vrighter

well it's because the AIs are trained on what they find online. And what they find online is now ai generated bullshit.

a hashset was used... then ".Where(x=>x == key).Count() == 0"

i got over 800 this way. O(n^2) everywhere and n+1 queries

i have a nintendo in only a couple of kilobytes of x64 instructions!

cloud native emedded devices are an oxymoron

just put it in the square hole!

SP3

It has delivered negative financial payoff here. Everyone is using it, nobody knows how anything works anymore, and I am a caveman for refusing to NOT understand the code I write. And they still have to pay the developers. And they still have to pay for the AI. They pay more for an inferior result. And then of course microsoft changed the billing contract on them... :D

you mean the flying car that was demonstrated on stage as already literally not what he was talking about right in front of it? The one that clearly showed that the stuff that should have been on the underside just.... wasn't? And the demo was different cars doing the different stunts? The one with a 5 second fuel tank that takes 7 hours to refill? That flying car?

Lol I found that presentation so hilarious.

what about lost interest? 7 years' worth

so.... it's still a several days endeavor if you want to write a parser. It's only 20 mins if you have it written for you.

I still fail to see what jev does. ALL llms produce a probability for all tokens. All jev did was stop there. Its function could be replicating by adding one "break;" statement in the right place in any other LLM.

doing anything slows cognitive decline. Using your pbrain is what slows it down. Not the specific thing you're doing. So does knitting, playing an instrument, and a lot of other things.

isn't it obvious why? One is designed to process language. The other is designed to process images. duh!

"chosen by fair dice roll"

ALL llms work exactly like that.

I could be confident that I God 2.0. Doesn't change the fact that I am just a lowly person in front of a keyboard.

I could be 100% confident the earth is flat. Doesn't change the fact that it is round.

I could be 100% confident of a lot of things. Confidence says absolutely nothing.

and what are the savings in this case? the size of the decompressor must be included with the size of the compressed data. and "i lossily compressed a few kilobytes into a couple of gigabytes" sounds a lot less impressive

only if they couldn't brute force their private keys. one of which i can fully recite from memory right here: 4

why not? a couple of raspberry pis could probably serve all of them

why do they call them white noise machines? Why not just noise machinesc or an actual colour? If they did generate white noise, nobody using them would be able to sleep

which one of these lets me choose the hardware and promises to let me do so indefinitely? I run sailfish os, for example, and hsbc decided to cut off my internet banking for it

that was good advice back then. It is not good advice in 2026

because they're all becoming the same. Just like cars pretty much have the same shape nowadays

it still needs the full semantic "understanding" of natural language. So I don't see how a smaller model one-shotting a single token could be any better than other bigger models

since daytime running lights have been proven to have a positive effect on accident rates, some cars just turn them on and just don't give the option of turning them off anymore. Mine does that and so does my wife's, for example. You can't run the car with zero lights

ah of course the 2026 model comes out (lol yeah right) in 2028.

because native should mean using the tools provided by the os. Not using a reinvented wheel which defeats the entire point

that's great! cross platform means it doesn't fit in anywherec not it fits in everywhere

aren't decentralized systems using the same infrastructure and are also relying on ip to route around missing routes? Just like centralized ones?