1 00:00:00,450 --> 00:00:04,447 - Let's have faith in the pace of progress of the technology, 2 00:00:04,530 --> 00:00:07,177 because if the models get good enough to do it end to end, 3 00:00:07,260 --> 00:00:10,107 a year from now and only then you start deploying it. 4 00:00:10,190 --> 00:00:12,607 There will be another two year delay and that's, you know, that's two years 5 00:00:12,690 --> 00:00:14,257 during which all the work that you're doing 6 00:00:14,340 --> 00:00:16,687 to benefit patients is not happening. 7 00:00:16,770 --> 00:00:21,217 Hello everyone. My name is Diogo Rau and I'm Chief Information and Digital Officer 8 00:00:21,300 --> 00:00:24,427 of Eli Lilly and Company. I'm joined here with Dario, 9 00:00:24,510 --> 00:00:26,767 who is the Founder and CEO of Anthropic. 10 00:00:26,850 --> 00:00:28,867 Dario, thanks for joining me today. 11 00:00:28,950 --> 00:00:31,327 - Thanks for having me, Diogo. - I know 12 00:00:31,410 --> 00:00:33,577 you're spending a lot of time now thinking about 13 00:00:33,660 --> 00:00:34,837 how do you work better with enterprises? 14 00:00:34,920 --> 00:00:36,517 What's your enterprise strategy 15 00:00:36,600 --> 00:00:41,137 and how do you see Anthropic different from other providers? 16 00:00:41,220 --> 00:00:42,997 - Yeah, I mean, you know, I think we've made a number 17 00:00:43,080 --> 00:00:45,787 of choices that are different, right? 18 00:00:45,870 --> 00:00:49,777 So if I think about the incentives given by consumer AI, 19 00:00:49,860 --> 00:00:53,767 their folks are in a competition for engagement and growth, right? 20 00:00:53,850 --> 00:00:57,487 And so that drives a lot of behaviors of the AI 21 00:00:57,570 --> 00:01:00,337 that I think are not ideal from an enterprise perspective, 22 00:01:00,420 --> 00:01:02,617 For example, there's this idea of model sycophancy 23 00:01:02,700 --> 00:01:05,337 where the model tells you whatever you say is a good idea, - Right. 24 00:01:05,420 --> 00:01:08,947 right, and even on the consumer side that can, you know, cause problems. 25 00:01:09,030 --> 00:01:10,597 We've seen stories of people who are like, oh yeah, 26 00:01:10,680 --> 00:01:12,777 I've discovered a new fundamental theory of physics. 27 00:01:12,860 --> 00:01:16,387 - And that's right. - Models, like, that's great and maybe you don't want it to, 28 00:01:16,470 --> 00:01:18,817 maybe you don't want it to say that, 29 00:01:18,900 --> 00:01:22,327 but I think, you know, of course on the enterprise side, 30 00:01:22,410 --> 00:01:27,637 you know, the problems are much greater and clearer with that, where, you know, 31 00:01:27,720 --> 00:01:30,787 you really don't want the model to say, oh yeah, this drug compound's great. 32 00:01:30,870 --> 00:01:32,767 Spend millions of dollars to, you know, 33 00:01:32,850 --> 00:01:34,897 I just think this is, you know, I think your idea is great. 34 00:01:34,980 --> 00:01:37,447 I think it's really promising, like, you want truth. 35 00:01:37,530 --> 00:01:38,363 - Yeah. 36 00:01:39,240 --> 00:01:41,737 - And so I think that incentive has led us 37 00:01:41,820 --> 00:01:43,447 to design our models in a different way, right? 38 00:01:43,530 --> 00:01:47,707 I think it's more compatible with making the model smarter, 39 00:01:47,790 --> 00:01:53,497 making them better at a wide variety of economically valuable tasks. 40 00:01:53,580 --> 00:01:58,087 And it causes us to put a premium on accuracy and reliability. 41 00:01:58,170 --> 00:02:01,147 One experiment, you know, that I give to everyone, 42 00:02:01,230 --> 00:02:03,007 although it's particularly relevant 43 00:02:03,090 --> 00:02:05,077 because I'm talking to you, is I say, you know, 44 00:02:05,160 --> 00:02:08,107 let's say I improve the model's knowledge of biochemistry 45 00:02:08,190 --> 00:02:11,107 from undergraduate level knowledge to graduate level knowledge. 46 00:02:11,190 --> 00:02:12,277 You know, if I go to consumers 47 00:02:12,360 --> 00:02:15,217 and say that, 99% of them are going to say, you know, 48 00:02:15,300 --> 00:02:16,837 I didn't know what you were talking about before. 49 00:02:16,920 --> 00:02:20,737 I don't know what you're talking about now, but if I go to you, like you really, 50 00:02:20,820 --> 00:02:22,657 you care about that, a lot. - Appreciate that. 51 00:02:22,740 --> 00:02:25,477 - Like, that's very important. 52 00:02:25,560 --> 00:02:27,697 - That's exactly right. Well, actually, that gets into something else 53 00:02:27,780 --> 00:02:30,067 that you've launched as well, which is Skills, right? 54 00:02:30,150 --> 00:02:32,047 There are a lot of skills that you want in biology 55 00:02:32,130 --> 00:02:34,950 or even just skills, like, as an enterprise, 56 00:02:35,810 --> 00:02:38,947 how you want to operate. Is that part of the future for you as well? 57 00:02:39,030 --> 00:02:40,597 - Yeah, I definitely think so. 58 00:02:40,680 --> 00:02:46,177 I mean, things ranging from skills to, you know, we're in the process of launching 59 00:02:46,260 --> 00:02:49,447 various specialized Claudes which are, you know, in some cases 60 00:02:49,530 --> 00:02:52,957 will be improvements to the model itself, fine tunings of the model. 61 00:02:53,040 --> 00:02:54,997 But in some cases it'll be something 62 00:02:55,080 --> 00:02:58,657 that looks more like wrapping the model with access 63 00:02:58,740 --> 00:03:00,097 to particular types of information. 64 00:03:00,180 --> 00:03:02,257 So when we did Claude for Financial Services, 65 00:03:02,340 --> 00:03:07,807 you know, we connected to a lot of the usual kind of indices and ratings. 66 00:03:07,890 --> 00:03:13,147 And so, you know, you'd be surprised to connect Claude to those things 67 00:03:13,230 --> 00:03:16,687 and kind of use it in a way that's aware of that knowledge is valuable. 68 00:03:16,770 --> 00:03:18,637 So I think, you know, we're working on a Claude 69 00:03:18,720 --> 00:03:20,887 for Life Sciences that will be some mixture 70 00:03:20,970 --> 00:03:25,867 of making the model inherently smarter and wrapping it with various things, right? 71 00:03:25,950 --> 00:03:28,147 I don't know exactly what the analogy will be here, 72 00:03:28,230 --> 00:03:30,967 but like, geez, there are zillions of databases of, 73 00:03:31,050 --> 00:03:34,117 you know, proteins, compounds, assays, like, you know, 74 00:03:34,200 --> 00:03:36,607 you probably want that at the model's fingertips. 75 00:03:36,690 --> 00:03:39,697 - Well, any parting advice for those of us that are working in this world 76 00:03:39,780 --> 00:03:41,917 of drug discovery and development? 77 00:03:42,000 --> 00:03:44,677 - You know, I would say there's a temptation 78 00:03:44,760 --> 00:03:48,157 and it's, I think it's hard to avoid starting this way 79 00:03:48,240 --> 00:03:51,097 of, you know, what are the small things we can do with AI? 80 00:03:51,180 --> 00:03:53,707 Like, and in a way you just kind of have to start there. 81 00:03:53,790 --> 00:03:56,557 I think one of my pieces of advice is be very, 82 00:03:56,640 --> 00:03:59,917 very ambitious in terms of where the models are going. 83 00:04:00,000 --> 00:04:04,177 I think you can get caught in a mode where there's an existing process, 84 00:04:04,260 --> 00:04:09,877 it has 20 parts you want to swap in AI, to part five and part 12. 85 00:04:09,960 --> 00:04:13,567 And, you know, that can actually be hard 86 00:04:13,650 --> 00:04:16,537 because part 12 has to, you know, intersect with part 13 87 00:04:16,620 --> 00:04:19,657 and part 11, which are not being done with AI. 88 00:04:19,740 --> 00:04:23,887 And you know, you look at it and you're like, well, the AI models aren't 89 00:04:23,970 --> 00:04:28,027 where they could do, you know, part zero to part 20, end to end, 90 00:04:28,110 --> 00:04:30,277 But in a year they might be. - That's right. - And so, 91 00:04:30,360 --> 00:04:34,057 you should start thinking now, don't get too seduced by, oh, 92 00:04:34,140 --> 00:04:36,727 we can make these little hill climbing gains 93 00:04:36,810 --> 00:04:39,337 by, you know, doing this part and that part, 94 00:04:39,420 --> 00:04:43,207 let's start preparing to do the whole thing end to end. 95 00:04:43,290 --> 00:04:47,377 Let's have faith in the pace of progress of the technology. 96 00:04:47,460 --> 00:04:50,077 Because if the models get good enough to do it end to end, 97 00:04:50,160 --> 00:04:52,837 a year from now and only then you start deploying it, 98 00:04:52,920 --> 00:04:55,207 there will be another two year, - That's right. - There'll be 99 00:04:55,290 --> 00:04:56,947 another two year delay. And that's, you know, that's two years 100 00:04:57,030 --> 00:04:58,657 during which all the work that you're doing 101 00:04:58,740 --> 00:05:00,877 to benefit patients is not happening. 102 00:05:00,960 --> 00:05:05,527 Whereas if you go in parallel, if you start preparing now for the large change 103 00:05:05,610 --> 00:05:09,607 as the models are getting better, then you know, you may save years of time. 104 00:05:09,690 --> 00:05:12,157 - That's right. So don't do two year long projects 105 00:05:12,240 --> 00:05:16,077 and expect that it's gonna be exactly the same way in two years from now as this, 106 00:05:16,160 --> 00:05:19,627 - Yes, if you do two year long projects, plan for where the AI is gonna, I mean, 107 00:05:19,710 --> 00:05:23,737 that sounds like an obvious thing to say, but I think it actually takes 108 00:05:23,820 --> 00:05:25,177 a lot of courage and foresight to do that. 109 00:05:25,260 --> 00:05:27,907 - It does. For sure. Well, thanks a lot for taking the time to chat today. 110 00:05:27,990 --> 00:05:32,500 Really appreciate it. - Yeah, yeah. Thank you.