On 2026-08-17 17:33, Aigars Mahinovs wrote:
> On Mon, 17 Aug 2026 at 17:20, Joerg Jaspert <[email protected]> wrote:
> 
>> You must have your head very deep down in the sand to not see all
>> the 
>> news about the various datacenter / ai companies planning ever
>> larger 
>> "factories". With extra coal/gas power plants just for them. Larger 
>> values of those for single data centers than needed to run whole 
>> countries on this world (and notan under-developed country even).
>> Lots 
>> of them companies entirely ditching their "green it" plans (however
>> they 
>> named their plans to go co2-neutral, whatever one may think about
>> those 
>> plans in general).
> 
> Companies bragging about crazy plans is about as realistic as Elon
> bragging about coast-to-coast full-self-driving ride no later than
> 2018 with zero human interaction (including charging). Or Elon
> planning for at least 20 million Tesla electric cars sold per-year by
> 2028. With production speed "mainly" concerned by air friction of the
> completed cars leaving the factory. Or how pets.com [1] planned to
> make revenue ten times larger than entrire worlds pet food and product
> supply volume at the time. Fairy tales for gullible investors.
> 
> 95% of those plans will never see completion. None of them have an
> actual business plan on how they *actually* expect to make profit on
> this given current efficiency and what people are actually ready to
> pay for AI. Taking those hyperscaler plans seriosly is a complete
> waste.
> 
> Meanwhile serious companies have provided 100% green datacenters for
> *ages*. You can run your own AI in 
> https://www.hetzner.com/dedicated-rootserver/matrix-gpu/ - completely
> green. And a lot of companies runnign their own AI do just that. All
> of the EU AI initiatives (including the internaltional ones that are
> training frontier-scale models) use 100% renewable power by default.
> 
> Focusing on a few, very loud, USA-based startups looses the real
> forest of AI that is being grown all over the world on completely
> different sets of values.
> -- 

I'm pretty dormant from Debian for quite some time, and I try to avoid
reading things here in general and even less reply when I feel partially
triggered but here I am, taking the bait. And this is nothing personal,
as I know you personally, but it is a reply on basically all your mails
regarding the entire AI debate.

I'll start with statement that I worked on AI before it became this
massive thing (in 2018) and in regards of just the software stack of
this tech, I'm more or less neutral, so this is not an anti-AI campaign
from my side.

Now with that out of the way - your emails regarding this topic are in
theory partially true (you throw some small facts and links here and
there which one can just easily find opposite ones that are valid) but
in practice they are bullshit.

AI consuming all the world information and spitting out some "answers"
is only valuable if you are an expert so you can decode and know when it
fails, and in majority of cases it is not even that useful. If this
trend continues for majority (which it will, specialized smaller models
will not really be something an average Joe is using), the average
quality of the said "answer" will be dropping down because most of newer
content will just be AI generated one, reinforcing its "answer" and not
really looking beyond/being innovative, and more and more average people
will be using it because only to them it will look like miracle, which
will further reduce the overall quality but the consumption will
increase for no real benefit whatsoever (on contrary, our collective, on
average, knowledge will decrease because there will (for most) be no
need to know something, to memorize something, to try out something, to
think about something or even be creative because that average answer
will either be more than good or good enough).

Second - talking about how AI can be used the "green" way, and that we
should ignore those who lead and absolutely dominate the marketshare
because they are "loud, stupid" or whatever the reasons is - is just
bonkers statement where it becomes hard to take your words seriously, if
not even slightly annoying. They are deciding what is the future, not
some obscure developer playing around with some models on some hardware.

Because of AI the demand for energy consumption increased multiple
times, it doesn't matter how efficient it will get, the demand will
continue increasing until the bubble burst, someone stops it or until we
as civilization collapse. People will get their homes destroyed because
AI will rob them of water, of the land where they belong, of the climate
where they live (some cities reported already an average increase of 4C
because of AI datacenters), of the energy and resources, but hey, let's
discuss and debate of all the possibilities that 99% of datacenters will
never even try to do. I will not even start going into ethics of how
this is used for privacy erosion, surveillance and war crimes (and no,
it doesn't matter that you say that you will not use it, any use of it
is used to reinforce their propaganda).

Hundreds of thousands of people lost their jobs because of AI. I guess
you don't care because it doesn't affect you. Even things like RAM and
disk prices increasing means that it got out of the hand for majority of
people, but it is hard to think from a privileged position why this is
bad.

AI is not going away, that ship sailed. It can be useful. But for it to
be good for civilization and humanity as a product, I would rather have
us vote against it and maybe, just maybe, we add to some effect that
might cascade and accelerate the downfall of this madness currently
ongoing. I personally think that whatever we vote here (and I think it
is a good time for this GR at the current state), we will need to revise
our stance in a year or two anyway.

As a last thing here let's have some comedy:
https://www.youtube.com/watch?v=kON2ZI2BNj8

Cheers,
Z

P.S. Fun fact, partially connected with this, Luddites weren't against
technological progress, they were actually highly skilled workers and
they smashed the machinery of those "owners" who exploited their workers
and ignored safety regulations. Sounds... close.
-- 
It's not the COST, it's the VALUE

Reply via email to