this post was submitted on 02 Jul 2025
271 points (94.4% liked)

Fuck AI

3516 readers
495 users here now

"We did it, Patrick! We made a technological breakthrough!"

A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.

founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[–] theunknownmuncher@lemmy.world 19 points 2 weeks ago* (last edited 2 weeks ago) (3 children)

So firstly, shut up, nerd. You’ve clearly never worked a corporate job and don’t understand the compulsion to in-house nothing.

I WISH. Would have saved so much headache from rolling our own framework for the millionth time instead of using standard solution that already exists... Tell me you've never worked in corporate software without telling me you've never worked in corporate software.

Secondly, approximately 0.0% of chatbot users will ever run an LLM at home.

Wut. Individual models on huggingface have monthly downloads in the hundreds of thousands... Just Deepseek R1 was downloaded 550,000 times in the last month alone. Gemma 3n was downloaded over 120,000 times in the last 7 days.

Thirdly, it’s slow as hell even on a top-end home Nvidia card — and the data centre cards are expensive.

They literally run well on mobile phones and laptops lol.

The author is desperately grasping at straws at best, intentionally making up false claims at worst... either way, they aren't qualified to write on this subject.

[–] Blaster_M@lemmy.world 4 points 2 weeks ago (1 children)

LocalLMs run pretty good on a 12GB RTX 2060. They're pretty cheap, if a bit rare now.

[–] Valmond@lemmy.world -1 points 2 weeks ago (1 children)

So 12GB is what you need?

Asking because my 4GB card clearly doesn't cut it 🙍🏼‍♀️

[–] Blaster_M@lemmy.world 0 points 2 weeks ago (1 children)

4GB card can run smol models, bigger ones require an nvidia and lots of system RAM, and performance will be proportionally worse by VRAM / DRAM usage balance.

[–] theunknownmuncher@lemmy.world 3 points 2 weeks ago

require an nvidia

Big models work great on macbooks or AMD GPUs or AMD APUs with unified memory

[–] CrayonDevourer@lemmy.world 2 points 2 weeks ago* (last edited 2 weeks ago)

Not only that, but the "They spent $2 for every $1"...

Yeah, that doesn't account for the $20/hr middle management job that they don't have to pay for another 5 meetings this week with a poor powerpoint presentation. Of course they're spending to make their company as competitive as possible. And sure, AI hallucinates, but have you seen MAGA fanatics? I've seen a 1b parameter model hallucinate less than that. And they have jobs -- and VOTE...

[–] faberfedor@lemmy.world 0 points 2 weeks ago (1 children)

Yeah, screw this guy. I came across him on YT. He goes out of his way to paint all AI as if it were an ineffectual, hype-only, resource hog with absolutely no redeeming qualities being foisted on us by EvilCorps everywhere who are simultaneously too dumb to know that it’s a farce.

What told me he was just going for eyeballs (gotta admit, his marketing is good! I see this mofo everywhere these days) was a YT post about a grift — that predates AI — which used AI and he blamed AI for the grift! Not the human, but the technology.