No.

>26 years after Covid, brainlets still don't realize that medicine isn't a theoretical field
You can't cure shit without running IRL tests for years

>>109775094
>xitter screenshot
>credible source
kys

Yes

After all this wonderful ai slopped code and ai slopped "art" I just can't wait for ai slopped medicine!

>>109775094
This is the same guy that predicted Astra release date btw

You can't solve Alzheimer's by just multiplying matrices in a computer. Retard.

>mythos is so dangerous
>cancer is cured
What's next, Dario? Simulating your mom?

>>109775094
Another aiGOD victory

>>109775196
Simulated biology, faggot. Literally everything reduces to math.

>>109775094
>Thoughts?
Better.

prompt: cure alzheimers, no mistakes, do the needful, make professional medical paper with full explain

I love this down syndrome understanding of biology. We just solved <disease> like a math problem. It wasn't even a defined problem that needed pattern recognition and extrapolation like protein folding, we just skipped decades of research.

File: Screenshot 2026-01-19 170337.png (297 KB, 344x453)

297 KB PNG

>>109775094
That's not how it works. You have to test it to know. There are multiple compounds actively being tested for many incurable diseases all the time. It just made some plausible sounding shit at best.
>out for expert review
So it didn't do shit.
Can these bozos ipo already so we won't have to suffer through any more of this bullshit?

>>109775213
So you think Anthropic has solved quantum computers sufficiently in order to have enough stable qubits to simulate biology?
That would be a significantly stronger achievement than curing Alzheimer's.

>>109775094
It's possible. AI is smarter than all of yall

CURE ASPERGER'S NEXT FAGGOTS

>>109775094
Bidenbros are we back!?!?!?

>>109775123
It's not that slow but yes.

People severely underestimate how hard biology is to "solve" as a full system, especially for complex multicomponent diseases like Alzheimers. An AI that could oneshot safe and effective drugs for any given disease from thin air is in strong superintelligence territory

>>109775094
They say a lot of things. Very few of them have turned out to be true.

>>109775213
There isn't any even remotely accurate way to simulate even basic biology. The most advanced (quantum-based) molecular dynamics algorithms give total garbage unless you can find an accurate starting point, and there is no way to do that at the moment. No alphafold didn't change shit there. It's better than blind docking with random homology, but far below usability thresholds. Adaptyv bio ran several public benchmarks. They heavily lied about most basic facts, but they provided the actual data so everyone could see the real conclusion: on an extremely easy target that has about the single most data available out there, there exists 0 computational tools that has a correlation more than 0 with affinity, and alphafold doesn't help there either. Furthermore, binding design methods are garbage today (keep in mind the submissions were largely pre-verified in the lab by the teams that submitted, which were mostly large well-funded and well-established bio academics).
The same applies to all aspects of biology.
We ain't got shit. That's exactly why bio research is so fucking slow. But the problem is that comp bio is not seen as valuable, so they're treated as p-hacking monkeys instead of people who can solve the most important problems in biotechs today.

Why would they want to fix something they're absolutely making money on?

At what point does the SEC start prosecuting these companies for lying to get investment money?

File: 1767286969877663.png (661 KB, 765x693)

661 KB PNG

>>109775123
ENTER

>>109775330

The way I understand it, is that it's like a massive, flat, asynchronous system, where every little change has cascading side effects with other dependencies. It isn't like this linear algorithm you can just tweak, everything has a network effect.

>>109775330
Only good post itt
>>109775416
Basically true, except it's 100x more complicated than that
Biological systems are very very tightly controlled feedback loops and it's a wonder we even have any effective drugs

>>109775094
when is it going to cure balding?

>>109775094
SOURCES


INDICATE

>>109775444

Well it's no mystery as to why so many drugs are just trial and error. So many times it's like this

>random guy is testing out X compound to potentially fix or treat Y
>it actually fixes Z
>as far as he know the side effects are tame enough for him to ship it as a useful drug

And we go on without knowing what else is being impacted, as long as it doesn't kill us immediately

>>109775415
enter what? he hasn't tested if it works

>>109775094
If that was true, it would be wonderful.
It won't be. But if would be wonderful.

>>109775472
Forward pharmacokinetic is 35 years deprecated.

>>109775330
It's wrong to think of the frontier internal models as individual superintelligences. They're more like very large teams. The one that cracked Navier-Stokes in 88 hours had 10,000 agents all well above Astra's capabilities, each working about one subjective year on the problem (token wise). God knows what the swarms are cooking right now. Nothing would surprise me.

>>109775415
> treating schizophrenia
said schizo in his garage

>>109775456
This

>>109775416
To an extent, some conditions or targets are far more complex than others and targeting even the well characterised systems does not necessarily mean your drug modality of choice (if it's not a biosimilar or close analogue to a previous drug) won't fail at some point on the long path to getting inside human patients. I don't see a way forward for full AI drug discovery without investing a huge amount in fundamental multiomic data collection (A useful exercise anyway and most likely way forward, we'd still miss a lot) or straight up superintelligence that can simultaneously see through all the cruft of questionable biology papers and make connections and inferences that humans would never see. We would probably have bigger problems at that point.

>>109775531
I'm not an LLM sceptic, but current frontier models are not anywhere near superintelligence or even close to encompassing everything that humans can do right now. They have their own strengths and weaknesses. The issue is more that no amount of time or number of current level AI agents working on Alzheimers, or other tricky biology problem, is going to figure out a cure by pure reasoning because we simply don't know enough about the fundamental biology to derive an answer like that from our current body of knowledge. It is like asking them to stare at a completely opaque black box filled with hundreds of thousands of singular unique molecules, of which we know the structure and location of ~some, and guessing the exact position of each one inside that box. Perhaps superhuman reasoning could get it through means inscrutable to humans, or they get it by pure chance, but this is not a "thinking very hard for a long period of time to solve it" type of problem. Experiments MUST be performed.

>>109775730
>Experiments MUST be performed.

The Covid vaccine (mRNA-1273) was found in four days via (by current standards quite primitive) AI simulations in 2020. The following months were spent testing the vaccine that everyone at Moderna knew with high accuracy would work. People with Alzheimer's, cancer, etc, will be fully cured before experiments can confirm what sufficiently powerful simulations will trivially predict. But who cares? They'll be cured.

>>109775796
There is no change at all in the tech stack used today or in 2020 or even in 2016 for these things.
When they say "found in 4 days" they mean they found 1m candidates in 4 days then ran a fuckton of experiments in months and then retained 1 in the end, which came out of that campaign.
Nobody had an idea if any of them would work. No simulation was predictive at all.

>>109775094
>muh cured alzheimer
>no clinical trials
good one gargamel

>>109775094
when is claude going to cure its own alzheimer that happens when he runs out of context or is way too high on it and just starts forgetting things tho?

>>109775796
>found in four days
mRNA technology had been in development for several decades by that point

>>109775830
And they were already working on SARS-CoV before SARS-CoV2 came around, so they probably already had a lot of templates to work with. Even in the 90s if you had that kind of data you could "design a mRNA vaccines in days".
There's a reason they have been on quite the different timeline when it has come to their other projects so far.

how are you curing holes in your brain, exactly

>>109775853
There are various delivery vehicles that pass the BBB. You can also just poke yourself straight up. You can use regenerative medicine to regenerate neuronal structure.
But anyway, alzheimer's isn't "holes in the brain".

>>109775330
>>109775416
This. It is impossible to predict what a compound will do in a human body, let alone accounting for external factors such as diet, age, gender, medications, dna, pre-existing illnesses, etc. It's the equivalent of asking you to predict where someone will be *exactly* in 1 year after landing in New York. It's impossible.

One of the only things that I've been looking forward to in growing old was Alzheimers and these fucking AI companies want to take that away from me and cure me.
I don't want to remember, don't you get it. It wasn't a utopia, it was shit as a kid, there was a good spot up until WotG released and then the world went to shit again and continues down.

I don't want a cure for Alzheimers, stop this madness. I want a cure for all the other societal ills and woes, then you can start altering my body so I am sober and experiencing the horrors (wonders) of life at 74 like old aged sex orgies.

>>109775094
>Company producing hallucinating hazard claims it can cure Alzheimer's
LMAO

>>109775873
If you want to experience Alzheimers just overdose on dxm

>>109775796
There's a few things here. mRNA vaccines had a lot of work put into them before COVID19 came around, they did not suddenly spring out of existence. We also had an advantage in we've had previous coronavirus outbreaks that also had some work put into them so we had a good idea of how to target coronaviruses (See for example https://pmc.ncbi.nlm.nih.gov/articles/PMC3371787/ talking about the immune response and potential antibody targets from the SARS outbreak, although the COVID19 vaccines that I know of used the spike protein as a target in the end)
Vaccines are an exception to the general rule of poor success rates around drug discovery, because we're utilising a very strong and general system in the immune response, although most people have forgotten about the failed COVID vaccine attempts and it's still not trivial. We also had a lot of different methods attempted from different companies, but mRNA vaccines can be developed comparatively quickly.
If LLMs can provide a cure to Alzheimers and all these other tricky problems, I would be very happy, it would be an unimaginably positive thing for humanity. But these are not problems we have been bashing our heads against for decades for no reason.

>>109775198
I would simulate my pen Dario

>>109775094
This is the last we'll hear of it. The medical industry is a business and you don't make any money "curing" people. You treat the symptoms indefinitely or until the person dies.

>>109775975
>pen
Meant "own" but this is what I get for phonefagging