I’m curious when AI will be directly responsible for reducing the cost of anything.
Not “if you use it it can make you more money” but “real, normal people are seeing a benefit, not from using AI but from something it was able to accomplish that made normal people’s lives better.”
Making websites is now much easier. I don't know how many normal people in your life need websites, but small business owners I know are able to use Claude to build their website for much cheaper than it used to be. Instead of paying Squarespace/Wix/WordPress anything per month, they can just pay Claude $20 for a month and build a site, have Cloudflare host it via pages for free, and save money.
Curing cancer is impossible, because cancer isn't one thing. For example, I have a cure for cancer. And not just some benign thing but scary leukaemia, specifically chronic myelogenous leukaemia (CML) in which genes on two chromosomes get switched and use a molecule called ATP to activate other proteins that cause cancerous growths. The cure is to block ATP using a drug called imantinib. So there, I've cured cancer!
What's that? You don't have CML but some other type of cancer? Well I didn't say I could cure all cancer, you should have read the fine print. Now gimme more money.
The issue with “AI” is how toxic it feels. Feels grimy like social media. Just like social media, I haven’t introduced any of it to my young kids nor do I plan to until they’re teens. At most I give them access to local AI through home assistant. I have a coworker who have kids of similar age and had to get rid of all his echos when Alexa started feeding his son new Lego sets as “Christmas gift ideas” to tell his parents.
It's the circular saw on the job site that makes a bunch of manual jobs quicker. But at the same time it can feel inappropriate, like people are in love with their circular saws so much that they bring them to concerts, and try and do jobs that don't call for a circular saw, like painting walls or digging a hole.
when I was a child I read a book about a luddite family who avoided getting their kids neurolinks and the oldest child decides to go and get one put in at you know, coming of age, and then dies from the implant. I have no idea why I remember this but. Godspeed to you.
> I do agree that the public has a negative view of AI (and that this is a big problem), but I don’t think it is primarily caused by me or any other AI leader warning about AI’s risks. I think it is fundamentally a crisis of trust.
Maybe someone here who works for a living doing pharma research might chime in but to me "AI curing cancer" makes about as much sense as "datacenters in space".
The biggest barrier for pharma research AFAIK doesn't have anything to do with finding candidate targets and corresponding drugs, there's thousands and thousands of those and progress for finding candidates as it stands is quite good, but all that pales in comparison to the time and cost of developing the actual candidates, then engaging in medical trials at all levels and waiting for results and then developing the engineering for manufacturing at scale and at cost, as usually all modern monoclonal antibody therapies run in the tens of thousands of dollars.
How is AI supposed to help at accelerating Phase 3 trials? How is it supposed to help us find better bioreactors?
I’ve read Dario’s two long essays. A fair representation of his view is that he thinks fundamental discoveries like mRNA vaccination and CRISPR will happen 10x faster due to AI, and he hopes this means no more diseases. That’s how he goes from the natural timeline of most diseases being eradicated even without AI to them being eradicated in 10 years.
Inserting my own viewpoint now, it’s a very natural way of thinking for this guy because his entire success is due to assuming a graph will always go up when you increase the inputs and apparently being right about it.
Sam Altman made a similar remark, in the case of Sam Altman though, GPT has been making breakthroughs in mathematics which is rather impressive, I could see AI helping to unlock more about cancer.
And make jobs. If sama and Dario said they have plans to invest in whatever research they want to do with their money, creating jobs, people can train up for that and anticipate the future instead of it being this looming scary thing.
To conjure fire from the skies, in other words. But the way technology works is it solves a problem by introducing other problems, and if they plan to fix cancer by messing with DNA, guess what side effects there will be.
I commonly see people online say something along the lines of, “instead of curing cancer we got a stupid chatbot.” Dario is simply responding to what people say they want.
The way for AI to win over the public is for it not to be used to ruin peoples lives, kill people (autonomous weapons) and exacerbate the climate crisis just so a handful of rich people can get richer.
Even if cancer was solved, those problems would still exist because it's not about betterment of our rae. It's about rich people trying to get richer, everyone knows that, and that's why people hate it.
Edit, I should mention, it's being used to ruin the internet too with "slop".
I love that they're at this point not only adopting the apocalyptic worldview of Millenarian religious cults but also the faith healing marketing strategy. Waiting for the day when Claude makes someone walk out of a wheelchair after speaking in tongues
I concur with the other commenter and previously felt the need to write something similar, but since they did I’ll do it too. Somehow the messianic nature these AI people present themselves with had eluded me until reading your comment. Really remarkable insight here.
I had a conversation with Opus 5 about Amodei’s belief that most human diseases will be cured in 5-10 years. It said that technically “most diseases” are rare genetic conditions, and apparently that the science of these is basically already finished (and has been since a while before LLM based AI took off). These can be cured with gene editing therapies and what’s missing are “delivery vehicles and economic/regulatory models.” It predicted by the mid 2040s these will have treatments. It predicted the “big burden” diseases however won’t be cured due to increased intelligence but may become manageable chronic conditions by the mid 2040s. And it predicted that it would be until the 2060s until most of the disease burden is actually alleviated around the world.
Interestingly when I asked it how much of the speed up of curing most human diseases will likely be creditable to LLM based AI it said only a small portion (maybe a 5-10 year speed up) and possibly that LLM based AI will slow humanity down, essentially because it produces so much bs to comb through (“polluting the literature with plausible synthetic findings”). It says the AI technology which is actually helpful in biology is AlphaFold, RFdiffusion, etc..
Okay, you told us what it said. Now what do you think? Do you have any expertise in this field? Do you have reason to believe that an LLM has the ability to predict the future with any accuracy (and if so, why)?
If we want to hear from an LLM, we can ask it ourselves. What are you bringing to the conversation?
I felt it was obvious my post presented a deep criticism of Amodei and over-investment in LLM based technology for the purpose of curing cancer. I don’t think overexplaining the point makes for good communication when people can think for themselves.
Incase you need it explained, it is ironic that either you are a believer in AI and so trust the AI analysis here, and so should doubt Amodei’s claims, or you don’t think very much of Amodei and AI and so probably don’t trust his claims.
I’m curious when AI will be directly responsible for reducing the cost of anything.
Not “if you use it it can make you more money” but “real, normal people are seeing a benefit, not from using AI but from something it was able to accomplish that made normal people’s lives better.”
Making websites is now much easier. I don't know how many normal people in your life need websites, but small business owners I know are able to use Claude to build their website for much cheaper than it used to be. Instead of paying Squarespace/Wix/WordPress anything per month, they can just pay Claude $20 for a month and build a site, have Cloudflare host it via pages for free, and save money.
Curing cancer is evidently easier than simply not going on Fox News every 3 weeks and telling everyone they will lose their jobs.
Curing cancer is impossible, because cancer isn't one thing. For example, I have a cure for cancer. And not just some benign thing but scary leukaemia, specifically chronic myelogenous leukaemia (CML) in which genes on two chromosomes get switched and use a molecule called ATP to activate other proteins that cause cancerous growths. The cure is to block ATP using a drug called imantinib. So there, I've cured cancer!
What's that? You don't have CML but some other type of cancer? Well I didn't say I could cure all cancer, you should have read the fine print. Now gimme more money.
The issue with “AI” is how toxic it feels. Feels grimy like social media. Just like social media, I haven’t introduced any of it to my young kids nor do I plan to until they’re teens. At most I give them access to local AI through home assistant. I have a coworker who have kids of similar age and had to get rid of all his echos when Alexa started feeding his son new Lego sets as “Christmas gift ideas” to tell his parents.
Why does it feel toxic? It's literally like the computer in Star Trek.
When used correctly.
It's the circular saw on the job site that makes a bunch of manual jobs quicker. But at the same time it can feel inappropriate, like people are in love with their circular saws so much that they bring them to concerts, and try and do jobs that don't call for a circular saw, like painting walls or digging a hole.
when I was a child I read a book about a luddite family who avoided getting their kids neurolinks and the oldest child decides to go and get one put in at you know, coming of age, and then dies from the implant. I have no idea why I remember this but. Godspeed to you.
- i have an even better one
- stop ageing and make people jellyfish immmortal
- think you can pull that off dario?
People are working on this, but why would they share it with anyone for less than a billion dollars?
Amodei's original toot:
- https://twitter.com/darioamodei/status/2088758819304443967 / https://xcancel.com/darioamodei/status/2088758819304443967
> I do agree that the public has a negative view of AI (and that this is a big problem), but I don’t think it is primarily caused by me or any other AI leader warning about AI’s risks. I think it is fundamentally a crisis of trust.
Pairs well with "Dario's wife tried to raise money from Epstein for a porn studio" (https://news.ycombinator.com/item?id=49294219).
> but I don’t think it is primarily caused by me or any other AI leader warning about AI’s risks.
True, how about this:
> I think we might be six to twelve months away from when the model is doing most, maybe all of what software engineers do end-to-end.”
Or rather essentially tangible benefits for the average person
Maybe someone here who works for a living doing pharma research might chime in but to me "AI curing cancer" makes about as much sense as "datacenters in space".
The biggest barrier for pharma research AFAIK doesn't have anything to do with finding candidate targets and corresponding drugs, there's thousands and thousands of those and progress for finding candidates as it stands is quite good, but all that pales in comparison to the time and cost of developing the actual candidates, then engaging in medical trials at all levels and waiting for results and then developing the engineering for manufacturing at scale and at cost, as usually all modern monoclonal antibody therapies run in the tens of thousands of dollars. How is AI supposed to help at accelerating Phase 3 trials? How is it supposed to help us find better bioreactors?
dont worry, without any jobs available the pool of desperate human test subjects will be massive.
If we knew the answers to those questions, we would have a cure for cancer.
I’ve read Dario’s two long essays. A fair representation of his view is that he thinks fundamental discoveries like mRNA vaccination and CRISPR will happen 10x faster due to AI, and he hopes this means no more diseases. That’s how he goes from the natural timeline of most diseases being eradicated even without AI to them being eradicated in 10 years.
Inserting my own viewpoint now, it’s a very natural way of thinking for this guy because his entire success is due to assuming a graph will always go up when you increase the inputs and apparently being right about it.
Or promote Boris cherny to ceo.
Is anyone using it for that though? Or is it being used for double-tapping girl schools?
"Our product is so toxic to much of society, we literally have to cure one of the worst scourges on Earth to do right."
Prove it
Sam Altman made a similar remark, in the case of Sam Altman though, GPT has been making breakthroughs in mathematics which is rather impressive, I could see AI helping to unlock more about cancer.
Maybe he could convince some by picking up a cigarette habit.
And make jobs. If sama and Dario said they have plans to invest in whatever research they want to do with their money, creating jobs, people can train up for that and anticipate the future instead of it being this looming scary thing.
Related:
Young People Hate AI CEOs So Passionately That It's Almost Hard to Believe
https://news.ycombinator.com/item?id=49323932
We should cure cancer not because it will save lives but because it might stop people from hating me, and that is surely the most important thing.
To conjure fire from the skies, in other words. But the way technology works is it solves a problem by introducing other problems, and if they plan to fix cancer by messing with DNA, guess what side effects there will be.
Probably new cancers.
As Amodei goes getting more attention, he shows that he's as disconnected from the real world as Sam Altman.
He's at least vocalizing he understands there is a crisis of trust. https://news.ycombinator.com/item?id=49323245
Then goes to say people want the cure for cancer from AI. People wanna good jobs, a place to live without paying rent and free time to enjoy life.
I commonly see people online say something along the lines of, “instead of curing cancer we got a stupid chatbot.” Dario is simply responding to what people say they want.
Exactly and it's clear his lack of sincerity and empathy.
He's saying "people had enough of us saying AI will cure cancer, what we need to do to regain their trust is to deliver on our promises".
HN reaction: he makes absurd promises, he's disconnected from reality.
oh, is that all?
The way for AI to win over the public is for it not to be used to ruin peoples lives, kill people (autonomous weapons) and exacerbate the climate crisis just so a handful of rich people can get richer.
Even if cancer was solved, those problems would still exist because it's not about betterment of our rae. It's about rich people trying to get richer, everyone knows that, and that's why people hate it.
Edit, I should mention, it's being used to ruin the internet too with "slop".
I love that they're at this point not only adopting the apocalyptic worldview of Millenarian religious cults but also the faith healing marketing strategy. Waiting for the day when Claude makes someone walk out of a wheelchair after speaking in tongues
I concur with the other commenter and previously felt the need to write something similar, but since they did I’ll do it too. Somehow the messianic nature these AI people present themselves with had eluded me until reading your comment. Really remarkable insight here.
This is such a good comment I wish I could upvote you 10x. I actually laughed outloud at your comparison to religious fundamentalists haha
I had a conversation with Opus 5 about Amodei’s belief that most human diseases will be cured in 5-10 years. It said that technically “most diseases” are rare genetic conditions, and apparently that the science of these is basically already finished (and has been since a while before LLM based AI took off). These can be cured with gene editing therapies and what’s missing are “delivery vehicles and economic/regulatory models.” It predicted by the mid 2040s these will have treatments. It predicted the “big burden” diseases however won’t be cured due to increased intelligence but may become manageable chronic conditions by the mid 2040s. And it predicted that it would be until the 2060s until most of the disease burden is actually alleviated around the world.
Interestingly when I asked it how much of the speed up of curing most human diseases will likely be creditable to LLM based AI it said only a small portion (maybe a 5-10 year speed up) and possibly that LLM based AI will slow humanity down, essentially because it produces so much bs to comb through (“polluting the literature with plausible synthetic findings”). It says the AI technology which is actually helpful in biology is AlphaFold, RFdiffusion, etc..
Okay, you told us what it said. Now what do you think? Do you have any expertise in this field? Do you have reason to believe that an LLM has the ability to predict the future with any accuracy (and if so, why)?
If we want to hear from an LLM, we can ask it ourselves. What are you bringing to the conversation?
I felt it was obvious my post presented a deep criticism of Amodei and over-investment in LLM based technology for the purpose of curing cancer. I don’t think overexplaining the point makes for good communication when people can think for themselves.
Incase you need it explained, it is ironic that either you are a believer in AI and so trust the AI analysis here, and so should doubt Amodei’s claims, or you don’t think very much of Amodei and AI and so probably don’t trust his claims.
Most of the interesting responses come from interesting prompts... why didn't YOU come up with that interesting prompt or the parent??