Sebastian Guth, Chief Operating Officer Global Pharmaceuticals and President, Bayer U.S.
Terra Terwilliger, Director of Strategic Initiatives, Google DeepMind
Moderator: Kristin Stoller, Editorial Director, Fortune

COO Summit 2024: AI And The COO - Leading The Transformation To A Futuristic Workplace

Conferences

Sebastian Guth, Chief Operating Officer Global Pharmaceuticals and President, Bayer U.S., Terra Terwilliger, Director of Strategic Initiatives, Google DeepMind, Kristin Stoller, Editorial Director, Fortune

COO Summit 2024: AI And The Coo - Leading The Transformation To A Futuristic Workplace

Sebastian Guth, Chief Operating Officer Global Pharmaceuticals and President, Bayer U.S., Terra Terwilliger, Director of Strategic Initiatives, Google DeepMind

I've already spoken with a lot of you over breakfast. And I think the number one topic that came up was A I, which is why I'm very excited to be here with both of our panelists. But the number one thing I heard was people asking me how, what are the specific use cases? I need a case study for how other people use it so I can bring it back to my company. So Tara Sebastian, I want start by asking how are you each using A I both internally and externally and special? Let's start with you. Sure, you know, look for us A I would deploy very widely across the organization and we really deploy it in three different buckets. One is driving workflow efficiencies as probably many others in this room. And there's a couple of examples that come to my mind. Um um One I would highlight uh is uh medical coding. Uh So in our industry, when we run very large state scale clinical trials, uh we have to translate medical events that are often described in plain language into code. Um uh so that we can ultimately analyze the data. Uh and that's something we do quite extensively through the use of A I we've estimated in a, in a recent uh clinical study that, that has saved us about 100 and 70,000 hours of work. Um So it's very significant uh and, and drives efficiencies. Another example in that bucket of workflow efficiencies is regulatory submissions uh were mandated by the FDA and many other regulatory bodies around the world to submit very extensive documentation. Uh as we seek the approval of clinical um of products of medicines. Uh And we use generative A I to populate about 70 to 80% of those dossiers today. And if you translate that into efficiencies and savings, uh and ultimately acceleration is significant. The second bucket in which we use A I um is to advance patient care. Um So, um an example in that space is uh is mammograms. Uh Some of you may know that it's breast cancer Awareness Month. Uh And some of the audience may have gotten or may regularly get mammograms. Uh But what many individual people don't realize is is that it's in fact, not one image but hundreds of images that are being taken. Uh And we um provide software to radiologists that helps select the best image uh so that they can uh appropriately diagnose uh the patient and, and, and uh mammograms and breast cancer is one area we do this in other spaces as well. What's most exciting to me personally. As I think about A I though is the use of A I and technology in drug discovery. Uh because at the end, you know, we're in the business of bringing medicines to patients as quickly as we can. And I'd say that's a pretty challenging task. Um We spend about a billion and a half to 2 billion on each medicine that we bring to market. It takes us 12 to 15 years and we fail most of the time. Uh uh And, and, and technology can help us significantly accelerate uh that work. Um An example in that space uh for us is our partnership with a company called Recursion, uh A company in which we are an investor uh and a company that has arguably and likely the most advanced A I guided drug discovery platform in our industry. Um uh And we partner with them in them, bringing technology so deep expertise in technology, data sciences uh and, and use our uh library uh of uh of uh compounds uh to essentially identify and develop novel cancer therapeutics. And, and that just gives me chills because, hey, that's, that's the use of A I to drive real impact and ultimately develop medicines that otherwise would not see the light of day. Absolutely. And it touches everyone here too. Just like Google deep mind. Tell us what you're doing over there. I think probably everyone in this room is, is using the tech that you developed. So tell us about what you're doing. Thank you. So, Google deepmind is Google's Artificial Intelligence Unit. We create Gemini the foundational launch language model for Google as well as a host of other technologies uh which are while not in the lang large language model category are also being used for, for example, fundamental scientific discovery just as you described Sebastian. So we two are using internal and external uh the model in different ways internally. Uh legal is an example of where we have found great use for Gemini uh in our legal system, you can think of legal problems as having three different categories. There's retrieval of information, there's summarization and then there is reasoning and we are finding Gemini is valuable and across all of those categories, retrieval of information, of course, is finding the information that you know, you have somewhere, but you can't put your hand on it right there right at that moment. Uh summarization. And that's, that's really the basic I would say use case of simply retreating and finding that information. But it saves a remarkable amount of time. We're estimating that 40 to 50% of the request to our legal team could be handled simply by that kind of automation on summarization. Again, you're retrieving the information, but you're also pulling out what is most relevant. This is a more difficult problem particularly in the case of a legal context because you don't want to miss an item that might be important. So our team is increasingly using generative A I to to review for example, scientific papers and say what does need legal review in this case and what does not. Uh and then third reasoning as probably the most exciting category, legal problems as you know, are a good test case for gen A I because they are often unstructured, ambiguous and can have many different answers depending on jurisdiction, depending on business choices. So reasoning we think is the the next frontier here and the initial results are promising would not be complete without also adding to your point, Sebastian about scientific discovery. And this is an area. Our our mission at Google Deep Mind is to build A I responsibly responsibly to benefit humanity. And we are particularly proud of our work in scientific fields including alpha fold, which is the transformational protein folding technology. We created it. We won the international challenge of protein folding with the with this technology and in fact, released all of the 220 million plus proteins in the database for free for researchers to use in their own applications including drug discovery. Excellent. And I'm gonna come to the audience for questions in just a second. But I I have one for both of you. There's a lot of fear mongering when it comes to A I uh A I is gonna steal my job um both internally and externally. I feel like people are very, very scared to, to use it whether it's, it's bias. And again, people being scared about how it's gonna affect them. Um What would each of you say are the, is the biggest pitfall that you would advise C OS against when implementing A I into their business strategies to start with you? Sure. So first in our industry, I actually don't think that A I is going to steal jobs uh because hey, unpacking 3 billion years of evolution compressed in a cell is pretty damn difficult and that will continue to require also uh human capacity. Uh And, and the art of science uh which we're marrying up with technology. Uh big pitfalls, I mean, in our industry, we, we cannot and will not compromise patient safety. Uh So we're building a lot of redundancy into uh the way we work. I gave you the example of medical coding earlier. Uh And in that example, we use a four eye principle. So we in fact use technology, but then we still have humans double check that whatever data entry um we put into the system is accurate. Uh And that's only appropriate given that, that uh patient safety cannot be compromised. Um One piece of advice I would have from my own personal experience is that in large organizations and maybe that also resonates with some of the audience, there's at times a risk to use technology for technology's sake, you know, and in my mind, uh at the end of the day. Um It's a means to an end and the end in our case is to bring new medicines to the world at a much faster pace than otherwise than we would have otherwise been able to. And it's not about chasing shiny toys for the, for just for the sake of chasing them. What advice do you have? Yes, I would echo that in that A I is an unparalleled tool for fundamental scientific discovery. So how can we look beyond the immediate applications and think about how much there is yet to be discovered to solve fundamental human problems. Now that said, certainly we want to be mindful of how we're asking our people to use artificial intelligence internally because of those fears you just discussed to me the most important principle. There is co creation. So how can you with your team core? How the A I will be truly helpful to them in doing the job that they are already doing? They are the experts in doing that role. And I believe there's plenty of low hanging fruit. There is plenty of paperwork to be automated so that high skilled people can go do more high value tasks. Uh But that co creation piece is key. I'd say also making sure that, that you're valuing the time that people are putting in to learn how to use these systems and rewarding them for doing so in tech, we have a wonderful uh culture of dog fooding and using our own products internally. Uh And, and I would simply say that it's important to value the time people invest in that. And then third think forward about what this means for people's careers, both in reality and in perception, what new training might people need? How might this contribute to their training and learning in a way that's intrinsically valuable? I want to come back to training in a second. But does anyone in the audience have any questions here and, and raise your hand and we can run, run a mic around see one over there? Yeah, just in general. Um you talk about the growth of A I and the data, the need for data centers all over the place. What do you guys think about like power for the future and your needs power for the data centers. Uh Power for the data centers. Uh It's an important question. It's a question on many people's minds these days. Uh Again, we have to think about the fundamental scientific advances that are going to be possible with this technology. Uh We have, we actually at Google Deep mind work on novel power applications as well. So we have uh a piece on optimal power flow on how to increase efficiency in data centers to reduce the amount of power that is being consumed. We also uh we do work on fusion, nuclear fusion and how to optimize reactors. Uh And the, the fusion reaction inside of reactors. Again to get us closer to that place of renewable energy. Excellent. Thank you. Chair. Any other questions? He went over here and over here. Hi. Hi. So I work at an autonomous vehicle company, GEN A I based autonomous vehicle company and I found it really interesting with the release of chat GP T and the popularization of, of gen A I, there's been kind of an increased fear of, of A I and the potential impact it could have in society. I was wondering if you guys saw something similar and what that shift looked like between the A I conversation and gen A I conversation. Maybe I get us started, you know, in our industry. Um There's very few individuals that are ready to put their health into the, into the hands of machines. Uh and rightfully so, uh in my mind, uh so we've had a lot of conversations uh over the years more so in recent times to your point because it's suddenly in the public domain on, on, on how we use uh A I responsibly and how we ultimately a commend the work physicians and clinicians do, but don't necessarily aim to replace it. And that's a philosophy that keeps driving us because we continue to believe that uh that humans will for, for time to come play a very important role uh when it comes to advancing the health of those that we serve. And I would doubt. I think all of you play a role in that too. And how you deploy these technologies in the world in your workplace? Makes sense. I saw one over here. Thank you. Hi, thank you for the reminder on purpose driven A I, I think that was beautiful. So, thank you so much. So, my question to both of you is that what is the importance in your mind of learning about A I from other industries? And if that is important, how do you set your organizations up for being open to receiving ideas from other industries around us? Thank you. I believe this comes back to co creation again. So we want to build artificial intelligence with people, not just for people. So that intelligence comes into us from a variety of different routes and customers are a wonderful source of information. Uh We also work with outside organizations across a variety of uh represent constituencies, industries to try and understand what, what will create value for people. And I'd say at GDM, we are especially focused on the scientific community because again, we want to understand what will be of most value to scientists. How can we help them create the next fundamental breakthrough that will transform our futures and maybe just as a quick build, you know, in our industry, we learn quite extensively from others. Uh The promise of A I in the pharmaceuticals industry is massive, but we are for very good reasons. As an industry, probably progressing somewhat slower than others. Um You know, I, I sometimes described as jokingly as the race of the turtles. Um but that's a very conscious choice. Uh because I said before, you know, we're not going to compromise patient safety and the work we do has such a big impact on those that we serve, that we cannot take shortcuts and we probably, and possibly can't sort of experiment uh to the same extent other industries can. But that also gives us an opportunity which is to learn. Well, thank you for the incredible questions. Thank you, Sebastian and Tara for being here. I appreciate it. Thank you. Thank you.