This is honestly hilarious. Dario must think people are stupid.
First, cancer research charities are some of the most well funded on this planet. They all obtain donations on the basis of "together, we will cure cancer" messaging. They all fund the best science they can.
Second, the human body is a complicated thing. You can feed your fancy LLM as many textbooks and academic papers as you like, but the reality on the hospital ward will always be different. Why do you think student doctors have to spend so many years "doing the rounds" Dario ? They are all academically smart, they are all capable of memorizing text books ... but there is no substitute for seeing and doing the reality.
I barely trust Claude to write code, let alone find a cure to cancer.
It's like saying "we'll cure infectious diseases"
Don't worry, I'm sure that's his next blog post. Right after Claude has finished solving famine and poverty. ;)
How weird that all those humans seem not to trust the logic of their betters ;P
the ultimate utopia
We have more than enough food on the planet for everyone, and we have the ability to deliver bombs or soldiers or amazon packages to anywhere in the world in days or hours. Combine excess food with amazon or army level logistics, and famine is gone.
What people mean when they say “solve famine” is can we make people not starve without actually doing or changing anything - which is of course impossible.
We do this. Have you ever heard of UN World Food Programme? $14B funding in 2022 (since then dropped drastically, guess why).
Has it solved famine? No. Would it solve famine if it got the funding it needs? Unlikely.
Because you cannot just spend billions of dollars to send amazin food parcels across the world and expect that to fix climate change, draught, poverty, colonialism, racism, illiteracy, lack of education, and propably 100 other issues all contributing to the fact that we still have not end hunger. And no, AI won’t solve it either.
So one would expect $14b/year would solve about 14/93 = 15% of world hunger.
> The World Food Programme supports over 152 million people annually and delivers life-saving food assistance where needed most. (https://en.wikipedia.org/wiki/World_Food_Programme)
An estimated 757 million people are food insecure (https://www.who.int/news/item/24-07-2024-hunger-numbers-stub...)
152/757 = the world food program helps ~20% of people who are food insecure. That's well in the ballpark of the estimated 15% spend.
If we spend that extra $93b, we'd be well within the range to actually stop world hunger.
Yes, "climate change, draught, poverty, colonialism, racism, illiteracy, lack of education" are still issues. No, you don't have to solve those problems to feed people. Don't let perfect be the enemy of good.
Almost all recent famines have been caused (either deliberately or incidentally) by government policies, not lack of money. The UN spending 93 BB may (somewhat) improve the situation, but is unlikely to end food insecurity.
Eliminating corrupt leaders (i.e. Kim Jong Un, Vladimir Putin, etc.), and poor regulation (e.g. Bombay Fodder and Grain Control Act, 1939) would be much more effective.
Actually, they were caused by lack of food.
> Eliminating corrupt leaders (i.e. Kim Jong Un, Vladimir Putin, etc.)
Yes, which of the 757 million food insecure people is Putin responsible for their food insecurity?
North Korea has about 26mil residents, so if they were all starving that'd be 3.4% of the food insecure people. Gaza has about 2mil, so we can pin a further 0.03% on shitty governments.
Obviously it'd be great if putin, un and netanyahu were gone. They suck for a variety of reasons. But it doesn't change the fact that we could spend money and feed most of the people who are going hungry
This is a tautology; a lack of food can't be caused by a lack of food, as it is a lack of food.
>"Yes, which of the 757 million food insecure people is Putin responsible for their food insecurity?"
In Putin's case, he has caused the price of food to increase through his war on Ukraine, as the cost of fertilizer makes many agricultural products uneconomical to grow for customers in less-developed nations.
>"But it doesn't change the fact that we could spend money and feed most of the people who are going hungry"
There was food literally rotting away (elsewhere in India) during the 2013 drought in Maharashtra, but it wasn't possible/worth transporting it to the affected region (due to government interference).[1] In places like Bangladesh, recent governments have used famine as a tool for obtaining foreign aid while also exterminating or weakening opposition.[2] During the 1990s North Korean famine, plenty of food was offered to the regime, but it was mostly rejected, and that which was accepted was not widely distributed. [3]
[1] https://en.wikipedia.org/wiki/2013_drought_in_Maharashtra
[2] https://asianews.network/over-a-quarter-of-bangladeshis-suff...
[3] https://en.wikipedia.org/wiki/1990s_North_Korean_famine#Fail...
Mind you, UN’s Food -programme (together with education, progress of farming technologies etc) has been a huge success and hunger has dropped historically low. I couldn’t find data on how Trump’s cuts have affected the levels, but I doubt it has been a positive change by any regard.
Going back to the original allegory, just like cancer, hunger is a vastly complex problem with strong political ties (why have still have we not illegalised cigarettes for instance?) and no gee-wiz tech gadget, even as expensive as LLM model, will solve neither of them.
[1] https://www.feedingamerica.org/hunger-in-america/mississippi
I am arguing nothing except that it’s possible if we are willing to spend the money. No AGI is required to come up with a magic solution. It literally just requires spending a lot of money.
> Look at it this way, would Trump end hunger in Mississippi (about 600k people) if, say, Luxemburg would donate USA anout $400m required to end it[1]?
Probably not. But would trump step in and prevent luxemburg from stocking mississippi’s food banks? Could luxemburg send containers of non-perishable food to community centers in every city in the state?
> Has it solved famine? No.
Hoist by your own petard
You know that saving millions of lives every year is better than NOT saving millions of lives, right? Just because something doesn't solve the problem 100% doesn't mean it's not VERY VERY good.
It is possible to prevent most viruses by recoding the human genome (requires embryo engineering or pre-embryo engineering) to use alternate codons that existing natty viruses don't use. This wouldn't prevent synthetic viruses from being developed that use the alternate codons, of course. See (1) https://diyhpl.us/wiki/transcripts/hgp-write/2016-05-10/ultr... and (2) https://diyhpl.us/wiki/transcripts/hgp-write/2017-05-09/ultr... and (3) https://x.com/RokoMijic/status/1720875112541364535 for some background.
As for "curing cancer", I suspect that Anthropic or OpenAI will use their high status/reputation to launder the existing solutions (germline, p53, enhanced DNA damage repair pathways, higher tumor-suppressor-to-oncogene copy number, more robust tumor-suppressor networks, more self-anti-cancer apoptosis, CD47 suppression, etc) that society has refused to accept. They might have enough reputation to get people to look at these ideas. It won't cure people with stage 5 turbo-cancer a.k.a the walking dead, but hopefully we won't be moving the goal posts _that_ much on them.
"Synthetic genomes unveil the effects of synonymous recoding" https://www.biorxiv.org/content/10.1101/2024.06.16.599206v1
"Probing the limits of genetic recoding using multi-omics-guided evolution" https://www.nature.com/articles/s41467-026-74300-9
"Escherichia coli with a 57-codon genetic code" https://www.science.org/doi/abs/10.1126/science.ady4368
Is it a product available to humans at all?
A more sophisticated view is that cancer (one idea) is an problem of rates (like most things in biology) and there is a tumor burden per unit time from whatever causes vs. the immune system's capacity to detect and destroy them per unit time. As humans age, the balance tips in one direction. If the balance is too far off for too long, tumors accumulate, doctors take notice, and assign whatever labels to the condition.
Looking into particular causes of tumors (and calling each cause its own cancer) can create a vast research program that employs a lot of people, since there are so many mutations that lead to tumors. But interrupting the development of a particular mutation is not going to have the same society-changing effect that solving the fundamental issue of rates would.
Literally first phrase on Cancer Wikipedia entry.... [1]
"Cancer is a group of diseases involving uncontrolled cell growth typically resulting in tumors with the potential to invade or spread to other parts of the body...."
I'm inclined to believe that the thing - that everyone intelligent says is complicated - is actually complicated. I would be surprised if the thing - that everyone intelligent says is complicated - is actually simple.
This is almost like an application of the conspiracy theory gambit where you try to find out how many people would have to be "in on it" in order to disprove the theory's likelihood. Like flat earth.
Also, not even all cancers have tumors. You have solid tumor oncologists and then oncologists who specialize in blood and lymphatic cancers, “liquid cancers”.
In addition, cancer does not arise from one mutation. It’s a set of mutations that allow cancer to occur. You need mutations for growth, mutations for preventing cellular death, mutations for stealing energy, mutations for hiding from the immune system. Your body, right now, has probably millions of cells that could be candidates for cancer. They either kill themselves or your immune system nukes them. You really need the stars to align on this stuff to get actual cancer.
And then, it’s not like the cancer stops mutating. Cancer becomes resistant to chemotherapy through mutations. Cancer is “smart”, it will even temporary slow its own growth (metabolism) to intake less chemotherapy through treatment.
Here's another way of framing things that might encourage better discussion: Suppose in a few years, it was announced that an LLM did discover a cure for a cancer. What approach would you expect it to have taken?
1. Developing an intervention for every possible mutation that leads to loss of cellular identity, and uncontrolled growth.
2. Preserving immune function later into life.
3. Preserving genetic and epigenetic accuracy later into life.
Maybe it will be 1. But I would bet on 2 or 3. Which both assume that by reducing the difference between rates, the phenomenon of cancer could be reduced or eliminated.
Also: https://pivot-to-ai.com/2026/07/30/google-kills-alphafold-so...
Or, even better, they never even respond to surging memory costs by saying, "Don't worry, our brilliant AI will figure out how to produce cheaper memory." Because they know it won't.
I genuinely don't know how to interpret the bizarre clichés these people think in. Cure cancer. Make death optional! I can't tell if they're crazy, stupid, or liars.
Maybe I'm overly cynical, but I think charities exist to exist. They don't have a strong incentive to actually deliver on their mission. Everyone at the org may be fully bought it and obviously want to cure cancer, but as an organization it would fail to deliver on the promise. Like most organisms, their actual purpose is to continue to be influential, grow and continue to exist.
I think cancer will be solved by some group that could make a lot of money solving cancer. The probability of any one path working is very low so the payout would have to be very large for anyone to be willing to pursue.
I mean, yes, "I haven't looked into the specifics, but working from the first principles I came up with in 3 seconds, cancer research charities are in fact cancer on society" does come off a bit reductively cynical.
If you can find and patent a cure to cancer, you can charge whatever you want for it, at least in the US. I don’t think profit would be an issue.
And even if ending chemotherapy would discourage the incumbents, smaller companies would love the opportunity to kill them.
For example, Polaroid didn’t want to embrace digital cameras since they made tons of money selling physical film. But all the smaller camera companies didn’t care, since they weren’t benefiting from that and wanted to just take Polaroid’s market share. And that’s why digital cameras took off regardless.
If a big public charity is able to cure a cancer type (eg. lung cancer), then it's next aim would be to try cure another cancer type. Let's just say they are somehow able to cure all the cancers, and therefore they cured cancer, then there's distribution of the cure, with a focus on 1st world but then it would have to go to developing nations too. Along the way, there may be issues that arise in recovery from the cure, so money towards helping survivors there. Shift goal, shift goal, shift distribution/priority of donations, on and on and on. For the proper charities that do want to meet their goals, something like this is what they'd do.
In Canada, measles cases suddenly exploded in 2025 as well, and they don't have RFK Jr. there: https://health-infobase.canada.ca/measles-rubella/
The policies around measles vaccines have shifted a little since RFK became HHS Secretary, but not in a material way. Generally the messaging has focused on putting measles, mumps, and rubella in individual vaccines rather than combined, but at no point has the federal government recommended not to get vaccinated. That, combined with the Canadian data, implies that RFK or other changes to US health policy are not likely to be the cause of the measles outbreaks.
There are charities that specialize in analyzing climate data to see where outbreaks may occur next. There are charities that just do government outreach to leaders that don't believe medicine is real.
If you want a simpler example that doesn't involve biotech research, I would just look at the diversity of animal welfare organizations, and their relative quality. Some of the best animal welfare organizations in terms of impact are local rescues run by (in my experience) spouses of wealthy career-people (executives, lawyers, etc.) who have a ton of free time and love animals. But then there are also plenty of animal welfare organizations with questionable motives, and questionable spending patterns - bordering on fraud and embezzlement, if not outright fraud and embezzlement.
It will become victim of the conflict between its core search for truth, the current administration mandate to spy on its own prompts for approval, and secret orders to hide the mission true purpose, like that other computer we have heard about.
All of those guys think people are stupid. That theme has been repeating over and over again.
Nixon was 24% when he resigned, which is amusingly not actually the lowest, as Truman seems to have twice achieved a spot rating of 22, though Nixon remains the lowest rating for a full year. I wonder what Truman did to be less liked than Nixon. Those were both well after dropping bombs on Japan and before the first H-bomb test.
> quite stupid
No. It's partisanship and polarization, which has little to do with intelligence.
But there's one thing that I know is stupid: assuming the reason opposite pole doesn't do what you think they should is because they're full of stupids.
Liberal or conservative, left or right, authoritarian or libertarian, it doesn't matter: if you (still) think Trump is the right man for the job, then you are either a criminal or other subversive agent, a glassy-eyed cultist, or stupid.
Failure to call out stupidity when we see it is no virtue, to paraphrase Goldwater. It is costing us everything we built as a society over the last 250 years.
No. I personally don't like Trump, and looking forward to the day he's gone, but:
1. Thinking of your opponents are stupid (or some kind of cartoon villain) is itself a really stupid attitude to take. It might feel righteous, but it blinds you. Some of them may actually be literally stupid, but there's no value in even thinking about them.
Looked at objectively, there's also lot of stupid on all sides. See: much of the Democrats'/progressives' reaction to Trump.
2. Approval is complicated. It's not a choice made in a vacuum. American politics nowadays is (for many people) a choice between the best of bad options. Partisanship exacerbates it, because it makes many of the alternatives seem even more loathsome.
3. Not everyone is a news junkie, and expressing approval of Trump (for instance), may not be even meant to be an evaluation of his job performance.
4. etc.
What he is saying makes sense to me.
Is it that you don't trust a single word he says and that he lies for financial gain?
Or do you think he is wrong and that applying AI to this kind of research cannot possibly produce a breakthrough and lead to a cure?
> Do you not think that Amodei, who worked on research about cancer biomarkers at Stanford, knows a thing or two about how to apply AI to the problem?
It certainly shows he is ambitious.
> Is it that you don't trust a single word he says and that he lies for financial gain?
Yes.
> Or do you think he is wrong and that applying AI to this kind of research cannot possibly produce a breakthrough and lead to a cure?
It is possible, but both could be true, that's not an exclusive or situation.
I am not familiar with Amodei's personal history, but I generally do not assume the worst of people just because they are extremely successful, like many others here seem to do.
Between Musk, Altman, Schmidt, Amodei, Gates and a bunch of others I have a really hard time figuring out who is the least ethical. Probably Musk. But it may well be that the long term damage done by Amodei and Altman is larger. Ask me again in 10 years, if I'm still around.
I see the conflict of interest of course, but so far I have not seen anything that convinced me Amodei works towards any other goal than benefit for humanity as a whole, like he claims.
if many other see it as substantial ground, and you just say "I don't see anything", how do you think this discussion can go forward?
It's amazing that I even have to ask for it, as if I am just supposed to believe so.
Alternatively, one could also engage with my argument that Amodei's position seems consistent with his beliefs, and point out flaws with it.
As for "bad person" is concept you brought to discussion, it could mean many things, but you were provided with enough ground from opposite point of view.
> It's amazing that I even have to ask for it, as if I am just supposed to believe so.
toxicity only adds to your character of bias and denial.
https://news.ycombinator.com/item?id=49175827
The sbf vibes are on the nose. Remember that it was a big deal that sbf bought into Anthropic, and many if not most Anthropic employees are EA, the tribe of people I'd rather call "rationalizists". The main thing they rationalize, is of course themselves taking on power and what sbf made famous as uh the linear approach to risk
Between Altman and Amodei, I'd say that Amodei is the more dangerous just because he seems even more "reasonable" to nonjerks. You should ask the mods which one of the two they think are more reasonable. It's clear to lobbyists and politicians --- people familiar with the human jungle that is DC--- that Amodei is weirder, though.
I think I have a good metric for who is least ethical. Which one on your list is most unlikely to write good satire (or most likely to write unintentional self-satire?)
I'd say its Schmidt, followed by Amodei, and then Musk. See Musk on SNL. I think Gates had some TV appearances where he was reasonably funny, almost self-deprecating.
If you ask me, SBF ran an obvious scam in a sector that is as a whole a scam.
With Anthropic's product, I experience a real benefit today.
This reminds me of the myth of Prometheus. I think most Greek thinkers would agree that Prometheus was the hero, and Hephaestus was a complete jerk, even though one created fire and the other merely stole it. Of course they had to make up a punishment for Prometheus so that he could earn his status the meritocratic way..
Probably much against most people here, I separate the product from the person creating it. As another Anthropic example, there was the math guy at Anthropic who disproved a long running conjecture.. it doesn't look like he's getting Fable and Anthropic the respect he thinks they deserve. The process matters.
People are sensitive to disrespect hardcoded in the process, especially against "tried and true" norms, and in most cases they are right. The ability to write satire could be a kind of razor to tell apart the influencers who are likely to be righter than their detractors
The Fable quota on the $200 personal plan could be higher, but the difference to Opus 5 or 5.6 Sol is not large enough to gain any real benefit from unlimited access to Fable, for my use.
From what I heard their internal 'Model 2' is hardly much better, but hopefully they will have something newer soon.
You don't become a CEO of a company like Anthropic (or any sufficiently large company) without lying for financial gain
And they’ll be using detailed medical records. Personalized medicine and the use of ML in medicine has been around longer than LLMs (and will continue regardless). There is a lot more data available than just studies and textbooks.
https://www.cancerresearchuk.org/about-cancer/treatment/pers...
"If you or a loved one have taken a class of drugs known as GLP-1, you may be entitled to compensation..."
A) verifiable proof generation (a limited closed set of axioms and a minimal algorithm like Metamath's are enough to verify correct vs incorrect machine intuition, plus feeding back the rejection/acceptance to adjust the machine intuition)
B) optimizing an illposed problem (how do we define the inaccessible fitness function)
Yes the field of biology or medicine will advance faster, but only because the mathematics and physics will advance faster, any non-axiomatic or non-postulate ramifications actually depend on unknown variables like fitness functions etc. How do you model the environment? That model will necessarily be oversimplified. What makes us believe that a long tail of ever less frequent distinct failure modes (like cancers, cancer is not a single problem) can not be guessed. It would require oracles and crystal balls: if global warming brings malaria mosquito's up north, then "eliminating sickle cell anemia" can actually turn into a death sentence instead of a life-long disease, but perhaps malaria will not come to higher latitudes, it depends on the future!
> This is honestly hilarious. Dario must think people are stupid.
I mean, if Claude actually cures cancer, I do think he'll be right.
The only problem is that you need to actually cure cancer.
Just like the utilitarians in the thought experiment tossing everything positive in their life into the bottomless pit of suffering, the effective altruist AI thought leaders are tossing our entire economy to the gambit that the singularity will bring about the end of human suffering.
Capitalism has always been a tension between capitalists, who want maximum return on capital and workers, who want pesky things like a living wage, sick leave or safe work conditions.
For the longest time, the only power labor has had to get those things was the power of collective bargaining. No agreement with your workers meant no production happened.
Now, there's finally a chance at salvation. The capitalist Messiah is AGI and it will finally deliver them from those annoying laborers.
This ideological bent is why they're putting everything they have into AI. It's why VCs and their fellow capitalists are going so crazy.
They see it as a way to finally solve the contradictions of their ideology, but in reality it'd only create a new one: If everyone's out of a job, who will buy their products?
AI will make human thought economically useless. In the new economy, the one economically useful thing we will be able to do is legal property ownership.
Permanent underclass is ideological goal of the above people. AI is an impersonal technology with no goals or wishes.
There's an inevitable effect that it has, and the only lever you have as an individual is being on the right side of that.
Even if the US federal government refuses to act, smaller government entities can deploy their own compute clusters running strong open-weights models like Kimi K3, and provide their own citizens with free access to strong AI models. There may come a time when even Chinese providers no longer release strong model weights, but then the previous generation of strong open-weights models can be used to accelerate the development of data pipelines and training setups to build the next generation. California or Australia (or many others) could fund university-led R&D efforts to build completely open models.
Despite the angry noises that leading providers make about their model outputs being used to train competing models, that's not illegal. States can use Common Crawl, Anna's Archive, Books3, commercial model outputs, and whatever other data sources that are useful for building models. States have sovereign immunity to ignore copyright complaints, if they decide to pursue model development.
The big AI companies don't have much of a moat. Let the big companies spend a fortune figuring out what works best, then openly build imitations having 95% of the capability at 5% of the cost.
AI is the least of the problem.
This is correct. But we don't need to ban AI to avoid the permanent underclass nightmare.
It's time for the means of production to be owned communally.
It's either a cybersocialist utopia or a neofeudalist dystopia.
Rosa Luxemburg formulated it pretty well more than a hundred years ago: Socialism or Barbarism!.
What does this mean in practice? That an individual cannot own a machine (which is a means of production)?
As a first intuition, I feel like its mostly server farms and data centers that are the concern, large scale means of production.
It feels right that those would be communally owned, be that via democratic government or maybe some sort of worker owned coop?
Nothing prevents the government or coops today from building these things.
Today, people have leverage because removing them has real economic consequences. With AI, we need to be on the useful side of the divide to have any leverage.
And then we'd better hope AI has respect for property rights.
It's probably a lot easier to convince the military AIs we're building to do it.
I don't particularly like the situation, but I plan to do what I can to be on the winning side. The world is going to be what we're all in on building, and for people on the winning side it may be a utopia. Either way, I know which side of the gun I want to be standing behind.
Also note that an AI well aligned with the people who initially trained it wouldn't need to actually kill anyone that didn't take action against the owners, it would just need to outcompete people for resources, and let them fend for themselves in extreme poverty. There's no need to act one way or the other beyond that.
I'm talking about morality here, not practical ability. The ultrarich have their circle of friends and family, who themselves have their own social circle, on and on until suddenly you're not looking at wealthy people anymore. Six degrees of separation and all that. How many people in the social graph can know about this extermination plan before the whole thing collapses?
It's just disenfranchisement. Only the people who try to use violence to replace the political leverage they lose when they lose economic leverage would need any action taken against them, and that's easy to justify as self defense.
I think it's inevitable.
When AGI becomes reality, that reasoning disappears. When everyone's worthless, no one is. Thus the scenario you speculate about would need everyone in the upper echelons of society to be consciously, openly sadistic power-hungry people, which is not the case even when taking into account the higher dark triad rates among them.
Why did you think there's a massive uptick in surveillance and policing technology and automation? It's to make the French solution unfeasible.
Can't help wondering if they've really thought this through.
That was the exact line of thought, coupled with resources are limited (marketed as climate change), that led to the 1 child policy in China and many other decisions that shall not be spoken of.
It doesn't matter that I can't do it. They'll pay me lots of money if they think I can.
The underlying purpose of AI is to allow wealth to access skill while removing from the skilled the ability to access wealth. (read on HN, not my words)
It's anti-human and evil. What a shame that it's illegal for me to say what ought to happen to them.
BTW, not sure what gets more anti-human and evil than wishing violence on whole groups of people!
Can you find any explicitly EA organization that is at all aligned in perspective with anything you suggested?
Working to build AI at the company that is most publicly discussing the risks posed by AI to humanity and most advocating for guardrails (while getting an immense amount of flack for implementing them independently) doesn't seem to support anything about what you, or rather the other person said, about sacrificing humanity for anything AI related.
Anthropic are explicitly building ASI, using recursive self improvement, with no knowledge of how to control it. They spend only a minimal fraction of their funds on safety work, none of which relates to controlling ASI.
I am open to being wrong about Anthropic, but it will be hard to convince me that a company developing ASI is prioritising safety. Do you really believe what they are doing is safe?
It's kind of like saying you can't kill everyone with weapons when you're looking at what you can buy at the local Walmart and not what's in the 1000 foot deep missile silo.
Also, it's kind of a once you build it, it's too late kind of issue. No model will be good enough, the problem is still in the future, we're all dead.
Instead, he is turbocharging development, otherwise said choosing to cause harm he warns about.
That is literaly example of causing harm now and in near future, not caring about people now, only about vision.
He works on the safety team and their responsible scaling policies. He literally may be working on slowing down AGI progress at Anthropic.
Hiring own husband to safety team while you raise money for (their words!) doom machine, while you do your utmost to scare people while you try to raise even more money from investors on doom ... does not show OP was not correct in their assessment.
I think typical bubble behavior the leaders have set up the whole promise to fail. Everyone is expecting some faux super intelligence to come and find a cancer solution everyone else missed. However, it is just as likely that vanilla current LLM's will create enough of a productivity boost for back office automations in research heavy hospitals to create the space for regular humans to create these breakthroughs, but LLM's won't be able to claim that for themselves and inevitably “fail”.
P.S: I am not saying applications will go away but LLM's are clearly massively effective here at the rote parts of it all.
It's hard to keep track of the frontier on bio ML, but it seems that we're going slower than what Demis Hassabis said in 2024 with 5 years to full cell molecular simulation. We still aren't able to reliably model a tiny surface of the cell membrane.
And of course there's Derek Lowe's takes on the drug discovery pipeline waiting for the proof in the pudding.
To me the only reasonable bullish position is that there is a very non-linear AGI threshold for accelerating progress that we haven't hit yet.
For me personally, I'm looking at other more tractable fields as a proxy to measure this kind of progress. The best modest evidence is from the agentic coding area, (modest because these kinds of gains may not translate to bio progress). Other soft-ish fields to like legal/law/tax are also interesting to watch, as a small amount of people are now trusting AI for these areas that were considered totally unusable a year ago. Another proxy is being able to generate generally entertaining media.
Turns out that “give a reasonable probability of being close enough such that you can bootstrap a solution out of experimental data” gives a very high utility and effectively obsoleted several experimental techniques overnight; pretty much “Molecular replacement” is about the only technique for phasing resolution anyone bothers with any more.
But again, “Bio” is an _extremely_ broad term; for every part of the field Alphafold had a big effect on there are a thousand different parts of the field that it did nothing for.
They made huge progress, but I would say that the vast majority of work on this problem was designing the harness for the model. That's a lot of work for each and every domain.
Isn't that so far only static folding?
[0] https://www.science.org/content/blog-post/so-how-ai-drug-dis...
Curing cancer sounds insane, but it's also a research problem, not a societal level coordination problem. And one AI has already proved to help with breakthroughs (alphafold). IMO it makes sense for them to shoot for something like that as proof of AI's beneficial sides.
This is not true. There are two companies at the center of AI direction: Anthropic and OpenAI. If there were anyone on the plant who has the ability to influence our direction then it would be Dario Amodei.
Acknowledging the grievances is a good step, but it's not enough. There needs to be a clear explanation of actions to address them, a plan to enact those actions, and commitments with consequences in failure of those actions. Tell people how you're going to make them more employable and effective and needed. Tell people how your datacenters will be carbon neutral. Tell people how financial actions resulting in a frothy market will be coming to an end. He and Sam Altman alone have this power and their inaction says everything we need to know about their intent.
Possibly, but not necessarily in a better direction. And that's the whole problem here. They should all go watch Phantasia.
Hoping that they solve those huge problems themselves is basically hoping that they will have more power.
It would be like expecting Craigslist to rescue journalism. The amount of power needed to disrupt an industry is nowhere near what's needed to fix it.
If you don't tackle the societal level problems, nobody will care about research breakthroughs.
I don't think anyone will care about the largely made up issues about electricity prices or water use in case AI literally cures cancer.
His body was healthy, aside from the bullet in his skull.
But something tells me either they can't or they won't, so no trust will be built.
I just don't buy the AI labs approach to this stuff. Like, unless we can basically simulate the entirety of human biology, I don't really see how LLMs can make progress here. Maths is different as it doesn't require a real-world interface, and programming already (by definition) can be simulated on a computer.
Without that, I can't see much (if any) progress being made on domains like biology.
Drug discovery is similar AFAIK. The space of possibilities is even larger than protein folding, but it's structurally similar enough that I think AI will help to make progress on the discovery side. Actually getting the drug tested and approved is another matter though for sure.
For one, the question for Anthropic is whether LLMs, specifically, not AI techniques more generally, can help significantly with cancer research. And here, all experience so far is that LLMs only really work when they can easily automatically verify their own outputs and self correct - such as in math (using automatic proof verifiers) or programming (using compilers and unit tests).
The second problem is that biological research speed is highly dependent on slow biological processes, such as cultures and long term studies. In programming, if an LLM could provide excellent insights and research suggestions 100x faster than a human, it would speed up the work roughly 100x. But in biology, it would only speed up the total work by a small amount - as any insight, even if absolutely brilliant and spot on, would still require months and years of actual experimentation.
If I was Dario, and I was after a cynical sales grab to go with the "actually curing cancer!" spiel, I'd probably aim for a drug repurposing strategy. I'd use the above approach to find existing drugs that might target known pathways that drive incurable cancers. I'd only screen compounds with extensive safety data and easy delivery mechanisms, and from the in silico hits, I'd throw a tonne of money at rapidly experimentally screening all those candidates in parallel, and then rapidly push those that worked into clinical trials. I'd assume that, with a small but non-negligible prior, and the money to push through hundreds and hundreds of candidate compounds through at once, I'd have a reasonable chance of getting one drug through to a "Claude Cured Cancer!" show-stopper headline.
But I think even then, with all of Anthropic's wealth, you'd need 2 years minimum to move from initial screening targets to a Phase 3 trial. And it would be an almost criminal waste of research funding to get there -- the dollars for discoveries ratio would be appallingly bad.
(If I was Dario, and I actually wanted to cure some forms of cancer? I'd just use my obscene profits to fund actual cancer research, step back, and let the researchers get on with it.)
I do agree that this kind of targeted approach makes sense.
However, discovery is not really the issue here. Running the clinical trials (1/2/3) is much much more difficult, and consumes basically all of the time in drug development, so even if LLMs perfectly automate this, the speedup will not be particularly large.
This is not correct, Alphafold predicts crystal structures of proteins. It is a known issue in the field that these folding models do not generate ensembles of protein conformations like the ensembles generated by DE Shaw Research (who built a super computer to simulate proteins). And these folding models fail when we don't have crystal structures, so they are certainly not generalizing. A great paper on the subject: https://www.biorxiv.org/content/10.1101/2025.02.03.636309v1
Drug discovery is significantly more complicated than protein folding. Small molecules with similar chemistries can adopt novel binding poses, can end up binding to off targets (PXR, hERG, etc) or simply never make it to the protein due to solubility or permeability.
I will always defer to Pat Walters who seems like the most careful and sane person in this field: https://patwalters.github.io/
Anthropic is literally creating the bubble. It is not beyond their scope of influence, it is literally what they are consciously achieving.
As for peoples jobs, same actually applies. Anthropic is selling itself on dream of replacing jobs, even or especially where they are well aware AI does not perform that well. They are actively trying to replace people quickly before management notices it does not work well.
And also, they can influence how much their data centers contribute to global warming.
Curing cancer is a societal problem. We have lots of cures for common diseases but resource allocation means people don't actually receive the treatment they need. For example, prohibitively expensive gene therapies, or HIV treatments in developing countries. A disease may have a "cure" but if people who need it don't receive it then from their perspective it may as well not exist.
It signals to people who have money and/or influence that the speaker is "committed to doing the right thing" and "eager to find a workable path forward" that "meaningfully addresses the 'very real' concerns and misunderstandings that non-experts have about what the company the speaker is associated with are doing".
Once the speaker has signaled that, what more could they possibly do? They've signaled their willingness to open the channels of dialog with the rabble who are non-experts and -thus- cannot possibly be engaged with as equals. By doing this, the speaker tried their best, didn't they?
Or, to put it another way, it's PR and damage control... soothing theater for the folks who can actually affect them.
I know people in positions of power often see lots filtered information that skews their perceptions of reality, but it seems like the level of pure hype Dario's pushing exceeds even the typical sycophantic filter bubble. Cure cancer indeed.
The way he's coming off here though gives me SBF vibes.
But instead of doing that, I'm going to drop everything and become a linux kernel maintainer. Learning about the internals of how RCU concurrency is implemented across subsystems is surely what I need to do right now!
(/j before I get crucified)
So Anthropic are going to spin up a medicinal chemistry lab and start mouse experiments?
I mean, it would be nice, but I don't think that the guys at Anthropic know what they're talking about here, or what they might be getting themselves into. (If indeed this is more than just PR.)
https://www.writingruxandrabio.com/p/intelligence-is-not-the...
Didn't and don't mean to disparage anyone's medical struggles, but "a cure for cancer" is a well-worn strawman. One that's been achieved for the low hanging fruit, the higher ones are seeing steady progress (already before LLM chatbots, even!), the bottleneck isn't "intelligence" and to the extent there are socioeconomic (access to screening, treatment) or environmental/lifestyle factors involved the AI boom is likely just making things worse!
I'm sure Dario knows this, and it's anyway too pedestrian compared to the usual list of fruits of ASI. The text probably originally read "nanobots eating you alive and uploading to the cloud" or something, but they figured that wouldn't go over with the intended audience. "What do the peasants care about? Oh I know! Curing cancer!"
How else are you going to work towards a dream? It's how Elon Musk managed to get reusable rockets when everyone said it's unfeasible.
Granted, most ideas don't work, this is why you need testing, but I am a bit surprised to see this attitude on hacker news.
But they said that they want to cure cancer. That's far outside the core competencies of any LLM, and it requires a lot of real-world wetwork with liquids, chemicals, cell line experiments, animal experiments, etc. They can't merely analyze existing data -- they'd need to generate vast amounts of new data, which isn't really the case in physics. And then regulatory approvals and so forth.
"We're working on curing cancer" sounds more like a poor PR attempt than an actual effort, though I'd love to be wrong.
As if we didn't know the challenges in translating research into clinical treatments.
I wonder if many of the commenters here read the post, where Amodei specifically references his essay with suggestions for streamlining the FDA approval processes: https://darioamodei.com/post/policy-on-the-ai-exponential
Writing a paper on streamlining the FDA approval process is nice, and I agree with all of it, but it's one thing to do that (about 4 hours of work) and quite another thing to cure cancer.
My point was more that LLMs are exceptionally weak at developing drugs. They're strong at every math/logic/language task, but curing cancer is something of an entirely different nature. Again, if he wanted to "solve physics," or "develop a universal real-time language translator," or even "solve global warming," I wouldn't be nearly so skeptical.
...The first is something that can be credibly attempted with a next-gen super-LLM and a ton of compute. The second is something that LLMs are exceptionally good at, and a dedicated engineering effort might get you there. The third is a modeling/simulation problem with solutions that are, at least in principle, known but haven't yet been thoroughly weighted. (i.e., how bad really is ocean seeding or launching SO2 into the atmosphere?)
But curing cancer? It's not even clear how to approach the problem. Cancer is also heterogeneous... so when I say that LLMs are weak at developing drugs, the problem is compounded, as curing cancer might require dozens or hundreds of distinct new drugs. It's like Dario picked the least likely, most difficult problem to solve. So it's weird.
Did you know that Amodei was working on research about cancer biomarkers at Stanford?
Given his background in biology, it is hardly like he is a random tech guy with no knowledge on the topic, as many here seem to imply.
Amodei mentioned "actually curing cancer" as an example of undeniable public benefit, not as an exclusive research focus or anything like that. He is clearly talking about generally intensifying biomedical efforts, not promising a cure for cancer specifically.
“The detection solution may be made available in the Union as one or more of the following: (i) a public, ideally standardised, specification allowing any third party to implement a detection mechanism; (ii) a piece of software (e.g., a standalone executable or library); (iii) a cloud-based service accessible to users in the Union through an API.”
Option 3: "a cloud-based service".
And accessible "to users". NOT to the public.
Note: this is not the actual law, it is the code of practice, ie. guidance for model providers. The start of that document clearly states that you can comply with the law in other ways if you want, you'll just have to justify yourself. So you have a choice to not even do this.
The actual law is here:
https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-5...
As to who decides the law clearly states who decides if someone is legal, like in most EU legislation. It's not the courts, it's "National market surveillance authorities designated by each EU Member State", and there is an EU office as well (and it is explicitly stated that they are not allowed to override each other). So every EU country has the right to provide exceptions to the law, just like they do for the GPDR. You do not have any rights under this legislation as an individual. Only these "National market surveillance authorities" get rights under this legislation.
Secondary: article 7 additionally gives the EU commission the power to declare any code of conduct they want that declares what compliance with the AI act actually means.
(and, of course, this is yet another attempt at declaring math illegal. The only way to actually enforce this legislation is for all models to comply with this, all over the internet. Obviously the EU does not remotely have the power to make that happen)
Where do I find that in the Act (or code of practice etc.)? IANAL, but Article 50 reads different to me but if there is a comment or guide how to read it - also fair enough.
> puts in it's regulations that you're not allowed to use AI to communicate with them (we all know this is coming),
.. do we?
> You get a mail from the government about taxes. You want to check if that text is generated by an AI
No, actually, I want to know whether it's correct and whether it's legally binding. Using AI makes it less likely to be correct, sure, but ultimately whether it's human, spreadsheet, or AI the important thing is the legal right to correct process.
Which is mostly a good thing because most of those processes were cooked up to be exclusionary but with plausible deniability
God forbid some upstart get a big contract they "shouldn't" have gotten because AI lets them create a submission that's on par with a big guys.
God forbid Joe Schmo be able to prepare the paperwork and dot his I's and cross his T's and be able to take advantage of some stupid zoning exemption that you spec'd out to essentially be only available to big moneyed interests.
The ability to suffer one's way through beurocratic process without having to suffer through paying the lawyers and accountants and engineers and whatnot to get you through that process is a huge part of the value proposition of AI. But people don't frame it like for obvious reasons.
These are adversarial processes. You might get a win in the short term, but everyone else is playing the same game with access to the same tools, and developing a game-breaking strategy in a PVP multiplayer game is going to both escalate and get patched out.
A colleague of mine gave the example of using AI to snipe the UK driving test queuing system, which currently has a massive backlog. Which is great .. for them. Someone else gets pushed down by automated sniping. Then everyone else will have to start sniping. Then the website falls over because it wasn't designed to deal with 100m requests from the few hundred thousand people who are trying to bot their way to victory.
(I can see one likely patch out of AI spam being "you must submit requests in person 9-5, online system is closed down"!)
Anything big enough always is, no? The next scale of order in the hierarchy starts concerned itself with what's good for it, at the expense of the smaller parts.
The fact that government is like a giant organization that everyone lives in, the fact that it mostly doesn't fuck us all over in terrible ways, that's a testament to the amazing alignment of regulation and law imho
The bio/cyber/alignment risk rhetoric is nonsense. Somewhat counterintuitively, all of those risks are lower/non-issues if frontier labs release their weights and do all of their research in the open, as explained in a past comment: https://news.ycombinator.com/item?id=49313739
Again: if we were to regulate anything, it would be to force frontier companies to release their model weights and do all of their research in the open. This is a necessary yet insufficient condition for the intelligence explosion to benefit the vast majority of people.
No snark here from my side, just curious about your vision - which concrete actions by Anthropic could improve trust for the concerns you listed?
Considering the vast amount of money, world-class talent and political and industry vonnections Anthropic has access to, it feels a bit unfair to ask me to solve this for them.
My answer to their proposal for what our society should look like, and what kind of technologies we should use is "no". It's not a trust issue, as in I would not accept the proposal from anyone, even if I did otherwise trust them. And if someone I trusted insistently proposed this, I would stop trusting them.
In any case, they don't accept refusal, but try this grotesque and condescending maneuver about "trust", as if I and many others were not really just saying "no".
No one who refuses to take "no" for an answer can ever deserve trust. If you are committed to convincing me to go along with your desires, and have no intent of accepting my refusal, you will never deserve trust, and I will never trust you.
These companies may have the power to do what they want anyway. But they will never have a clear conscience, or trust and support from me or anyone like me.
Thiel et al. have described technological change as a means of forcing unilateral changes on an unwilling public. There is no basis for trust unless there exists a justice system that can appropriately discipline and restrain such people and their coconspirators, and force them to abandon their plans to inflict unwanted conditions on masses of people.
Surely it will be quite dystopian future - AI will dictate how the money allocated, where there is only one private company left - Anthropic (as per Dario says)
At the same time, the best Anthro models refuse to answer biology questions, keeping the public in the dark not only about new AI-assisted results but also about conventional and health-protective research.
D.A. >> we hope to have incredible results in the coming years and some early glimmers in the coming months.
So, what changed? He went from a promise of AI bonanza any minute now to a promise to maybe cure cancer in the coming years, that's right, an unspecified number of years. Until that time, we're supposed to eat glimmers of hope because the AI-caused inflation will make food unaffordable.
D.A. >> When we’ve actually accomplished something real, the whole world will hear about it, as loudly as possible, you have my word on that.
And don't forget to stock on ear protection, because when something happens, it's going to be "as laud as possible", you've got Amo's word too - to marvel at in the meantime.
> I think that ordinary people don’t trust companies, governments, or the tech industry and always suspect that we are cooking up some new way to screw them over.
Well, yeah Dario. You keep PUBLICLY TALKING about ways you want to fuck over people and everyone has noticed. I'm sorry, but when a lot of people are living paycheck-to-paycheck and watching layoffs everywhere, being like "well it'll take all the jobs and maybe kill humanity" is going to set off a LOT of alarm bells, despite your attempts to make concerned looking faces. You're trying to solve hypothetical problems in unproven ways, while worsening real problems people actually face. These people are SO disconnected from the society they live in.
People have a problem with the efforts, not that they haven’t reached utopia yet. Doubling the efforts will not solve the problem, it will only deepen it.
Other than the fact I doubt that will happen, if it did, I think it would just underline the trust issue.
If some independent group cures cancer using Anthropic’s models as critical support, that would be great—but the nature of biology and medicine is that guardrails to prevent misuse will also cripple research.
If someone with an tight special relationship cures cancer with special privileged access to Anthropic’s models cures cancer making significant use of the models, well, the cure for cancer is nice, but people are going to notice that Anthropic has the control over who gets to try, and the power to set terms, which underlines the stakes for the trust issue.
[...]
> I don’t think that a glitzy marketing campaign with a positive spin (which some have advocated that Anthropic do) is the way to win back that trust — at this point, saying that AI will cure cancer is more a cliche than it is inspiring, and most people think it is deceptive. The thing that will work is actually curing cancer. I think by far the most accurate criticism of AI companies including Anthropic is that we haven’t yet delivered on our big promises to benefit the world. That is totally on us, and I think it’s the criticism you should be making, instead of all this stuff about messaging and marketing.
> We are however doing our best to fix this: Anthropic is ramping up its efforts very quickly in biology and medicine, and we hope to have incredible results in the coming years and some early glimmers in the coming months. When we’ve actually accomplished something real, the whole world will hear about it, as loudly as possible, you have my word on that.
Turns out "winning back trust" doesn't have anything to do with any actual concerns people may have re: employment, electricity prices, stock market bubble, intellectual property, scams, cybersecurity, environmental issues etc. Rather we'll just do all of that even harder and the miracles ("curing cancer", lol) we've so far failed to deliver are bound to arrive in short order!