00:33:27 · Josh Meier
[00:33:27] S0: Well, people often look at drug approval as the thing that people often think about. They're like, it's going to take ten years for these drugs to get approved. But the reality is the clinical trials will start way sooner than that, and patients are actually being helped already at that point. We're both doing an experiment with a trial, but we're also hopefully looking for some efficacy and actually trying to help people at that stage. And I think that's, you know, very possibly going to happen in a very small number of the next few years, It's one of the biggest industries. It's something where a ton of money is spent on any single, given program. And pharma's very competitive as well, right? If you work on a drug project and, you are even a year ahead of your competitors, that can turn into billions of dollars of additional value downstream, it's so an industry where people are very incentivized to use the best. I think it's almost different than something like cogeneration, for example, where it's like, well, maybe this model is like a little bit worse, but it's like a 100 times cheaper, so, you know, maybe I should try that instead. The dynamics of this industry are actually quite different, and I think that's why as a technologist, it's actually one of the most exciting things to work on, because, the cost equation is very different here. People are less worried about the costs, and they are actually much more focused on the upside that can be created, And as a team that wants to be very ambitious in terms of how we can push, you know, medicines and technology forward, it's it's a very good industry to be playing in as a result.
00:34:46 · Adam Miller, Josh Meier, Sachin
[00:34:46] S2: So we're also gonna be making a docker on sci quantum, and in quantum computing, one of the promises is that it'll be able to enable drug discovery to also be a lot faster. If that becomes true, does Chai then become obsolete? [00:35:00] S0: So I don't know a ton about about quantum, so I'm very excited to watch the episode, but I think that if we could really accelerate computation, it's actually hugely impactful to what we're doing here, right? We're trying to turn compute into molecules, and if we can get more of that or we can solve these problems in a more meaningful way, I think that could hugely accelerate what we're doing here. [00:35:19] S2: It's probably very good for you, given your probable very high NVIDIA bills right now. [00:35:24] S0: That's right. [00:35:25] S2: Yeah. How do you [00:35:27] S1: hypothesize that three is going to be received by the market and what are people going to think about these results?
00:35:33 · Josh Meier
[00:35:33] S0: Yeah, it's, I'm very excited to see how it lands. If we look at our last release with Chai two, people almost didn't believe it. They were like, how is this possible? And there was like even skepticism and then yeah, people started to, it started to soak in, people realized, okay, yeah, this is, I guess this is physically possible and now someone has achieved it. And then, you know, companies started using it, right? You have the Lilly announcement, for example, and people are like, okay, I guess, I guess this is the new way, that, that molecules might be made. With three, I think we're at this point now where people think of, AI and drug discovery as as sometimes being a point solution. Like, I'm able to get that starting point, but I still need to do bunch of work to make the molecules better. And with Chai three, we've taken a a massive leap forward, not just in the success rates, but in quality of the molecules that are coming out, and I think it's going to be another one of these like, oh wow moments of like, oh wow, I guess like it doesn't just stop at what we saw before, these molecules can keep getting better, and I think we might even get to a place where people almost like, they're just like suspend disbelief, right, where it's like things are moving so quickly in this space. It feels like any time people have like a critique or a question about the models in this space, it just gets solved a couple months later with the next model release. So at some point, I think people, I hope it doesn't get too far where people's expectations just get too high because of the progress is being made so far, but actually that might be a good thing because it means people are starting to think about what problems... Okay, we want people to think about what problems they can solve with Chai three. But it's actually pretty valuable if pharma starts thinking about what are they going to solve with Chai four and Chai five, because the ideas can start percolating, and then once the models are here, then we can get really fast adoption with those technologies. So I think it's almost great for people to be following that curve and to start to even now start extrapolate where things might go from here.
00:37:11 · Sachin, Josh Meier, Adam Miller
[00:37:11] S1: I think what's so exciting about this is our whole mission of this channel is based around tech optimism. And I think when we think about what's happening in LLM land, there is real layoffs happening now and there's real kind of a turning point in society whether or not we want to keep continuing this progress, right? Like there's no matter what side of that equation you sit on, there is like a real conversation going on. But with what you guys are doing, it's so objectively good for the world that it's like, let's throw as much compute at this as possible because the world could look profoundly different. Maybe there's a world in which anyone with any disease can design the perfect drug specifically for them and have it in their hands really, really quickly. Like that's just such a objectively amazing world. [00:37:51] S0: This is what I tell our cloud providers whenever we can't get compute. You should move this off of some entertainment app and move it over to actually making medicines that can help cure and treat diseases. [00:38:02] S2: Do you think there will still be skeptics into the future? Like, is there going to be an element of storytelling here where you need to speak to regulators, consumers, different sort of health providers, and really, like, sort of guide them along the journey of how this all works, or do you think the results are simply just gonna speak to themself? Because if all of this is successful, it is a complete shift change for humanity.
00:38:02 · Adam Miller, Josh Meier
[00:38:02] S2: Do you think there will still be skeptics into the future? Like, is there going to be an element of storytelling here where you need to speak to regulators, consumers, different sort of health providers, and really, like, sort of guide them along the journey of how this all works, or do you think the results are simply just gonna speak to themself? Because if all of this is successful, it is a complete shift change for humanity. [00:38:23] S0: I think it will be the latter. I think people will see the results, and then they'll be convinced. The same thing happened in the rest of AI. You don't really have skeptics of chat GPT. It's like the thing works. Try it out for yourself. And the thing with drug discovery is it is, you know, it's not as simple as a chat GPT and looking at the answer that comes out. You really need to do some lab testing to understand whether these molecules work, but those experiments are being done, and, and turns out it doesn't take that long, right, maybe two weeks to see if the thing is working. So I I think we're gonna move away from this world of skepticism towards one of more, okay, how can I use this to solve my problem? And that's that's a really exciting place to be.
00:38:23 · Josh Meier, Adam Miller
[00:38:23] S0: I think it will be the latter. I think people will see the results, and then they'll be convinced. The same thing happened in the rest of AI. You don't really have skeptics of chat GPT. It's like the thing works. Try it out for yourself. And the thing with drug discovery is it is, you know, it's not as simple as a chat GPT and looking at the answer that comes out. You really need to do some lab testing to understand whether these molecules work, but those experiments are being done, and, and turns out it doesn't take that long, right, maybe two weeks to see if the thing is working. So I I think we're gonna move away from this world of skepticism towards one of more, okay, how can I use this to solve my problem? And that's that's a really exciting place to be. [00:38:58] S2: Do you think philosophically there might be skeptics about curing everything almost on demand? Like, for example, with ChatGPT, you've still got a large percentage of the population that are just plainly anti AI. They think it might erode our cognitive abilities. They think it's going to hurt labor markets, etc. Do you think there'll be these anti health skeptics eventually in the future?