However drugs fail because we don’t understand the biology well enough. Drug researchers develop drugs based on all their knowledge of biology, and when they fail it’s because the biology didn’t work the way they thought it did.
So if the current understanding is the cause of failure how can AI help? It can chug through vast amounts of data, but the data doesn’t explain the biology well enough.
Now what?
Yes, biology is different and experimental evidence is very different than mathmatics, but the ability to pull from a lot of known information and formulate possible problems that biologists didn't anticipate isn't really a stretch from where we are on the capability curve right now.
Look at the showcase AI proof of the Naiver-Stokes existence. It basically took an equation and proved it was valid across all possible variables.
That’s a problem in a closed system. There is no question how the function works. The tools needed to solve it are known.
Biology is like solving a math equation we only have a crude approximation for. Parts of the equation are missing and there are variables we aren’t aware of. You also have a limited set of tools to test the equation.
All AI has to do is catch problems earlier in the pipeline to outcome success rates.
You just have to think about the system wholisticly, the ways that AI could make trials more consistently positive are many. Moving the needle on any process upstream of the bottleneck benefits greatly. Currently something like 80% of trials fail, so without changing trial rate you have 5x potential sitting on the table.
Which it has shown no ability to do, so far.
you have inputs and outputs mapped for the trials and know historically how trials tend to fail
not a leap to have ai assist in proposing or reviewing trials to reduce failure rates. even a 10% reduction is massive
All AI has to do to prove n=NP is to just prove n is equal to NP
All ai has to do to make space travel possible is find a way around the laws of thermodynamics
Which is demonstrably false, since e.g. the US and Europe have different requirements and neither is a subset of the other. If we accept that they both require adequate proof of "safe and effective" that means they also both impose unnecessary requirements, since otherwise the requirements would fully overlap.
2) I didn't say there are no additional requirements. For example, you have to fill out some paperwork. I said they are trivial compared to actually making a drug that works and proving that it works. Drugs do not die because of the approval processes, they die because it's unfathomably hard to make drugs that work and then to prove that they work.
This is circular logic. It's just defining "prove" to be coterminous with what a given set of regulations require in order to claim that it's the "proving" which is the hard part and not the complying with the regulations.
For it not to be that, "prove" would need an independent objective meaning, and then we're back to non-overlapping circles.
> I said they are trivial compared to actually making a drug that works and proving that it works.
If this was the case then there wouldn't be drugs that are prohibited in the US but not other first world countries (or vice versa) specifically because they can't be patented or the patents have already expired, and as a result no one pays for the prohibitively expensive regulatory process to get them approved.
They've already been proven in the other countries -- much higher numbers of patients than you have in normal clinical trials -- but because the approval rules are excessively rigid and unreasonable, that doesn't count.
You are wrong. You absolutely can get drugs approved in the US based on real-world studies or observational data. The reason this happens so infrequently is actually because it's virtually impossible to prove a drug is safe and effective using real-world studies or observational data.
We don't rely on (super expensive) RCTs because FDA says you need RCTs. We rely on them because FDA says you need to prove your drugs are safe and effective, and no one knows a better/cheaper/faster way to do that than RCTs.
You can absolutely bundle up a bunch of observational data and file for approval. You likely won't get it, not because observational data isn't allowed, but that observational data almost certainly cannot prove the claims you're making.
This is back to the thing where you define the requirement in terms of the regulations.
If you have 10,000 people taking a drug for years and it's "unsafe" then it would be conspicuous. A third of them would be dead or their children would have a significantly higher proportion of birth defects than the overall population and the people with the same condition who took a different drug, etc.
Whereas if it has some minor side effects, it might not be able to pick that up, but that's because they're minor, so why should that be a reason to prohibit it by law?
The thing that really makes the point is how different the process is for food. Their hands are tied there, because food is full of chemistry and biology. You're cooking a plant. Heat triggers chemical reactions. If you use a different temperature for a different amount of time, you get different chemical reactions. The same thing if you store it for a different period of time under different conditions. If you grow the same plant in a different place or use a different variety. If you change the proportions of different ingredients.
It would be totally infeasible to require RCTs every time a restaurant wanted to add a new item to their menu or change their supplier of tomatoes. We don't even do that with drugs -- different brands of the same drug have "inactive ingredients" and the requirement is essentially that they themselves be on the safe list but not that anybody has done a comprehensive study to prove that they're inert and unable to affect e.g. the metabolism of the active ingredient. "Sugar is a drug" but the patient doesn't even get a choice about whether it's that or something else they have to ingest along with their prescription.
So every time you bite into a sandwich you're putting a million unknown chemical compounds into your body. We're not even talking about herbicides or preservatives here. The chicken feed was fish meal from the Pacific rather than the Atlantic and that affects what's in the chicken, which results in a different chemical reaction when you cook it in the sauce. And for the most part that doesn't matter. The food is more likely to be bad for you because it's full of sodium and high fructose corn syrup on purpose than because it contains <enigmatic organic compound> unintentionally. We have more problems when something contains e.g. heavy metals (which we already know are bad for you) than when it contains something novel.
So we're running the scenario where you don't require RCTs for everything in parallel because we have to, and it seems to be fine. What does that say about the claim that not de facto requiring RCTs is "unsafe"?