WEBVTT - The Bloomberg Technology Summit 0:00:00.080 --> 0:00:02.120 I'm Caroline Heinde a Bloomberg's WORLDEAD quarters in. 0:00:02.080 --> 0:00:04.960 New York and Imed Ludlow in San Francisco. This is 0:00:05.000 --> 0:00:08.720 a special edition of Bloomberg Technology coming to you live 0:00:09.039 --> 0:00:11.120 from the Bloomberg Technology Summit in Caroline. 0:00:11.119 --> 0:00:11.960 What a way to start. 0:00:12.280 --> 0:00:15.680 Man of the moment, really, Sam Altman open AI CEO, 0:00:15.800 --> 0:00:18.960 so much to recap, but at a top level, saying 0:00:19.000 --> 0:00:23.439 we need global regulation with a capability threshold. Things will 0:00:23.440 --> 0:00:26.040 go wrong in AI, but the future is bright from 0:00:26.040 --> 0:00:27.160 a technology. 0:00:26.640 --> 0:00:32.279 Perspective, ultimately seemingly optimistic. Also some clarity really around his 0:00:32.360 --> 0:00:37.000 own motivating forces, wanting to make impact, wanting to further 0:00:37.479 --> 0:00:42.199 technology for good, and really sort of seemingly at odds 0:00:42.240 --> 0:00:44.960 with why humanity, why most people really want to know 0:00:44.960 --> 0:00:47.200 why he has not much skin in the game financially. 0:00:47.280 --> 0:00:48.839 He seems to be saying like he's going to make 0:00:48.880 --> 0:00:50.960 money in the future, don't you worry about it? Also, 0:00:51.000 --> 0:00:53.720 I thought the interesting perspective on China and Russia and 0:00:53.760 --> 0:00:56.000 the fact that ultimately he doesn't have great insight there 0:00:56.040 --> 0:00:56.800 at the moment either. 0:00:57.280 --> 0:00:59.720 Me too that the theme of the Bloomberg Technology Summit 0:00:59.760 --> 0:01:02.600 is the turning point in Sam Outman talks about how 0:01:02.600 --> 0:01:05.200 little we know about what's going on in China in 0:01:05.240 --> 0:01:08.280 the field of artificial intelligence. But he also framed this 0:01:08.400 --> 0:01:12.759 is the biggest step for mankind in terms of technology, 0:01:13.040 --> 0:01:14.680 and I really think that's going to be a discussion 0:01:14.680 --> 0:01:15.399 here all day long. 0:01:15.720 --> 0:01:18.160 Yeah, we've got more AI discussion coming right up. Bluing 0:01:18.200 --> 0:01:20.759 Bog Technology Summit is still going on. Emily chang Am 0:01:20.760 --> 0:01:23.680 Bradstone And now sitting down, we've read Hoffmann of Graylock 0:01:23.800 --> 0:01:26.640 and co founder of DeepMind, now CEO of Inflection AI 0:01:26.959 --> 0:01:30.880 stuff ustlument In brands for digital influencers. 0:01:31.440 --> 0:01:33.800 Our belief is that everyone is also going to have 0:01:33.840 --> 0:01:37.560 a personal AI, one that is aligned to your interests, 0:01:38.120 --> 0:01:41.440 on your team, in your corner, gets to know you 0:01:41.680 --> 0:01:45.240 and really forms a trusted relationship with you over time. 0:01:45.360 --> 0:01:49.200 It'll be like a confidant, a conciliary there for you 0:01:49.240 --> 0:01:51.720 when you need to make tough decisions, but also a 0:01:51.800 --> 0:01:57.200 chief of staff, you know, scheduling, organizing, prioritizing, booking. Your 0:01:57.240 --> 0:02:00.680 AI is really going to be your digital representation, you know, 0:02:00.800 --> 0:02:05.280 negotiating on your behalf, interacting with other sales AIS that 0:02:05.360 --> 0:02:07.720 are trying to, you know, encourage you to buy something 0:02:08.120 --> 0:02:10.079 and helping you to you know, get a great deal, 0:02:10.680 --> 0:02:13.519 and like for example, it's the whole set of things. 0:02:13.520 --> 0:02:14.880 So, like example, one of the things I just found 0:02:14.880 --> 0:02:17.000 out yesterday is one of our team members at Greylock 0:02:17.600 --> 0:02:21.120 actually in Factor is using Pie for parenting advice. Oh cool, right, 0:02:21.120 --> 0:02:22.640 which is really awesome. And that's the kind of thing 0:02:22.639 --> 0:02:24.120 where you're like, great, but. 0:02:24.200 --> 0:02:26.440 Read why is this an opportunity for a startup? 0:02:26.560 --> 0:02:29.680 Isn't a Microsoft or a Google Bar just a little 0:02:29.720 --> 0:02:32.640 closer to the customer and in a better position to 0:02:32.720 --> 0:02:33.200 offer that. 0:02:33.160 --> 0:02:36.200 Personal digital assistant? Well, part of what's I mean? 0:02:36.240 --> 0:02:38.840 You know, obviously I am on the board of Microsoft 0:02:40.320 --> 0:02:41.920 and you know it is on the board of open Ai. 0:02:42.520 --> 0:02:44.280 And part of the thing is what I love about 0:02:44.280 --> 0:02:46.799 startups is that you have a unique vision where you're 0:02:46.880 --> 0:02:50.120 unfettered by the other aspects of your business and you're 0:02:50.120 --> 0:02:53.639 building it. So part of the idea that Mistafa the 0:02:53.800 --> 0:02:57.160 entire Inflection team, uh you know, kind of came up with, 0:02:57.240 --> 0:02:59.520 and you know, I, you know, have a little bit 0:02:59.560 --> 0:03:02.320 of fingerpas here too, is that IQ is not the 0:03:02.360 --> 0:03:03.440 only thing that matters here. 0:03:03.520 --> 0:03:04.600 EQ matters as. 0:03:04.520 --> 0:03:07.880 Well, And so how do you have this personal intelligence PI? 0:03:09.200 --> 0:03:10.520 Be helpful to you. 0:03:10.720 --> 0:03:11.840 Obviously very high. 0:03:11.760 --> 0:03:14.919 End you know, IQ, but also EQ And that was 0:03:14.960 --> 0:03:17.120 one of the reasons I use the parenting example because 0:03:17.200 --> 0:03:18.679 obviously there's a question of how you do it, but 0:03:18.720 --> 0:03:21.400 it's also how you connect with people what you do 0:03:21.480 --> 0:03:23.600 in order to amblify that, and I think that's one 0:03:23.600 --> 0:03:27.639 of the things that you know, the inflection folks are 0:03:27.680 --> 0:03:29.399 doing better than everyone else. 0:03:29.480 --> 0:03:31.079 So can we talk about the EQ thing? 0:03:31.120 --> 0:03:34.080 Because I was asking, I asked PI what we should 0:03:34.080 --> 0:03:37.400 ask you? It's questions, We're fine, And then I was like, 0:03:37.560 --> 0:03:39.880 are you going to tell them that we had this conversation? 0:03:40.760 --> 0:03:45.440 And it responded, ha ha, I'm just a computer program. 0:03:45.440 --> 0:03:47.520 I can't tell anyone anything. But I assure you our 0:03:47.520 --> 0:03:50.600 conversations are confidential. I do think it's pretty cool that 0:03:50.640 --> 0:03:53.600 we're having this meta conversation about your interview with my creators. 0:03:54.040 --> 0:03:59.360 Amazing point, And it was actually the personality that shocked me. 0:04:00.280 --> 0:04:02.800 Can you explain how you design that? 0:04:03.400 --> 0:04:08.520 I mean, we've deliberately designed it to be patient, curious kind. 0:04:09.120 --> 0:04:11.920 One of the things that first struck us like a 0:04:12.000 --> 0:04:13.680 year and a half ago when we started working on 0:04:13.720 --> 0:04:16.880 this is like, what makes for great conversation when you 0:04:16.920 --> 0:04:20.520 feel that sense of flow and that sense of energy 0:04:20.520 --> 0:04:24.120 and connection with another person. And I think most of 0:04:24.160 --> 0:04:27.680 the time it's when you feel heard and understood. You've 0:04:27.720 --> 0:04:30.760 received a little bit of affirmation. But you know, it's 0:04:30.760 --> 0:04:34.400 not completely sycophantic, right, It doesn't just agree with you 0:04:34.480 --> 0:04:35.279 at every moment. 0:04:35.680 --> 0:04:37.120 It can be a little bit challenging. 0:04:37.279 --> 0:04:39.960 It has boundaries, so you can push it on certain 0:04:40.000 --> 0:04:42.920 topics and it'll take a position, and that's really healthy. 0:04:43.400 --> 0:04:45.359 But also it's just curious, you know. 0:04:45.440 --> 0:04:48.760 I mean, I think many people are excited by the idea, 0:04:48.839 --> 0:04:51.720 especially the users that we have of someone asking them 0:04:51.720 --> 0:04:54.320 lots of questions about the topic that they're interested in. 0:04:54.680 --> 0:04:56.880 You know, we don't always have someone in our lives 0:04:56.880 --> 0:05:00.919 who is as knowledgeable and as passionate about all favorite 0:05:00.960 --> 0:05:02.920 topics as. 0:05:02.720 --> 0:05:03.279 We might like. 0:05:03.520 --> 0:05:05.880 And so that's where pie comes in because it's super 0:05:05.920 --> 0:05:07.240 knowledgeable and engaging. 0:05:07.560 --> 0:05:10.240 So it sounds like a competitor to chat gept. 0:05:10.440 --> 0:05:11.840 Is it a competitor to chat chept? 0:05:11.960 --> 0:05:13.560 And how does that make sense if you're also an 0:05:13.560 --> 0:05:14.680 investor in open ai? 0:05:15.160 --> 0:05:17.760 Well, so I think there's going to be a number 0:05:17.760 --> 0:05:19.440 of agents. I don't think there's going to be one 0:05:19.480 --> 0:05:22.520 agent to rule them all. You know, a ring and mortar. 0:05:23.320 --> 0:05:25.280 But is chatchipt going to rule most of us? 0:05:25.320 --> 0:05:27.240 I mean no, I don't think so. I think that. 0:05:27.400 --> 0:05:28.320 I mean it's partially. 0:05:28.400 --> 0:05:31.960 It's it's it's like the same reason we talk to 0:05:32.000 --> 0:05:34.839 different people for different aspects of our lives. Right, So 0:05:35.320 --> 0:05:38.039 you know, this person we talked to about our passion 0:05:38.080 --> 0:05:41.200 about snowboarding, this person we talked to about what's going 0:05:41.240 --> 0:05:43.159 on in the country, this person we talk to, like 0:05:43.200 --> 0:05:45.480 we have a pantheon, same thing, We'll have a pantheon 0:05:45.560 --> 0:05:49.279 of different kinds of AI. One will go You know, 0:05:49.520 --> 0:05:52.680 you ask chat GPT, you know, how do you know 0:05:52.839 --> 0:05:56.080 comfort a friend who's lost the treasured pet? 0:05:56.120 --> 0:05:58.200 And is us here are five possible ways you might 0:05:58.240 --> 0:05:58.560 do that? 0:05:59.000 --> 0:06:02.159 You ask pie, Oh, that sounds like it's really hard 0:06:02.240 --> 0:06:04.760 for you and your friend. You know your friend, Well, 0:06:04.800 --> 0:06:06.960 what would count as being present there for your friend? 0:06:07.000 --> 0:06:09.120 And it still it has the IQ, it has the 0:06:09.120 --> 0:06:11.840 five reasons, but it goes through that style of going 0:06:11.880 --> 0:06:14.800 through it. And I think these are different interaction experiences 0:06:14.839 --> 0:06:18.040 And the question is which one do you want this minute? 0:06:18.320 --> 0:06:20.400 Which one do you want more in your life? You 0:06:20.400 --> 0:06:23.599 know that kind of thing and part of the design 0:06:23.640 --> 0:06:26.119 of pies. How do we help you be the best 0:06:26.160 --> 0:06:30.160 you right like? And that doesn't mean by being sycophantic, 0:06:30.240 --> 0:06:32.839 for example, that's the question of you know, like, ask 0:06:32.880 --> 0:06:34.960 you questions about what do you think about this, and 0:06:35.440 --> 0:06:37.400 how do you go through it and help you navigate 0:06:37.440 --> 0:06:39.760 your life by being helpful in that way. 0:06:40.120 --> 0:06:42.919 So Sam Allman was just on stage talking as. 0:06:42.720 --> 0:06:47.280 He sometimes does, about the theoretical dangers of AGI and 0:06:47.960 --> 0:06:52.080 chat GPT nine and I'm just curious if you guys 0:06:52.120 --> 0:06:56.160 think generally that's a discussion worth having and where you 0:06:56.200 --> 0:07:00.800 each are on this spectrum of existential data from AI. 0:07:01.800 --> 0:07:03.640 Well, let's start with the fact that one of the 0:07:03.680 --> 0:07:08.000 things that I think is is dangerous about the existential 0:07:08.040 --> 0:07:11.000 discussion is that it blinds us from some of the 0:07:11.600 --> 0:07:15.560 nearer things, which is AI's amplification intelligence as for the 0:07:15.600 --> 0:07:18.440 impromptu book, and so it amplifies human beings, and so 0:07:18.680 --> 0:07:20.480 we see a bunch of good things in human beings. 0:07:20.800 --> 0:07:23.040 AI tutor is AI doctor is a bunch of other things. 0:07:23.280 --> 0:07:26.680 There's also bad human beings, and so what to do 0:07:26.800 --> 0:07:30.040 about bad human beings? Doing you know with AI is 0:07:30.080 --> 0:07:33.360 also part of the portfolio mix of how you shape 0:07:33.360 --> 0:07:36.560 this technology. That's one of the reasons why Mustafa's a book, 0:07:36.600 --> 0:07:40.480 The Coming Wave, I highly recommend, and I will hand 0:07:40.520 --> 0:07:41.960 the next part of the answer. 0:07:41.680 --> 0:07:44.120 Over to you and wait, let me frame it them 0:07:44.160 --> 0:07:47.760 with that. Sam probably left the building right now. 0:07:48.440 --> 0:07:51.800 Frank, like, where do you kind of disagree with some 0:07:51.840 --> 0:07:54.440 of the alarmism notes that he has sounded. 0:07:54.640 --> 0:07:58.040 Look, I think it's easy to speculate around what a 0:07:58.120 --> 0:08:01.120 GPT nine might look like in you know, six more 0:08:01.280 --> 0:08:05.120 orders of magnitude of compute would be eyewatering. And I'm 0:08:05.240 --> 0:08:08.840 absolutely with those concerns around existential risk. But if you 0:08:08.880 --> 0:08:11.480 play that out, that is many, many years, and it is, 0:08:11.760 --> 0:08:14.600 you know, unclear what actually happens at that scale. 0:08:14.640 --> 0:08:16.440 What's much more clear and what you know. 0:08:16.480 --> 0:08:18.080 A lot of the themes I explore in the new 0:08:18.080 --> 0:08:21.120 book is that in the near term, we're about to 0:08:21.240 --> 0:08:26.240 empower many many people, if not ultimately everybody, with access 0:08:26.280 --> 0:08:29.960 to the ability to amplify their existing power. Right, this 0:08:30.160 --> 0:08:33.640 is not just a knowledge engine, but in time it'll 0:08:33.960 --> 0:08:36.480 allow you to take actions. Right, You'll be able to 0:08:36.480 --> 0:08:40.120 make recommendations, buy things, book things, and that's going to 0:08:40.160 --> 0:08:43.720 get smaller and cheaper and therefore proliferate far and wide. 0:08:44.240 --> 0:08:47.560 That is going to cause a dramatic instability and potentially 0:08:47.640 --> 0:08:50.040 a threat to the nation state because anyone who has 0:08:50.040 --> 0:08:52.800 an agenda or is trying to sort of push a 0:08:52.800 --> 0:08:56.400 political outcome is suddenly going to see the barrier to 0:08:56.600 --> 0:09:00.200 entry to that kind of scaled impact lowered. And so 0:09:00.240 --> 0:09:02.679 there's going to be a real question around how states 0:09:02.800 --> 0:09:05.560 kind of manage that distribution of power. And I think 0:09:05.600 --> 0:09:08.800 there'll be a tendency to lunge towards, you know, slightly 0:09:08.840 --> 0:09:13.280 more authoritarian, more surveillance based mechanisms to prevent that proliferation, 0:09:13.360 --> 0:09:17.080 which would both be bad for innovation and obviously dystopian. 0:09:16.559 --> 0:09:18.520 From a political governance perspective. 0:09:18.640 --> 0:09:21.160 Well, hang on, because Sam and I were talking backstage 0:09:21.200 --> 0:09:25.800 as well, and he's much more confident that AI will 0:09:25.880 --> 0:09:29.840 lead us to a more equal world than an unequal world. 0:09:29.960 --> 0:09:33.120 Are you saying that there's a better chance of economic 0:09:33.160 --> 0:09:37.880 dislocation and social dislocation as a result of this technology? 0:09:38.160 --> 0:09:40.880 So both are likely to be true actually, which is 0:09:40.920 --> 0:09:43.960 a little bit surprising. But if you look back over 0:09:44.000 --> 0:09:47.840 the last fifty years, the transistor, the last wave which 0:09:47.920 --> 0:09:51.480 enabled the personal computer, has clearly made us in many 0:09:51.480 --> 0:09:54.120 ways more equal whether you're a billionaire or you earn 0:09:54.200 --> 0:09:57.200 twenty thousand dollars a year, we all get access to 0:09:57.280 --> 0:10:00.680 the same cutting edge hardware. The smart phone and the 0:10:00.760 --> 0:10:04.240 laptop is broadly at the cutting edge, and everyone else 0:10:04.280 --> 0:10:06.760 will catch up over the next five to ten years. 0:10:07.240 --> 0:10:11.040 We're on the same trajectory with respect to access to intelligence, 0:10:11.360 --> 0:10:15.200 and that is an unbelievable idea. Over the next decade, 0:10:15.480 --> 0:10:17.959 hundreds of millions of people and then billions of people 0:10:18.000 --> 0:10:21.560 will get access to the same expert doctor, the same 0:10:21.720 --> 0:10:26.480 expert educator, the same tool for scheduling, prioritizing, organizing your life. 0:10:26.840 --> 0:10:30.000 That is going to be the most meritocratic moment in 0:10:30.080 --> 0:10:31.360 the history of our species. 0:10:31.600 --> 0:10:32.079 For sure. 0:10:32.840 --> 0:10:36.079 It is a question around how individuals and groups and 0:10:36.200 --> 0:10:39.760 organizations use that power right, because clearly we all have 0:10:39.960 --> 0:10:44.080 conflicting agendas and priorities and goals. And that's basically where 0:10:44.120 --> 0:10:46.040 I think that we end up with significant disruption. 0:10:46.200 --> 0:10:47.800 Okay, read, but that's the class have. 0:10:47.880 --> 0:10:51.000 Full I mean, what is the potential thread to the 0:10:51.040 --> 0:10:55.600 millions of coders, of customer service representatives, the job dislocation, 0:10:56.000 --> 0:11:00.280 the people who are serving jobs whose functions can be 0:11:00.360 --> 0:11:01.600 replaced by AI. 0:11:01.800 --> 0:11:02.840 Your LinkedIn hat on. 0:11:05.040 --> 0:11:08.800 So the closest metaphor that I've come to use to 0:11:08.800 --> 0:11:12.360 describe this moment is it's a steam engine of the mind. 0:11:12.960 --> 0:11:15.560 If you look at this, the former steam engine, you know, 0:11:15.600 --> 0:11:20.559 current steam engine, although obviously majorly amplified. Industrial evolution give 0:11:20.600 --> 0:11:23.960 us superpower of muscles, super of transport, superpower of construction, 0:11:24.440 --> 0:11:24.920 all these ends. 0:11:24.960 --> 0:11:26.520 Now we're going to have superpowers of the mind. 0:11:27.000 --> 0:11:30.640 Now there's a whole bunch of very positive things that 0:11:30.679 --> 0:11:31.200 come out of that. 0:11:31.280 --> 0:11:34.240 I mean, all of the things we have of. 0:11:36.040 --> 0:11:39.280 The increase in wealth that allows medicine, general education, everything 0:11:39.280 --> 0:11:41.520 else comes out of the industrial evolution. I think the 0:11:41.520 --> 0:11:43.120 same thing will be coming out of the steam engine 0:11:43.160 --> 0:11:46.720 of the mind now. But the transition is going to 0:11:46.720 --> 0:11:50.200 be difficult. The transition is going to be okay, well, 0:11:50.520 --> 0:11:54.160 customer service jobs are going to change a whole lot now, engineers, 0:11:54.200 --> 0:11:56.400 I think, you know, look, if you roughly kind of 0:11:56.400 --> 0:11:58.720 go through a company and you go we ten x 0:11:58.720 --> 0:12:01.679 each function, you say you ten x salespeople, great, so 0:12:01.720 --> 0:12:03.920 I want to hire more salespeople want our sales ten 0:12:04.000 --> 0:12:07.120 XT marketing people. It's like, okay, maybe the marketing functions 0:12:07.120 --> 0:12:10.160 will change some so less order entry, more thinking about it. 0:12:10.160 --> 0:12:13.080 But you're still in a business competitive ecosystem. You don't 0:12:13.120 --> 0:12:16.880 want to stop marketing or not be doing it. When 0:12:16.880 --> 0:12:19.040 you get to customer service, you get more replacement. But 0:12:19.160 --> 0:12:21.120 here's part of the thing that I part of the 0:12:21.160 --> 0:12:23.040 reason why I did the book impromptu and are trying 0:12:23.040 --> 0:12:25.640 to orient people, is like, okay, so customer service people. 0:12:25.400 --> 0:12:27.840 There's going to be a transition. Well, AI can help 0:12:27.880 --> 0:12:28.080 with that. 0:12:28.440 --> 0:12:31.000 You can build AIS that help figure out other kinds 0:12:31.000 --> 0:12:33.360 of work and jobs, that can help you find them, 0:12:33.400 --> 0:12:35.000 that can help you learn them, that can help you 0:12:35.080 --> 0:12:37.679 do them. And so that's the thing that we as 0:12:37.679 --> 0:12:40.120 a society need to be doing. So it's like less 0:12:40.160 --> 0:12:42.720 like how do we slow down AI? How do we 0:12:42.800 --> 0:12:46.439 shape it to help the broad swath of humanity in 0:12:46.480 --> 0:12:49.760 this transition. That's where I'm trying to get the dialogue 0:12:49.760 --> 0:12:52.440 and discussion to Moustafa. 0:12:52.800 --> 0:12:55.640 We were just talking to Sam about Google and whether 0:12:55.720 --> 0:12:57.360 or not it's still scary, and he's like, yeah, they're 0:12:57.400 --> 0:13:03.720 still formidable. You quit Mind and Google AI and obviously 0:13:03.760 --> 0:13:06.840 a lot of people have have have left Google. Is 0:13:06.880 --> 0:13:09.119 there something wrong at Google or some sort of innovator's 0:13:09.120 --> 0:13:13.040 dilemma there that will prevent it from truly succeeding on 0:13:13.120 --> 0:13:16.120 generative AI because it underminds its own business model. 0:13:17.000 --> 0:13:22.520 Look, fundamentally, AI stands intention with Google's existing business model. 0:13:22.840 --> 0:13:26.440 It's very hard to eat yourself from within and adapt 0:13:26.600 --> 0:13:29.160 and respond to the coming wave. So for a while 0:13:29.760 --> 0:13:34.319 Google was just idling. You know, I was there on Lambda. 0:13:34.440 --> 0:13:37.080 We had chat GBT before chat gibt. It was just 0:13:37.080 --> 0:13:41.199 like a remarkable feeling internally playing with this incredible tool 0:13:41.360 --> 0:13:43.840 and not really being able to get it out into products. 0:13:44.360 --> 0:13:46.760 It took, you know, the launch of chat CBT to 0:13:46.920 --> 0:13:49.920 kind of threaten Google and really shake everything up. And 0:13:50.320 --> 0:13:53.400 you know, Google's are very formidable and you know, you know, 0:13:53.640 --> 0:13:57.760 organization full of super smart people. So I'm sure they'll 0:13:57.800 --> 0:14:01.080 be just fine. But I can understand why they're stuck 0:14:01.360 --> 0:14:05.520 because nobody wants a personal AI in your pocket that 0:14:05.640 --> 0:14:08.920 is actually funded by ads. Right, you don't want a 0:14:08.960 --> 0:14:11.120 salesman in your pocket that is trying to persuade you 0:14:11.200 --> 0:14:13.040 to go and buy more or do xyz. 0:14:13.640 --> 0:14:15.199 You need fiduciary alignment. 0:14:15.559 --> 0:14:18.280 Right, Your AI has to be on your team, and 0:14:18.280 --> 0:14:20.440 that means ultimately you have to pay for it. And 0:14:20.440 --> 0:14:22.760 if somebody else is paying for it, you have to 0:14:22.800 --> 0:14:25.800 ask yourself what are they trying to persuade you of? 0:14:26.080 --> 0:14:28.840 Are they trying to influence you in some way? And 0:14:28.880 --> 0:14:31.080 I think people are alive to that now and that's 0:14:31.120 --> 0:14:33.840 a real challenge for Google because it's very unclear how 0:14:33.880 --> 0:14:35.480 they're going to manage this transition. 0:14:35.720 --> 0:14:39.160 So we asked Sam if Google scares him, does OpenAI 0:14:39.280 --> 0:14:39.760 scare you? 0:14:40.520 --> 0:14:41.880 Wastaf a first many. 0:14:43.240 --> 0:14:48.200 So no, No, it definitely doesn't. I mean, actually, this 0:14:48.320 --> 0:14:53.400 morning we've announced our new large language model called Inflection one, 0:14:54.080 --> 0:14:56.200 and we set out to build a model that was 0:14:56.560 --> 0:14:59.200 fast enough and more capable than every other model on 0:14:59.240 --> 0:15:03.040 the market for our compute class. So we're very proud that, 0:15:03.160 --> 0:15:05.440 you know, a year on from our launch, we're now 0:15:05.480 --> 0:15:10.240 better than lama Chat, GPT, Palm five forty, Chinchilla, all 0:15:10.240 --> 0:15:13.120 of the other models of our size. And that's super 0:15:13.160 --> 0:15:15.880 important because it powers PI and ultimately it will be 0:15:15.880 --> 0:15:19.520 available as a conversational API, so you can obviously play 0:15:19.520 --> 0:15:22.120 with it at pie dot ai now. And you know, 0:15:22.520 --> 0:15:25.400 I think that demonstrates that with a team of thirty 0:15:25.400 --> 0:15:28.520 five people, you know, in a pretty short order, we're 0:15:28.560 --> 0:15:31.680 able to exceed the cutting edge and now build an 0:15:31.760 --> 0:15:34.920 absolute best in class AI, which is very exciting. And 0:15:34.920 --> 0:15:37.560 obviously we've also managed to do that because we've been 0:15:37.600 --> 0:15:40.320 able to gather a huge amount of capital and some 0:15:40.400 --> 0:15:44.000 great investors and train. In fact, what we now have 0:15:44.200 --> 0:15:47.920 is the largest operational cluster in the world of h 0:15:48.000 --> 0:15:52.200 one hundreds in Vidia's latest chip and that's a huge advantage. 0:15:52.320 --> 0:15:52.640 Okay. 0:15:52.880 --> 0:15:55.240 And one of the things, because I was listening to 0:15:55.280 --> 0:15:57.600 your conversation with Sam, and you know, it's obviously an 0:15:57.640 --> 0:15:59.800 awkward question when you ask him, is like, well, should 0:15:59.800 --> 0:16:02.440 we trust you? Because it's like, yes, trust me. It's like, well, 0:16:02.440 --> 0:16:03.360 that's weird because if. 0:16:03.240 --> 0:16:05.720 Someone says no, you shouldn't. 0:16:05.480 --> 0:16:08.120 Someone says yes, trust me, You're like, well wait a minute, 0:16:08.440 --> 0:16:10.560 but look, I uh, the Open Eye people are really 0:16:10.600 --> 0:16:12.280 great people. I think it's it's governed by a five 0:16:12.320 --> 0:16:14.480 to one C three. There's frequently a bunch of fud 0:16:14.920 --> 0:16:19.160 social media stuff that that that that that that kind 0:16:19.160 --> 0:16:23.440 of obscures that I you know, spent a number of 0:16:23.480 --> 0:16:26.160 years on the board and in the team. I still 0:16:26.160 --> 0:16:30.040 work with them and help them. They are fully paying 0:16:30.080 --> 0:16:33.000 attention to every serious question. I mean, like, for example, 0:16:33.080 --> 0:16:35.480 there Sam went and spent a month. It's crazy, You're 0:16:35.480 --> 0:16:38.200 doing a startup. It spent a month doing a world tour. 0:16:38.040 --> 0:16:38.560 Talking to people. 0:16:38.600 --> 0:16:40.480 Say look, I'm here, you can talk to me about 0:16:40.520 --> 0:16:42.640 your concerns. I want to make sure it's good for humanity. 0:16:42.720 --> 0:16:45.320 I don't just care about this kind of US San 0:16:45.360 --> 0:16:46.320 Francisco text scene. 0:16:46.360 --> 0:16:48.080 I care about what it's impact on humanity is. 0:16:48.360 --> 0:16:50.360 And so I have come here to talk to you, 0:16:50.440 --> 0:16:52.040 So that kind of like. 0:16:52.040 --> 0:16:53.120 I'm engaging conversation. 0:16:53.280 --> 0:16:56.040 I care about this is the kind of reason in 0:16:56.040 --> 0:16:58.680 addition to a five ones three emission structure and everything else. 0:16:58.880 --> 0:17:01.240 And so no, I'm I'm delighted with Hope. 0:17:01.280 --> 0:17:02.840 I don't see how he's getting any work done with 0:17:02.880 --> 0:17:07.360 all this, But I want to change gears quickly. We're 0:17:07.359 --> 0:17:09.240 going to throw some Twitter polls up on the screen 0:17:09.359 --> 0:17:13.000 today and we asked Twitter users where the greatest advancements 0:17:13.080 --> 0:17:14.240 in AI would come from. 0:17:14.280 --> 0:17:16.320 We'll see the results, But I want to ask. 0:17:16.200 --> 0:17:19.800 About maybe retarting with you are the US China policy 0:17:20.040 --> 0:17:24.399 of divestment is, can we reasonably hope that you know, 0:17:24.520 --> 0:17:28.960 doing that will maybe prolong America's advantage in AI? 0:17:29.160 --> 0:17:32.520 Is it a smart strategy? I don't think divestment is 0:17:32.560 --> 0:17:35.720 a smart strategy. I think staying connected is better for 0:17:35.920 --> 0:17:39.920 both the US and China and the world. I think 0:17:40.280 --> 0:17:41.760 competition is very good. 0:17:41.880 --> 0:17:44.840 I think the fact that there's various ways that we 0:17:45.760 --> 0:17:46.040 in the. 0:17:46.040 --> 0:17:48.679 US and the West Coast have a AI lead, I 0:17:48.680 --> 0:17:50.719 think that's a great thing for kind of the values, 0:17:50.760 --> 0:17:53.520 the ecosystem and the kind of the great world order 0:17:53.560 --> 0:17:55.399 that the US should be very proud of in the 0:17:55.480 --> 0:17:57.720 last seventy years. Like, there's lots of things to be 0:17:57.720 --> 0:18:02.200 critical of two, of course, but as a really peaceful time, 0:18:02.359 --> 0:18:04.520 trying to make it business and interconnection is a really 0:18:04.560 --> 0:18:08.000 good thing. And I'm a very strong believer and proponent 0:18:08.040 --> 0:18:11.160 of that. But I think investments not the. 0:18:11.200 --> 0:18:12.879 Ends half of You write in the book that China 0:18:12.920 --> 0:18:16.000 has an explicit national strategy to be the AI leader 0:18:16.040 --> 0:18:16.800 by twenty thirty. 0:18:16.840 --> 0:18:21.000 So do you agree with read that withdrawal divestment as 0:18:21.040 --> 0:18:21.920 a smart strategy. 0:18:22.040 --> 0:18:24.600 I mean, China is already ahead of its own schedule. 0:18:24.720 --> 0:18:28.560 I mean it's publishing significantly more papers on AI than 0:18:28.680 --> 0:18:31.359 we are collectively in the rest of the world. So 0:18:31.760 --> 0:18:35.600 you know, I do think that our export controls were 0:18:35.640 --> 0:18:39.080 effectively a declaration of economic war on China. They were 0:18:39.200 --> 0:18:42.240 very firm and very aggressive and I think it sets 0:18:42.320 --> 0:18:45.199 us out on a hyper adversarial footing. They have a 0:18:45.280 --> 0:18:48.160 huge number of leavers too, and we should expect them 0:18:48.200 --> 0:18:50.360 to use them against us pretty soon, unfortunately. 0:18:50.800 --> 0:18:53.240 So we talked a lot about regulation with Sam, and 0:18:53.280 --> 0:18:56.760 obviously President Biden was just in town here this week 0:18:56.840 --> 0:19:00.640 talking about AI. He met with you know, the President 0:19:00.680 --> 0:19:06.840 and Sindar and Satya in DC. What is the prospect 0:19:06.920 --> 0:19:12.840 for real government oversight that protects users, protects the economy 0:19:14.040 --> 0:19:15.280 and our political process. 0:19:15.840 --> 0:19:18.159 So, look, I think the good news is that the 0:19:18.240 --> 0:19:22.040 administration is taking a very Let's learn how to be 0:19:22.080 --> 0:19:25.280 smart about this. Let's assemble a set of thoughtful people 0:19:25.320 --> 0:19:30.919 from industry yesterday, assembling a thoughtful people from outside of industry, academia, 0:19:30.920 --> 0:19:31.640 other kinds of places. 0:19:31.720 --> 0:19:33.080 We're doing it, and let's learn about it. 0:19:33.359 --> 0:19:37.000 I think they have a number of good people tasked 0:19:37.040 --> 0:19:41.639 with this, and so I'm optimistic in that. But on 0:19:41.680 --> 0:19:43.320