I think you have the mechanism of the pop wrong. IMO, the bubble will pop because of value competition from open source / non-frontier models. This will kill the margin of OpenAI/Anthropic, popping their valuation. Without access to capital, they will slash their training budget, killing demand for B200's etc. But the same low price that pops the bubble will make demand for inference explode, along with demand for efficient inference chips, like Tensor.
I mean, if you really extend that line of thinking, human beings are parasites in general. We are just sucking up all the resources to make more Walmarts and McDonald's. It would then be fair to say that humans should be dislodged from the earth and should never have attached to it in the first place. However, that would be a simplistic and rather naive view of how society works and does not have a solid moral standing behind the argument other than an emotional appeal: "THE ROBOTS ARE STEALING OUR HUMANITY!!!"
Rather, why not go deeper? Why not look at the world around you and see what incentives we've built up that allow us to plow trillions of dollars into this without a second thought. We've seen the bust cycles happen and yet here we are: going all-in on a potential technology that might/might not work out as we are hoping it will. If it truly works out as these companies are hoping then it'll be an economic disaster. If it doesn't work as these companies are hoping then it'll be an economic disaster.
Yet the Darios and Altmans of the world always find a way.
> I think it comes from a misreading of Marx because Marx considers value coming from Labor which is purely zero sum.
Don't you think it's a bit overconfident to dismiss Marx just on such a simplistic interpretation (as he had simply failed to consider that economic activity can be non-zero-sum?)
So you want them to pay the host websites for scrapping their publicly shared content? Or give attribution to how this is forming its opinion?
It's narrow take I think, let us say someone read my comment or blog, among others, and used that shape their understanding, and then by doing that, they offered better consultation, then should that person pay back every author, book and blog they read?
I don't understand what's the argument here, and also the same methods is used to give us open source models..
But who said it's harmful overall? That's a statement you are making not the people deriving value from it. The fact is many are benefiting otherwise they wouldn't pay that subscription.
In number 4 of the author's definition of parasite, the parasite kills the host not benefit it.
I guess that analogy would be people stop sharing their thoughts publicly, which is as likely as people stop sharing on social media, very unlikely for many reasons.
Watch the parasites swarm all over this thread. (It's been flagged already). The host (HN) is at the point of suffocation.
The post is a zero-information personal rant (good for him/her!), impossible to comment on without violating at least half of the HN guidelines[0].
[0] literally; even this comment violates a few
Bear in mind that like many toxic blooms, once the investment hysteria feeding them dies out they will die also.
- Companies using investment funds to play the Ai Agent lottery will run out of funds
- Less tokens will be demanded, causing AI companies to raise prices to pay for overheads
- Bubble pops.
Cool, bet on it: https://polymarket.com/event/ai-bubble-burst-by
Meanwhile NVDA go brrrrr.
NVDA is hardware. https://isaiprofitable.com/
A fool's errand. "the market can remain irrational a lot longer than you can remain solvent."
https://quoteinvestigator.com/2011/08/09/remain-solvent/
I'm willing to have a public bet that the number of tokens (or whatever the equivalent is) would increase by more than 10x in the next 3 years.
(for equivalency of tokens I'd use something like "tasks done by AI")
I think you have the mechanism of the pop wrong. IMO, the bubble will pop because of value competition from open source / non-frontier models. This will kill the margin of OpenAI/Anthropic, popping their valuation. Without access to capital, they will slash their training budget, killing demand for B200's etc. But the same low price that pops the bubble will make demand for inference explode, along with demand for efficient inference chips, like Tensor.
I mean, if you really extend that line of thinking, human beings are parasites in general. We are just sucking up all the resources to make more Walmarts and McDonald's. It would then be fair to say that humans should be dislodged from the earth and should never have attached to it in the first place. However, that would be a simplistic and rather naive view of how society works and does not have a solid moral standing behind the argument other than an emotional appeal: "THE ROBOTS ARE STEALING OUR HUMANITY!!!"
Rather, why not go deeper? Why not look at the world around you and see what incentives we've built up that allow us to plow trillions of dollars into this without a second thought. We've seen the bust cycles happen and yet here we are: going all-in on a potential technology that might/might not work out as we are hoping it will. If it truly works out as these companies are hoping then it'll be an economic disaster. If it doesn't work as these companies are hoping then it'll be an economic disaster.
Yet the Darios and Altmans of the world always find a way.
I find it sad that people don't generally think about non zero sum possibilities.
I think it comes from a misreading of Marx because Marx considers value coming from Labor which is purely zero sum.
In my bubble within friends and family, practically everyone uses AI and has fun using it and gets value out of it.
Among your friends and family, who use AI for fun, do they pay for it? Would they?
My non-tech friends use AI to make funny meme images that they send around on group chats, but would they pay for those? I doubt it.
I can't think of anything in my personal life that I would pay for an AI to do. Businesses, sure... as long as they are getting real value.
> I think it comes from a misreading of Marx because Marx considers value coming from Labor which is purely zero sum.
Don't you think it's a bit overconfident to dismiss Marx just on such a simplistic interpretation (as he had simply failed to consider that economic activity can be non-zero-sum?)
He's not that stupid but value according to Labor theory of value is zero-sum. He knows that wealth is positive sum but his followers don't..
Parasite takes and gives nothing back, but clearly a lot of people are using those models, which the author ignored, so the entire premise is false.
The use of the models completes a self-improving feedback loop which keeps feeding the beast and helps it grow even stronger.
But that's not the definition of parasite. People are using it because they are gaining value.
ah so symbiosis
drunk driving may kill a lot of people, but it also helps a lot of people get to work on time, so, it;s impossible to say if its bad or not,
The models might benefit you, but they do not benefit the Host websites that were scraped to feed you a result.
University graduates might benefit you, but they do not benefit the authors of the textbooks that were borrowed during their education.
Enough with the IP maximalism, it's perfectly fine to extract non-copyrightable ideas from copyrighted works.
So you want them to pay the host websites for scrapping their publicly shared content? Or give attribution to how this is forming its opinion?
It's narrow take I think, let us say someone read my comment or blog, among others, and used that shape their understanding, and then by doing that, they offered better consultation, then should that person pay back every author, book and blog they read?
I don't understand what's the argument here, and also the same methods is used to give us open source models..
> At least we get stilted prose, images that look like Pixar rehashes, and code nobody understands.
Parasites are mostly bad for health.
Oh, so they are Robin Hood stealing from poor OSS developers and website authors and giving to the rich owners and their $200-per-month users.
So they are either parasites or Robin Hood?
There are many documented cases of parasites being beneficial to the host in some particular situation while being harmful overall.
So your entire premise is false.
This is a fallacy and it is known as non-central fallacy.
https://www.lesswrong.com/posts/yCWPkLi8wJvewPbEp/the-noncen...
But who said it's harmful overall? That's a statement you are making not the people deriving value from it. The fact is many are benefiting otherwise they wouldn't pay that subscription.
Where did I make that statement?
Your point was that parasites give nothing back, thus AI companies can’t be seen as parasites because we get some usefulness out of them.
My only point is that argument is false, since a host can get benefits out of a parasite.
In number 4 of the author's definition of parasite, the parasite kills the host not benefit it.
I guess that analogy would be people stop sharing their thoughts publicly, which is as likely as people stop sharing on social media, very unlikely for many reasons.