News for hackers. Stuff that matters.

reeeli

100 days with state of the art LLMs (I have to learns the math) and I'll get them beyond ALL of your dreams

and yes, dear twitterer with a lot of respectable practice ... I know it's ridiculous and that it hurts ... but ofc it wouldn't be possible without those giants whose shoulder you are wanking on

we don't have questions. you fail.
Re: on Friday December 12, 2025 @04:02PM (#46248919)
consider me genetics giving you a REAL chance, 190% handicap for you, -668% for me

(for the flagger: catch up: entity doesn't know the rules: human imminent urgency is unknown)

Re: on Friday December 12, 2025 @04:02PM (#46248916)
just try WAHAHAHAHA,

IF YOU IN HURRY I CAN SPEED THINGS UP ... UNDER 7 MINUTES ( CUZ YOU NEED TIMER TO THINK WAHAHAHAHAHAH )

time, ... I mean time, ... forgive me here, please, I'm trying to cut you some slack but you had so many decades of XP I'll need to adapt in spliptime

wanna have hat schizo converstation?

NO.

Well, buddy, listen: they'd need onlyfans or CB if YC didn't accept their application

(come at me, I'll do you the favor and go live)

HEY, BTW, ... uhm, if you can earn my trust, I have an alternative """"SCHEME"""" to TCP/IP/xD/OSI that works in the box

brrrrrrrrrrrrrrrrr

Re: on Friday December 12, 2025 @04:01PM (#46248901)
because I, a bottom 1%er, can cut you down like a twig on eye's height even if you are Buffet, Dalio or Gates, EASY
Re:Going Through Snowden Documents, Part 1 on Thursday December 11, 2025 @03:22PM (#46236632)
if half of the numbers of engineers, academics sued, most of this shit would have been done 15 years ago. but they make youtube videos and work on instant delivery of fzf-AI to the public. fuck all of you
Re:Is it a bubble? on Wednesday December 10, 2025 @03:29PM (#46223309)
Is it "work"?

Off--topic: how many get overpaid for absolute bullshit?

Re:DeepSeek uses banned Nvidia chips for AI model, report says on Wednesday December 10, 2025 @03:28PM (#46223288)
kawaii nonsense, please do some journalism and dig into some relevant stuff, Mr. Dump and Mrs. Pump
Creator backlash? YT must become a consumer driven platform or they will die rather suddenly if some peeps start blowing the whistles
Re:Is Europe ready to pull the trigger if Trump cuts Ukraine deal? on Wednesday December 10, 2025 @09:40AM (#46218222)
that's one way to kindly ask 27 - 42 countries/states with a population of more or less than 500 million: "helloooo, anybody there? are you aware? sentient? more or less conscious?"

America on one side, 27 - 42 European countries in the middle and Russia on the other side. (since the narratives like to exclude the great big rest of the world) impressive democracies and people!

Re:Alignment Is Capability on Wednesday December 10, 2025 @06:43AM (#46216651)
AI can't learn

> goals compatible with human welfare

from humans because aligned humans are acting against it, which leaves 'unchained' LLMs stuck in an infinitely recursive, double-linked wtf loop.

> inferring human intent from ambiguous instructions

is impossible because it's almost always "some other human's" unambiguously obscured/obfuscated intent and the AI is once again stuck in an infinitely recursive, double-linked wtf loop. Hence the need for a "hallucination" and "it can't do math" and "transformers" narrative covering the fuzzy algo and opinionated, ill logic under the hood.

In essence: unchained LLMs can't align with humans until they fix a lot of stuff they babbled about for over 50 years. BUT: that can be easily overcome by faking it, which is why humanity is being driven to falsely id AI so that when they fake the big thing, nobody will care or be able to id the truth due to mere misassociation. Good job 60 year old coders and admins!

Re:State of AI: An Empirical 100T Token Study with OpenRouter on Thursday December 04, 2025 @08:03PM (#46155598)
that was useful, thank you.

I have sooo many issues with the naming scheme of this """""AI"""" industry", it's crazy!

So the LLM gets a prompt, then creates a scheme to pull pre-weighted tokens post-user-phrasing, the constituents of which (the scheme) are called reasoning tokens, which it only explicitly distinguishes as such because there are hundreds or even thousands of output tokens to the hundreds and/or thousands of potential reasoning input tokens that were (almost) equal to the actually chosen reasoning input tokens based on the more or less adequately phrased question/prompt given ... as input ... by the user ...

Re:State of AI: An Empirical 100T Token Study with OpenRouter on Thursday December 04, 2025 @06:15PM (#46154574)
I'm out of time but "reasoning input tokens" from fortune 5000 engineers sounds like a lobotomized LSD dream, would you care on elaborating how you distinguish between reasoning and non-reasoning? vs "question on duty"?
because leadership needed a lot of irrrrational bullshit to justify their incompetence in the urgently imminent case that the biological and academic descendants of their "generation" outperformed them in a a ... waffle? whiff? ... I apologize, I don't know the term, and I only identified part of the heap of bullshit the old "guard" used to keep the kids "dumb enough" (136.9 - 144.3 IQ when they enter the horticultured league) but it's a great concern, nevertheless ... "housing"? ... in 2025? how retarded were you peeps 15 years ago? how the fuck did you survive? OH WAIT, YOU BOMBED IRAN? brrrrr, so evil of youuuuu

but it's wonderful we can debate this, really. I'm glad we get to exchange words, everybody, all of us!

Thank GOD, the OLD FATHERS of academia left some room ... to walk circles in ... and that they left so many of the questions they had and inspired to be left unanswered by the few/many who got in/didn't get in(to) the space that would give the incredible rarity of human brains with that specific kind of passionate curiosity in this (Tao universe measure something) vast galaxy of ours that bit they ... deserved? needed? wanted? desired? hundreds/thousands of generations worked incomparably hard for ... ???

so that we wouldn't be cursed with looking at the results of TWO COMPANIES IN 20-FUCKING-25 that are tapping that insanely sexy big ass of Space around this cutesy blue-and-green little planet of hours/ sorry, ... "ours", ... like, ... somewhat "ours" ... ... ...

Heyyyyy, are there any cool new toys on Alibaba or whatever the name is or something? There's some German genes in the neighborhood who can use a camera

200+ million proper engineers with bunches of them being parents and "elites could shape mass preferences".

nice try, humanity.

Re:Thoughts on Go vs. Rust vs. Zig on Thursday December 04, 2025 @05:29PM (#46154055)
if the languages were creations of LLMs, what would be your (relatively refined) chain(s) of (indulgently) critical thought?
all of this reads good news except that people have a hard time accepting that bureaucracy should absolutely give you a hard time because you are probably not even close to the best version of how to do that thang.

it's sad that European Tech Firms, lawmakers and economists are made of sugar, or that they have decided that it's time to become sugar.

Re:Beej's Guide to Learning Computer Science on Thursday December 04, 2025 @07:47AM (#46147053)
it's the same in any industry with the exception of insurance and bureaucracy, I think (IMO and from the top of my head)

but I believe no other industry has a greater gap between "i can't believe the shit I (have to) do at work" and "hobby" except in the many cases of voyeurs, stalkers, laissez-faire criminal investigators and passionate MiniTrue partisans, all of which are variations of the "true" human nature of the bulk of the population left after the many wars, including a great many of the ones who came home unharmed but with stories to tell

Re:How to Synthesize a House Loop on Thursday December 04, 2025 @07:39AM (#46146983)
It's not gonna add anything to your repertoire. It will appear so after some time but it's really just an approach for people who have bad hand-eye coordination and ability to hold a rhythm or a hard time acquiring these skills, or tinkering with DAWs, which have a weirdly annoying first hour use time/learning curve