This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.drvinayprasad.com/subscribe
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
SPEAKER_00Dario Amade says that cancer is going to be cured in less than 10 years. Is that realistic? Is it plausible? We'll get to the bottom of it in today's Medicine Unpacked. Well, Dario Amade is back in the news. He's got a bold new prediction that just as AI has made unprecedented advances in mathematics, moving from being unable to do even high school mathematics just a couple of years ago, to already solving Fields Medal level science in math, That AI will do the same thing for cancer biology and it will be less than a decade
SPEAKER_00before we have unprecedented advances in things like cancer. He's even referenced Alzheimer's. But for this video, we're going to talk about cancer. Is this plausible? Is this realistic? Is this science fiction? So, you know, I'm willing to concede for the sake of argument that AI is already doing what human beings do. When you get to trillions and trillions of complicated calculations and predicting your train of thought, it's very likely that human brains and AI's underlying algorithms are actually achieving the exact same thing.
SPEAKER_00Let's just say that for sake of argument. But even if you believe that's true, there are a number of reasons to doubt that cancer is as amenable to AI-guided advances as mathematics. When it comes to mathematics, all of the facts and axioms that are well accepted in mathematics are widely known, and these are known indisputably among all mathematicians. Mathematicians apply logic, they bring some portions of mathematics to weigh in on other portions of mathematics,
SPEAKER_00and they're able to do this in a very logical and regimented way. But in cancer biology, the algorithms have been trained on things that may or may not be true. They're trained on experiments that may not be reproducible, that if you actually ran it a hundred times, you would find that the experiment that you're hanging your hat on is nothing more than chance variance or noise. They've been trained on experiments where the reagents are mislabeled or perhaps incorrect, where there's selective reporting bias among the user. So the fundamental input in cancer biology is often misleading.
SPEAKER_00Couple this with the fact that in a human body, the interactions between the cancerous part of the body and the non-cancerous part of the body may be a hundred trillion or even more than that. So how is AI going to go into the middle of this space and solve this conundrum? And I'm willing to stipulate for the sake of argument that the AI is as smart as all people's brains put together. But that's the problem. It's not smarter than experiments that don't exist. In other words, it can only know the result of all the experiments that have been done to date.
SPEAKER_00It doesn't know what would happen if you were to Insert the blank. Do this next experiment in science. I think Anthropic gets that. That's why they have a South San Francisco lab that's actually being run by Anthropic. It's the PI and it's running experiments. But I don't think they fully appreciate how little we know about cancer biology. You know, I'm reminded of Plato's allegory of the cave. In Plato's Republic he talks about the allegory of the cave where you can imagine a prisoner chained in a cave and all he sees are his shadows,
SPEAKER_00the shadows of the men around him thrown on the cave walls and he never sees anything else in his whole existence. You can imagine how confused this person will be if you bring him out of the cave. He's come to understand the world in terms of two-dimensional shadows and he suddenly realizes there's a three-dimensional component to every person. Similarly, the way we understand biology in 2026 is as primitive as that man in the cave. I know we think we can sequence the entire genome and sequence all the mRNA and tRNA. We can understand how ribosomes work and we can predict how proteins fold.
SPEAKER_00But the three-dimensional complexity of cells is far beyond just this. When you sequence DNA, you just have the shadow on the wall. You have no idea how the molecule is folding and undulating and changing in three-dimensional space, how RNA is implicating and acting upon the DNA with complex chemical reactions, how protein is acting upon that. You know someone once told me that in biology we have some things we can hang our hat on like the central dogma that DNA becomes RNA and RNA becomes protein and protein does the action. Oh really?
SPEAKER_00Is that how it works? The central dogma is no such thing. It's been contradicted many times as the complexity of DNA and RNA are much more than just merely copying a message forward for the protein to translate and to carry out. So to Dario, I guess I understand why you want to be bullish about this. Your IPO is about to debut and maybe it's going to go for $1.4 trillion or maybe it's going to go $1.2 and maybe a few months ago it had gone for $2 trillion. And so how much people believe your science can alter
SPEAKER_00the future is incredibly important for you and your friends and your business interests. At the same time, I think caution is warranted and we have learned the painful lesson that cancer biology is something that still evades our grasp. Even the greatest advances in cancer biology have only been sort of half fixes. We can improve some tumor types and not others. In the case of pembrolizumab, we improve a number of cancer types where evading T cell evasion is a mechanism of cancer.
SPEAKER_00Promulgation or Resistance and Pembro and Nevo work in that space, but you know, LAG3 didn't work as well. Some of the other T-cell evasion antigens. CTLA4, the thing Jim Allison won the Nobel Prize for didn't work as well. The complexity of just something so small as how does a cancer cell tell a T-cell, do not kill me, that's something we haven't even fully unraveled. And you mean to tell me you're going to come up with a panacea that treats all cancers? As if all cancers are even fully understood, the range in which human cancer exists.
SPEAKER_00You know, there are at least thousands or tens of thousands of different cancers and they all behave differently. And we haven't even fully understood that, but you're going to come in and within 10 years, make it all non-lethal. I find that a bit hard to believe. So while I'm bullish about AI and I'm as bullish as the next person that it's going to have tremendous advances in medicine and it's going to fundamentally disrupt how we get care and give care, I'm pessimistic that it's going to be this panacea that's going to solve cancer in the next decade. And that's why I'm not, unlike other people in San Francisco, I'm not.
SPEAKER_00Picking up smoking cigarettes because that's something that the researchers are doing. The people who really believe they're smoking cigarettes. That's an according to an article I read in the newspaper. Well, whether or not it's true or not, that's a discussed on a prior video. But I will say I would not be so confident that it will cure lung cancer in the time span you need it to. So. If you like this video, you know what to do. Like, subscribe, comment, leave a message. I'm on Instagram. I'm going to post that here. You can follow me on Instagram and I'll be back with more videos in the weeks to come.
SPEAKER_00I'm doing the VPZD show and I'm going to be doing something new coming very soon. So until next time.
Live episode from Youtube This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.drvinayprasad.com/subscribe
Elizabeth Holmes/ AI researchers start smoking/ Moderna vax by Plenary Session This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit...
Pope has cancer; Phase 1 trials for AI; Can AI hack cancer trials__1 by Plenary Session This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit...