Behold, an outcast so slimey he got thrown out of office and shunned.

…And he’s trending on Twitter. He’s getting clicks and into people’s feeds.

This is why whe don’t use Twitter. Don’t feed the trolls.

Not sure I understand you but I think I get it?

Like, most of what AI bad is the cultism and corporate shit. Like literally shaving 2% off costs to drain a town’s water or something, or proselytizing scaling up transformers while ignoring the efficiency/scaling papers that keep coming out (because that would break the Tech Bro grift).

…At the same time, the absolute energy cost is ridiculously overstated compared to, say, global aluminum or steel production.

And then you have the ridiculous politicization. An example I often cite is a TV series that was ‘fan remastered’ and (as one component in a long chain) upscaled with an oldschool GAN that cost peanuts to train. Beloved years ago, but all of a sudden the fandom hates it because it has something to do with ‘AI’.

…At the same, you can’t ignore how irresponsibly its presented, where these companies are making pennies from spam/slop literally destroying everything. It’s quite reasonable to say “The idiots making this put no effort into it” or “I just don’t like it, yuck” when 99.99% of user-visible AI generation is slop/spam.

I use this, it’s good.

See this comment for math and specifics: https://lemmy.world/post/38090104/20233592

But the TL;DR version:

• Launching anything into space is heinously expensive. And CO2 emissive.

• With very generous math, you’d need a radiator like a mile across to cool a space data center, but practically? Larger.

• Datacenter hardware is unreliable and goes obsolete quickly, and any kind of maintenance in space is basically cost prohibitive.

• There are other smaller yet still crippling engineering challenges, like bit flips from radiation (which gets move severe as lithography shrinks; look up Nvidia’s research on this), assembling large structures in space reliably, cooling loops for such gigantic structures, and extremely difficult/expensive networking (with distinct issues in LEO or geosynchronous).

And most of all… Solar is dirt cheap on Earth, compared to that.

So is just sticking a pipe in the ground for a geothermal loop, or ambient radiative cooling. We literally have tons of mass to dissipate heat into for free, instead of having to radiate it thermally, yet that’s too expensive for ground data centers, apparently.

That’s the joke.

It’s like saying “air conditioning is difficult” and proposing “I know! Let’s live under the Antarctic ice sheet!” That’s not hyperbole. It might be more practical, actually, as getting mass there is waaaay cheaper…

To answer your question, it’s very hard at that scale/temperature: https://projectrho.com/public_html/rocket/heatrad.php

The really hilarious thing is evaporative cooling (that takes so much water) is simple penny pinching over a closed loop system. That’s all.

…Yet Bezos and Musk are talking orbital datacenters?

Pick a lane?

Speaking as one of those “I self isolated and stop responding to friends” folks years later:

Definitely reach out. Emphasize not to worry about the ghosting; I guarantee she’s literally worried sick over it. Do it, even if you have to do it through family. They’ll all be glad.

Pasty faced millennial here.

Also EE dropout (working on that).

I hope I am as enlightened and put together as you when I’m 50. Cheers.

But I do have one answer for you:

I lost hope for this nation when Trump was elected this last time. I couldn’t understand how people could not see the obvious.

Facebook.

My ‘moment of clarity’ was when I saw a local news station running clips of college girls’ enthusiasm for Trump, just after the election. The sentiment was along the lines of “He’s so strong!” or “I really feel like he gets us, you know?” and almost directly quoting some meme’s I’d seen.

And seeing die hard libertarians I know swallow Trump’s accumulation of executive power, via their perception of events, drove it home.

…Walk the street, and look around. I’m using ‘Facebook’ as a catch-all here: algorithmic feeds and influencers are warping reality for smart folks, both ways. Every generation is caught up in it, but I feel the age extremes (Boomers and Z) are most vulnerable here.

Transhumanism is cool though! I adore the OA universe, for example: https://www.orionsarm.com/xcms.php?r=oaeg-front

…I think its important to keep ‘machine learning as a neat tool’ distinct from cultish ‘AI’ futurology. Hence some folks might interpret this questions as “why aren’t you using AI yet, bro!? AGI’s inevitable, get on board!” The latter is what a lot of news/advertising is blasting.

I think you double posted OP.

Fuck Facebook, Twitter.

…That’s the big one. Folks will always get in cults, but never has it been so supercharged and weaponized by algorithmic machines that will, literally, hook into peoples’ brains to make a penny on mis/disinformation: https://www.reuters.com/investigations/meta-is-earning-fortune-deluge-fraudulent-ads-documents-show-2025-11-06/

4GB VRAM

Mmmmm… I would wait a few days, and try a GGUF quantization of Kimi Linear once its better supported: https://huggingface.co/moonshotai/Kimi-Linear-48B-A3B-Instruct

Otherwise you can mess with Qwen 3 VL now, in the native llama.cpp UI. But be aware that Qwen is pretty sycophantic like ChatGPT: https://huggingface.co/unsloth/Qwen3-VL-30B-A3B-Instruct-GGUF/blob/main/Qwen3-VL-30B-A3B-Instruct-UD-Q4_K_XL.gguf

If you’re interested, I can work out an optimal launch command. But to be blunt, with that setup, you’re kinda better off using free LLM APIs with a local chat UI.

Really raises the question of how profitable these scam ads have become.

It does, doesn’t it?

If Meta is literally making billions from ads, how much are the scammers buying them making?

People keep coming back to these obviously booby-trapped websites to claw at a thin veneer of simulated friendship because they’ve got nothing better to do with their lives.

Eh, part of it is FOMO. I can’t blame older folks if all their friends and kids and grandkids are on (say) Facebook, and that’s the most advanced means they have of communicating with them that they’re capable of using. What are they supposed to do?

Yeah, exactly… In other words, ‘general server buildout.’

That’s (as I understand it) a misconception.

Apple attaches their laptop RAM the same way all smartphones do. It’s a wide bus with LPDDR, which makes it an unusual configuration amongst laptops, but it’s technically conventional. And relatively cheap.

AMD’s Strix Halo chips are the same. Apple could use LPCAMM to make the memory upgradable if they wanted, they just… don’t.

When we talk ‘packaging’, we’re talking putting chips on advanced substrates with denser wires than one could possibly get on a motherboard (or a ‘mini’ motherboard which is kinda what Apple/smartphone RAM is packaged on), stuff silicon fabs have to do:

https://www.tsmc.com/english/dedicatedFoundry/services/advanced-packaging

And HBM falls into this bucket. The way its hooked up to the processor is physically different than PC RAM sticks, or Apple’s RAM. This is mostly not done on consumer stuff because its very expensive, and most of TSMC’s advanced packaging production capacity is reserved for server stuff.

This feels like how Bitcoin was neat for a time, before 99% of the space became pyramid schemes.

…And if “AI” is on the same trajectory as crypto, that’s not great, heh…

The key is which model, and how.

For the really sparse MoEs, you might be better off trying ik_llama.cpp, especially if you are targeting a ‘small’ quant. But the dense Gemma models (as good as they are) are probably not the best choice for 8G RAM these days.

Eh, maybe that’s fair. Especially this:

doesn’t automatically mean he’s never been an engineer.

Maybe he was quite knowledgeable… On the other hand, I thought his old ‘Mars Colonization’ talk was bonkers too. I can’t remember what year it started, but it sounded like he had little understanding of the impracticality.

Ah. Yeah. And its on the fab to do that.

I always though it’d be cool for CPUs to switch to packaged RAM, too. Samsung apparently tried to do it with Wide I/O for mobile ARM stuff, but it never caught on.