It's basically talking about the impact on the environment and which class of society will end up benefiting from AI. I thought the article was pretty clear, and a message we as a society should be discussing more, especially with the current state of things.
With that said, the article reads like it's been through an AI pass, and not as concise and direct as it could be.
"GPU compute powering crypto for the last decade..."
I keep hearing this lately. Also, that GPU compute has been re-purposed from crypto to AI.
I had the impression that crypto was powered by ASICs for the last... many years. And that was why GPU mining (let alone CPU mining) was not profitable
GPUs were used primarily for ethereum and to a lesser extent, other shitcoins. ETH switched to PoS in Sept 2022 and 18+ million GPUs were turned off almost over night. Nothing else has the liquidity to sustain a large operation any longer. ASIC's are used for bitcoin and a few smaller shitcoins, GPUs hadn't been used for bitcoin in many years.
GPUs used for mining don't translate well to AI at any large scale. It is all newly created (modern) hardware. Some miners who ran GPUs (and ASICs) have transitioned to running AI compute (CoreWeave, IREN, NScale, myself...), but that's not the same thing as taking the hardware and using it for AI. It is new hardware.
CPUs for mining crypto are a niche edgecase with very little liquidity and generally a bad idea, since it just incentivizes botnets.
I don't understand how you could compare crypto, which has no utility and no fair value, with AI which is used by hundreds of millions of people every day to achieve real outcomes.
Not OP, but I have, and I'm not impressed. Perhaps it would be good if you are doing things in the training data (like churning out yet another bland website). But if you are doing things that are not well represented in publicly available code, or in unusual languages, then LLMs are not great currently.
They are great for finding the right ffmpg command line arguments though, and not having to deal with that mess must count for something.
“You ingested the vast majority of available human knowledge to build your product, and intend to change the entire world with it, and in fact are doing so incredibly rapidly at a rate that dodges scrutiny. Could we know more?” Would be the expanded version I guess
The OPs framing is better. Where does anyone say there’s a requirement to share data as demanded because you used a publicly available resource. These companies have created something world changing that everyone uses now, that you can get for free or for a very small amount of money. They certainly don’t need to indulge anyone’s nosing or whatever this is, the argument is ridiculous.
No, AI tokens aren’t generic, they by definition cannot be used as a currency because they aren’t fungible. Generated tokens have completely different value depending on their meaning. You can spend 1M tokens doing nothing but wasting time and energy, then spend 100K tokens that bring a lot of value. It’s even weirder because we don’t have a way to know the ROI of tokens in advance
Quite a few of the usual suspects are involved in Crypto, NFT’s, “Metaverse”, and now AI. You’d think people would figure it out, but there’s the old Upton Sinclair quote I guess.
How much you spend on something does have an indication on what the utility is for you. Though people spent on all the other hype things so at the time those things also had some utility for that user.
the metaverse went from being about owning your identity and digital possessions to mostly being about buying jpegs. once collectibles and nft speculation became the headline, it was hard to explain why any of the underlying technology mattered. after that, most people just associated the whole thing with speculation rather than the original vision.
which is a shame, I still feel like VR is full of untapped potential
Yeah - they not just VRChat but many other VR projects are alive and very active, doing verz innovative and almost marvelous things most people don't know anything about.
It might not be a mainstream thing any time soon or ever, but that is not necessary as long as the people invovled woth VR are having fun. :)
Does this article being unreadable prove that it was at least written by a human? Frustrating tradeoff.
Not the solution you where hoping AI to solve?
This is a pretty confusing article. It seems to string together some loosely related facts to make the current AI boom look bad.
It is confusing. But I enjoyed the read, it’s refreshing to read a string of rants from a human rather than something perfectly polished by Opus.
I didn’t read it, primarily because of the tulip thing in the title which seems anything but a refreshing take.
One sentence in, it mentions that the tulip reference is from a book on botany and perhaps not just a reference to Wallstreet and bubbles.
Nice. What other stuff haven’t you read?
I also enjoyed it. Something about the writing really made me enjoy it, it was so different from the llm text i read all day long
Unless you subscribe to the view that This Time It's Different(tm), they seem pretty pertinent.
tulips do not have intelligence, tokens do
That's an unintelligent thing to say about tulips.
Agreed. It has nothing to do with either tulips nor mania. Just some random musings.
It's basically talking about the impact on the environment and which class of society will end up benefiting from AI. I thought the article was pretty clear, and a message we as a society should be discussing more, especially with the current state of things.
With that said, the article reads like it's been through an AI pass, and not as concise and direct as it could be.
"GPU compute powering crypto for the last decade..."
I keep hearing this lately. Also, that GPU compute has been re-purposed from crypto to AI.
I had the impression that crypto was powered by ASICs for the last... many years. And that was why GPU mining (let alone CPU mining) was not profitable
Bitcoin mining is powered by ASICs. Random shitcoins are not.
Random shitcoin mining surely isn't a large fraction of industrial GPU usage?
Not anymore, but around 10 years ago it absolutely was, because for a short period you could literally print money with them.
Ah, but any 10 year old GPU now is e-waste, not useful LLM hardware. So it might as well have been mining ASICs.
GPUs were used primarily for ethereum and to a lesser extent, other shitcoins. ETH switched to PoS in Sept 2022 and 18+ million GPUs were turned off almost over night. Nothing else has the liquidity to sustain a large operation any longer. ASIC's are used for bitcoin and a few smaller shitcoins, GPUs hadn't been used for bitcoin in many years.
GPUs used for mining don't translate well to AI at any large scale. It is all newly created (modern) hardware. Some miners who ran GPUs (and ASICs) have transitioned to running AI compute (CoreWeave, IREN, NScale, myself...), but that's not the same thing as taking the hardware and using it for AI. It is new hardware.
CPUs for mining crypto are a niche edgecase with very little liquidity and generally a bad idea, since it just incentivizes botnets.
https://en.wikipedia.org/wiki/Template:Cryptocurrencies
Major cryptocurrencies are based on Proof of Work or Proof of Stake. There are other consensus algorithms.
https://en.wikipedia.org/wiki/Scrypt#Introduction
Scrypt has been introduced to stave off ASIC and FPGA solutions.
I don't understand how you could compare crypto, which has no utility and no fair value, with AI which is used by hundreds of millions of people every day to achieve real outcomes.
Crypto bros claim that crypto is super useful, revolutionary tech too. The proof is in the pudding, and AI has yet to show any real utility.
> and AI has yet to show any real utility
Ummm ... have you not tried using LLMs for coding work?
Not OP, but I have, and I'm not impressed. Perhaps it would be good if you are doing things in the training data (like churning out yet another bland website). But if you are doing things that are not well represented in publicly available code, or in unusual languages, then LLMs are not great currently.
They are great for finding the right ffmpg command line arguments though, and not having to deal with that mess must count for something.
> But if you are doing things that are not well represented in publicly available code, or in unusual languages, then LLMs are not great currently.
Have you tried since November 2025?
We created the content their systems are built on, yet they won’t even share how much energy each query uses, how much water goes to cooling
"You read my blog so now show me your utility bill" is... an interesting argument.
“You ingested the vast majority of available human knowledge to build your product, and intend to change the entire world with it, and in fact are doing so incredibly rapidly at a rate that dodges scrutiny. Could we know more?” Would be the expanded version I guess
Take advantage of open information, provide little yourself. Seems like a consistent enough line of questioning.
The OPs framing is better. Where does anyone say there’s a requirement to share data as demanded because you used a publicly available resource. These companies have created something world changing that everyone uses now, that you can get for free or for a very small amount of money. They certainly don’t need to indulge anyone’s nosing or whatever this is, the argument is ridiculous.
Aren't human skills the tulips? That's why we're recording all kinds of movements; tokens become generic execution currency.
No, AI tokens aren’t generic, they by definition cannot be used as a currency because they aren’t fungible. Generated tokens have completely different value depending on their meaning. You can spend 1M tokens doing nothing but wasting time and energy, then spend 100K tokens that bring a lot of value. It’s even weirder because we don’t have a way to know the ROI of tokens in advance
In the meantime tulips prove math theorems that humans couldn't solve. Ridiculous article title (I wouldn't read it)
this article reminds me of Greta for some reason .. wonder what she's upto these days ?
Or the 1990s Beanie Babies.
Or hell, 2 years ago with "those fucking apes", aka NFTs.
Or, the Meta-Verse. Yeah, that really took.... Off.
Quite a few of the usual suspects are involved in Crypto, NFT’s, “Metaverse”, and now AI. You’d think people would figure it out, but there’s the old Upton Sinclair quote I guess.
Tulips, NFTs, and the Metaverse all revolved around exclusivity and collectibles. Art with no real utilitarian value.
I pay OpenAI $100/mo, twice what I pay for my mobile phone service.
It’s a useful service that I use every day.
I have never even heard of anyone in my social circle expressing even a mild interest in tulips, NFTs, or the metaverse.
This bubble has much more in common with the rail road boom and the 2000 fibre build out bubble.
You start with "real utilitarian value" and your example is... how much you pay for it. That sounds more like a drug habit.
How much you spend on something does have an indication on what the utility is for you. Though people spent on all the other hype things so at the time those things also had some utility for that user.
I think their point is that is pretty cheap for such a useful service. The opposite of a bubble price.
How did metaverse revolved around exclusivity or collectibles?
the metaverse went from being about owning your identity and digital possessions to mostly being about buying jpegs. once collectibles and nft speculation became the headline, it was hard to explain why any of the underlying technology mattered. after that, most people just associated the whole thing with speculation rather than the original vision.
which is a shame, I still feel like VR is full of untapped potential
Yeah - they not just VRChat but many other VR projects are alive and very active, doing verz innovative and almost marvelous things most people don't know anything about.
It might not be a mainstream thing any time soon or ever, but that is not necessary as long as the people invovled woth VR are having fun. :)