cowardly to use genai for ideation when you could simply go to sleep and have a dream

working from home

Looked into it and that got removed years ago, back in the GPT-3 days.

This app is 1000x better than any LLM chatbot

spiggiecode.github.io/rubberduck/

I'm also actively working on improving it so you can do wilder stuff without making the fill algo flip out

I hate Al and I refuse to use it Peter, if you don't use Al a bunch of super rich men will lose a lot of money I already said I hated Al Harry, you don't need to sell it to me

this is so beyond surreal. I can't believe they were just, out there. god that's so unbelieveable.

I didn't think the moon was just so... brown!!

"My stance on LLMs is this: if you freeze a turd in liquid nitrogen, you can use it to hammer nails but this doesn't change the fact that your tool is a piece of shit." - mycomancy

it always surprises me when people react really viscerally to the idea of destroying systems and institutions that either no longer serve us (and can't be reformed), or never existed to serve humanity in general. putting the destruction of a system on the table shouldn't be a radical position

look at palantir (and other AI) dipshits talk about the work they do - ie identifying, monitoring, & facilitating the extrajudicial murder of people all over the world - and tell me you defend a corporate structure's right to exist, or a "kill chain's" right to exist.

and MY work scandalizes you?

"A DDoS attack? At this time of day, at this time of year, localized entirely within your internal service?"

I was so glad to see the opposite in Monogame's contribution guidelines.

Again, it's interesting to point out that since LLM's stole literally every scifi book and convo ever written, they contain lots and lots and lots of training data about the statistical correlation of words like "AI" and "superhuman genetics" and "spaceship" and "immortality" and "I must scream."

I go full 😍 if I encounter a problem that requires some research to solve, looking up ds&a papers, wiki dives, etc to find out what cool solutions ppl have thrown at it before me. Just to have another solution tactic on my toolbelt

It's supposed to be anyway 😂
But yeah, when done well, the hard part is figuring out exactly what it needs to do and how to make small, easy to test steps to get there

I don't get how the dopamine hit of "finally getting working code out of the machine" isn't massively outweighed by the dopamine hit of "figuring out how to solve a hard problem and feeling your coding skills expand in real time". That's the dragon I want to chase.

Evergreen quote that I have trouble sourcing right now: "Any spec detailed enough to fully detail what a program needs to do is a program in and of itself." The problem with this is that genAI is not deterministic(or ever provably correct) so not a good basis for a programming language :p

Finding out that something that looked complicated to add to your program on its face could be pulled off by a 2-liner because of the work you put into architecting the code earlier is one hell of a drug, I'll tell you

"it doesn't matter what the code is all that matters is that it runs and produces the desired result"

liar. what you're saying is bullshit. you know it's bullshit and you're saying it anyway

One person tried to gotcha me, a person who literally codes purely for enjoyment, by asking the supposedly rhetorical question of "on your deathbed will you think back fondly on all the code you wrote and frameworks you used?"

Uh, yes, actually, absolutely. Thanks for playing!

How many fishers with 6 magikarps you think jump you on this thing

bsky.app/profile/pook...

"Why would you think this was the fault of vibecoding?!"

Why *wouldn't* you assume the dogshit-extruding machine is responsible for the dogshit all over the place

it's that easy

Why are you not here then ._.

;)

A literal curse on anyone who replies to a question with "did you ask AI?" or "AI said..."

You are sad pathetic people and just because you don't actually ever want to learn from humans who *know* things doesn't mean the rest of us are in your sad little anti-intellectualist boat.

A post that reads, "I'm spiting my addicting to ai by actively doing creative projects without any ai. ever. i hate ai and im now almost 4 months clean :)"

People are dunking on this guy but we really need to support and encourage people who make the decision to step away from AI. Like this is the goal!! For folk to realize and stop using it, even if they were to begin with!! This is a good thing!!

LLMs reviewing code is just the Slow Path with extra steps.

You’re still verifying a system that can’t verify itself, and now you have to verify its verification too. Recursively to taste.

The gap doesn’t shrink, the audit surface grows.

Ultimately, validating something made to seem correct but incapable of guaranteeing correctness (because it’s a specification gap) seems like the perfect cognitive torture.