EP 477: Putting patients first with medical AI

This is the Everyday AI show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business and everyday life.

This is one of those episodes that I think relates to actually everyone, because even if you don't work in the medical field, we all, especially here in the us, have run ins with the medical system, right? We all want better care. We want more affordable medical care. We want things to go faster. We want access to better, better drugs and more effective medicines. And maybe what you don't realize is generative AI is actually, I think, rewriting the rules of how the medical industry has traditionally worked for many decades, even when we had traditional machine learning and artificial intelligence. I think that generative AI has completely changed not just things like drug discovery, but also how we can put patients first. So I am excited to talk about that and a lot more today on Everyday AI. What's going on, y'?

All?

My name is Jordan Wilson and welcome to the show. This is Everyday AI and we do this every single day. This is a daily live stream, podcast and free daily newsletter helping everyday people like you and me not just keep up with all the advancements in the industry, but how we can all make sense of them and leverage them to grow our companies and our careers. Because I want you, by listening and reading our newsletter, I want you to be the smartest person in AI at your company or in your department. And you do that not just by listening and learning, but from our guest today, who I'm excited to bring on here in a minute, but also by going to our website@your everydayai.com because we're going to be recapping the highlights and putting more information from today's conversation into that newsletter. So you want to make sure to check that out. And while you're at our website, we have like almost 500 episodes. We've interviewed hundreds of the world's leading experts in AI from every single field that you can think about. So go get yourself educated on the website. All right, so I'm excited today to talk a little bit more about how we can actually put patients first with medical AI. If, if you're tuning in for the news, don't worry, that's going to be in the newsletter. This is technically pre recorded, so. All right, let's bring on our guest for today. I'm excited. So live stream audience, please help me welcome to the show. There we go. We have Lena Nilsen, the SVP and head of platform and product at Recursion. Lena thank you so much for joining the Everyday AI Show.

Thanks for having me, Jordan. Super excited to be here.

All right, so we've actually had one of your colleagues on maybe a year or so ago. I know there's been a lot happening at Recursion and in the industry, but just catch us up. What does recursion do? And tell us a little bit about your role.

Yeah. So I'm leading the team that builds our machine learning models and our huge robotic laboratories, generating over 2 million experiments every week, then creating scientific insights to drive our drug discovery pipeline and get better drugs to patients faster. That's what I do at Recursion together with about 800 other colleagues, split about evenly across tech and biological and chemical sciences over the last year or so. Since you heard less time from my colleague Chris Gibson, the founder CEO of the company, we have really evolved our pipeline so that we have lots of potential drugs, potential treatments in clinical trials. That means that we have the confidence and the FDA have the confidence that we're now testing these drugs for safety and efficacy in human patients. And we're just really excited for the readouts we've had to date and the ones coming up over the next 18 months or so.

Awesome. Yeah. And give us just like high level of recursion. Like, are you just in the drug discovery business? Like, what else do you all do?

Yeah, I mean, the main thing we do is just drug discovery.

Yeah, that small thing that changes the world, right?

Yeah. This is huge industry, but we believe that the data sets that we're building can fundamentally be used in much broader areas. Right. We're not going after drug discovery one disease, one gene at a time. Instead, we're modeling biology and chemistry. That's really broadly so. For example, when we look at a genetic disease, we don't model just that disease. We induce CRISPR modifications across the whole genome in human cells so that we get a holistic view. And you can imagine that there's lots of interesting things that you can do, insights you get when you take this broad floodlights view instead of a narrow flashlight view. Right.

And you know you dropped that word in there. Crispr. Explain that for our audience. That may not keep up with cutting edge medical tech.

Yeah, sure. CRISPR is this Nobel Prize winning, super cool technology that allows you to edit individual human cells very precisely without any unintended side effects that were common with prior technologies. It's like a really way cool and powerful way to interrogate the human genome. Maybe a decade or so ago, we had the first full sequences of the human genome, but we weren't able to manipulate it in the lab to really understand what was going on, to intervene. And now that is possible to do just a large scale, and we do exactly that at recursion to. To build our large machine learning models. So maybe a pitch here is also outside of medical, right. If you're thinking about how to approach machine learning in a big way, where you are, what are the big, powerful new data sets and how can they be generated? Right. And CRISPR in the medical field has been such an unlock for many companies.

And I do want to eventually talk a little bit more about new powerful data sets and what that means in terms of large language models or small models that companies of all, all types can use. But maybe I first want to zoom out a little bit. Can you tell us, like, where are we at? Because even as someone that follows AI literally every single day, I can't keep up with every industry. Right. What's the latest? Where are we at with, you know, medical AI, AI and drug discovery, especially, you know, with the, the, the last couple years of the generative AI boom.

Yeah, yeah. So maybe just putting drug discovery in context, right? You look at kind of the numbers today, it takes a bit. We're two mil, two billion or two and a half billion, depending on how you count, to get a new drug to market, and it will take you over a decade. Right. It's really slow and it's getting slower and harder. Right. The trajectory is the wrong direction. And at the same time, we're at this inflection point of new biological and chemical technologies like CRISPR that we just talked about. And also these amazing new opportunities in a plethora of new machine learning technologies, transformers and LLMs, in new scientific agents, in representation and reinforcement learning and so on. And so you're starting to see companies like recursion really capture this moment to bring, we hope, a change in, in the inflection of that curve to get cheaper and better drugs to patients faster. And maybe if I can say last thing here on that, Jordan, you know, it might feel like there's tons of drugs out there. Right. You can keep track of all the ads on tv, for example. Right. But just to give one example data point, in the rare disease space, as where recursion started, there's 7,000 rare diseases that cumulatively have a huge disease burden, and the vast, vast majority of them have no treatment whatsoever. So there's just a gigantic unmet medical need. Whether you're talking about oncology, neurology, rare disease just across areas and we need to do better and we need to do it now.

So can you quickly tell us why is the process, why does it take two and a half billion dollars in over a decade? Right, Like I'm sure there's a lot of like red tape and things that maybe we can't go too deeply in, but like overall, why does it take so long and why is it so expensive? Are you still running in circles trying to figure out how to actually grow your business with AI? Maybe your company has been tinkering with large language models for a year or more, but can't really get traction to find ROI on Genai. Hey, this is Jordan Wilson, host of this very podcast. Companies like Adobe, Microsoft and Nvidia have partnered with us because they trust our expertise in educating the masses around generative AI to get ahead. And some of the most innovative companies in the country hire us to help with their AI strategy and to train hundreds of their employees on how to use Gen AI. So whether you're looking for ChatGPT training for thousands or just need help building your front end AI strategy, you can partner with us too. Just like some of the biggest companies in the world do. Go to your everyday AI.com partner to get in contact with our team or you can just click on the partner section of our website. We'll help you stop running in those AI circles and help get your team ahead and build a straight path to ROI. On genius.

Yeah, I mean it's complicated, right? Many reasons, but maybe fundamentally one part is that we often start historically in the industry with very narrow hypothesis and that's pursued and if you hit a dead end, that data is tossed, it sits somewhere in a notebook and you're done and you start again, right? And what new generation of companies are doing like recursion is to say no, no, no, no, no. We want to keep the data and we want to generate it in a way where it's relatable over time, right. With much higher quality control, with tracking of the metad was happening in the laboratory, what were the other reagents? So that rather than starting over again and again and again when you hit a dead end, because you will, that data can be compared and joined together quantitatively over time. So I'll give a concrete example here, right. Every time recursion screens compound a potential new drug that results of that, what it did to human cells gets put in databases where models can find it years later. And that will happen, right? We're working on a new program and the Models will say, oh, already have this compound in your library. It's sitting there. You should pick that up. You tested it for something totally different. It didn't work there. But here, look, the hook is here. So that's like an example of like a totally different, unsiloed way to think about drug discovery where insights kind of get a flywheel effect of accumulation rather than this narrow one. And done.

It's, it's, it's interesting because, you know, I always read stories and it's hard to separate and, you know, what's real and what's just grabbing headlines. But, you know, I've, I've read a lot over the last year or so that, you know, some industry experts believe that maybe every single drug in the future will start from an AI discovery. Is, is that the case? Is that where we're headed? Like, are there still going to be, you know, offline scientists and offline researchers not using AI to discover new drugs?

I mean, like, look, there's tons of hype, no doubts, right? There's tons of headlines that aren't real. There's tons of stuff being tried that isn't going to work out, like, for sure. At the same time, it's also fundamentally true, I believe, that we're going to not even say AI drug versus not. It's just going to be used everywhere, like maybe to kind of draw a parallel with the genetics. Right. There was a time when people said, like, is genetics useful in the life sciences? Will we use it in drug discovery? And now it's just a given component of every drug discovery program in different places in different ways. But there's no such thing as do drug discovery and forget that genetics exists. Right. And the same thing is, I think probably almost already today, true, for, for, for drug discovery, right. The degree will increase, but it's just going to be kind of part of the fabric of drug discovery.

You know, one thing, one person that I do trust on this is, you know, Nvidia CEO Jensen Wong said a while ago that he thought that AI driven digital biology would be the next kind of amazing revolution. And, you know, he said that it's going from a science to more of an engineering discipline. Is that where the industry is kind of headed? That is this going to be more of an engineering versus more of a, you know, traditional scientific discovery?

Yeah, I mean, I think it's going to be more data driven. Right. And I think we see that even within biology and chemistry themselves. Right. When I was doing my PhD, you could get away with quantifying using some plus signs in your paper. This is two plus signs. So it was four plus signs. Right. And now of course, you know, no one does that. It's precise analytics on any microscopy image, for example. Right. So it's certainly going to become much more data driven and structured and many more engineering and data science tools being deployed. Yeah, for sure, no doubt. I'm with Jensen.

Looking at the bigger kind of topic for today that I want to shift into a little bit. You know, how does all of this ultimately help companies, you know, help medical institutions put patients first? Right? Is it, is it just, you know, oh, we're going to have so much data or we'll have better drugs. But like, how do patients ultimately win from the type of work that you're doing?

Yeah, totally. I mean, one is timelines, right? Moving faster. People are waiting. People are dying today. They don't have a treatment. Right. We just had someone come in. We have patients come in and give, give talks, both people waiting for treatments and people who, who have been matched with treatments. And there was someone in just the other week sharing this powerful story about a rare disease that he has and that if he had been diagnosed two years before, there would have been no treatment. Right. He would not have been there telling that story. So, so one parameter is just time, right. Real people waiting for real cures cost. You touched on that at the beginning. I think the medical system is complicated. Things are expensive in one lever. There are many pieces to this, but one piece is how, how crazy expensive it is to get a drug to the market. Right. So if you can bring costs down through better data driven decision making along the way, that's a huge component. And then ultimately, you know, one day we hope that we can make we in the industry as a whole better medicines. Right. Where today there's lots of studies showing, for example, that many medicines do things in the body addition to the ones we know about. Like the better we understand things as a systems level hard to grasp within the human mind, but can be modeled with machine learning models, with knowledge graphs, et cetera. We hope in the future in a way where more of these unintended consequences of medicines can, can be factored out. But yeah, long journey and I think a meaningful one.

All right, you said, you said one of my favorite keywords there, models. Let's talk about that. How, you know, because large language models aren't necessarily new, but obviously they're growing, they're exploding. You know, all of a sudden they're being used in every domain imaginable. But Even within your industry, how have large language models both changed from you know, how you were maybe previously using traditional machine learning in artificial intelligence and you know, where do you see right now, you know, the excitement with, with using these domain specific models for, you know, highly specific use cases?

Yeah, totally, totally. And I'll give you two what I think cool examples with the large language models. Large language models, but I put it in context to the underlying architectures of large language models. Transformers generally are, we're now used by Recursion and many other companies to solve many problems outside of of language. So in chemistry, in biology, in imaging, et cetera. So there's been a huge unlock which I think maybe you don't see kind of in the everyday conversations about AI that the same watershed moment that we all appreciate with Chat, GTP and kind of other equivalent tools, it's not just in language. We're seeing similar transformations in all of these other technical fields. So super cool. And then the second piece there, along with Transformers, there's all just richness of other new models. It used to be kind of one AI model to rule them all maybe half a decade ago and now, you know, you have your pick depending on what it is that you want to work with. And we have some proprietary ones in how to recursion. Many other people do too, and many open source ones with that. You want still to LLM examples?

Yes. Let's do it.

Let's do it. Let's do it. One is just reduce toil. And so this is just this like it's not just about the future, it's not just about hope that the drug is better later on. We're saving our scientists material time on things like literature review. So you know, we just put out a statement a little bit ago on. We've reduced the time that our scientists need to review literature by 60% before we go into expensive, more time consuming, it's called hit to lead activities. It's kind of a standard step in drug discovery that's really key and core. And so that's a big deal, right? That's time that they're spending thinking about the details of the science, not searching for the right paper, not kind of doing arbitrage on what it is you're focused on versus not right. So like just a huge win in putting human minds on kind of the most thorny problems, not on kind of the simpler tasks. The second example is using LLMs to compare what we know internally with our huge data sets and experiments and using the LLMs to compare that with what the world knows doing that then also a massive, massive scale imagining that we can do that. We do over trillions of trillions of different kind of relationships and in combinations how genes interact with genes, how potential drugs interact with genes, how genes relate to diseases, et cetera. And so this is just these human huge combinatorial space that you couldn't go after manually just wouldn't be feasible. It'd be you, you know, you'll get some.0001 percentage with a lot of effort. And this allows us then to make sure that we as a company focus not on what the LLM knows from the outside world, but on what recursion uniquely knows. Right. We don't want to be working on the exact same things as every other pharma company because we want to make sure that patients have shots on goal for treatments in areas that currently aren't being worked on, on using angles and ideas that haven't already been tied tried 10 times. Right. And so that's a cool version of LLMs, I think that focusing not on what is known, but uncovering what's unknown.

And speaking of, you know, uncovering new things, you know, I'm not going to ask you to, you know, look into your crystal ball but you know, you've already kind of walked us through very briefly how these models have changed. Right. It was just first this, you know, one giant model and everyone was enamored. And then you figure out, okay, it's actually not that great for what we need it for. And then you just said hey, right now, you know, some of your researchers are, are, you know, saving 60% of time to, to review literature. What's that next big phase or next iteration, you know, that maybe you're excited about or maybe your colleagues are exciting about or you know, the industry. What's that big next jump for generative AI?

Yeah, I mean a big thing that, that we really have our eye on for our models at recursion is, you know, for the last decades we have been generating huge data sets fit for purpose of machine learning to then build the best models we can, industry leading models. We're starting to see a shift where we think on the horizon is a shift on that world of data first, then models to having so called world models, models good enough to be your starting point. And then the experimental world is really focused on validating with depth the most important insights from the models. So models first, then data.

Yeah. And let's talk about the concept of world models because we haven't talked about It a lot on the show, which is crazy after like 500 episodes. You know, generally when we talk about world models, it's more for like AI video generators. Right. It's like oh, open, right? It's like oh, can open AI understand physics? Right. And can Google understand, you know, how a person moves from point A to point B? What does it mean in your field when we talk about world models? Can you explain that a little more?

Yeah, good call. Out. Not video games, it's. It's models that aren't created to, to tune to a in single individual predetermined problem, but are created to understand a large domain. And so in our case, this is about human health and disease and it's about doing it at multi scales from what proteins do. Some of you might have seen the alphafold news on the news a few years ago, being able to predict protein structures up to what cells do into data sets from patients so that insights can be linked between these different domains and reasoned about how do not just one protein, but networks of proteins and so on.

So you've, you've given us, you know, quite a few good examples and illustrations on, you know, what you all are doing and how you're using generative AI to, you know, put patients first. Right. And just, you know, the ability that now, you know, certain patients have drugs that they can use. Right. And they've been cleared and how generative AI can help expedite the drug discovery and drug trial process. But what about if we take the word patience out of it? And how can this help in the future? Maybe humans not become patients per se. Right. So on the predictive end, how can this help maybe, you know, people not get diseases or people not get sick. Is that another area that you're looking into? And if so, how does AI help with that?

Yeah, totally. Totally. Yeah. I mean, we believe by understanding like the different, you can imagine that this same model concept, world model concept could be used, can be used. Right. To model any state of, for example, a human cell. Right. From all the way to deceased, all the way to healthy. And we already do this in a certain degree at the company. You can imagine also identifying molecules or treatments or behaviors that stop the progression into a disease state. Right. In fact, we've had some exploratory work, work in senescence, in the idea of stopping, unwanting aging pathways on a cellular level. Right. And so conceptually I think that is totally something that these approaches are going to be used for, whether at recursion or at other companies.

Yeah. And you know, speaking of aging, I think it was the. Which I was shocked when I read this headline kind of recently, actually. The anthropic CEO said he could see with advancements in AI, the human lifespan doubling, I think he said, in like as little as 10 years. Which I'm like, okay, that sounds nuts, right? I mean, is that a potential future? Right. Like, I think these, these models, you know, as they become more powerful, as you have world models have you, like, as you have the equivalent of, you know, the world's best model, but for protein, you know, you know, building and duplicating proteins and all these things. Is that, like, a realistic thing to say in the future? Lifespans might be way more than they are now.

I mean, I am not of a biologist to know about what kind of the inherent limitation of human aging are. Right. But, but kind of more fundamentally about, like, the potential on the AI side. Right. I think often about this, there's this famous Andrew in quote, he's kind of one of the big names in AI, right? And he, he around 2016 or so had this quote everyone was using. You know, that if a normal person can do it in less than a second, then you can automate it with AI. And lots of people were like, no way. That can't be done. You're overstating it. Right? And now you look back and you say, well, that feels like really undershooting it. Right. And kind of the famous Andrew Inc. Quote from last year is, AI is the new electricity. Right. So just to say, like, I think we're in this, like, big inflection phase where it's kind of hard to know where the big directions is going to go. Right?

Yeah. Yeah. So, you know, we've, we, we've covered a lot in today's conversation. This has been a fun one. I, I just decided to take it in some, some weird places. But, you know, when I think about what you all are doing and, and, you know, the future of what new drugs mean and, you know, like, aging and all these things, I just, you know, start to think outside of the box. But, you know, maybe as we wrap up today's show, you know, what is the one most important takeaway that you want people to know and understand about, you know, the type of work that you're doing and how that, you know, in that intersection of generative AI and drug discovery, how can that ultimately put patients first?

Yeah, I mean, I mean, maybe like, zooming out on that. I think there's a lot of conversations about the risks, the regulatory aspects of AI. And I think for good reason, right? I think those are conversations we should be having as a society. And at the same time, I hope that maybe these conversations get a few people really interested. Also, the flip side of this, right, really powerful AI technologies are coming. So how do we incentivize people? How do we align on really leveraging those to the best of our abilities against some of humanity's biggest problems? And I would argue that human health is pretty high up there.

Yeah, I don't, I don't think anyone would argue with you on that. And I don't think anyone would argue that this has been an extremely valuable conversation. So, Lena, thank you so much for joining the Everyday AI Show. We really appreciate your time.

Thanks for having me, Jordan. All right.

And as a reminder, y', all, we covered a lot. That was a fun one, but there's a lot more to learn. We're going to be breaking down today's episode in a lot more detail. So if you missed something or if you want to know more about something that Lino talked about, don't worry. We're going to be recapping it in our newsletter. So if this was helpful, please let someone know about it. If you're listening on LinkedIn, go ahead and repost this to someone who needs to know. Also, please go to your everydayai.com Sign up for that free daily newsletter where we're going to be recapping this episode and a whole lot more. Thank you for tuning in. Hope to see you back tomorrow and every day for more Everyday AI. Thanks, y'.

All.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going for a little more AI magic. Visit your everydayai.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.