Comment by compass_copium
17 hours ago
I'm a hobbyist IT enthusiast, not a professional computer guy, but from my point of view there's a big difference between a bunch of servers hosting something people want, like Netflix, and a warehouse full of GPUs running around the clock to make Lobster Jesus or whatever.
It's also a proxy for people's feelings on AI--essentially the only way people have to fight back against the new technology that makes eternal garbage and is supposedly coming for their jobs is to push for banning "data centers."
Totally agree that it's just a proxy for people's feelings on AI, but my Lobster Jesus is much more important to me than a Netflix show! All of the datacenter usage (for inference) is just serving something people want. Time has already shown that the datacenter for training is also just creating something people want.
> All of the datacenter usage (for inference) is just serving something people want.
I agree with this in principle, but I wonder how asymmetric the usage is. When I hear about developers dropping $200/mo (subsidized!) on AI coding, and having multiple agents on supergigaultramax effort churning on stuff for hours... Versus your average person maybe asking a few chatgpt questions every day or week... IDK.
Like others have said, yes, OpenAI probably isn't literally going to sell you more than $200 of electricity for $200/mo. But data centers get industrial rates, and $200 of power is still a decent bit of power just to be churning out slop! (It's about a monthly home bill and EV charging cost combined) And some companies spend way more than that on API pricing for tokens... It's just hard to get a handle on where the usage is actually going.
I just wish we could pass a common sense law to require them all to build renewables. It's literally electricity, it's the easiest energy to decarbonize.
i think this sentiment comes from a deep misunderstanding of how our modern society is the product of transitive satisfaction of wants. the developers churning out agent work on supergigaultramax effort are fulfilling the transitive preferences of people like supply chain managers, etc. who are in turn fulfilling the transitive preferences of a school teacher for cheap groceries, who are in turn....
This is how economies work and the alternative strategy of enumerating morally correct wants/needs and assigning the correct price from on-high has been a demonstrable failure. It is a cognitive failure of humanity that this strategy is so intuitively appealing to us.
7 replies →
Lobster Jesus FTW.
> there's a big difference between a bunch of servers hosting something people want, like Netflix, and a warehouse full of GPUs running around the clock to make Lobster Jesus or whatever.
ChatGPT has been the most downloaded app every month for a year and a half.
It's gotta be something. The actual apps I use to do work on my phone either were included in the OS or downloaded long ago.
And yet, AI is broadly unpopular:
https://techstrong.ai/features/public-sentiment-sours-as-ai-...
The fact that people try out ChatGPT does not change that. People are complicated!
That doesn't mean people want LLMs to replace their jobs so they can (hopefully) go on welfare, and to have ever more data enters built with that goal in mind.
They don't want them to replace their job, they want them to replace other peoples jobs.
2 replies →
There's also different building pressures. Before, companies would want to build / power / maintain the smallest amount of data center hardware they could while still serving their needs. Overprovisioning just meant worse opex number.
Now, because supply is so constrained relative to demand, there's no limit. More is always better, which leads to a crazy amount of construction (as its negative effects).
How is there a big difference between those two things? People want Netflix, people also want the output of AI whether that is lobster Jesus or something else. This seems more like substituting what people want, which clearly includes AI, for what some people think others should want. I can guarantee you there are people who don’t want Netflix and think it’s a waste of resources that encourage people to binge TV, become a couch potato, and end up being drags on the healthcare system. Should their preferences override those who do want it?
Democracy, even at its most ideal, is often about overriding preferences. Netflix helps to float an entire industry of creators and a tangible pop culture. It's not entirely clear (or softened, maybe not clear to communities) what AI build out floats currently except AI companies and the tangible negatives of generative AI.
Look, I'm not entirely convinced AI will be as revolutionary as some believe but it's still pretty clear it's able to provide capabilities we've never had before.
Dismissing it entirely is pretty hard to justify.
I think that potentially the difference is that, while people do in some sense want AI, they don't want it in the same way that they want Netflix. I'd compare it to Amazon: it perfectly possible for people to like the ability to like 2 day shipping while also liking that the site is flooded in cheap crap.
So maybe it is that people like AI in certain domains but HATE the way that it is being used overall?
The specific examples are not the point. Obviously some people hate Netflix and some people love Lobster Jesus. There is a rising sentiment against AI:
https://techstrong.ai/features/public-sentiment-sours-as-ai-...
>Should their preferences override those who do want it?
This is rather the point of government, balancing the competing amd incompatible desires of different groups of people for the benefit of everyone. Yes, if a community decides they don't want a GPU warehouse, using whatever means that community uses to make decisions (referenda, petitioning representatives, etc.), their desire should overrule the desire of people to have one. That's what society is!
> How is there a big difference between those two things?
Netflix just serves existing data so their biggest need is data storage ala CDN for closer retrieval to the user. Very low requirements. Creating sweet baby lobster jesus requires incredible amounts of computer/power/cooling. The requirements are totally different. Yes, both ultimately deliver pixels to consumer's eyeballs, but how they get there are so different that it's laughable that people cannot see the difference.
The point is they don't use nearly the same amount of energy.
It's weird how this point is being ignored, or that you even have to say it.
The conflating of a data center and an AI/GPU farm is disingenuous at best, and the AI people know what they are doing. None of the "traditional" data centers are being powered on site with gas turbine generators. New GPU farms are also being designed with much larger footprints with much more needs from the public utilities.
You really think Lobster Jesus is what most of this is doing?
It's doing code, analytics, training, research, operating services. The silly slop stuff is a small rounding error of usage.
Ah, the job-killing stuff, right. Don't forget selecting Iranian girls' schools as targets for missile strikes, AI is doing that, too.
[dead]