Etienne
Joachimthis is how you all fucking sound
by stillvreni : https://bsky.app/profile/vreni.bsky.social/post/3mms5baylac2p
ArchiveScribe
ケレムさん 🏴☠️ イラスト 
Nahuel Lofeudo
Chu 朱Lol. The nft made me actually laugh out loud. Thanks.
enoch_exe_inc@nahuellofeudo @joachim The difference, though, is that AI is far more than just chatbots and porn. There’s a lot of hype (i.e. Silicon Valley propaganda, as I like to call it), but it *is* a genuine field of study that has more applications than in selling you innovation theatre.
Kg. Madee Ⅱ.@enoch_exe_inc @nahuellofeudo @joachim ¯\\_(ツ)_/¯
machine learning is. LLMs are not, and they're the only thing this hype is about
machine learning is. LLMs are not, and they're the only thing this hype is about
Samy
Fairness4All@joachim I hope the bubble pops already. Spending 8 hours a day prompting and reviewing code is making me question my sanity and life choices
Jean-Baptiste Fidélia@joachim Yes, and as long as AI doesn't unionize, trigger a revolution and fight against AI exploitation, it's not going to stop.
😉
😉
Beggar Midas@joachim So what are you proposing there, John Brown? Gonna go all bleeding kansas affair over server farms? I mean, when you're done making a false equivalent between human suffering and a fucking datacenter? Jay-sus H fecking christ. Go touch grass. You've been cooped up too long somewhere & your losing touch.
Keith@Beggarmidas @joachim what to do is stop investing into AI companies, and stop buying shit produced by AI. Stop hyping the shitty text prediction machine that just sort of “works well enough” for a glitzy demo, then fails spectacularly when put into production use. And stop trying to shoehorn it into everything.
@keydelk @joachim There's more to it than that. Has it occurred to you yet that the pushback has become just as irrational as it's hyper promotion? Both sides are depending on emotive appeals that supersede rational analysis. That is not an accidental association. It is in fact the goal. Weaponized LLM's have been deployed to create pockets of mutually adversarial grievances. Primarily but not exclusively China & Russia originated.
https://stratcomcoe.org/publications/beyond-spam-bots-the-rise-of-ai-powered-disinformation-machines-and-the-imperative-for-strategic-response/342
https://stratcomcoe.org/publications/beyond-spam-bots-the-rise-of-ai-powered-disinformation-machines-and-the-imperative-for-strategic-response/342
Process ParsnipYes, it's very emotional to want to keep climate change from accelerating, clean drinking water available to living creatures, and electricity prices from spiking *all to put people out of work.*
Super weird, v irrational of us to be concerned about a known and quantifiable threat. Much amaze that we would object. Definitely the same as people who think they're dating autocorrect.
@ProcessParsnip @keydelk @joachim recent findings have already ruled out worst case scenarios. You've had cleaner drinking water for a century now than any of your ancestors have EVER enjoyed since our species first evolved. By relative to total water usage datacenters are only occupying .2% of *total* USA withdrawal. The numbers only seem big if you don't compare them to any other sector consumer in society. Which as you know is bad science. https://apnews.com/article/climate-change-future-worst-case-best-danger-cc7a20fba4f5b42ce33024e1b781e7c9?utm_source=copilot.com
Scientists revise global warming projections, say small gains not enough
Both the old best and worst case future scenarios in the fight against climate change are being jettisoned by the world’s top scientists as they prepare the next series of huge United Nations reports. That means scientists are acknowledging the Paris climate treaty goal of limiting warming to 1.5 degrees Celsius warming is now implausible without it getting warmer and then being artificially cooled. It also means the worst case scenario of 4.5 degrees Celsius warming — often used in scientific studies and once the path the world was close to — is also implausible.
@ProcessParsnip @keydelk @joachim The electrical consumtion issue is valid, but highly locally variable and too nuanced to discuss in this format.
Bottom line is hysterics aren't helping. Not you. Not any cause you care about.
@ProcessParsnip @keydelk @joachim ONe final note. If you've calmed down enough to listen. We are making progress on reducing & sequestering carbon emissions, but we're not going to meet Paris accord goals. That was always a mythical target. That's kind of what happens when you take one group of scientists priorities as gospel while excluding input from energy sector specialists, economists, buisness concerns, and industry reps.
"ONe final note. If you've calmed down enough to listen."
Please kindly fuck off.
@ProcessParsnip @keydelk @joachim Theres a lot of various people now working on the problem and there's a lot of genuine reasons to be hopeful. There was never going to be a workable 100% solution. Instead like other events its just going to take us a bunch of 1% & 2% & 3% solutions that'll collectively carry us over the finish line.
@Beggarmidas @ProcessParsnip @joachim possibly. I think the solution isn’t going to involve widespread use of LLMs though. It’s a terribly expensive, inefficient and unreliable technology. I think we’ll look back on them a bit like zeppelins or jet-packs; a technology that sounded cool, but had too many impracticalities to really be useful. And jetpacks and zeppelins never consumed more resources than a midsized nation.
@keydelk @Beggarmidas @ProcessParsnip zeppelin accidents killed more people than chatbots did, for now (and it’s not counting AI use in war)
@joachim @Beggarmidas @ProcessParsnip Also not counting the deaths from increased emissions from data centers (https://hsph.harvard.edu/news/analyzing-air-pollution-health-economic-risks-from-ai-data-centers/ ). And probably also excluding the deaths from suicide by the poorly compensated labors who have to weed out the CSAM and other internet filth from training data (https://restofworld.org/2025/the-ai-con-book-invisible-labor/ ).
Analyzing air pollution health, economic risks from AI data centers | Harvard T.H. Chan School of Public Health
Michael Cork has conducted analyses showing that air pollution from power sources at AI data centers could have serious health and economic impacts.
@keydelk @joachim @ProcessParsnip Compared to other imminent & existential risks we're facing datacenters are kinda chump change, wouldn't ya say? We've got a tyrant insistent on staying in the white house by any means, with all the inevitable holocaust ver. 2 that'll follow comes bundled with it. WW III ready to kick off anytime from five minutes to two years from now. A bleeding kansas style civil war fermenting in the, heh, wings.
As things go the datacentersis are piffling small change
As things go the datacentersis are piffling small change
@keydelk @joachim The thing about techies is they are typically alphalogicals who believe they can't be manipulated. Which of course means they are among the easiest TO manipulate. They don't tend to score high on EQ indexes, fail routinely at accurately identifying emotive states & triggers in themselves. Making them especially vulnerable to false consensus environments.
@keydelk @joachim
You don't have to agree or disagree. But you need be aware that this sort of thing is rampant. One in nine conversations you've had online, on average, in the last six months has been with a synthetic identity you mistook as human. No blame there. I have too. Just with far less frequency because I know what to be on the lookout for.
You don't have to agree or disagree. But you need be aware that this sort of thing is rampant. One in nine conversations you've had online, on average, in the last six months has been with a synthetic identity you mistook as human. No blame there. I have too. Just with far less frequency because I know what to be on the lookout for.
Jhooper@Beggarmidas @keydelk @joachim Now you just sound like an amalgamation of a centrist and a Republican. "Both sides are the same". ""That's what the enemy wants you to believe".
I'm assuming it never occurred to you that people have actively used these systems and are pointing out extremely obvious issues that are getting worse after every update? The costs are enormous, the companies running it have no solutions except to build bigger data centers and consume even more water and electricity with no solid goals on what progress even looks like.
And don't tell me "oh, the Russians or the Chinese will beat us at this!" They can't even get indoor plumbing to their rural populace, they don't give a solitary shit as to what sacrifices their population will have to pay. And for what? A barely working system that is guaranteed to make mistakes that will be more difficult to correct than if they were never used at all?
@jhooper @keydelk @joachim Lets start with I am not a winger in any sense. I regard them as binary failures to ramrod 19th century social policies into not-solving 21st century problems. I made that very clear in my bioblurb. You aren't identical in outlook or culture or character. But you do share many qualities you might find uncomfortably in common. You share problems with bandwagons mythologies. A tendency for ideological zealotry. A bad habit of trying to recast moderates as enemies.
@jhooper @keydelk @joachim Next. I have considered the technology carefully and explored it considerably. I regard it as immature, at best. There is a mania fueled by the last century of science fiction that has drastically overvalued its stock value & expectations. Not to say it's lacking use. I think it'll show great promise with some iterative reprioritization, especially for neurodivergent peoples. However it's now become the largest bubble in world history. AI and crypto have now
<CONT>
<CONT>
@jhooper @keydelk @joachim <cont> absorbed an inordinate share of world GDP investment. Not just the usual suspects either. Old money has gone all in on both, so when one inevitably tanks it'll pop the other. Taking an astronomical collective sum down a fiscal black hole triggering a world recession lasting decades. Paired with the high probability of WW III in the next 18 months it'll reorganize the geopolitical world. These are the last days of a golden age. Go try to enjoy them while you can.
@Beggarmidas @jhooper @joachim oh yeah, the world is going to shit, and AI is just the latest tanker of fuel on the fire. But going out and enjoying it is a lot easier without the slop being shoveled in our faces.
bit
Nathan Salwen@joachim More like, "guns are already here, there's no going back."
</𐒴ⴹⲦЕᏟᎻ>
canleaf08✅ 加拿大葉子 🏳️🌈⚧️🇪🇺🍁@joachim The same sadly with climate change. The same with RTO. The same with high rents. The same with loss of human rights. The same with surveillance since 9/11.
Levka
Cogito Ergo Disputo@joachim the internet is already here, there's no turning back
bluestarultor@Disputatore @joachim The Internet brought us pictures of cats. A.I. brought us pictures of cats with three ears. These things are not the same. XD
bor[x]a 🇵🇸 ボルシャ 
@joachim "Vi jobbar inte så här".
Wulfy—Speaker to the machinesOh yeah, let's fucking (your word) insult folks who don't huff your ideology
That's going to end well.
Hot Dog Water, roilingMy favorite part is them believing electricity is guaranteed in the smoky world of sweltering swamp-ass that they are incubating.
No power, no ai.
VANTABlack@joachim At least to me I think Child Labor is the best example.
Because "AI" is here doesn't distinguish the difference between what was "AI" 10 years ago to now. People don't have nearly as many problems with auto-correct or GPS even though those are "AI" GenAI is where the problem stems.
So like with Child Labor it would like if someone had a problem with Child Labor and the response was "Labor is here to stay" okay, but the problem is with ONE SUBSET of Labor not Labor in general IMO.
Lord Caramac the Clueless, KSC@joachim Machine learning isn't going away. Open source AI models anybody can download and run on a local machine for free aren't going away. There is no need for these humongous computing centres to run "AI" unless you are chasing the dream of having one universal AI system that does it all. Instead of training models to do everything (sloppily), we can train them to do get really good at only one thing.
The whole idea of using natural language to tell the computer what you want it to do is completely absurd IMHO, yet LLMs are trying to do that. We already have a lot of beautiful languages in which we can tell computers exactly what to do, they are strict and unambiguous, the computer will always do exactly the same thing when you combine the same instructions with the same data. The idea of talking to a computer like you talk to a human may be an old dream, but it is a very silly one. If you have a powerful machine, you want it to behave in a reliable and reproducible manner. Human languages aren't made for machines, they're made for upright walking apes with swollen mutant brains.
The whole idea of using natural language to tell the computer what you want it to do is completely absurd IMHO, yet LLMs are trying to do that. We already have a lot of beautiful languages in which we can tell computers exactly what to do, they are strict and unambiguous, the computer will always do exactly the same thing when you combine the same instructions with the same data. The idea of talking to a computer like you talk to a human may be an old dream, but it is a very silly one. If you have a powerful machine, you want it to behave in a reliable and reproducible manner. Human languages aren't made for machines, they're made for upright walking apes with swollen mutant brains.
Matt Jordan@joachim I just want to make it clear that we're making fun of that first speaker because of what they're saying. It has nothing to do with their five teats.
manic pixie donkey kong@joachim not so sure about this one chief I think this is rather minimizing of the unconscionable multigenerational oppression of chattel slavery
ECHAEA@Teratogenese that’s what I’m asking myself. The block button has been working overtime :)
Analog AI
MėMėNTō MōRi📍🇪🇺🇩🇪@joachim I’m saying eliminate the billionaires and above and CEOs they are the very worst minority and the main threat to humanity and the planet
Jessie Kirk@joachim I really am quite uncomfortable comparing chattle slavery with AE data centres. I assure you that the enslaved people had it waaaaaaay worse.
Winnie The Poodge@joachim its not what most people mean when they say "its already there" . Apples and oranges.
Most people i know mean well by reminding everyone that u cant wish away a box that has been openened.
No matter what happens, the technology itself will remain.
Dmitry Tantsur@joachim good that nothing bad in the history had ever come to stay. Otherwise, one could think you were cherry-picking convenient examples by using things that were actually over, right?
Wait.
None of these are even over.
@joachim now a serious answer to a less than serious take. "AI" is a bunch of matrix multiplications, it can't "be over". The idea of LLMs specifically fits into one rather short paper.
OpenAI may be making its last breaths now, but Gemini and Copilot belong to companies that are too large to fail. Anthropic will be saved by acquisition, most likely.
Open weight models are just files and won't disappear, and that's a good thing. We could use them to take power from the big tech. If we cared.