0:01:50.640 --> 0:01:54.240 and as a kid he became really interested in programming. 0:01:54.360 --> 0:01:57.279 According to The New Yorker, he was programming as early 0:01:57.400 --> 0:02:00.240 as age eight, and that he went so far to 0:02:00.240 --> 0:02:02.840 take a part a Macintosh computer in order to learn 0:02:02.840 --> 0:02:06.840 how it worked. He also challenged social restrictions and taboos. 0:02:07.240 --> 0:02:10.640 When a Christian group announced a boycott for an assembly 0:02:10.680 --> 0:02:14.000 that was supposed to be focused on sexuality, Altman came 0:02:14.040 --> 0:02:17.280 out to his community as gay, and he challenged them 0:02:17.280 --> 0:02:20.720 to adopt an open attitude toward different ideas. And he 0:02:20.760 --> 0:02:23.880 was just a teenager at the time, so very much 0:02:24.639 --> 0:02:32.400 someone who is curious, motivated, and by most accounts I've read, fearless. 0:02:33.080 --> 0:02:36.640 Altman attended Stanford, but he was only there for two years. 0:02:36.800 --> 0:02:39.280 He studied computer science while he was there. In fact, 0:02:39.280 --> 0:02:43.160 he studied artificial intelligence by some of the leading thinkers 0:02:43.360 --> 0:02:46.440 in the discipline at Stanford, but he dropped out of 0:02:46.440 --> 0:02:48.560 college in order to work on an app and a 0:02:48.600 --> 0:02:51.600 business idea. So in many ways, it was the stereotypical 0:02:52.040 --> 0:02:55.520 founder story of Silicon Valley. Right. You go to Stanford, 0:02:55.840 --> 0:02:58.760 and when you're there, you're really there to make connections. 0:02:59.120 --> 0:03:02.280 You don't bother completing your studies, You drop out of school, 0:03:02.760 --> 0:03:05.839 you make a company, and then you get rich. It's 0:03:05.960 --> 0:03:08.119 kind of like, you know, the whole idea of step one, 0:03:08.760 --> 0:03:12.360 go to Stanford, step two, drop out, step three profit. 0:03:13.639 --> 0:03:15.720 I guess you could argue that if you can make 0:03:15.720 --> 0:03:19.639 a successful tech business, there's no real point to completing 0:03:19.639 --> 0:03:21.720 your studies. I mean, if your studies are all about 0:03:22.360 --> 0:03:25.480 learning the technology and it turns out you already have 0:03:25.520 --> 0:03:27.680 a good mastery of that and you can make a 0:03:27.720 --> 0:03:31.480 profitable business, why would you continue to spend money going 0:03:31.520 --> 0:03:35.400 to school, unless, of course, maybe you would grow more 0:03:35.400 --> 0:03:38.880 as a person and develop a deeper appreciation and understanding 0:03:38.920 --> 0:03:42.480 of things that could perhaps help you when you make 0:03:42.880 --> 0:03:45.600 decisions in the future. But don't listen to me. I 0:03:45.640 --> 0:03:48.240 graduated with a degree in the humanities, so I have 0:03:48.320 --> 0:03:52.520 these wacky ideas about how the experience of college is 0:03:52.840 --> 0:03:56.800 about more than just learning a subject. But that's beside 0:03:56.800 --> 0:03:59.120 the point. Let's get back to Sam Altman. So the 0:03:59.240 --> 0:04:02.240 app he was working on with a few friends was 0:04:02.280 --> 0:04:07.440 called Looped loopt, and it was meant to let users 0:04:07.480 --> 0:04:11.760 to share their location data with selected other users, So 0:04:11.800 --> 0:04:14.160 you can make friends with people on the app, and 0:04:14.200 --> 0:04:18.200 then you could share your precise location with that person, 0:04:18.680 --> 0:04:21.039 kind of a shorthand way of saying here I am. 0:04:21.400 --> 0:04:24.360 And it could facilitate stuff like real world meetups. Like 0:04:24.480 --> 0:04:27.920 imagine that you're heading to a concert venue and it's 0:04:27.920 --> 0:04:30.560 a big venue. There's a lot of different interests and stuff. 0:04:30.640 --> 0:04:32.440 You get there, you're going to meet up with your friends. 0:04:32.800 --> 0:04:35.720 You use this app to say this is specifically where 0:04:35.720 --> 0:04:37.920 I am, so that you can find each other. That's 0:04:38.000 --> 0:04:41.360 kind of a use case. The Looped team applied to 0:04:41.760 --> 0:04:46.799 the Why Combinator accelerator company to become part of their program. 0:04:47.160 --> 0:04:50.600 So let's talk about Why Combinator for a moment. It is, 0:04:50.720 --> 0:04:55.039 as I said, a startup accelerator organization. So Why Combinator 0:04:55.400 --> 0:04:59.680 its purpose is to provide early funding in promising startup 0:04:59.680 --> 0:05:03.119 ideas in order for them to start to get off 0:05:03.279 --> 0:05:07.640 the ground. So it's an early investment part of a 0:05:07.720 --> 0:05:10.479 startup so that they can at least get a chance 0:05:10.560 --> 0:05:14.839 to mature into an actual business. Now, in return for 0:05:14.920 --> 0:05:19.760 this early investment, why Combinator takes a small percentage of 0:05:19.800 --> 0:05:24.960 ownership in the startup. So let's say that why Combinator 0:05:25.000 --> 0:05:28.799 provides you know, a fairly modest sum in the early days, 0:05:28.839 --> 0:05:30.880 Like it was a lot of money, don't get me wrong, 0:05:30.960 --> 0:05:33.200 Like maybe like one hundred thousand dollars or maybe one 0:05:33.240 --> 0:05:35.719 hundred and twenty thousand dollars. That's a lot of money, 0:05:35.960 --> 0:05:38.159 but it's a tiny amount when you think about what 0:05:38.279 --> 0:05:40.920 a company needs to actually run. So this is really 0:05:41.120 --> 0:05:44.440 just to get a startup to go from idea to 0:05:44.560 --> 0:05:50.120 something slightly more you know, coherent. But in return, why 0:05:50.160 --> 0:05:55.080 Combinator gets like, you know, seven percent ownership of that startup. Now, 0:05:55.160 --> 0:05:58.560 let's say that startup is something like Dropbox, and then 0:05:58.920 --> 0:06:00.960 years down the road it's worth you know, more than 0:06:01.000 --> 0:06:03.839 ten billion dollars. Well, that becomes a heck of a 0:06:03.880 --> 0:06:07.919 return on investment. Right, you can have a huge profit 0:06:08.080 --> 0:06:11.320 as this startup accelerator, even if just a few of 0:06:11.320 --> 0:06:14.880 the startups really hit it big. Ideally, you want them 0:06:14.960 --> 0:06:17.479 all to hit big and then you get huge payouts 0:06:17.520 --> 0:06:22.279 down the line, and you become an important part of 0:06:22.360 --> 0:06:25.839 the whole tech startup ecosystem, which is exactly what why 0:06:25.920 --> 0:06:28.919 Combinator said up to do now why Combinator launched in 0:06:28.920 --> 0:06:30.760 two thousand and five, that was the same year that 0:06:30.880 --> 0:06:36.760 Looped would become part of its inaugural class of startups. Essentially, now, 0:06:36.800 --> 0:06:40.880 not all startups make it. Obviously, Looped at least appeared 0:06:40.960 --> 0:06:44.400 to do well initially, at least on paper. Altman and 0:06:44.440 --> 0:06:48.360 his colleagues secured a couple of major rounds of investment funding, 0:06:48.440 --> 0:06:52.280 Series A and Series B. The company's valuation hit more 0:06:52.320 --> 0:06:55.480 than one hundred and seventy million dollars. But they were 0:06:55.520 --> 0:06:59.679 running into a tiny little problem. They had developed this app, 0:07:00.200 --> 0:07:03.279 and they couldn't convince people to use it. They were 0:07:03.880 --> 0:07:06.000 all thinking that, you know, Looped was going to be 0:07:06.000 --> 0:07:09.000 this really useful and popular tool. Everyone was going to 0:07:09.080 --> 0:07:12.600 download it. But turns out the general public didn't seem 0:07:12.640 --> 0:07:17.120 to agree with that, and so in twenty twelve, Looped 0:07:17.600 --> 0:07:21.240 that team accepted an offer from the company green Dot, 0:07:21.760 --> 0:07:25.200 and they sold Looped for around forty three million dollars. 0:07:25.760 --> 0:07:27.840 That's also a lot of money, but it did not 0:07:28.040 --> 0:07:32.040 cover the amount of money that venture capitalists had invested 0:07:32.240 --> 0:07:35.600 into looped, so it was a negative return for investors. 0:07:36.200 --> 0:07:38.239 You know, sometimes you bet on the ponies and you lose. 0:07:38.720 --> 0:07:41.880 But Altman walked away with around five million bucks, so 0:07:42.160 --> 0:07:45.200 it was a pretty decent return for him, even though 0:07:45.840 --> 0:07:48.960 the app that he had worked on for several years 0:07:49.000 --> 0:07:52.560 never really gained traction. Now, in the wake of this disappointment, 0:07:52.720 --> 0:07:56.280 Altman founded a venture capital company of his own, and 0:07:56.320 --> 0:07:59.800 it was called Hydrazine Capital. So he sunk most of 0:07:59.800 --> 0:08:02.400 his personal wealth that he had earned from this sale 0:08:03.080 --> 0:08:06.440 into this new venture capital company, and he also raised 0:08:06.480 --> 0:08:10.520 millions more from other investors. He focused on investing in 0:08:10.560 --> 0:08:14.240 companies that were in the y Combinator program, and he 0:08:14.360 --> 0:08:17.960 was largely successful in this. He was picking some really 0:08:18.000 --> 0:08:20.920 good startups, and he was backing ones that would become 0:08:20.960 --> 0:08:22.760 a big deal in the tech space a few years 0:08:22.840 --> 0:08:25.520 down the road, and so he was seeing big returns 0:08:25.520 --> 0:08:29.280 on those investments. Within just a few years, his venture 0:08:29.320 --> 0:08:32.920 capital firm increased in value by an order of magnitude. 0:08:33.679 --> 0:08:37.959 But Altman wasn't super happy doing this work. He didn't 0:08:38.000 --> 0:08:41.960 find it rewarding on a personal level. Financially sure, but 0:08:42.080 --> 0:08:44.680 on a personal level, he didn't really like the work, 0:08:45.280 --> 0:08:49.520 so he then extricated himself from the venture capital company 0:08:49.679 --> 0:08:53.440 to try and do something else. Now, around this same time, 0:08:53.480 --> 0:08:56.680 the folks who were behind why Combinator were looking to 0:08:56.840 --> 0:08:59.839 hand off the whole accelerator program to someone else to 0:09:00.120 --> 0:09:03.640 lead it, and that someone else ended up being Sam Altman. 0:09:04.440 --> 0:09:07.720 The guy had gone through the process with looped and 0:09:07.760 --> 0:09:10.480 now he would be running it, and he reportedly agreed 0:09:10.520 --> 0:09:13.120 without hesitation. He was eager to do this job. It 0:09:13.160 --> 0:09:15.800 was something that he didn't even necessarily know he wanted 0:09:15.800 --> 0:09:18.400 to do, but once he was offered it, he was 0:09:18.600 --> 0:09:22.280 really enthusiastic about doing it. So he really ramped up 0:09:22.400 --> 0:09:26.000 y Combinator, and Altman began recruiting, you know, startups that 0:09:26.000 --> 0:09:29.680 were focused on science and technology, so we're talking about 0:09:29.720 --> 0:09:34.240 bleeding edge stuff like quantum computing or nuclear power or 0:09:34.600 --> 0:09:39.360 AI that kind of stuff. Around this time, he started 0:09:39.440 --> 0:09:43.560 to be part of a group that included Elon Musk, 0:09:44.280 --> 0:09:48.720 and this group ostensibly wanted to develop artificial intelligence in 0:09:48.760 --> 0:09:53.240 a responsible, accountable way while being extra careful not to 0:09:53.320 --> 0:09:58.440 do anything foolish to make safe artificial intelligence. You know, 0:09:58.520 --> 0:10:01.080 they didn't want to go down the wrong path and 0:10:01.120 --> 0:10:04.959 do something like accidentally unleash Skynet and terminators all over 0:10:05.000 --> 0:10:10.640 the place, so this group would become open ai. Now, 0:10:10.640 --> 0:10:15.240 the original open ai was a non profit organization, and 0:10:15.280 --> 0:10:18.080 the whole idea was to help in this effort to 0:10:18.280 --> 0:10:22.520 foster the development of artificial intelligence in a responsible and 0:10:22.600 --> 0:10:26.600 safe way. It wasn't some for profit company pushing a 0:10:26.720 --> 0:10:30.600 generative AI chatbot at that time, and it was not 0:10:30.720 --> 0:10:36.160 yet a partner to massive companies like Microsoft. So Altman 0:10:36.320 --> 0:10:40.000 was running y Combinator, and he also began to work 0:10:40.040 --> 0:10:43.920 with the open ai folks and tried to recruit various 0:10:44.080 --> 0:10:49.440 leaders in artificial intelligence to join open ai. In twenty fifteen, 0:10:49.800 --> 0:10:53.199 Sam Altman published a two part blog post about machine 0:10:53.200 --> 0:10:57.880 intelligence and quote why you should fear it end quote, 0:10:58.480 --> 0:11:01.160 And it starts off with the comfort sentence and I 0:11:01.320 --> 0:11:08.559 quote development of superhuman machine intelligence SMI is probably the 0:11:08.600 --> 0:11:13.520 greatest threat to the continued existence of humanity end quote. 0:11:13.720 --> 0:11:18.240 Now Altman allows that other massive threats, like say, an 0:11:18.320 --> 0:11:23.280 asteroid hitting the Earth, are possibly more likely to happen 0:11:23.640 --> 0:11:27.480 than super human intelligent machines run amok, but he also 0:11:27.559 --> 0:11:30.520 points out that a lot of the other threats that 0:11:30.600 --> 0:11:34.840 we think of, like super volcanoes and climate crisis might 0:11:35.000 --> 0:11:38.160 end up having a massive impact on the human population, 0:11:38.360 --> 0:11:43.079 but probably wouldn't just totally wipe out humans in totality. 0:11:44.160 --> 0:11:51.080 But superhuman machine intelligence, he argued, did have that potential, 0:11:51.960 --> 0:11:54.000 and that this is why he thought of it as 0:11:54.040 --> 0:11:59.480 being the most important or perhaps most dangerous threat. Almand 0:11:59.520 --> 0:12:01.839 goes on in those blog posts to point out that 0:12:02.000 --> 0:12:04.840 SMI doesn't have to be malevolent to be a threat. 0:12:05.640 --> 0:12:09.000 Just setting an SMI to complete a task like trying 0:12:09.000 --> 0:12:12.720 to manage resources could end up causing massive human harm. 0:12:13.520 --> 0:12:17.079 The SMI might determine that the biggest cause of resource 0:12:17.080 --> 0:12:19.920 depletion is the human race, So presto, you get rid 0:12:19.960 --> 0:12:21.959 of the people, and now you don't have to worry 0:12:22.000 --> 0:12:25.760 about these resources running out anymore. Now that's an oversimplification, 0:12:25.840 --> 0:12:28.800 but you get the idea. Altman's point is if you 0:12:29.000 --> 0:12:33.840 don't develop artificial intelligence in a way that is safe, 0:12:34.400 --> 0:12:37.640 you can get terrible consequences, whether that was your goal 0:12:37.840 --> 0:12:41.439 or otherwise. And of course there are malicious ways to 0:12:41.520 --> 0:12:44.800 use AI. Right You could develop AI in an effort 0:12:44.880 --> 0:12:48.360 to try and come up with new biological weapons. For example, 0:12:48.440 --> 0:12:53.439 that's often a scenario that's cited by concerned critics of 0:12:53.520 --> 0:12:58.080 artificial intelligence. That's certainly something that could potentially happen. So 0:12:58.120 --> 0:13:01.200 again Altman saying, well, you need to have the right 0:13:01.280 --> 0:13:05.520 team responsible to develop AI in a way that is 0:13:05.559 --> 0:13:09.320 most likely to benefit people and to protect people from 0:13:09.880 --> 0:13:15.320 malicious or badly designed AI. So Altman makes an argument 0:13:15.400 --> 0:13:19.840 that machine intelligence could hit an inflection point once recursive 0:13:19.880 --> 0:13:25.200 self improvement becomes a real possibility. That means, if we 0:13:25.240 --> 0:13:28.080 get to the point where we can create machines that 0:13:28.160 --> 0:13:32.559 are smart enough to reprogram themselves, and to reprogram themselves 0:13:32.600 --> 0:13:35.480 in a way that is better than what humans could do, 0:13:36.160 --> 0:13:40.920 so to program these machines at a higher than human 0:13:41.000 --> 0:13:45.200 level of capability, a superhuman capability, if you will, then 0:13:45.360 --> 0:13:48.520 machines suddenly engage in self improvement and can do so 0:13:48.600 --> 0:13:52.440 at increasingly shorter intervals. They get better at doing the 0:13:52.480 --> 0:13:55.840 thing that they're doing, so they get better at improving themselves, 0:13:56.080 --> 0:13:58.880 and they improve themselves over and over, and this becomes 0:13:58.920 --> 0:14:02.920 a version of the singularity, which is a moment where 0:14:03.080 --> 0:14:06.200 change is so sudden and it's happening all the time 0:14:06.400 --> 0:14:09.800 that effectively it becomes impossible to even describe the present, 0:14:10.600 --> 0:14:13.640 everything will change and continue to change at a rate 0:14:13.720 --> 0:14:18.440 that's beyond our ability to describe. Altman says, we might 0:14:18.440 --> 0:14:21.240 be creeping toward that now, and maybe we're creeping toward 0:14:21.280 --> 0:14:23.080 it at a rate that's just impossible for us to 0:14:23.120 --> 0:14:26.560 notice because it's so gradual. That makes it really tricky, 0:14:26.680 --> 0:14:30.120 because it could be that it goes from it's happening 0:14:30.160 --> 0:14:33.720 so slowly that we can't notice it, to it's happening 0:14:33.800 --> 0:14:38.360 so quickly that we are unable to describe it, and 0:14:38.400 --> 0:14:40.640 that there's no point in the middle where we can say, 0:14:41.560 --> 0:14:45.080 wait a second. So in part two of his blog posts, 0:14:45.120 --> 0:14:48.440 Altman makes a clear argument. He says, quote, the US 0:14:48.560 --> 0:14:53.680 government and all other governments should regulate the development of SMI. 0:14:54.080 --> 0:14:57.240 In an ideal world, regulation would slow down the bad 0:14:57.280 --> 0:15:00.720 guys and speed up the good guys. It seems like 0:15:00.760 --> 0:15:03.600 what happens with the first SMI to be developed will 0:15:03.600 --> 0:15:08.360 be very important end quote. Essentially, what Altman is arguing 0:15:08.360 --> 0:15:12.320 here is that if ethical researchers develop a superhuman machine 0:15:12.360 --> 0:15:16.360 intelligence first, they can employ that SMI to prevent the 0:15:16.400 --> 0:15:21.640 development or deployment of malevolent or poorly built SMIs. So 0:15:21.680 --> 0:15:26.280 we unleash our good guy Superman against their bad guy 0:15:26.440 --> 0:15:31.560 General Zod or you know whatever whichever superhero supervillain pairing 0:15:31.600 --> 0:15:35.120 you happen to like. Interestingly, this is going to come 0:15:35.160 --> 0:15:37.720 back again when we talk about Altman and his appearances 0:15:37.760 --> 0:15:41.240 around the world while talking about the potential for AI regulations. 0:15:42.160 --> 0:15:44.520 Before we dive any further into this, let's take a 0:15:44.560 --> 0:15:47.320 quick break to thank our sponsors and we'll be right back. 0:15:56.720 --> 0:15:59.360 So Altman went so far in his blog posts to 0:15:59.400 --> 0:16:03.640 say that he thinks, generally speaking, that tech is often overregulated, 0:16:04.040 --> 0:16:06.280 but on the flip side, he doesn't want to live 0:16:06.320 --> 0:16:09.080 in a world that has no regulation at all. In 0:16:09.120 --> 0:16:12.080 some cases, you can see regulation as a necessary evil 0:16:12.600 --> 0:16:16.920 that maybe it does slow down innovation or it has 0:16:17.160 --> 0:16:21.960 unintended consequences, but in the absence of regulation you can 0:16:22.040 --> 0:16:25.280 have some really poorly thought out deployments that can cause 0:16:25.320 --> 0:16:29.200 a lot of harm. From twenty fifteen to twenty eighteen, 0:16:29.800 --> 0:16:35.680 Open Ai operated as a nonprofit organization. The organization championed 0:16:35.760 --> 0:16:39.920 the open part of its name, claiming that it would 0:16:39.960 --> 0:16:44.360 freely share research and its patents with AI researchers all 0:16:44.400 --> 0:16:47.680 around the world, all in an effort to ensure safety 0:16:47.720 --> 0:16:51.320 in AI development. Greg Brockman, one of the co founders, 0:16:51.600 --> 0:16:55.600 identified a short list of top AI researchers, and the 0:16:55.960 --> 0:16:58.560 organization as a whole began to recruit several of them 0:16:58.560 --> 0:17:02.560 to join open ai as the first employees of the organization. 0:17:03.040 --> 0:17:07.119 The talent helped attract more talent. Some folks said they 0:17:07.160 --> 0:17:10.520 actually joined open ai because it was where you could 0:17:10.520 --> 0:17:13.560 work on really exciting research with the most brilliant and 0:17:13.760 --> 0:17:17.600 talented people in the discipline, even though it would mean 0:17:17.640 --> 0:17:20.359 you wouldn't be making as much money there as you 0:17:20.400 --> 0:17:23.960 could somewhere else. Even high paid individuals at companies like 0:17:24.040 --> 0:17:27.160 Google found themselves switching jobs for the chance to work 0:17:27.160 --> 0:17:31.239 on something that they saw as important and challenging and 0:17:31.280 --> 0:17:35.680 potentially critical to the survival of humanity. One of those 0:17:35.720 --> 0:17:38.840 people who would also be listed as a co founder 0:17:38.920 --> 0:17:44.040 of open ai was Ilia Sutzkiver, who would become chief 0:17:44.080 --> 0:17:47.959 Scientist at open ai and would join the board of directors. 0:17:48.600 --> 0:17:52.840 Musk reportedly played a critical role in recruiting Sutzkiver over 0:17:52.840 --> 0:17:55.120 to open Ai. Like it went back and forth between 0:17:55.560 --> 0:17:58.080 open Ai and Google, which really wanted to hold on 0:17:58.119 --> 0:18:01.320 to him, and reportedly the Musk was a big reason 0:18:01.400 --> 0:18:05.440 why Sutskever eventually moved over to open ai, and also 0:18:05.480 --> 0:18:08.639 Ilia Sutzkever is one of the people who would ultimately 0:18:09.359 --> 0:18:12.320 be part of the decision making group that fired Sam 0:18:12.359 --> 0:18:16.200 Altman this past weekend. Anyway, we're up to twenty eighteen, 0:18:16.240 --> 0:18:19.280 and behind the scenes there was drama a bruin, and 0:18:19.400 --> 0:18:23.240 much of it was in the cauldron known as Elon Musk. 0:18:24.200 --> 0:18:26.960 Not a big surprise there, right, because Elon Musk is 0:18:27.080 --> 0:18:30.040 kind of a magnet for drama in the tech sphere 0:18:30.200 --> 0:18:33.440 and the business sector. So Musk was on the board 0:18:33.440 --> 0:18:36.320 of directors for open Ai, but in twenty eighteen he 0:18:36.600 --> 0:18:40.800 left open ai entirely, and the official story was that 0:18:40.960 --> 0:18:44.000 Musk chose to step down because of a potential conflict 0:18:44.040 --> 0:18:46.679 of interest because there he was on the board of 0:18:46.680 --> 0:18:50.960 directors for an organization working on artificial intelligence, but he 0:18:51.119 --> 0:18:55.119 also was CEO of Tesla, a car company that was 0:18:55.119 --> 0:18:59.439 pushing hard to develop and deploy autonomous driving capabilities to 0:18:59.480 --> 0:19:03.240 the market, and autonomous driving is of course a subset 0:19:03.520 --> 0:19:07.560 of artificial intelligence, so stepping down was the responsible thing 0:19:07.600 --> 0:19:10.680 to do because of this potential conflict of interest between 0:19:10.680 --> 0:19:15.040 the two companies. There was, however, more to his decision 0:19:15.119 --> 0:19:19.280 than just that. So, according to Business Insider, Musk was 0:19:19.359 --> 0:19:23.760 not happy with open Ay's progress. He compared it negatively 0:19:23.840 --> 0:19:27.000 to Google. He was saying Google is spending huge amounts 0:19:27.000 --> 0:19:31.080 of money and is getting ahead in artificial intelligence research, 0:19:31.520 --> 0:19:37.160 and he argued Musk argued that Google, and he specifically 0:19:37.200 --> 0:19:42.120 targeted Larry Page in this criticism, was not paying any 0:19:42.160 --> 0:19:45.439 attention to safety, that safety was not a factor when 0:19:45.480 --> 0:19:49.479 it came to Google's approach to artificial intelligence, and so 0:19:49.720 --> 0:19:52.479 that was one of the things he said that it 0:19:52.560 --> 0:19:55.600 was critical of open ai, saying you're not doing enough, 0:19:56.119 --> 0:19:59.240 and he was kind of pointing at Sam Altman as 0:19:59.320 --> 0:20:01.919 the reason for them, that Altman's leadership was the reason 0:20:02.000 --> 0:20:08.040 why open ai was lagging behind. So Musk then reportedly 0:20:08.880 --> 0:20:14.600 went to other co founders of open ai, including Sam Altman, 0:20:15.080 --> 0:20:18.000 and essentially he said, I want to run open ai, 0:20:18.640 --> 0:20:21.040 and he was told in no uncertain terms that this 0:20:21.160 --> 0:20:24.600 would not happen, and so, again, according to Business Insider, 0:20:24.720 --> 0:20:27.960 Musk decided to take his ball and leave. His ball 0:20:28.000 --> 0:20:31.640 also included a sizable investment or donation to open ai, 0:20:32.680 --> 0:20:35.480 so when he left, he left with a whole bunch 0:20:35.520 --> 0:20:38.119 of money that otherwise was going to go to the 0:20:38.280 --> 0:20:42.520 organization and didn't. Musk would later say he disagreed with 0:20:42.560 --> 0:20:45.720 the direction of open ai and that the company wasn't 0:20:45.920 --> 0:20:50.560 nearly as open as its name would suggest. That last 0:20:50.600 --> 0:20:54.080 criticism happened after open Ai would create a for profit 0:20:54.119 --> 0:20:58.679 company in twenty nineteen. Musk actually leveled the lack of 0:20:58.720 --> 0:21:03.040 openness at open Ai that critique. Around twenty twenty, Musk 0:21:03.119 --> 0:21:06.960 also was founding his own ai research organization and would 0:21:06.960 --> 0:21:10.879 occasionally throw shade at open Ai and Sam Altman. And 0:21:11.160 --> 0:21:14.200 I am not an Elon Musk fan. Most of y'all 0:21:14.200 --> 0:21:16.640 know this. I'm not a huge fan of Elon Musk. However, 0:21:17.000 --> 0:21:19.360 at least some of the criticisms he had toward open 0:21:19.400 --> 0:21:23.160 Ai I actually agree with, or at least I think 0:21:23.320 --> 0:21:26.040 they were true, like the fact that open ai was 0:21:26.080 --> 0:21:30.400 becoming less open. I think that criticism has merit. Meanwhile, 0:21:30.440 --> 0:21:33.440 open Ai was in a pretty tough position because, as 0:21:33.480 --> 0:21:39.199 it turns out, artificial intelligence research is expensive, so you 0:21:39.280 --> 0:21:42.479 need access to a whole lot of compute power and 0:21:42.560 --> 0:21:45.800 that's not cheap, and then you also need to have 0:21:45.880 --> 0:21:49.280 the money to attract the best talent, especially if your 0:21:49.320 --> 0:21:55.360 goal is to be the first to develop superhuman machine 0:21:55.400 --> 0:22:00.440 intelligence that is ethically sound. Like, if that's your goal 0:22:01.040 --> 0:22:04.600 and you need to outpace everybody else who's also working 0:22:04.680 --> 0:22:08.960 on developing superhuman machine intelligence, you got to spend the 0:22:08.960 --> 0:22:12.000 big bucks to get the top of the class to 0:22:12.080 --> 0:22:16.399 come over to your organization. And a nonprofit organization is 0:22:16.520 --> 0:22:18.680 just not the fastest way to gather huge amounts of 0:22:18.720 --> 0:22:21.879 money needed to fund research and operations. It would be 0:22:21.920 --> 0:22:25.240 way easier if you could get investors to pour money in, 0:22:25.520 --> 0:22:29.080 but investors want a return. Meanwhile, a nonprofit is a 0:22:29.080 --> 0:22:33.040 place where you donate money. You're not expecting return on 0:22:33.080 --> 0:22:36.520 your donation. It's no an investment. This is what led 0:22:36.560 --> 0:22:39.840 to the decision to create a for profit arm of 0:22:39.880 --> 0:22:43.800 open ai, which in turn would generate money that could 0:22:43.880 --> 0:22:47.960 be theoretically at least used by the nonprofit part of 0:22:48.040 --> 0:22:52.320 open ai to further the original organization's goals and mission. 0:22:52.920 --> 0:22:56.840 So the result was open Ai LP, which open Ai 0:22:56.960 --> 0:23:00.800 called a capped profit company. So what the heck is 0:23:00.840 --> 0:23:04.399 a capped profit company? That's actually a really good question, 0:23:04.520 --> 0:23:10.320 because I've found two somewhat conflicting answers from various sources. 0:23:10.400 --> 0:23:13.560 Like they lay it out in two different ways that 0:23:13.640 --> 0:23:17.199 are similar but distinct. So I'm going to give you 0:23:17.280 --> 0:23:19.840 both of the ways that it has been explained in 0:23:19.960 --> 0:23:22.840 various sources, because I'm gonna be honest with y'all. I'm 0:23:22.840 --> 0:23:25.040 not a business person. Despite the fact that I have 0:23:25.200 --> 0:23:28.120 hosted a business podcast of the Best, I'm not really 0:23:28.119 --> 0:23:30.760 a business person, so I can't pretend like I have 0:23:31.240 --> 0:23:34.280 a firm grip on this. And also, open ai was 0:23:34.359 --> 0:23:37.159 kind of charting new territory while they were announcing this. 0:23:37.680 --> 0:23:42.120 But here are the two ways that it is frequently described. 0:23:42.400 --> 0:23:46.840 So version number one means that open ai would accept 0:23:46.880 --> 0:23:50.880 investments from venture capitalists and that would pay out returns 0:23:50.880 --> 0:23:54.600 on those investments from profits, but only up to a 0:23:54.600 --> 0:23:58.240 certain amount. So in open AI's case, the early backers, 0:23:58.280 --> 0:24:00.640 the people who first poured money in to open ai, 0:24:01.280 --> 0:24:04.800 would have a cap of one hundred times their initial investment. 0:24:05.359 --> 0:24:08.880 So let's take a very simple scenario. Let's say that 0:24:08.920 --> 0:24:11.200 some kids in your neighborhood want to start a lemonade 0:24:11.240 --> 0:24:15.959 stand and you invest one dollar into their lemonade stand. Now, 0:24:16.040 --> 0:24:18.320 let's say the kids running the stand turn out to 0:24:18.320 --> 0:24:23.160 be business geniuses and your dollar investment helps lead that 0:24:23.280 --> 0:24:29.600 stand into making tens of thousands of dollars in profits, Like, 0:24:29.680 --> 0:24:32.560 even after the expenses, these kids are raking in tens 0:24:32.560 --> 0:24:37.520 of thousands of dollars. However, when you invested, you did 0:24:37.560 --> 0:24:41.359 so knowing there was a one hundred time cap on returns, 0:24:41.760 --> 0:24:43.560 So that means the most you're ever going to get 0:24:43.600 --> 0:24:46.440 from the stand is one hundred dollars. It's a one 0:24:46.480 --> 0:24:50.240 hundred times return on your investment. Meanwhile, those snot nosed 0:24:50.280 --> 0:24:52.560 kids who never could have made the stand without your 0:24:52.640 --> 0:24:56.800 dollar are pocketing thousands of bucks and they're franchising across 0:24:56.840 --> 0:25:01.000 the town, those rotten kids. Anyway, that's one version of 0:25:01.040 --> 0:25:05.400 how the capped profit structure work. It works, investors can 0:25:05.400 --> 0:25:07.680 make a return, but only up to a certain amount, 0:25:08.359 --> 0:25:11.080 the early backers being one hundred times whatever they put in, 0:25:11.480 --> 0:25:13.920 and that means if they put in ten million dollars, 0:25:14.240 --> 0:25:16.920 they could potentially make as much as a billion dollars 0:25:16.960 --> 0:25:20.919 in returns if open ai profited that much. So you 0:25:20.960 --> 0:25:25.600 know it does add up. However, there is a second 0:25:25.680 --> 0:25:29.840