AI will do a lot of good things. But it won't cure cancer. Here are a few reasons why
1. AI is trained on the best available published literature. If you have AI read all of the papers in cancer, you have just had AI read 50% or more of literature that cannot be reproduced or may even be fraudulent. If you have AI restrict itself to reading nature science cell JBO, AI has still trained on 50% of material that will not replicate, and is not true. How can AI solve the problem when you pump it full of false information?
2. There is no way to separate the false information from the true information unless you replicate the experiments. That requires beakers pipettes dishes and hands. Now Maybe if AI has a robot form and starts replicating things now you're talking.
3. The limit with cancer cures is is not that we are not reading and synthesizing all the available materials. There are a lot of very smart people working in this space, and they do read widely and think about always all possible ways to combine different strategies. Instead, the limit is biology itself. Understanding what is happening in the cell that has not been seen or detected in current experiments. AI has no way to do this unless it again picks up a pipette.
4. Some of the things people celebrate today to like AI can find more lesions on mammograms. It just reveals the ignorance of the technology people. What is a lesion on a mammogram? It's something that when you stick a needle in it according to a pathologist it has basement invasion. So it looks like cancer under the microscope. this is what AI is being trained on. But it has nothing to do with whether or not the lesion you see is going to progress in your natural life, and kill you, whether or not it's going to stay dormant, whether or not it's spread already. You want to find lesions that are going to kill you. But you can cut out today. We don't know which lesions those are! AI is being trained on a faulty data set yet again. I described this at length in a recent plenary session lecture.
5. 20 years ago they said Judah folkman was going to cure cancer. Then they said immunotherapy was going to cure all cancer. I guess the difference between folkman and immunotherapy is at least immunotherapy has cured some people but folkman cured about zero people. 10 years ago the hype was all about personalized sequencing. Now the hype is AI. There's always going to be some hype. You're better off not smoking, keeping your weight down, exercising and then just accepting that immortality is not going to be happening.
Side note: it's probably a good thing that people are not immortal. People need to die so that their terrible ideas can be replaced with slightly less terrible ideas. Certainly there are backslides, but such is the nature of progress. Reducing untimely death is a much better goal than making people live to 120.
sam altman: “we will see diseases get cured at an unprecedented rate … what this will do to cure the diseases at a rapid rapid rate”
immortality is not too far ahead. we live in the most incredible of times