WEBVTT - Smart Talks with IBM - Generative AI: Its Rise and Potential for Society 0:00:04.519 --> 0:00:12.680 Welcome to Tech Stuff, a production from iHeartRadio. Today we 0:00:12.720 --> 0:00:15.680 are witnessed to one of those rare moments in history, 0:00:16.079 --> 0:00:19.279 the rise of an innovative technology with the potential to 0:00:19.440 --> 0:00:24.160 radically transform business and society forever. That technology, of course, 0:00:24.640 --> 0:00:28.200 is artificial intelligence, and it's the central focus for this 0:00:28.360 --> 0:00:32.360 new season of Smart Talks with IBM. Join hosts from 0:00:32.400 --> 0:00:36.120 your favorite Pushkin podcasts as a talk with industry experts 0:00:36.120 --> 0:00:39.760 and leaders to explore how businesses can integrate AI into 0:00:39.760 --> 0:00:43.080 their workflows and help drive real change in this new 0:00:43.200 --> 0:00:46.879 era of AI. And of course, host Malcolm Gladwell will 0:00:46.880 --> 0:00:49.200 be there to guide you through the season and throw 0:00:49.280 --> 0:00:52.040 in his two cents as well. Look out for new 0:00:52.080 --> 0:00:54.960 episodes of Smart Talks with IBM every other week on 0:00:55.000 --> 0:00:59.240 the iHeartRadio app, Apple Podcasts, wherever you get your podcasts 0:00:59.440 --> 0:01:03.680 and learn more are at IBM dot com slash smart Talks. 0:01:06.680 --> 0:01:12.840 All Right, Welcome everybody, you guys excited, Here we go. 0:01:13.160 --> 0:01:16.640 Hello, Hello, Welcome to Smart Talks with IBM, a podcast 0:01:16.680 --> 0:01:22.560 from Pushkin Industries, iHeartRadio and IBM. I'm Malcolm Gladwell. This season, 0:01:22.600 --> 0:01:26.600 we're continuing our conversations with new creators visionaries who are 0:01:26.640 --> 0:01:30.160 creatively applying technology and business to drive change, but with 0:01:30.240 --> 0:01:34.560 a focus on the transformative power of artificial intelligence and 0:01:34.600 --> 0:01:37.479 what it means to leverage AI as a game changing 0:01:37.560 --> 0:01:42.240 multiplier for your business. Today's episode is a bit different. 0:01:42.560 --> 0:01:46.480 I was recently joined on stage by Dario Gill for 0:01:46.520 --> 0:01:48.880 a conversation in front of a live audience at the 0:01:48.920 --> 0:01:53.000 iHeartMedia headquarters in Manhattan. Dario is the senior vice president 0:01:53.240 --> 0:01:56.760 and director of IBM Research, one of the world's largest 0:01:56.840 --> 0:02:01.160 and most influential corporate research labs. We discussed the rise 0:02:01.160 --> 0:02:04.640 of generative AI, what it means for business and society. 0:02:04.960 --> 0:02:08.480 He also explained how organizations that leverage AI to create 0:02:08.560 --> 0:02:13.239 value will dominate in the near future. Okay, let's get 0:02:13.280 --> 0:02:19.840 on to the conversation. Hello everyone, welcome, and I'm here 0:02:19.880 --> 0:02:23.560 with doctor Dario Gil and I wanted to say before 0:02:23.560 --> 0:02:25.160 we get started. This is something I said backstage that 0:02:25.639 --> 0:02:31.920 I feel very guilty today because you're the you know, 0:02:32.680 --> 0:02:34.720 you know, arguably one of the most important figures in 0:02:35.080 --> 0:02:37.799 AI research in the world, and we have taken you 0:02:37.880 --> 0:02:42.160 away from your job for a morning. It's like, if 0:02:42.520 --> 0:02:46.359 you know, Oppenheimer's wife in nineteen forty four said let's 0:02:46.680 --> 0:02:50.440 go and have a little getaway in the Bahamas. It's 0:02:50.520 --> 0:02:52.320 that kind of thing. You know, What do you say 0:02:52.320 --> 0:02:55.400 to your wife, I can't We have got to work 0:02:55.440 --> 0:02:57.680 on this thing I can't tell you about. She's like 0:02:57.760 --> 0:03:00.160 getting me out of Los Alamos. No, So I do 0:03:00.200 --> 0:03:05.040 feel guilty. We've set back AI research by by about 0:03:05.080 --> 0:03:10.080 four hours here. But I wanted to you've been up 0:03:10.080 --> 0:03:13.320 with with ibo for twenty years, twenty years this summer. 0:03:13.680 --> 0:03:15.280 So and how old were you when you Not to 0:03:15.320 --> 0:03:17.040 give away your age, but you were how old when 0:03:17.080 --> 0:03:17.560 you started? 0:03:17.840 --> 0:03:18.560 I was twenty eight? 0:03:18.639 --> 0:03:20.360 Okay, yeah, So I want to go back to your 0:03:20.400 --> 0:03:23.280 twenty eight year old self. Now, if I asked you 0:03:23.360 --> 0:03:26.960 about artificial intelligence, I asked twenty eight year old Dario, 0:03:27.840 --> 0:03:31.080 what does the future hold for AI? How quickly will 0:03:31.520 --> 0:03:35.000 this new technology transform our world? Et cetera, et cetera. 0:03:35.080 --> 0:03:37.080 What would twenty eight year old Darigo have said? 0:03:37.520 --> 0:03:39.760 Well, I think the first thing is that even though 0:03:39.800 --> 0:03:42.160 AI as a feel has been with us for a 0:03:42.200 --> 0:03:45.280 long time since the mid nineteen fifties, at that time, 0:03:45.680 --> 0:03:49.120 AI was not a very polite word to say, meaning 0:03:49.200 --> 0:03:52.600 within the scientific community, people didn't use sort of that 0:03:52.720 --> 0:03:54.800 term that would have said things like, you know, maybe 0:03:54.800 --> 0:03:58.640 I do things relate to machine learning, right, or statistical 0:03:58.680 --> 0:04:00.760 techniques in terms of classific fires and so on. 0:04:01.120 --> 0:04:03.320 But AI had a mixed. 0:04:03.000 --> 0:04:06.000 Reputation, right, it had gone through different cycles of hype, 0:04:06.440 --> 0:04:10.240 and it's also if moments of you know, a lot 0:04:10.280 --> 0:04:15.040 of negativity towards it because of lack of success. And 0:04:15.120 --> 0:04:16.560 so I think that would be the first thing we 0:04:16.640 --> 0:04:19.120 probably say, like AI is like what is that? 0:04:19.240 --> 0:04:21.960 Like, you know, respectable scientists are not working. 0:04:21.720 --> 0:04:25.239 On AI the fin as such, and that really changed 0:04:25.279 --> 0:04:27.640 over the last fifteen years. Only, right, I would say, 0:04:27.640 --> 0:04:30.560 with the advent of deep learning over the last decade 0:04:30.720 --> 0:04:33.679 is when that re enter again the lexicon of saying 0:04:33.680 --> 0:04:36.839 AI and that that was a legitimate thing to work on. 0:04:36.960 --> 0:04:38.440 So I would say that that's the first thing I 0:04:38.440 --> 0:04:40.400 think we would have noticed a contrast twenty years ago. 0:04:40.640 --> 0:04:44.559 Yeah, So what point in your twenty year tenure at 0:04:44.760 --> 0:04:48.520 IBM would you say you kind of snapped into present 0:04:49.160 --> 0:04:50.440 kind of wow mode. 0:04:51.800 --> 0:04:59.680 I would say in a late two thousands, when IBM 0:05:00.520 --> 0:05:05.800 was working on the Jeopardy project and just seeing the 0:05:05.839 --> 0:05:09.279 demonstrations of what could be done in question answering. 0:05:09.720 --> 0:05:12.880 It's literally Jeopardy is this crucial moment in the history 0:05:12.920 --> 0:05:14.680 of Yeah. 0:05:14.160 --> 0:05:16.520 You know, there had been a long and wonderful history 0:05:17.240 --> 0:05:21.000 inside IBM on AI. So so for example, like you know, 0:05:21.080 --> 0:05:23.920 in terms of like these grand challenges, at the very 0:05:23.960 --> 0:05:26.839 beginning of the field founding, which is this famous Dartmouth 0:05:26.880 --> 0:05:31.200 conference that actually IBM sponsored uh TO to create, there 0:05:31.240 --> 0:05:35.800 was an IBM and there called Nathaniel Rochester and there 0:05:35.800 --> 0:05:38.640 were a few others who right after that they started 0:05:38.640 --> 0:05:41.800 thinking about demonstrations of this field. And then for example, 0:05:41.800 --> 0:05:45.039 they created the first you know game to play checkers 0:05:45.320 --> 0:05:48.640 and to demonstrate that you could do machine learning on that. 0:05:49.320 --> 0:05:51.960 Obviously we saw later in the nineties like chess that 0:05:52.040 --> 0:05:54.640 was a very famous example of that, deep Blue with 0:05:54.680 --> 0:05:57.960 Deep Blue right and playing with Caspar and then but 0:05:58.040 --> 0:06:01.120 I think the moment that was really the ones felt like, 0:06:01.240 --> 0:06:03.440 you know, kind of like brute force anticipating, sort of 0:06:03.480 --> 0:06:06.200 like moves ahead. But this aspect of dealing with language 0:06:06.200 --> 0:06:10.560 and question answering felt different, and I think for for 0:06:10.680 --> 0:06:12.919 us internally and many others, was when a moment of 0:06:13.000 --> 0:06:16.000 saying like wow, you know, what are the possibilities here? 0:06:16.440 --> 0:06:19.000 And then soon after that connected to the sort of 0:06:19.040 --> 0:06:22.520 advancements in computing and with deep learning. The last decade 0:06:22.600 --> 0:06:24.520 has just been an all out, you know, sort of 0:06:24.520 --> 0:06:26.960 like front of advancements and that, and I just continued 0:06:27.000 --> 0:06:28.960 to be more and more impressed. And the last few 0:06:29.040 --> 0:06:30.280 years have been remarkable too. 0:06:30.480 --> 0:06:34.960 Yeah, So I ask you three quick conceptual questions before 0:06:34.960 --> 0:06:36.760 we dig into it, just so I sort of get 0:06:36.760 --> 0:06:41.040 a we all get a feel for the shape of AI. 0:06:41.839 --> 0:06:45.880 Question Number one is where are we in the evolution 0:06:46.000 --> 0:06:50.040 of this? So you know the obvious question. We we're 0:06:50.040 --> 0:06:52.640 all suddenly aware of it, we're talking about it. Can 0:06:52.680 --> 0:06:54.760 you give us an analogy about where we are in 0:06:54.800 --> 0:06:57.479 the kind of likely evolution of this is a technology? 0:06:59.640 --> 0:07:03.320 So I think we're on a significant inflection point that 0:07:03.760 --> 0:07:07.840 it feels the equivalent of the first browsers when they 0:07:07.880 --> 0:07:12.080 appear and people imagine the possibilities of the Internet or 0:07:12.120 --> 0:07:16.040 more imagined experience the internet. The Internet had been around, 0:07:16.200 --> 0:07:18.560 right for quite a few decades. AI has been around 0:07:18.840 --> 0:07:21.520 for many decades. I think the moment we find ourselves 0:07:21.560 --> 0:07:24.679 is that people can touch it, and they can. Before 0:07:24.680 --> 0:07:26.840 they were I systems that were like behind the scenes, 0:07:26.880 --> 0:07:31.040 like your search results or translation systems, but they didn't 0:07:31.040 --> 0:07:32.960 have the experience of like, this is what it feels 0:07:32.960 --> 0:07:34.160 like to interact with this thing. 0:07:34.840 --> 0:07:35.960 So that's what I mean. 0:07:36.000 --> 0:07:38.080 I think maybe that analogy of the browser is appropriate 0:07:38.160 --> 0:07:40.360 because it's all of a sudden, it's like whoa, you know, 0:07:40.560 --> 0:07:44.160 these network of machines and content can be distributed and 0:07:44.200 --> 0:07:46.840 everybody can self publish, and there was a moment that 0:07:46.880 --> 0:07:49.000 we all remember that, and I think that that is 0:07:49.080 --> 0:07:51.560 what the world has experience over the last nine months 0:07:51.640 --> 0:07:55.360 or so on. So but fundamentally, also what is important 0:07:55.520 --> 0:07:57.680 is that this moment is where the ease of the 0:07:57.760 --> 0:08:00.640 number of people that can build and u use AI 0:08:01.280 --> 0:08:06.240 has skyrocketed. So over the last decade, you know, technology 0:08:06.280 --> 0:08:10.440 firms that had large research teams could build AI that 0:08:10.560 --> 0:08:14.000 worked really well, honestly, but when you went down into 0:08:14.160 --> 0:08:16.880 say hey, can everybody use it? Can a data science 0:08:16.880 --> 0:08:20.320 team in a bank, you know, go and develop these applications. 0:08:19.920 --> 0:08:22.240 It was like more complicated. Some could do it, but 0:08:22.280 --> 0:08:24.120 it was more the barrier of entry was high. 0:08:24.440 --> 0:08:27.800 Now it's very different because of foundation models and the 0:08:27.840 --> 0:08:29.160 implications that that has. 0:08:28.960 --> 0:08:32.640 For at the moment where the technology is being democratized. 0:08:32.000 --> 0:08:37.480 In demarketized, frankly, it works better for classes of problems 0:08:37.480 --> 0:08:40.120 like programming and other things, is really incredibly impressive what 0:08:40.160 --> 0:08:42.520 it can do. So the accuracy and the performance of 0:08:42.520 --> 0:08:45.440 it is much better, and the ease of use and 0:08:45.480 --> 0:08:47.720 the number of use cases we can pursue it much bigger, 0:08:47.720 --> 0:08:49.440 So that democratization is a big difference. 0:08:49.480 --> 0:08:51.480 But when you say, when you make it an analogy to 0:08:51.520 --> 0:08:55.600 the first browsers, if you if we do another one 0:08:55.640 --> 0:08:58.360 of these time travel questions back at the beginning of 0:08:58.360 --> 0:09:02.800 the first browsers, to say, many of the potential uses 0:09:02.920 --> 0:09:05.880 of the Internet and such we hadn't even begun. We 0:09:05.880 --> 0:09:08.600 couldn't even anticipate, right, right, So we're at the point 0:09:08.600 --> 0:09:11.640 where the future direction is largely unpredictable. 0:09:11.920 --> 0:09:14.719 Yeah, I think that that is right, because it's such 0:09:14.720 --> 0:09:19.080 a horizontal technology that the intersection of the horizontal capability, 0:09:19.160 --> 0:09:23.400 which is about expanding our productivity and tasks that we 0:09:23.440 --> 0:09:26.280 wouldn't be able to do efficiently without it, has to 0:09:26.360 --> 0:09:29.160 marry now the use cases that reflect the diversity of 0:09:29.200 --> 0:09:32.240 human experience, our institutional diversity. So as more and more 0:09:32.280 --> 0:09:34.920 institutions said, you know, I'm focused on agriculture, you know, 0:09:35.040 --> 0:09:38.160 to be able to improve seeds, you know, in these 0:09:38.200 --> 0:09:41.200 kinds of environments. They'll find their own context in which 0:09:41.240 --> 0:09:43.040 that matters that the creators of a I did not 0:09:43.040 --> 0:09:46.000 anticipate at the beginning. So I think that that is 0:09:46.040 --> 0:09:48.480 then the fruit of surprises will be like why I 0:09:48.480 --> 0:09:49.680 wouldn't even think that it could be. 0:09:49.720 --> 0:09:50.199 Used for that? 0:09:50.480 --> 0:09:52.679 And also clever people will create. 0:09:52.400 --> 0:09:55.400 New business models as associated with that, like it happened 0:09:55.440 --> 0:09:58.120 with the Internet of course as well, and that will 0:09:58.120 --> 0:10:01.280 be its own source of transformation, unchanged in its own right. 0:10:01.360 --> 0:10:03.559 So I think all of that is yet to unfold. 0:10:03.679 --> 0:10:03.839 Right. 0:10:03.880 --> 0:10:06.640 What we're seeing is this catalyst moment of technology that 0:10:06.679 --> 0:10:08.680 works well enough and it can be democratized. 0:10:09.600 --> 0:10:11.559 What next sort of conceptual question? 0:10:12.200 --> 0:10:12.360 You know? 0:10:12.640 --> 0:10:19.000 We can loosely understand or categorize innovations in terms of 0:10:19.040 --> 0:10:23.360 their impact on the kind of balance of power between 0:10:23.679 --> 0:10:28.720 haves and have nots. Some innovations, you know, obviously favor 0:10:28.760 --> 0:10:32.920 those who already have will make the rich richer. Some 0:10:32.920 --> 0:10:35.360 some it's a rising tie the lift cell boats and 0:10:35.400 --> 0:10:39.640 some bias in the other direction, they close the gap between. 0:10:40.840 --> 0:10:44.080 Is it possible to say to predict which of those 0:10:44.120 --> 0:10:46.160 three categories AI might fall into. 0:10:47.160 --> 0:10:48.880 It's a great question, you know. 0:10:49.000 --> 0:10:53.880 A first observation I would make on your first two categories. 0:10:53.920 --> 0:10:57.720 Is that it will be both likely be true that 0:10:57.880 --> 0:10:58.680 the use. 0:10:58.520 --> 0:11:00.839 Of AI will be highly demarketized, meaning the number of 0:11:00.880 --> 0:11:04.239 people that have access to its power to make improvements 0:11:04.240 --> 0:11:07.119 in terms of efficiency and so on will be fairly universal, 0:11:07.920 --> 0:11:11.319 and that the ones who are able to create AI 0:11:12.080 --> 0:11:15.319 UH may be quite concentrated. So if you look at 0:11:15.320 --> 0:11:18.920 it from the lens of who creates wealth and value 0:11:19.280 --> 0:11:23.280 over sustained periods of time, particularly it's saying a context 0:11:23.360 --> 0:11:26.840 like business, I think just being a user of AI 0:11:26.960 --> 0:11:31.160 technology is an insufficient strategy and UH. And the reason 0:11:31.240 --> 0:11:33.120 for that is, like, yes, you will get the immediate 0:11:33.120 --> 0:11:36.080 productivity boost of like just making API calls and you 0:11:36.120 --> 0:11:38.760 know that would be a new baseline for everybody, but 0:11:39.200 --> 0:11:42.959 you're not accruing value in terms of representing your data 0:11:43.040 --> 0:11:44.640 inside the AI in way that it. 0:11:44.679 --> 0:11:46.599 Gives you a sustainable competitive advantage. 0:11:47.000 --> 0:11:49.360 So I always try to tell people is don't just 0:11:49.400 --> 0:11:52.200 be an AI user, be an you know, AI value creator. 0:11:52.679 --> 0:11:55.040 And I think that that will have a lot of 0:11:55.120 --> 0:11:58.360 consequences in terms of the haves and have nots as 0:11:58.400 --> 0:12:01.000 an example, and that will apply both to institutions and 0:12:01.040 --> 0:12:04.360 regions and countries, et cetera. So I think it would 0:12:04.360 --> 0:12:07.920 be kind of a mistake, right to just develop strategies 0:12:07.920 --> 0:12:09.200 that are just about. 0:12:09.040 --> 0:12:13.199 Usage, but to come back to that question from them 0:12:13.200 --> 0:12:16.400 and to give you a specific suppose I'm a I'm 0:12:16.400 --> 0:12:21.199 an industrial farmer in Iowa with ten million dollars of 0:12:21.240 --> 0:12:24.880 equipment and move and I'm comparing it to a subsistence 0:12:24.960 --> 0:12:29.040 farmer somewhere in the developing world who's got a cell phone. 0:12:29.840 --> 0:12:34.559 Over the next five years, who's whose well being rises 0:12:34.600 --> 0:12:36.000 by a greater amount. 0:12:37.080 --> 0:12:39.800 Yeah, I think, I mean, it's a it's a good question, 0:12:39.840 --> 0:12:41.480 but it might be hard to do a one to 0:12:41.600 --> 0:12:44.160 one sort of like attribution to just one variable in 0:12:44.200 --> 0:12:48.400 this case, which is AI. But again, provided that you 0:12:48.520 --> 0:12:51.360 have access to a phone, right and some kind to 0:12:51.600 --> 0:12:54.000 you know, be able to be connected. I do think 0:12:54.080 --> 0:12:57.080 so for example, in that context we've developed, we don't 0:12:57.120 --> 0:12:59.600 work with NASA as an example, to build you spatial 0:12:59.640 --> 0:13:03.080 models using some of these new techniques, and I think 0:13:03.120 --> 0:13:06.079 for example, or ability to do flood prediction, I'll tell 0:13:06.080 --> 0:13:08.559 you an advantage of why it would be a democratization 0:13:08.640 --> 0:13:11.680 force in that context. Before to build a flowed model 0:13:12.200 --> 0:13:15.640 based on satellite imagery was actually so. 0:13:15.520 --> 0:13:17.840 Onerous and so complicated and difficult. 0:13:17.400 --> 0:13:19.880 That you would just target to very specific regions and 0:13:19.920 --> 0:13:23.240 then obviously countries prioritize their own right. But what we've 0:13:23.280 --> 0:13:26.040 demonstrated is actually you can extend that technique to have 0:13:26.280 --> 0:13:29.000 global coverage around that. So in that context, I would 0:13:29.040 --> 0:13:32.000 say it's a four stores democratization that everybody sort of 0:13:32.000 --> 0:13:34.840 would have access if you have some connectivity, as. 0:13:34.679 --> 0:13:37.640 Today, Iowa farmer might have a flood model. The guy 0:13:37.720 --> 0:13:40.480 in the developing world definitely didn't, and now he's a 0:13:40.520 --> 0:13:41.200 shot at getting one. 0:13:41.240 --> 0:13:42.679 Yeah, but now it has a shot of getting one. 0:13:42.720 --> 0:13:44.560 So there's aspects of it that so long as we 0:13:44.640 --> 0:13:47.600 provide connectivity and access to it, that they can be 0:13:47.760 --> 0:13:49.040 democratization forces. 0:13:49.160 --> 0:13:49.640 But I'll give you. 0:13:49.640 --> 0:13:53.320 Another example that that can be quite concerning, which is language. Right, 0:13:53.440 --> 0:13:57.559 So there's so much language in the you know, in English, 0:13:58.160 --> 0:14:01.400 and there is sort of like the reinforcement loop that 0:14:01.520 --> 0:14:03.960 happens that the more you concentrate because it has obvious 0:14:03.960 --> 0:14:07.559 benefits for global communication and standardization, the more you can 0:14:07.679 --> 0:14:11.200 enrich like base AI models based on that capability. If 0:14:11.200 --> 0:14:15.319 you have very resource cars languages, you tend to develop 0:14:15.400 --> 0:14:18.840 less powerful AI with those languages and so on, So 0:14:19.000 --> 0:14:23.080 one has to actually worry and focus on the ability 0:14:23.440 --> 0:14:27.160 to actually represent that in that case is language as 0:14:27.160 --> 0:14:29.920 a piece of culture. Also in the AI sets that 0:14:30.080 --> 0:14:33.240 everybody can benefit from it too. So there's a lot 0:14:33.320 --> 0:14:36.600 of considerations in terms of equity about the data and 0:14:36.680 --> 0:14:39.920 the data sets that we accrue and what problems are 0:14:39.920 --> 0:14:42.120 we trying to solve. I mean, you mentioned agriculture or 0:14:42.120 --> 0:14:44.640 healthcare and so on. If we only solve problems that 0:14:44.640 --> 0:14:47.240 are related to marketing as an example, that would be 0:14:47.280 --> 0:14:49.960 a less rich world in terms of opportunity that if 0:14:49.960 --> 0:14:52.680 we incorporate many many other broader server problems. 0:14:53.680 --> 0:14:55.600 Who do you think what do you think are the 0:14:55.640 --> 0:15:01.040 biggest impediments to the adoption of a as you would 0:15:01.120 --> 0:15:03.520 like as you think aire to be adopted. I mean, 0:15:03.560 --> 0:15:05.400 if you look, what are the sticking points that. 0:15:05.400 --> 0:15:08.240 You would look In the end, I'm going to give 0:15:08.280 --> 0:15:10.840 a non technological answer as a first one has to 0:15:10.840 --> 0:15:13.760 do with workflow, right, So even if the technology is 0:15:13.880 --> 0:15:19.000 very capable, the organizational change inside a company to incorporate 0:15:19.000 --> 0:15:21.720 into the natural workflow of people or how we work 0:15:22.320 --> 0:15:24.800 is It's a lesson we have learned over the last 0:15:24.800 --> 0:15:26.320 decade is hugely important. 0:15:26.920 --> 0:15:29.920 So there's a lot of design considerations. 0:15:29.960 --> 0:15:33.000 There's a lot of how do people want to work right, 0:15:33.240 --> 0:15:34.280 how do they work today? 0:15:34.320 --> 0:15:36.400 And what is the natural entry points for AI? So 0:15:36.440 --> 0:15:37.760 that's like number one. 0:15:38.000 --> 0:15:40.760 And then the second one is, you know, for the 0:15:40.800 --> 0:15:45.120 broad value creation aspect of it is the understanding inside 0:15:45.120 --> 0:15:49.480 the companies of how you have to curate and create 0:15:49.560 --> 0:15:53.000 data to combine it with external data says that you 0:15:53.040 --> 0:15:56.280 can have powerful AI models that actually fit your need. 0:15:56.760 --> 0:16:00.400 And that aspect of what it takes to actually and 0:16:00.440 --> 0:16:05.800 curate the data for this modern AI, it's still working progress, right. 0:16:06.120 --> 0:16:08.640 I think part of the problem that happens very often 0:16:08.680 --> 0:16:11.960 when I talk to institutions is that they say yah, yeah, yeah, yeah, 0:16:11.960 --> 0:16:14.520 I'm doing it. I've been doing it for a long time. 0:16:15.240 --> 0:16:17.640 And the reality is that that answer can sometimes be 0:16:17.680 --> 0:16:19.760 a little of a cop out, right, is like, I 0:16:19.800 --> 0:16:22.400 know you were doing machine learning, you were doing some 0:16:22.480 --> 0:16:23.359 of these things. 0:16:23.080 --> 0:16:25.560 But actually the leader's version of AI. 0:16:25.480 --> 0:16:28.200 What was happening with foundation models, Not only is it 0:16:28.320 --> 0:16:31.680 very new, it's very hard to do and honestly, if 0:16:31.720 --> 0:16:34.440 you haven't been you know, assembling very large teams and 0:16:34.480 --> 0:16:36.960 spending hundreds of millions of dollars of compute, and so 0:16:37.120 --> 0:16:39.880 you're probably not doing it right. You're doing something else 0:16:39.920 --> 0:16:42.560 that is in the broad category. And I think the 0:16:42.720 --> 0:16:45.480 lessons about what it means to make this transition to 0:16:45.560 --> 0:16:48.400 this new wave is still in early phases of understanding. 0:16:48.560 --> 0:16:50.160 So what would you say? I want to give you 0:16:50.200 --> 0:16:53.720 a couple of examples of people with kind of real 0:16:53.720 --> 0:16:56.800 work in real world positions of responsibility. Imagine I'm sitting 0:16:56.880 --> 0:16:59.720 right here, So imagine that I am the president of 0:16:59.720 --> 0:17:02.120 a small liberal arts college and I come to you 0:17:02.120 --> 0:17:04.240 and I say, Dario, I keep hearing about the AI 0:17:04.920 --> 0:17:07.680 my college has. You know, I don't make it. You know, 0:17:07.800 --> 0:17:11.040 I'm I'm I'm making this much money. If that every 0:17:11.080 --> 0:17:15.320 year my enrollments declining, I feel like this maybe is 0:17:15.359 --> 0:17:18.720 an opportunity. What is the opportunity for me? What would 0:17:18.720 --> 0:17:19.520 you say? 0:17:20.640 --> 0:17:22.960 So it's probably in a couple of segments around that. 0:17:23.200 --> 0:17:24.840 Right one has. 0:17:24.680 --> 0:17:27.480 To do is well, what is the implications of this 0:17:27.600 --> 0:17:31.919 technology inside the institution itself instead of the college, and 0:17:31.960 --> 0:17:35.800 how we operate and can we improve for example, efficiency, 0:17:35.960 --> 0:17:39.160 like if you have in very low levels of sort 0:17:39.200 --> 0:17:41.959 of margin to be able to reinvest, is you know 0:17:42.160 --> 0:17:46.440 you run it, You run you know, infrastructure, you run 0:17:46.480 --> 0:17:49.000 many things inside the college. What are the opportunities to 0:17:49.000 --> 0:17:52.800 increase the productivity or automate and drive savings such that 0:17:52.880 --> 0:17:55.159 you can reinvest that money so into the mission of 0:17:55.240 --> 0:17:56.840 education right as an example. 0:17:56.520 --> 0:17:58.240 So number one is operational efficiency. 0:17:58.240 --> 0:18:01.399 Operational efficiencyage is is a big one. I think the 0:18:01.400 --> 0:18:03.560 second one is within the context of the college, there's 0:18:03.560 --> 0:18:06.600 implications for the educational mission on its own, right, how 0:18:06.640 --> 0:18:08.400 will you know how. 0:18:08.359 --> 0:18:09.919 Does a curriculum need to evolve or not? 0:18:10.359 --> 0:18:12.840 What are acceptable use policies or of some of these 0:18:12.840 --> 0:18:15.119 AI I think we've all read a lot about like 0:18:15.119 --> 0:18:17.520 what can happen in terms of exams and so on, 0:18:17.560 --> 0:18:19.639 and cheating and not cheating, or what are the actually 0:18:19.720 --> 0:18:22.480 positive elements of it in terms of how curriculum should 0:18:22.480 --> 0:18:25.840 be developed and professions sustain around that. And then there's 0:18:25.840 --> 0:18:28.879 another third dimension, which is the outdoor oriented element of it, 0:18:28.920 --> 0:18:32.760 which is like prospect students right, so, which is frankly speaking, 0:18:32.800 --> 0:18:34.960 a big use case that is happening right now, which 0:18:35.000 --> 0:18:37.440 in the broader industry is called customer care or client 0:18:37.480 --> 0:18:39.840 care or citizen care. So in this question will be education, 0:18:40.000 --> 0:18:42.600 like you know, hey are you reaching the right students 0:18:43.280 --> 0:18:46.200 around that that may apply to the college. How can 0:18:46.240 --> 0:18:48.679 you create them, for example, an environment to interact with 0:18:48.680 --> 0:18:50.720 the college and answering questions that could be a chat 0:18:50.760 --> 0:18:53.800 bought or something like that to learn about it, and personalization. 0:18:54.200 --> 0:18:55.320 So I would say there. 0:18:55.240 --> 0:18:57.360 Is like at least three lenses with which I would 0:18:57.359 --> 0:18:59.959 give advice right the positive. 0:19:00.040 --> 0:19:04.239 Second, because it's really interesting. So I really can't as 0:19:04.280 --> 0:19:07.720 sign an essay anymore? Can I? 0:19:07.720 --> 0:19:09.000 Can I sign an essay? 0:19:09.160 --> 0:19:12.000 Can I say, rend me a research paper and come 0:19:12.040 --> 0:19:13.840 back to being three? Can I do that anymore? 0:19:13.880 --> 0:19:14.520 I think you can? 0:19:14.640 --> 0:19:15.400 How do I do that? 0:19:15.560 --> 0:19:16.240 And then you can that. 0:19:16.560 --> 0:19:21.040 Look, there's there's two questions around that. I think that 0:19:21.320 --> 0:19:22.640 if one goes. 0:19:22.520 --> 0:19:24.440 And explains in the context like what is it? 0:19:24.560 --> 0:19:25.160 Why are we here? 0:19:25.160 --> 0:19:27.159 Why are in this class? What is the purpose of this? 0:19:28.040 --> 0:19:31.560 And and one starts with assuming like an element of 0:19:31.600 --> 0:19:33.760 like decency and people are people are there like to 0:19:33.840 --> 0:19:36.200 learn and so on, and you just give it disclaimer. Look, 0:19:36.400 --> 0:19:38.959 I know that one option you have is like just 0:19:39.000 --> 0:19:41.200 you know, put the essay question and click go and 0:19:41.240 --> 0:19:43.200 like and give an answer. You know, but that is 0:19:43.240 --> 0:19:45.359 not why we're here, and that is not the intent 0:19:45.400 --> 0:19:46.960 of what we're trying to do. So first I would 0:19:47.000 --> 0:19:51.320 start with the sort of like the norms of intent 0:19:51.440 --> 0:19:54.600 and decency and appeal to those as step number one. 0:19:55.119 --> 0:19:57.240 Then we all know that there will be a distribution 0:19:57.320 --> 0:19:59.359 of use cases of people like that will come in 0:19:59.359 --> 0:20:01.199 one year and come one or the other and do that. 0:20:01.760 --> 0:20:04.280 And so for a subset of that, you know, I 0:20:04.280 --> 0:20:06.240 think the technology is going to have all in such 0:20:06.240 --> 0:20:08.679 a way that we will have more and more of 0:20:08.680 --> 0:20:11.440 the ability to discern right, you know, when that has 0:20:11.480 --> 0:20:14.240 been AI generated right and uncreated. 0:20:14.400 --> 0:20:15.960 It won't be perfect, right. 0:20:15.840 --> 0:20:18.399 But there's some elements that you can imagine in putting 0:20:18.400 --> 0:20:20.399 the essay and you say, hey, this is likely to 0:20:20.440 --> 0:20:23.560 be generated right around that. And for example, one way 0:20:23.600 --> 0:20:24.959 you can do that, just to give you an intuition, 0:20:25.080 --> 0:20:27.399 you could just have an essay that you write with 0:20:27.480 --> 0:20:30.840 pencil and paper. At the beginning, you get a baseline 0:20:30.840 --> 0:20:33.520 of what you're writing is like. And then later when 0:20:33.560 --> 0:20:37.440 you you know generated, there will be obvious differences around 0:20:37.480 --> 0:20:39.000 what kind of writing has been generated. 0:20:39.040 --> 0:20:42.360 On the other way, but you've turned it's everything you're 0:20:42.400 --> 0:20:46.280 describing makes sense. Put it greatly in this In this respect, 0:20:46.320 --> 0:20:49.000 at least it seems to greatly come complicate the life 0:20:49.000 --> 0:20:51.920 of the teacher, whereas the other two use cases seem 0:20:52.000 --> 0:20:56.879 to kind of clarify and simplify the role. Right suddenly, 0:20:57.200 --> 0:21:00.959 you know, reaching student perspective students, sounds like I can 0:21:01.000 --> 0:21:03.080 do that much more kind of efficienty a lot. Yeah, 0:21:03.119 --> 0:21:05.720 I can bring out administration costs, but the teaching thing 0:21:05.800 --> 0:21:06.760 is tricky. 0:21:07.880 --> 0:21:11.639 Well, until we developed the new norms, right, I mean again, 0:21:11.800 --> 0:21:14.120 I mean, I know it's not abuse analogy, but calculators 0:21:14.119 --> 0:21:17.680 we deal. We've done with that too, right, And it says, well, calculator, 0:21:17.680 --> 0:21:18.680 what is the purpose of math? 0:21:18.720 --> 0:21:19.520 How are we going to do this? 0:21:19.600 --> 0:21:22.840 And so can I tell you my dad's calculator story? 0:21:22.960 --> 0:21:23.520 Yes? Please. 0:21:23.800 --> 0:21:27.760 My father was a mathematician, taught mathematics at University of Waterloo, Agada, 0:21:28.480 --> 0:21:31.800 and in the seventies when people started to get pocket calculators, 0:21:32.240 --> 0:21:34.840 his students demanded that they'd be able to use them, 0:21:35.000 --> 0:21:36.760 and he said no, and they took him to the 0:21:36.800 --> 0:21:42.600 administration and he lost. So he then changed completely throughout 0:21:42.640 --> 0:21:45.679 all of his old exams and introduced new exams where 0:21:46.080 --> 0:21:50.320 there was no calculation. It was all like deep think, 0:21:50.440 --> 0:21:52.760 you know, figure out the problem on a conceptual level 0:21:52.760 --> 0:21:56.040 and describe it to me. And they were all students 0:21:56.080 --> 0:22:00.520 deeply unhappy that he'd made their lives for complicated. But 0:21:59.760 --> 0:22:03.800 to your point, I mean he probably the result was 0:22:03.840 --> 0:22:07.960 probably a better education. He just removed the element that 0:22:08.000 --> 0:22:10.919 they could game with their pocket calculators. I suppose it's 0:22:10.920 --> 0:22:11.560 a version. 0:22:11.359 --> 0:22:13.119 Of I think it's a version of that, And so 0:22:13.200 --> 0:22:14.919 I think they will develop the equivalent of what your 0:22:14.920 --> 0:22:16.520 father did. And I think people say, you know what 0:22:16.600 --> 0:22:18.920 if like these kinds of things, everybody's doing it generically 0:22:19.000 --> 0:22:21.200 and none of us have any meaning because all you're 0:22:21.240 --> 0:22:23.240 doing is pressing buttons, and like the intent of this 0:22:23.400 --> 0:22:24.960 was something which was to teach you how to write 0:22:25.160 --> 0:22:26.119 or to think or something. 0:22:26.400 --> 0:22:28.160 There may be a variant of how we do all 0:22:28.160 --> 0:22:29.520 of this. I mean, obviously some. 0:22:29.560 --> 0:22:31.720 Version of that that has happened is like Okay, we're 0:22:31.720 --> 0:22:33.240 all going to sit down and doing with pencil and 0:22:33.280 --> 0:22:35.800 paper and computers in their classroom. But there'll be other 0:22:35.880 --> 0:22:38.680 variants of creativity that people will put forth to say, 0:22:38.680 --> 0:22:40.760 you know what, you know, that's a way to solve 0:22:40.760 --> 0:22:41.400 that problem too. 0:22:41.480 --> 0:22:44.520 But this is interesting because to stay on this analogy, 0:22:45.160 --> 0:22:50.119 we're really talking about a profound rethinking just using a 0:22:50.200 --> 0:22:54.080 college as an example, a real profound rethinking of the 0:22:54.119 --> 0:22:57.000 way there's no part of this college it's unaffected by 0:22:57.240 --> 0:23:02.280 aia B. In one case, I've made everyone's job easier. 0:23:02.320 --> 0:23:05.280 In one case, I've made I'm asking us to really 0:23:05.320 --> 0:23:09.560 rethink from the ground up what teaching means. In another case, 0:23:09.560 --> 0:23:11.600 I've automated systems that I didn't think of it. I mean, 0:23:11.640 --> 0:23:14.680 it's like, that's right, that's it's not that's a lot 0:23:14.720 --> 0:23:17.399 to ask someone who got a pH d in medieval 0:23:17.480 --> 0:23:19.800 language literature, you know, forty years ago. 0:23:20.440 --> 0:23:22.640 Yeah, but you know, I'll tell you a positive sort 0:23:22.640 --> 0:23:25.320 of development that I'm seeing the sciences around this, which 0:23:25.400 --> 0:23:28.760 is you're seen as you see more and more examples 0:23:29.080 --> 0:23:33.320 of applying AI technology within the context of like historians 0:23:33.320 --> 0:23:36.800 to as an example, Right, you have archival and you know, 0:23:36.800 --> 0:23:38.439 and you have all these books and being able to 0:23:38.520 --> 0:23:41.600 actually help you as an assistant right around that, but 0:23:41.640 --> 0:23:44.880 not only with text now, but with diagrams, right. And 0:23:45.240 --> 0:23:48.960 I've seen it in anthropology too, write and archaeology with 0:23:49.119 --> 0:23:52.359 examples of engravings and translations and things that can happen. 0:23:52.720 --> 0:23:56.280 So so as you see in diverse fields people applying 0:23:56.320 --> 0:23:58.800 these techniques to advance and how to do physics or 0:23:58.840 --> 0:24:02.160 how to do chemistry. They inspire each other, right, and 0:24:02.160 --> 0:24:04.000 this said, you know, how does it apply actually to 0:24:04.040 --> 0:24:07.680 my area? So once as that happens, it becomes less 0:24:07.680 --> 0:24:09.239 of a chore of like, my god, you know, how 0:24:09.240 --> 0:24:11.359 do I have to deal with this? But actually it's 0:24:11.359 --> 0:24:15.040 triggered by Curiosity's triggered by you know, there'll be like, 0:24:15.200 --> 0:24:16.840 you know, faculty that will be like, you know what, 0:24:16.960 --> 0:24:19.240 you know, let me explore what this means for my area, 0:24:19.560 --> 0:24:21.879 and they will adapt it to the local context, to 0:24:21.920 --> 0:24:25.480 the local you know, language, and the professional itself. So 0:24:25.760 --> 0:24:28.560 I see that as a positive vector that is not 0:24:28.600 --> 0:24:29.880 all going to feel like homework. 0:24:30.119 --> 0:24:31.760 You know, it's not going to feel like, oh my god, 0:24:31.760 --> 0:24:33.560 this is so overwhelming. 0:24:33.200 --> 0:24:35.520 But rather to be very practical to see what works, 0:24:35.680 --> 0:24:37.800 What have I seen others to do that is inspiring? 0:24:38.000 --> 0:24:40.000 And what am I inspired to do? You know what? 0:24:40.000 --> 0:24:42.000 What is how is this going to help my career? 0:24:42.280 --> 0:24:44.040 I think that that's going to be an interesting question 0:24:44.160 --> 0:24:47.119 for you know, those faculty members, for the students. 0:24:46.720 --> 0:24:49.720 The professionals. Sorry, I'm gonna stick with this example along 0:24:49.760 --> 0:24:52.439 because it's really interesting. I'm curious following up on what 0:24:52.520 --> 0:24:55.919 you just said, that one of the most persistent critiques 0:24:55.960 --> 0:25:00.320 of academic but also of many of many corporate institutions 0:25:01.320 --> 0:25:05.280 in recent UITs, has been siloing. Right, different parts of 0:25:05.320 --> 0:25:08.480 the of the organization are going off on their own 0:25:08.480 --> 0:25:11.280 and not speaking to each other. Is a potential is 0:25:12.359 --> 0:25:17.520 a real potential benefit to AI the kind of breaking 0:25:17.560 --> 0:25:21.440 down a simple tool for breaking down those kinds of barriers, 0:25:21.440 --> 0:25:23.320 is that a very is that an elegant way of 0:25:23.359 --> 0:25:25.200 sort of saying what. 0:25:24.640 --> 0:25:25.159 I really think? 0:25:25.200 --> 0:25:28.119 And I was actually just having a conversation with Provos 0:25:28.119 --> 0:25:30.840 stuff and very much on this topic, very recently exactly 0:25:30.840 --> 0:25:34.000 on that, which is all these this you know, this 0:25:34.119 --> 0:25:37.040 appetite right to collaborate across disciplines. There's a lot of 0:25:37.520 --> 0:25:41.960 attempts stores a goal, right, creating interdisciplinary centers, creating dual 0:25:42.000 --> 0:25:46.240 degree programs or dual appointment programs. But actually in a 0:25:46.320 --> 0:25:50.600 lot of progress in academia happens by methodology too. Write 0:25:50.640 --> 0:25:53.680 like a new you know, when when some methodology gets adopted, 0:25:53.880 --> 0:25:56.040 I mean the most famous example of that is a 0:25:56.080 --> 0:25:58.679 scientific method as an example of that. But when you 0:25:58.760 --> 0:26:01.840 have a methodology that gets adopted, it also provides a 0:26:01.920 --> 0:26:05.879 way to speak to your colleagues across different disciplines. And 0:26:05.920 --> 0:26:08.879 I think what's happened in AI is linked to that 0:26:08.880 --> 0:26:12.120 that within the context of the scientific method as an example, 0:26:12.680 --> 0:26:17.359 the methodology about we about what we do discovery, the 0:26:17.480 --> 0:26:20.119 role of data, the role of these neural networks, of 0:26:20.200 --> 0:26:23.040 how we actually find proximity to concepts to one another 0:26:23.480 --> 0:26:28.240 is actually fundamentally different than how we've traditionally applied it. 0:26:28.640 --> 0:26:31.960 So as we see across more professions, people applying this 0:26:32.080 --> 0:26:35.439 methodology is also going to give some element of common 0:26:35.520 --> 0:26:38.959 language to each other. Right And in fact, you know, 0:26:39.119 --> 0:26:42.399 in this very high dimensional representation of information that is 0:26:42.400 --> 0:26:46.439 pressent to neural networks, we may find amazing adjacencies or 0:26:46.440 --> 0:26:50.240 connections of themes and topics in ways that the individual 0:26:50.280 --> 0:26:53.800 practitioners cannot describe, but yet will be latent. 0:26:53.720 --> 0:26:55.240 In these large calural networks. 0:26:56.080 --> 0:26:58.480 We are going to suffer a little bit from causality, 0:26:58.520 --> 0:27:00.640 from the problem of like, hey, what's the root cause 0:27:00.720 --> 0:27:04.600 of that? Because I think one of the unsatisfying aspects 0:27:04.640 --> 0:27:07.560 that this methodology will provide is they may give you 0:27:07.640 --> 0:27:10.480 answers for which they don't give you good reasons for 0:27:10.600 --> 0:27:13.840 where the answers came from, and then there will be 0:27:13.880 --> 0:27:16.639 the traditional process of discovery of saying, if that is 0:27:16.680 --> 0:27:19.440 the answer, what are the reasons? So we're going to 0:27:19.560 --> 0:27:22.720 have to do this sort of hybrid way of understanding 0:27:22.760 --> 0:27:25.320 the world. But I do think that common layer of 0:27:25.359 --> 0:27:26.960 AI is a powerful new thing. 0:27:27.280 --> 0:27:30.560 Yeah, well, a couple of random questions. I couldn't manage 0:27:30.600 --> 0:27:33.560 you talk in the In the Writer's strike that just 0:27:33.680 --> 0:27:36.680 ended in Hollywood, one of the sticking points was how 0:27:36.720 --> 0:27:40.920 the studios and writers would treat AI generated content. Good 0:27:40.960 --> 0:27:44.880 writers get credit if their material with somehow the source 0:27:44.960 --> 0:27:49.679 for A but more broadly, did the writers need protections 0:27:49.720 --> 0:27:51.720 against the use of I could go on? You know what, 0:27:51.920 --> 0:27:54.400 I'll be worth familiar with all of this. Had you been, 0:27:54.440 --> 0:27:57.240 I don't know whether you were, but had either side 0:27:57.280 --> 0:28:01.159 called you in for advice? During that the writers, the 0:28:01.200 --> 0:28:03.920 writers called you and said, Daria, what should we do 0:28:04.119 --> 0:28:06.879 about AI? And how should we that should be reflect 0:28:06.880 --> 0:28:10.080 should that be reflected in our contract negotiations? What would 0:28:10.080 --> 0:28:12.200 you have told them? 0:28:12.880 --> 0:28:14.919 The way I think about that is that I divided. 0:28:15.000 --> 0:28:16.440 I would divide it into two pieces. 0:28:16.600 --> 0:28:20.840 First, is what's technically possible right and anticipate scenarios like 0:28:21.680 --> 0:28:24.200 you know, what can you do with voice cloning for example? 0:28:24.520 --> 0:28:27.240 You know, now, for example, it is possible this being 0:28:28.480 --> 0:28:31.280 dubbing right legis take that topic right around the world. 0:28:31.280 --> 0:28:34.159 There was all these folks that would dub people in 0:28:34.240 --> 0:28:38.080 other languages. Well, now you can do these incredible renderings. 0:28:38.160 --> 0:28:40.200 I mean, I know if you've seen them, where you 0:28:40.240 --> 0:28:42.800 know you match the lips is your original voice, but 0:28:42.880 --> 0:28:45.400 speaking any language that you want. As an example, so 0:28:45.480 --> 0:28:47.640 busy that has a set of implications around that. I mean, 0:28:47.720 --> 0:28:49.320 just to give an example, So I would say, create 0:28:49.360 --> 0:28:53.560 a taxonomy that describes technical capabilities that we know of 0:28:53.640 --> 0:28:58.440 today and applications to the industry and to examples of 0:28:58.520 --> 0:29:00.000 like hey, you know I could film you for five 0:29:00.080 --> 0:29:02.080 minutes and I could generate two hours of content of 0:29:02.120 --> 0:29:04.000 you and I don't have to you know, then if 0:29:04.040 --> 0:29:06.120 you'll get paid by the hour, obviously I'm not paying 0:29:06.120 --> 0:29:08.320 you for that other thing. So I would say technological 0:29:08.320 --> 0:29:12.280 capability and then map with their expertise consequences of how 0:29:12.320 --> 0:29:14.560 it changes the way they work or the way they 0:29:14.600 --> 0:29:16.840 interact or the way they negotiate and so on. 0:29:16.880 --> 0:29:19.120 So that would be one element of it, and then the. 0:29:19.080 --> 0:29:22.000 Other one is like a non technology related matter, which 0:29:22.040 --> 0:29:25.000 is an element of almost distributed justices like who deserves 0:29:25.000 --> 0:29:27.320 what right and who has the power to get what? 0:29:28.160 --> 0:29:31.920 And then that's a completely different discussion. That is to say, well, 0:29:31.960 --> 0:29:34.760 if this is the scenario of what's possible, you know, 0:29:34.840 --> 0:29:37.560 what do we want and what are we able to get? 0:29:38.000 --> 0:29:40.200 And I think that that's a different discussion, which. 0:29:40.040 --> 0:29:42.200 Is which all that life? Which when do you do? 0:29:42.320 --> 0:29:42.600 First? 0:29:43.840 --> 0:29:47.600 I think it's very helpful to have an understanding of 0:29:47.640 --> 0:29:50.800 what's possible and how it changes a landscape. 0:29:51.200 --> 0:29:53.400 As part of a broader. 0:29:53.560 --> 0:29:58.239 Discussion, right, and a broad negotiation, because you also have 0:29:58.280 --> 0:30:00.280 to see the opportunities because there will be a lot 0:30:00.360 --> 0:30:03.240 of ground to say, actually, you know, if we can 0:30:03.280 --> 0:30:05.760 do it in this way and we can all be 0:30:05.840 --> 0:30:08.480 that much more efficient in getting these piece work done 0:30:08.600 --> 0:30:12.120 or these filming done, but we have a reasonable agreement 0:30:12.280 --> 0:30:16.360 about how we both sides benefit from it, right, then 0:30:16.480 --> 0:30:16.800 that's a. 0:30:16.760 --> 0:30:17.760 Win win for everybody. 0:30:17.840 --> 0:30:18.040 Yeah. 0:30:18.160 --> 0:30:21.160 Right, so that's I think that would be a golden triangle. 0:30:21.240 --> 0:30:21.360 Right. 0:30:21.440 --> 0:30:24.040 Here's my reading, and I would like you to correct 0:30:24.040 --> 0:30:25.800 me if I'm wrong, and I'm likely to be wrong. 0:30:26.920 --> 0:30:29.000 When I looked at that strike, I said if they're 0:30:29.040 --> 0:30:33.040 worried about AI. The writers are worried about AI. That 0:30:33.160 --> 0:30:35.720 seems silly. It should be the studios who are worried 0:30:35.760 --> 0:30:38.200 about the economic impact of AI. Does it? In the 0:30:38.200 --> 0:30:41.280 long run AI put the studios out of business long 0:30:41.320 --> 0:30:43.240 before it puts the writers out of business. I only 0:30:43.240 --> 0:30:47.239 need the studio because the costs of production are as 0:30:47.320 --> 0:30:49.920 high as the sky, and the cost of production are overwhelming. 0:30:49.960 --> 0:30:53.200 And whereas if I don't, if I have a tool 0:30:53.240 --> 0:30:58.600 which brings introduces massive technological efficiencies to the production of movies, 0:30:59.040 --> 0:31:02.360 then why don't studio? Why would they the scared ones? 0:31:02.520 --> 0:31:04.560 Or maybe maybe you need like a different kind of 0:31:04.560 --> 0:31:06.520 studio or a different kind of different kind of studies. 0:31:06.560 --> 0:31:09.680 But I mean the in the but in the in 0:31:09.720 --> 0:31:12.840 this strike the fright the frightened ones with the writers 0:31:12.880 --> 0:31:17.080 and the you know, with the studios. Wasn't that backwards? 0:31:18.520 --> 0:31:19.480 I haven't thought about it. 0:31:19.640 --> 0:31:21.719 Uh, it can be about the implications of it. 0:31:21.720 --> 0:31:23.960 It goes back to we're talking before the implications because 0:31:23.960 --> 0:31:26.600 are so horizontal, it is right to think about it, 0:31:26.640 --> 0:31:27.200 like what does. 0:31:27.080 --> 0:31:28.560 It do to the studios as well? Right? 0:31:28.720 --> 0:31:29.920 Yeah? 0:31:29.960 --> 0:31:32.560 But then you know the reason why that happens is 0:31:32.600 --> 0:31:36.640 that it's the order of either negotiations or or who 0:31:36.720 --> 0:31:40.760 first got concerned about it and did something about it, right, 0:31:40.800 --> 0:31:43.560 which is in the context of the strike. You know, 0:31:43.680 --> 0:31:46.080 I don't know what the equivalent conversations are going inside 0:31:46.080 --> 0:31:47.960 the studio and whether they have a war room saying 0:31:47.960 --> 0:31:49.719 what this is going to mean to us? Right, but 0:31:50.000 --> 0:31:52.600 it doesn't get exercise through a strike, but maybe through 0:31:52.600 --> 0:31:54.960 a task force inside you know, the companies about what 0:31:55.000 --> 0:31:55.560 are they going to do? 0:31:55.640 --> 0:31:55.800 Right? 0:31:56.040 --> 0:31:58.280 Well, and to go back to your thing you said, 0:31:58.280 --> 0:31:59.560 the first thing you do is you make a list 0:31:59.600 --> 0:32:04.320 of what techological capabilities are. But don't technological capabilities change 0:32:04.320 --> 0:32:07.880 every I mean they do. You're racing ahead so fast, 0:32:08.080 --> 0:32:11.040 so you can't Can you have a contract? I'm sorry 0:32:11.040 --> 0:32:13.000 for begetting in the little weeds here, but this is interesting. 0:32:13.440 --> 0:32:16.239 Can you You can't have a five year contract if 0:32:16.280 --> 0:32:20.000 the contract is based on an assessment of technological capabilities 0:32:20.000 --> 0:32:22.240 in twenty twenty three, because by the time we get 0:32:22.240 --> 0:32:28.080 to twenty eight twenty three eight, it's totally different. Right. 0:32:28.200 --> 0:32:30.280 Yeah, But like you know, I mean where I was 0:32:30.320 --> 0:32:34.200 going is like there are some abstractions around that is like, 0:32:35.240 --> 0:32:37.680 you know, whate can we do with my image? Right, Like, 0:32:37.720 --> 0:32:40.239 if I generally get the category that my image can 0:32:40.280 --> 0:32:43.360 be reproduced, generated contents and so on, it's like, let's 0:32:43.360 --> 0:32:45.760 talk about the abstract notion about who has rights to 0:32:45.840 --> 0:32:48.520 that or do we both get to benefit from that? 0:32:48.720 --> 0:32:51.520 If you get that straight, Yes, the nature of how 0:32:51.520 --> 0:32:54.880 the image gets alter created as something will change underneath, 0:32:55.080 --> 0:32:57.640 but the concept will stay the same. And so I 0:32:57.680 --> 0:32:59.880 think what's important is to get the categories right. 0:33:00.200 --> 0:33:04.160 Yeah. Yeah, if you had to you to think about 0:33:04.200 --> 0:33:10.480 the biggest technological revolutions of the post war era last 0:33:10.600 --> 0:33:14.040 seventy five years, you can all come up with a list. Actually, 0:33:14.040 --> 0:33:15.479 it's really fun to come up with a list. I 0:33:15.520 --> 0:33:18.920 was thinking about this when we were you know, containerized 0:33:18.920 --> 0:33:25.719 shipping is my favorite, the green revolution, the internet is 0:33:25.960 --> 0:33:27.320 Where is the I in that list? 0:33:29.960 --> 0:33:32.760 So I would put it first in that context that 0:33:32.800 --> 0:33:36.920 you put forth over since World War Two, undoubtedly, like 0:33:37.080 --> 0:33:41.320 computing as a category is one of those trajectories that 0:33:41.440 --> 0:33:43.040 has reshaped. 0:33:42.560 --> 0:33:43.320 Right or world. 0:33:43.760 --> 0:33:47.600 And I think we think computing, I would say the 0:33:47.720 --> 0:33:52.160 role that semiconductors have had has been increwdly defining. I 0:33:52.160 --> 0:33:56.160 would say AI is the second example of that as 0:33:56.200 --> 0:33:59.560 a core architecture that is going to have an equivalent 0:33:59.640 --> 0:34:02.400 level of impact. And then the third leg I would 0:34:02.400 --> 0:34:03.520 put to that equation. 0:34:03.200 --> 0:34:04.880 Will be quantum and quantum information. 0:34:05.400 --> 0:34:07.400 And that's sort of like I like to summarize that 0:34:07.440 --> 0:34:09.960 the future of computing its bits, neurons, and cubits, and 0:34:10.040 --> 0:34:13.319 it is that idea of high precision computation, the world 0:34:13.360 --> 0:34:15.960 of neural networks and artificial intelligence, and the world of 0:34:16.040 --> 0:34:19.400 quantum and the combination of those things is going to 0:34:19.440 --> 0:34:21.880 be the defining force of the next hundred years in 0:34:21.920 --> 0:34:23.160 that category of computing. 0:34:23.200 --> 0:34:24.640 But it makes the list for sure. 0:34:24.880 --> 0:34:27.239 If it's that high up on the list. This is 0:34:27.280 --> 0:34:31.080 a total hypothetical. Would you if you were starting over 0:34:31.560 --> 0:34:35.080 if you're starting IBM right now, would you say, oh, 0:34:35.360 --> 0:34:38.839 our AI operations actually should be way bigger, Like how 0:34:38.840 --> 0:34:40.880 many how many thousands of people working for you? 0:34:41.760 --> 0:34:45.080 So within the research division it's about like three thousand, 0:34:45.120 --> 0:34:46.040 five hundred scientists. 0:34:46.040 --> 0:34:48.200 So in a perfect world, would you if it's that big, 0:34:48.280 --> 0:34:51.280 isn't that too small as a group? 0:34:51.719 --> 0:34:53.839 Yeah, Well that's like in the ricer division. I mean 0:34:53.880 --> 0:34:57.879 IBM overall, but I mean, like. 0:34:58.000 --> 0:35:01.080 So starting from first, so you have a we've got 0:35:01.120 --> 0:35:05.880 a technology that you're ranking with compute and you know, 0:35:06.160 --> 0:35:08.800 up there with as in terms of a world changer, 0:35:11.000 --> 0:35:13.359 are we So what I'm basically asking is are we 0:35:13.480 --> 0:35:16.440 underinvested in this huge you. 0:35:16.360 --> 0:35:18.919 Know, but so so yeah, it's a it's a good question. 0:35:19.000 --> 0:35:20.799 So like what I would say is that I think 0:35:20.800 --> 0:35:23.839 we should segment how many people do you need on 0:35:24.000 --> 0:35:27.680 the creation of the technology itself, and what is the 0:35:27.760 --> 0:35:30.480 right size of research and engineers and compute to do that, 0:35:31.080 --> 0:35:33.359 And how many people do you need in the sort 0:35:33.400 --> 0:35:38.239 of application of the technology to create better products, to 0:35:38.320 --> 0:35:41.960 deliver services and consulting and then ultimately to diffuse it 0:35:42.000 --> 0:35:44.719 through you know, sort of all spheres of society. And 0:35:44.760 --> 0:35:47.240 the numbers are very different, and that is not different 0:35:47.239 --> 0:35:49.279 than anywhere else. I mean, I mean, if you give 0:35:49.320 --> 0:35:52.000 examples of since you were talking about in context of 0:35:52.000 --> 0:35:54.080 World War two, how many people does it take to 0:35:54.120 --> 0:35:57.440 create you know, an atomic weapon as an example, it's 0:35:57.480 --> 0:35:58.200 a large number. 0:35:58.280 --> 0:35:59.760 I mean, it wasn't just loss animals. 0:35:59.800 --> 0:36:01.600 There was a lot of people in Okay, it's a 0:36:01.640 --> 0:36:04.920 large number, but it wasn't a million people, right, So 0:36:05.440 --> 0:36:09.360 you could have highly concentrated teams of people that, with 0:36:09.560 --> 0:36:13.880 enough resources, can do extraordinary scientific and technological achievements, and 0:36:13.920 --> 0:36:16.160 that's always by definition, is going to be a fraction 0:36:16.320 --> 0:36:19.040 of like one percent compared to the total volume that 0:36:19.120 --> 0:36:20.240 is going to require to then. 0:36:20.120 --> 0:36:20.560 Deal with it. 0:36:21.200 --> 0:36:24.080 But the application side is infinite almost. 0:36:23.760 --> 0:36:24.520 That's exactly. 0:36:24.600 --> 0:36:27.040 So that is where like in the end, the bottleneck 0:36:27.080 --> 0:36:27.560 really is. 0:36:28.040 --> 0:36:31.680 So with you know, thousands of scientists and engineers, you 0:36:31.719 --> 0:36:35.480 can create world class AI, right, And so no, you 0:36:35.480 --> 0:36:37.719 don't need ten thousand to be able to create the 0:36:37.800 --> 0:36:39.799 large language model and the generatic model and some but 0:36:39.840 --> 0:36:43.480 you need thousands, and you need you know, very significant 0:36:43.480 --> 0:36:44.479 amount of computer and data. 0:36:44.520 --> 0:36:44.960 You need that. 0:36:45.640 --> 0:36:49.640 The rest is okay, I build software, I build databases, 0:36:49.760 --> 0:36:52.440 or I build a software product that allows you to 0:36:52.480 --> 0:36:55.320 do inventory management, or I build you know, a photo 0:36:55.440 --> 0:37:00.840 editor and so on. Now that product incorporating the AI, modifying, 0:37:00.960 --> 0:37:03.880 expanding it and so on. Well, now you're talking about 0:37:03.920 --> 0:37:06.800 the entire software industries. So now you're talking about millions 0:37:06.840 --> 0:37:09.480 of people right who are neted, you know, who are 0:37:09.560 --> 0:37:12.680 required to bring AI into their product. Then you go 0:37:12.760 --> 0:37:15.759 on a step beyond the technology creators in terms of 0:37:15.800 --> 0:37:18.600 software and you say, well, okay, now what the skills 0:37:18.640 --> 0:37:22.880 to help organizations go undeployed in the department of you know, 0:37:22.960 --> 0:37:25.279 the interior, right, And then I said, okay, well, now 0:37:25.320 --> 0:37:28.759 you need like consultants and experts and people to work 0:37:28.800 --> 0:37:31.160 they are to integer into the workflow. So now you're 0:37:31.200 --> 0:37:33.960 talking into the many tens of millions of people around that. 0:37:34.239 --> 0:37:36.600 So I see it as these concentric circles of it. 0:37:37.080 --> 0:37:40.040 But to some degree in many of these core technology areas, 0:37:40.080 --> 0:37:41.600 just saying like, well, I need a team of like 0:37:41.600 --> 0:37:43.960 one hundred thousand people to create like AI or a 0:37:44.280 --> 0:37:46.799 or a new transistor or a new quantum computer. It's 0:37:46.840 --> 0:37:49.040 actually a diminished in return right in the end, like 0:37:49.120 --> 0:37:50.920 too many people connecting with each other. 0:37:50.800 --> 0:37:53.480 Is very difficult. But on the application side of just 0:37:53.880 --> 0:37:58.600 to go back to our our example of that college, 0:37:59.160 --> 0:38:03.200 just a task of sitting down with a faculty and 0:38:03.320 --> 0:38:07.640 working with them to reimagine what they do with these 0:38:07.680 --> 0:38:10.319 new set of tools in mind, with the understanding that 0:38:10.360 --> 0:38:12.200 the students coming in are probably going to know more 0:38:12.200 --> 0:38:14.839 about it than they do that a lot. I mean, 0:38:14.880 --> 0:38:18.680 that's that is a hurriculeion people problem. 0:38:18.880 --> 0:38:19.720 It's a people problem. 0:38:19.800 --> 0:38:21.760 Yeah, that's why I started in terms of the barriers 0:38:21.760 --> 0:38:23.800 of adoption of that, I mean the context of IBM, 0:38:23.840 --> 0:38:28.680 an example That's why we have a consulting organization, Ivan Consulting, 0:38:28.719 --> 0:38:32.799 that complements ib and technology, and the Ivan Consulting Organization 0:38:32.880 --> 0:38:35.840 has over one hundred and fifty thousand employees because of 0:38:35.880 --> 0:38:38.279 this question, right, because you have to sit down and 0:38:38.320 --> 0:38:40.920 you say, Okay, what problem are you trying to solve? 0:38:41.239 --> 0:38:43.160 What is a methodology we're going to do, and here's 0:38:43.160 --> 0:38:45.200 the technology options that we have to be able to 0:38:45.200 --> 0:38:49.279 bring into the table. In the end, the adoption across 0:38:49.920 --> 0:38:54.239 or society will be limited by this part. The technology 0:38:54.280 --> 0:38:56.640 is going to make it easier, more cost effective to 0:38:56.719 --> 0:39:01.319 implement those solutions. You first have to think about what 0:39:01.360 --> 0:39:03.239 you want to do, how you're going to do it, 0:39:03.360 --> 0:39:04.719 and how are you going to bring it into a 0:39:04.760 --> 0:39:08.120 life of this in this context faculty member or you know, 0:39:08.200 --> 0:39:10.200 the administrator and so on in the college. 0:39:10.480 --> 0:39:13.279 With that Hollywood that that notion I thought, which was 0:39:13.320 --> 0:39:18.600 absolutely I thought really interesting that in a Hollywood strike 0:39:18.680 --> 0:39:22.160 you have to have this conversation about a distributive justice, 0:39:22.360 --> 0:39:25.200 conversation about how do we That's it's a really hard 0:39:25.239 --> 0:39:28.560 conversation right to have and a So this brings me 0:39:28.560 --> 0:39:30.080 to my netflie, which is that you we were talking 0:39:30.080 --> 0:39:34.360 a backstage you have. You have two daughters, one in college, 0:39:34.400 --> 0:39:37.239 one about to go to college. That's right, so they're 0:39:37.280 --> 0:39:41.719 both scients minded. So tell me about the conversations you 0:39:41.719 --> 0:39:44.560 you have with your daughter. You have a unique conversation 0:39:44.680 --> 0:39:47.680 with your daughters because your conversation, your advice to them 0:39:47.880 --> 0:39:51.280 is is influenced by what you do for a living. 0:39:51.560 --> 0:39:52.400 Yes, it's true. 0:39:52.560 --> 0:39:56.880 So did you warn your daughters away from certain fields? 0:39:56.920 --> 0:40:02.440 Did you say, whatever you do, don't be No, no. 0:40:01.440 --> 0:40:04.040 No, that's not my style. I mean, for me, no, 0:40:04.320 --> 0:40:06.439 I try not to be like you know, preachy about that. 0:40:07.400 --> 0:40:09.760 So for me, it was just about showing by example 0:40:09.840 --> 0:40:12.520 all things I love, right and things I care about, 0:40:12.920 --> 0:40:14.920 and then you know, bringing them to the lab and 0:40:14.960 --> 0:40:18.440 seeing things, and then the natural conversations of things working 0:40:18.480 --> 0:40:21.800 on or interesting people I meet, so to the extent 0:40:21.840 --> 0:40:23.759 that they have chosen that and obviously this has an 0:40:23.800 --> 0:40:27.920 influence on them. It has been through seeing it, you know, 0:40:28.000 --> 0:40:29.400 perhaps through my eyes, right. 0:40:29.280 --> 0:40:30.560 And what do you see me do? And that I 0:40:30.640 --> 0:40:31.600 like my profession? Right. 0:40:31.680 --> 0:40:34.600 But one of your daughters, you said, is thinking that 0:40:34.640 --> 0:40:37.920 she wants to be a doctor. But being a doctor 0:40:38.000 --> 0:40:40.520 in a post AI world is surely a very different 0:40:40.520 --> 0:40:43.160 proposition than being a doctor in a pre AI world. 0:40:43.600 --> 0:40:46.279 Do you think you have you tried to prepare her 0:40:46.320 --> 0:40:49.120 for that difference. Have you explained to her what you 0:40:49.200 --> 0:40:51.319 think will happen to this profession she might enter. 0:40:51.840 --> 0:40:54.760 Yeah, I mean not in like, you know, incredible amount 0:40:54.800 --> 0:40:57.640 of detail, but but but yes, at the level of 0:40:57.760 --> 0:41:02.000 understanding what is changing, like lens of the information, lens 0:41:02.040 --> 0:41:03.920 with which you can look at the world and what 0:41:04.080 --> 0:41:07.600 is possible and what it can do, Like what is 0:41:07.600 --> 0:41:09.960 our role and what is the role of the technology 0:41:09.960 --> 0:41:12.839 and how that shapes At that level of abstraction, for sure, 0:41:13.160 --> 0:41:15.760 but not at the level of like, don't be a radiologist, 0:41:15.880 --> 0:41:17.319 you know, because this is what we. 0:41:17.239 --> 0:41:19.320 Want for you. I was gonna say, if you'ren't happy 0:41:19.320 --> 0:41:21.359 with your current job, you could do a podcast called 0:41:21.360 --> 0:41:25.240 Parenting Tips with Dario, which is just an AI person 0:41:25.840 --> 0:41:28.080 gives you advice on what your kids should do based 0:41:28.120 --> 0:41:30.720 on exactly this, Like should I be a radiologist? Dario 0:41:30.920 --> 0:41:35.240 tell me, Like it seems to be a really important question. Yeah, 0:41:35.560 --> 0:41:37.360 let me ask this question in a more I'm joking, 0:41:37.400 --> 0:41:41.640 but in a more serious way. Surely it would If 0:41:41.719 --> 0:41:43.320 I don't mean to use your daughter as an example. 0:41:43.360 --> 0:41:45.560 But let's imagine we're giving advice to someone who wants 0:41:45.600 --> 0:41:49.920 to enter medicine. A really useful conversation to have is 0:41:50.440 --> 0:41:54.120 what are the skills that are will be most prized 0:41:55.040 --> 0:41:58.239 in that profession fifteen years from now, and are they 0:41:58.280 --> 0:42:00.799 different from the skills that are prized now. How would 0:42:00.840 --> 0:42:01.880 you answer that question? 0:42:02.719 --> 0:42:03.000 Yeah? 0:42:03.320 --> 0:42:07.160 I think for example, this goes back to how is 0:42:07.160 --> 0:42:10.080 the scientific method in this context, like the practice of 0:42:10.120 --> 0:42:12.839 medicine going to change. I think we will see more 0:42:12.920 --> 0:42:15.319 changes on how we practice a scientific method and so 0:42:15.360 --> 0:42:19.360 on as a consequence of what is happening with the 0:42:19.360 --> 0:42:22.600 world of computing and information. How we represent information, how 0:42:22.600 --> 0:42:26.120 we represent knowledge, how we extract meaning from knowledge as 0:42:26.120 --> 0:42:29.120 a method than we have seen in the last two 0:42:29.160 --> 0:42:32.880 hundred years. So therefore, what I would like strongly encourage 0:42:32.960 --> 0:42:35.080 is not about like, hey, use this tool for doing 0:42:35.120 --> 0:42:38.080 this or doing that, but in the curriculum itself, in 0:42:38.320 --> 0:42:41.920 understanding how we do problems solving in the age of 0:42:42.040 --> 0:42:45.120 like data and data representation and so on, that needs 0:42:45.120 --> 0:42:48.400 to be embedded in the curriculum of everybody you know. 0:42:48.440 --> 0:42:51.240 That is I would say, actually quite horizontally. 0:42:50.640 --> 0:42:53.360 But certainly in the context of medicine and scientists and 0:42:53.360 --> 0:42:56.600 so on for sure, And to the extent that that 0:42:56.800 --> 0:42:59.400 gets ingrained, that will give us a lens that no 0:42:59.480 --> 0:43:03.480 matter what what specialty they go within medicine, they will say, Actually, 0:43:03.880 --> 0:43:06.440 the way I want to be able to tackle improving 0:43:06.440 --> 0:43:08.799 the quality of care, the way to do that is 0:43:08.920 --> 0:43:11.320 in addition to all the elements that we have practiced 0:43:11.719 --> 0:43:14.120 in the field of medicine, is this new lens? 0:43:14.360 --> 0:43:14.920 And are we. 0:43:14.920 --> 0:43:17.160 Representing the data the right way? Do we have the 0:43:17.239 --> 0:43:20.080 right tools to be able to represent that knowledge? Am 0:43:20.080 --> 0:43:23.160 I incorporating that in my own so with my own 0:43:23.239 --> 0:43:25.320 knowledge in a way that gives me better outcomes? 0:43:25.400 --> 0:43:25.640 Right? 0:43:25.760 --> 0:43:29.080 Do I have the rigor of benchmarking too? And quality 0:43:29.400 --> 0:43:31.839 of the results? So that is what needs to be incorporated. 0:43:32.000 --> 0:43:37.759 How in a perfect world, if I asked you your 0:43:37.800 --> 0:43:42.320 team to rewrite curriculum for American medical schools, how dramatic 0:43:42.440 --> 0:43:45.640 a revision is that? Are we tinkering with ten percent 0:43:45.680 --> 0:43:47.880 of the curriculum or we tinkering with fifty percent of it? 0:43:50.080 --> 0:43:54.279 I think they would be a subset of classes that 0:43:54.400 --> 0:43:56.800 is about the method the methodology, what has changed? Like 0:43:57.280 --> 0:44:01.000 have these lens of it to understand? And then within 0:44:01.239 --> 0:44:05.799 each class that methodology will represent something that is embedded 0:44:05.880 --> 0:44:10.799 in it, right, so it will be substantive but not 0:44:11.000 --> 0:44:14.320 But but doesn't mean replacing the specialization and the context 0:44:14.320 --> 0:44:16.759 and the knowledge of each domain. But I do think 0:44:16.840 --> 0:44:20.839 everybody should have sort of a basic knowledge of the horizontal, right, 0:44:20.880 --> 0:44:23.480 what is it, how does it work, what tools you have, 0:44:23.840 --> 0:44:25.839 what is the technology, and like you know what are. 0:44:25.800 --> 0:44:27.160 The dos and don'ts around that? 0:44:27.640 --> 0:44:29.799 And then every area you say, and you know that 0:44:29.880 --> 0:44:32.920 thing that you learn, this is how it applies to anatomy, 0:44:33.160 --> 0:44:34.880 and this is how you know how it applies to 0:44:35.080 --> 0:44:37.600 you know, radiology if you study that, or or this 0:44:37.640 --> 0:44:39.960 is how you apply you know, in the context of discovery, 0:44:40.040 --> 0:44:42.000 right of self structure and this is how we can 0:44:42.080 --> 0:44:44.480 use it. Or protein folding and this is how it 0:44:44.800 --> 0:44:47.960 does so that way you'll see a connecting tissue through 0:44:48.360 --> 0:44:49.160 throughout the whole thing. 0:44:49.440 --> 0:44:52.359 Yeah, I mean I would add to that because I 0:44:52.400 --> 0:44:57.360 was thinking of the that it's also this incredible opportunity 0:44:57.400 --> 0:45:00.239 to do what doctors are supposed to do but don't 0:45:00.280 --> 0:45:03.160 have time to do now, which is they're so consumed 0:45:03.200 --> 0:45:07.799 with figuring out what's wrong with you that they have 0:45:07.880 --> 0:45:11.600 little time to talk about the implications of the diagnosisness 0:45:11.800 --> 0:45:14.720 and what we really want to if we can freedom 0:45:14.719 --> 0:45:17.600 of some of the burden of what is actually quite 0:45:17.600 --> 0:45:20.120 a prosaic question of what's wrong with you and leave 0:45:20.160 --> 0:45:23.160 the hard human thing of let make should you be 0:45:23.320 --> 0:45:26.880 scared or hopeful? Should you you know? What do you 0:45:27.000 --> 0:45:28.719 need to do? Or what? Let me put this in 0:45:28.760 --> 0:45:31.360 the context of all the patients I've seen. That conversation, 0:45:31.400 --> 0:45:33.040 which is the most important one, is the one that 0:45:33.760 --> 0:45:36.000 seems to me. So like, if I had to, I 0:45:36.040 --> 0:45:39.880 would add, if we're reimagining the curriculum of med school, 0:45:40.239 --> 0:45:43.879 I'd like with whatever, by the way, very little time. 0:45:43.880 --> 0:45:46.880 Maybe we have to add two more years to med school, 0:45:47.000 --> 0:45:52.120 but like a whole but the whole thing about bringing 0:45:52.200 --> 0:45:56.840 back the human side of you know, now, if I 0:45:56.840 --> 0:45:59.440 can give you ten more minutes, how do you use 0:45:59.480 --> 0:46:00.520 that ten more men? 0:46:00.640 --> 0:46:03.759 But in that, in that reconceptualization that you just did, 0:46:04.120 --> 0:46:06.080 is what we should be doing around that, because I 0:46:06.080 --> 0:46:08.920 think the debate as to like, well I'm not going 0:46:09.000 --> 0:46:10.719 to need doctors or not, it's actually a not very 0:46:10.760 --> 0:46:13.960 useful debate. But rather this other question is how is 0:46:14.000 --> 0:46:16.720 your time being spent? What problems are you getting stuck. 0:46:17.040 --> 0:46:19.880 I mean I generalize this by like the obvious observation 0:46:20.040 --> 0:46:22.160 that if you look around in your professions, in our 0:46:22.239 --> 0:46:24.839 daily lives, we have not run out of problems to solve. 0:46:25.120 --> 0:46:27.200 So as an example of that is, hey, if I'm 0:46:27.200 --> 0:46:29.319 spending all my time trying to do diagnosis, and I 0:46:29.320 --> 0:46:31.600 could do that ten times faster, and it allows me 0:46:31.680 --> 0:46:34.560 actually to go, you know, and take care of the 0:46:34.600 --> 0:46:36.560 patients and all the next steps of what we have 0:46:36.600 --> 0:46:38.719 to do about it. That's probably a trade off that 0:46:38.760 --> 0:46:42.160 a lot of doctors would take, right, And then you say, well, 0:46:42.200 --> 0:46:43.880 you know, to what degree does it allow me to 0:46:43.920 --> 0:46:45.839 do that? And I can do these other things and 0:46:45.880 --> 0:46:49.200 these other things are critically important for my profession around that. 0:46:49.640 --> 0:46:52.920 So when you actually become less abstract and like we 0:46:53.080 --> 0:46:56.160 get past the futile conversation of like, oh, there's no 0:46:56.200 --> 0:46:57.920 more jobs and I's going to take it all of it, 0:46:57.920 --> 0:47:01.040 which is kind of nonsense, is you go back to say, 0:47:01.080 --> 0:47:05.120 in practice in your context, right, for you, what does 0:47:05.160 --> 0:47:06.680 it mean, how do you work? 0:47:06.840 --> 0:47:08.480 What can you do differently around that? 0:47:08.600 --> 0:47:11.040 Actually that's a much richer conversation, and very often we 0:47:11.080 --> 0:47:13.279 would find ourselves that there's a portion of the work 0:47:13.320 --> 0:47:15.439 we do that we say I would rather do less 0:47:15.440 --> 0:47:17.880 of that. This is this other part I like a lot, 0:47:18.200 --> 0:47:21.040 And if it is possible that technology could help us 0:47:21.040 --> 0:47:22.360 make that trade off, I'll. 0:47:22.160 --> 0:47:23.080 Take it in a heartbeat. 0:47:23.800 --> 0:47:27.479 Now, poorly implemented technology can also create another problem. 0:47:27.640 --> 0:47:29.480 You say, Hey, this was supposed to solve. 0:47:29.200 --> 0:47:32.399 Me things, but the way it's being implemented is not 0:47:32.440 --> 0:47:35.200 helping me, right, it's making my life more more miserable, 0:47:35.320 --> 0:47:37.920 or so on, or I've lost connection in how I 0:47:38.000 --> 0:47:41.400 used to work, et cetera. So that is why design 0:47:42.080 --> 0:47:45.200 is so important. That is why I also workflow is 0:47:45.239 --> 0:47:48.000 so important in being able to solve these problems. But 0:47:48.840 --> 0:47:52.360 it begins by, you know, going from the intergalactic to 0:47:52.440 --> 0:47:54.840 the reality of it, of that faculty member in the 0:47:54.840 --> 0:47:57.759 liberal Arts college or you know, or a you know, 0:47:57.920 --> 0:48:00.560 a practitioner in medicine in a hospital and what it 0:48:00.680 --> 0:48:01.399 means for them. 0:48:01.520 --> 0:48:01.680 Right. 0:48:02.400 --> 0:48:06.640 Yeah. What struck me Daria throughout our conversation is how 0:48:06.719 --> 0:48:11.880 much of this revolution is non technical, as to say, 0:48:12.160 --> 0:48:14.120 you guys are doing the technical thing here, but the 0:48:14.160 --> 0:48:17.080 real the revolution is going to require a whole range 0:48:17.120 --> 0:48:20.840 of people doing things that have nothing to do with software, 0:48:21.080 --> 0:48:24.520 that have to do with working out new new human arrangements. 0:48:24.960 --> 0:48:27.600 Talking about that, I mean, does keep coming back to 0:48:27.640 --> 0:48:30.400 the Hollywood strike thing that you have to have a 0:48:30.400 --> 0:48:37.120 conversation about our values is creators of of of of movies? 0:48:37.160 --> 0:48:39.920 How are we going to divide up the exactly credit 0:48:39.960 --> 0:48:44.319 and the like. That's a that's a conversation about philosophy, 0:48:44.400 --> 0:48:46.319 and you know, you know it's it is. 0:48:46.280 --> 0:48:49.600 And is it's in the grand tradition of why you know, 0:48:51.080 --> 0:48:52.160 a liberal. 0:48:51.960 --> 0:48:55.359 Education is so important in the broadest possible sense. Right, 0:48:55.760 --> 0:48:59.480 there's no common conception of the good right that is 0:48:59.520 --> 0:49:03.799 always tested a dialogue that happens within our society, and 0:49:03.920 --> 0:49:06.239 technology is going to fit in that context too, right. 0:49:06.280 --> 0:49:08.520 So that's why I personally, as a philosophy I'm not 0:49:08.560 --> 0:49:12.280 a technological determinists, right, And I don't like when colleagues 0:49:12.280 --> 0:49:14.960 in my profession right starts saying like, well, this is 0:49:15.000 --> 0:49:17.560 the way the technology is going to be, and by consequence, 0:49:17.760 --> 0:49:19.719 this is how society is going to be. I'm like, 0:49:19.800 --> 0:49:22.920 that's a highly contested goal. And if you want to 0:49:23.000 --> 0:49:25.560 enter into realm of politics or the real other ones, 0:49:25.560 --> 0:49:28.120 go and stand up on a stool and discuss whether 0:49:28.200 --> 0:49:30.279 that's what society wants. You will find that it's a 0:49:30.400 --> 0:49:34.560 huge diversity of opinions and perspective and that's what makes 0:49:34.600 --> 0:49:37.200 you know, you know, in a democracy, the richness of 0:49:37.239 --> 0:49:39.440 our society, and in the end that is going to 0:49:39.440 --> 0:49:42.399 be the centerpiece of the conversation what do we want? 0:49:43.320 --> 0:49:45.879 You know, who gets what? And so on? And that 0:49:46.200 --> 0:49:48.480 is actually, I don't think it's anything negative. That's acid 0:49:48.520 --> 0:49:51.400 should be because in the end is anchored of who 0:49:51.480 --> 0:49:55.000 we want as humans, you know, you know, as friends, family, citizens, 0:49:55.160 --> 0:49:57.960 and we have many overlapping sets of responsibilities, right and 0:49:58.000 --> 0:50:01.120 as a technology creator my only version ponsibilities not just 0:50:01.160 --> 0:50:02.960 as a scientist and a technology creator. 0:50:03.320 --> 0:50:04.400 I'm also a member. 0:50:04.239 --> 0:50:06.000 Of a family, I'm a citizen, and I'm many other 0:50:06.080 --> 0:50:08.160 things that I care about. And I think that that 0:50:08.400 --> 0:50:13.560 sometimes in the debate of the technological determinists, they start 0:50:13.640 --> 0:50:18.200 now budding into what is the realm of you know, 0:50:18.600 --> 0:50:22.319 justice and you know, in society and philosophy and democracy, 0:50:22.680 --> 0:50:25.279 And that's where they get the most uncomfortable because it's 0:50:25.280 --> 0:50:28.120 like I'm just telling you, like, you know, what's possible, 0:50:28.400 --> 0:50:32.239 and when there's pushback, it's like, yeah, but now we're 0:50:32.280 --> 0:50:35.160 talking about how we live and how we work and 0:50:36.280 --> 0:50:36.839 how much I. 0:50:36.800 --> 0:50:38.040 Get paid or not paid. 0:50:38.360 --> 0:50:42.920 So that technology is important. Technology shapes that conversation, but 0:50:43.000 --> 0:50:45.759 we're going to have the conversation with a different language, 0:50:46.160 --> 0:50:49.360 as it should be, and technologists need to get accustomed 0:50:49.400 --> 0:50:51.319 to if they want to participate in that world with 0:50:51.400 --> 0:50:54.360 the broad consequences. Hey, get a custom to deal with 0:50:54.400 --> 0:50:59.160 the complexity of that world of politics, society, institutions, unions, 0:50:59.200 --> 0:51:01.279 all that stuff. And you know, you can be like 0:51:01.360 --> 0:51:03.840 whiny about it. It's like they're not adopting my technology. 0:51:04.000 --> 0:51:06.280 That's what it takes to bring technology into the world. 0:51:07.200 --> 0:51:13.759 Yeah, well, said thank you Dario for this wonderful conversation. 0:51:13.920 --> 0:51:17.600 Thank you to all of you for coming and listening, 0:51:17.800 --> 0:51:19.239 and thank you. 0:51:19.520 --> 0:51:21.360 Thank you. 0:51:23.360 --> 0:51:26.480 Dario gild transformed how I think about the future of AI. 0:51:27.200 --> 0:51:29.480 He explained to me how huge of a leap it 0:51:29.640 --> 0:51:33.000 was when we went from chess playing models to language 0:51:33.080 --> 0:51:36.279 learning models, and he talked about how we still have 0:51:36.360 --> 0:51:39.000 a lot of room to grow. That's why it's important 0:51:39.360 --> 0:51:42.640 that we get things right. The future of AI is 0:51:42.760 --> 0:51:47.040 impossible to predict, but the technology has so much potential 0:51:47.120 --> 0:51:50.880 in every industry. Zooming into an academic or medical setting, 0:51:51.080 --> 0:51:54.200 showed just how close we are to the widespread adoption 0:51:54.560 --> 0:51:58.200 of AI. Even Hollywood is being forced to figure this out. 0:51:58.840 --> 0:52:01.560 Institutions have all Soortz will have to be at the 0:52:01.600 --> 0:52:05.280 forefront of integration in order to unlock the full power 0:52:05.280 --> 0:52:10.239 of AI thoughtfully and responsibly. Humans have the power and 0:52:10.280 --> 0:52:14.279 the responsibility to shape the tech for our world. I 0:52:14.520 --> 0:52:17.399 for one, I'm excited to see how things play out. 0:52:18.719 --> 0:52:22.879 Smart Talks with IBM is produced by Matt Romano, Joey Fishground, 0:52:23.120 --> 0:52:27.520 David jaw and Jacob Goldstein. We're edited by Lydia Jane Kott. 0:52:27.840 --> 0:52:32.320 Our engineers are Jason Gambrel, Sarah Bruguier, and Ben Holliday. 0:52:32.960 --> 0:52:38.200 Theme song by Gramoscope. Special thanks to Andy Kelly, Kathy Callahan, 0:52:38.560 --> 0:52:41.399 and the eight Bar and IBM teams, as well as 0:52:41.400 --> 0:52:45.160 the Pushkin marketing team. Smart Talks with IBM is a 0:52:45.160 --> 0:52:49.759 production of Pushkin Industries and Ruby Studio at iHeartMedia. To 0:52:49.800 --> 0:52:54.880 find more Pushkin podcasts, listen on the iHeartRadio app, Apple Podcasts, 0:52:55.000 --> 0:53:00.279 or wherever you listen to podcasts. I'm Malcolm Gladwell. This 0:53:00.360 --> 0:53:07.239 is a paid advertisement from IBM.