Pfizer CEO Albert Bourla on AI and the future of medicine

I think the COVID period was very different because it represented an existential threat and zero opportunity. There was nothing good that you could expect out of COVID. The only good thing that you could expect was to find a solution and put it behind us. And I think we did well, and the world did well because there was tremendous collaboration. And we did put it behind us. Now with AI it's a very different set of animal here. There are threats. But there are tremendous opportunities that are ahead of us. And do I think that this is really something special or just another moment of technological innovation? I think it's really special. I think it's opening a completely new era that can only be compared to the wheel and the fire and the first engine, things that really changed society and the way that was operating. And that's what we are experiencing right now.

I think what COVID was, that it was disruptive. Because every major big technology is disruptive and changes the status quo. That's why we had all these events. There is no doubt that that can happen also now.

I think that AI, as I said, will be extremely disruptive, which means that it will create new realities in the world, which means that it will create new winners and new losers. And the winners will be those that get it right how to be able to deal with it. The losers will be those that either get it wrong or they are complacent. So, staying on the sidelines, it's mathematically certain that will bring you in a disadvantage. And what we do is something noble. We are making breakthroughs that change patients' lives. It's not that we are making cars that make life more pleasant or cell phones. We are saving lives of human beings. And for us, it's of paramount importance. I see it that winning means winning for patients. Now, what is the successful recipe to do that? Because as I said, some will get it right, some will — I don't think we know. But it's high time to make choices. So, even in the midst of uncertainty, we are making choices about how to go about it. For AI, I think we can distinguish two different areas. One is that AI cannot do today everything, but can do way more than what we are using it for. Way more. And what is the limitation? The bottleneck is not technological limitations. Technology is there and can do much more. The bottleneck is the ability of organizations like ours to transform themselves into an AI-native organization. To transform themselves into an organization that is completely different in the way that it structures and operates, because now we have the tools that we can do things much better.

Everybody is adopting AI. Very few have transformed, if any, so far because of AI. But that will be the secret to be able to transform yourself.

I think it resembles other technological advancements, but in a scale and speed that we haven't seen before. I couldn't imagine two years ago. I didn't know that we will reach a level like that. So I think it's, as I said before, it's something different. It's not just one more important moment in the way that technology is advancing in the world. It's really a transformational moment.

I think people are getting excited, but also are getting very worried about it, and rightly so, because as I said, the disruption will change everything, and people are concerned about it. And I think when I first heard Dario going out and speaking about it, clearly I got concerned. It particularly forced me to go and understand what really happened with the OpenAI incident and what the agents did. And I couldn't believe it when I realized what these agents did. And clearly there is significant risk because they can act on their own, they can act independently apparently from humans and those that program them, and they have a lot of power to do things. So, yeah, clearly that's something that is concerning. Still, I think that the risk-benefit ratio of this technology is very advantageous towards the benefit. There are risks, but they are very big. But the benefits are gigantic. I will do a distinction with nuclear power because many people are saying, 'Look, nuclear is the same thing. Imagine if when nuclear was there, we would let it just run without regulation. So that's equally powerful, should do the same.' I would say that with nuclear power, the risk-benefit was not that favorable. At best, nuclear could help us have cheaper energy. At worst, can kill us all. With AI, it's very different. At worst, there are risks that can happen, but the benefit, if we get it right, will be new medicines, cancer medicines, will be tremendous education — could be tremendous, the benefit. So why am I saying that? Because you can take a conservative stance and say, 'Let's put a lot of regulations to minimize the risks.' You can take a very aggressive stance that says, 'Let's deregulate everything so we can maximize the speed and the benefits that we will see.' And as with everything in life, the balance is somewhere in between. And I think that's what we need to do.

There are a lot of things that I've already seen, and I can speak about them. They are not maybe — I mean, depends how you see it. They are not as impressive, you know, 'we discovered a cancer cure,' but they are impressive as to how we liberated that many hours of scientists from bureaucracy to the bench and to the studies. But I'll tell you how I see that eventually happening, right? Of course, more medicines faster and better. Right now, the development of a medicine takes on average 10 years. And from that, four or five years is what we call preclinical. So this is — you don't test anything on humans. It is all in the bench, in laboratory animals, on the petri dishes, all of this phase. Obviously, this is the less regulated part of our work because once you go to humans, the regulations are very, very strict as to what you can and what you cannot do. The four to five years, I can see a path that will become one. Right? Because the pieces on humans, there are limitations that are, let's say, connected with a disease. Or let's say something which is you need to increase the survival of a cancer patient. That means that the study needs to take three years because that's how, when you have statistically significant numbers of people that survive, you can condense that, right? Actually, if anything, you can prolong it because you make the three years four and five. But the limitations, they do not exist if it is how can I design the molecule, how can I test it in animals, how can I do the pharmacokinetics, the toxicology. If I can do those things in silico, if I can develop the models that I don't have to, I can really speed up things. And then still test them very well before I go to humans because the last test will be on animals, but instead of doing for five years, it will be for a few months. That's a very tangible example. Is it going to make it less costly? No. Because in our drug discovery and development, the most expensive part — when I say the most expensive, 80% of the cost — is the last three or four years when you go to the big phase three studies in humans. This part of preclinical and animals costs a lot of money, but in the overall scheme of things is a small fraction. However, time is extremely important in that phase. And in the way that we speak in Pfizer, we say it's a very — that we introduced it during COVID — time is life. That's what we used to say to all our people when they were working on the vaccine, we were saying time is life. Every year that you're going to save makes the difference of a father or mother that will be able to see the graduation of the daughter or the wedding of the son or not. So it's extremely important, right? This one year. So we can save a tremendous amount of time in the drug development and discovery.