n00bmind

Software is eating the world alive

[n00bmind @n00bmind]

@LubaRaphael @SinkingStarGame @vkrajacic define "sane browser" xD

[ThePrimeagen @ThePrimeagen]

the masculine urge to use polymorphism to solve all problems is real

I am realizing my last 20 years were a lie and the strategy pattern isn't the best thing ever...

Maybe a function with 200 lines and a switch statement is just easier by all means

[Anna Rettberg @aerettberg]

Generative AI is just a giant billboard that says “I didn’t care about making the thing, but I still expect you to engage with it. I bet you can’t even tell, you idiot. Fuck you.”

I’ve never felt more deeply committed to making handmade things than I do now. ✏️

[Doomer @DoomerOnSol]

@TIME the dot-com comparison only works if you include the part where it crashed first

[n00bmind @n00bmind]

@lemire beyond delusional

[n00bmind @n00bmind]

@arr_ohh_bee @thestanduppod so, you're talking about shell scripts? Casey is talking about *native binaries* here.. what am I missing?

[Ray @raysan5]

I use Notepad++ to write things, locally saved text files usually don't disappear. Simple and easy. Why make it so hard? 🤔

[n00bmind @n00bmind]

@lemire so then? everything an LLM spits is pure "AI interpretation"??

[n00bmind @n00bmind]

@arr_ohh_bee @thestanduppod you seem to think "packaging" an app is a pre-requisite to running it? it's not, and that's precisely Casey's point here, that you should just be able to provide your compiled binary and users shouls just be able to download and run it on their system

[n00bmind @n00bmind]

@omninomsky @thestanduppod besides the obvious technical issues of what you're proposing, good luck making linux go beyond "the 1 percent" with that attitude!

[n00bmind @n00bmind]

@johncrickett @nicbarkeragain well, that'd pretty much be the "spec" I was talking about :)

[n00bmind @n00bmind]

@johncrickett @nicbarkeragain not necessarily so if each team member is working under some kind of spec, whether they themselves proposed and advanced this spec or somebody else did.. i.e. say we're working on an engine and I'm writing infra, someone's doing rendering and another's doing tools

[Hyper Reel Vibe @HyperReelVibe]

@Vivek4real_ Spending two years and trillions of dollars building an artificial super-brain just to realize it's still cheaper to pay a 23-year-old junior dev in cold brew and pizza is the funniest possible outcome of the AI bubble. 🍕

[John A De Goes @jdegoes]

You missed the point of my post by a light year.

An LLM is not a pure RNG -- it's a statistical lossy model of all text written by humans.

So of course the odds of it producing 2 pages of text that, when interpreted by a thinking human or fed into a (actually de novo reasoning) proof checker, leads to a proof, are insanely greater than an RNG.

That is both true and irrelevant, because I used reductio ad absurdum to demonstrate my point.

Rather, the point of my post is that as of today, LLMs have been used by billions of people in an effort to solve trillions of problems. They fail in almost all of these problems--if what we mean by success is a solution as factually correct and free of defects as a mathematical proof must necessarily be.

People don't write papers on these failures. They don't talk about them on X like the Second Coming. They don't make the front page of Hacker News.

But the few times the linear string of tokens spewed out of an LLM's probability tables can be fed into a proof checker, or interpreted by a thinking human as a proof, well, you hear about that EVERYWHERE, and a subset of humans who annoy me greatly fall down and worship at the feet of the probability tables, believing them to contain the essence of immortal gods.

This is not what AGI looks like. Not even close. AGI does not look like OpenAI press release announcing their LLM "disproved" a mathematical conjecture.

It looks like those trillions of problems being solved by billions of humans **correctly**, because the AGI actually knows what that means. It looks like a million, a billion, or a trillion mathematical proofs -- limited only by how much electrical power you want to throw at the task. It looks like AIs being so obviously capable of de novo reasoning that no one talks about it and no one writes about it anywhere, at any time.

Until then, feel free to worship your LLM gods as you wish (freedom of religion and all that), but please do so with the knowledge you are operating under a fundamental delusion.

[n00bmind @n00bmind]

@DJ_CURFEW you guys need to stop with that crack you're smoking

[Petra Schurenhofer @PetraSchur]

@tparsi These scenes reminded me of this image (from 2014).

"Sderot cinema": Israelis bringing chairs to the hilltop in Sderot to watch bombs being dropped on Gaza, clapping and cheering when blasts are heard.

History did not start on 7 October 2023.

[David__Osland @David__Osland]

I'm worried that Trump will launch a war on Cuba, to distract attention from his war on Iran, which distracted attention from his war on Venezuela, which distracted attention from the Epstein files

[Naz @nazbowling102]

If there is one job that shouldbe entirely replaced with AI, its Tech CEO. The AI is more than capable of coming up with vapid ideas and pitching them to venture capitalists with technobauble word salads, the AI can easily generate the shitty base code that will eventually be replaced by experienced developers, the AI will even do a better job at pretending to be human than soulless abominations like Altman, Andreessen and Karp do.

[𝕭𝖆𝖒 𝕾𝖚𝖑𝖑𝖎𝖛𝖆𝖓 @BamSullivan]

Can't raise minimum wage because it will kill jobs.

Can't raise taxes on the bourgeoisie class because it will kill jobs.

Can't ditch oil because it will kill jobs.

But when these companies replace 50% of their workforce with AI, it's "sorry, that's just the way it goes."

[Abdulsamet Altaş @SametAlta10]

Burası bir Arap ülkesi değil. Burası Londra👏

Gazze için binler sokaklarda.