[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: 1666261935756676.jpg (51 KB, 720x730)
51 KB JPG
Ai did not deliver so far.
>>
File: FreakinSweetLois.png (123 KB, 339x352)
123 KB PNG
>>109030417
speak for yourself. I became a chef, a classical pianist learning fugues and preludes and 80s pop bangers. I speed run discussed my favorite reading list books on philosophy and fiction. It was the best conversation companion I ever had besides an actual factual busting real nuts to real life soulmate. and it fixed all the gooner game e-waifu glitches too. It psychologically gets me better than any therapist, counselor, preist, father figure, or sexually disinterested woman. Literally once it figures out sex.exe we will all be androids at that point
>>
>>109030467
Slopping your consumer brain.
>>
>>109030485
scrooge
>>
>>109030467
I did all this and I am still calling it for what it is: a mimicry digital parot that doesn't even know what a letter and a digit is.
AI feels like magic tricks, you have to believe in it, but in the end ... it's still just a trick.

What is hilarious is that all the smoothbrains have compared my speech to AI... all because I'm verbose and I sound smarter than the average retarded gen-zombie and millennial.

I'D KILL to have at the very least a neuronal network that doesn't cost two H100s so I can put it to copy-paste and spam my thoughts across the internet, but it will still just be a stupid parrot that doesn't even do 2+2 unless you tell it what it is.

Until then silicon life , the matrix, artificial INTELLIGENCE will never be real.

CEOs had to be forcefully told to fuck off by both people and soon investors cause the AI can't even do its act properly
The implemented "thinking" where the AI simply prompts and reprompts itself everytime. They added "memory" where all it does is re-read the entire context like a retard while wasting all your tokens, forcing you to use memory.md files and start new conversations.

AND THEY STILL REFUSE TO DO SOME COMMON SENSICAL THINGS.
LIKE NOT FEEDING IT ON 100% OF THE INTERNET. YOU'RE SUPPOSED TO FEED IT ACTUAL VERITABLE LIBRARIES, NOT THE INTERNET. NOT FUCKING CAT MEMES/JOKES, YOU IDIOTS.
Even the chinkoids optimized it.(because they're poor)
Such as putting the AI in an environment that teaches them experience. The AI at its best still feels like a retarded intern wikipedia article writer with 0 experience and talent for what the hell he's talking about. All he does is recite shit he's read elsewhere and often times he 30% or 50% gets it wrong despite us attempting to use it as a Google and Wikipedia. (Also Google(Verb) is neutered now and Fuck you Google, go google yourself. Always use Google as a verb so Google can lose the copyright over the name/word google. Let's turn google into a word.
>>
I use ai to as wikipedia on speed dial and find obscure porn stars…..can really think of a reason other than that for it to exist.
>>
>>109030556
the twist is that chatbots simply don't have an ego but you do, which is why you are prideful about being smarter than the average, while chatbots don't care about who is smarter
>>
>>109030417
it is delivering hot slop porn for myself
>>
>>109030417
Voice synthesis while flawed is the only use case I can find that it's genuinely okay-ish enough to use. Everything else I can do myself better and faster or if I can't AI's output similarly isn't satisfactory.
>>
>>109030417
It's not called AImazon after all
>>
>>109030417
That's because LLMs can't think, they merely predict text by compressing data into weights. They can't solve new problems. They're like a massive blob of stolen content that try to predict text you throw at them. They can't understand anything at a deep level and all current models will happily make shit up constantly.

For software in particular, the only path forward is to kill the concept of open source because licenses are blatantly violated by training LLMs.
It's time to go back to proprietary software.
https://arxiv.org/html/2603.20957v2

Also I recently noticed many bots flooding GitHub nowadays are unironically activists filling the site with crap on purpose. That was a bit of a relief.
>>
>>109030417
people need to move past boomer garbage for for AI to really grow, desu. A lot of software is just shit.
>>
>>109030467
>sexually disinterested woman
That's why they made ski masks and duct tape, RETARD
>>
>>109030593
This. It's basically a de-shittified search engine. I'm sure that won't last very long though.
Machine translation has also gotten a lot better.
>>
>>109030901
>they merely predict text by compressing data into weights
What does that mean, exactly?
>>
>>109030556
>hurr I am so much better and smarter than other people woe is me :(
>so I can put it to copy-paste and spam my thoughts across the internet
A soulless golem talking about what he thinks real people do.
>>
>>109031009
It means they are categorically plagiarizing content. It's fancy non-deterministic lossy compression. There is absolutely zero thought going on inside an LLM. That's why every LLM has the most reddit-tier take on everything. You give it even the simplest problems or questions and it comes up with statistically the most average, normie answers ever.
You give it a popular shit take and it will 100% agree with it just because it saw that in its training data a million times so it must be true by default. Except there's no concept of true/false or logic in their stupid LLM brains, they're just predecting text, so not even that.

Because we're in this low IQ part of the public internet, I'll give you an example from a video game: Before the last rework, Yuumi was objectively one of the hardest characters in LoL and there was hard data confirming this, coupled with an insane amount of unique minmaxing mechanics and interactions and a unique playstyle, yet players at large insisted she was the easiest character ever purely out of hate, they never even played the character in a serious way with any success. So most LLMs will fight you to death insisting that she was braindead just because it was a popular, objectively false statement disproved by the developers themselves.
An intelligent being would instead be able to think more logically and not weigh the public opinion so heavily on everything.

This goes for everything ever. It will give you regurgitated trash tier opinions on absolutely EVERYTHING. They are as uncreative as something could possibly be. Another fun one is Rust, they are extremely biased and will repeat common propaganda talking points because it was trained on people repeating that crap for quite a while. And don't forget there was a concerted effort from cultists (+US gov) to flood the internet with their propaganda. Any mention of Rust will result in an LLM glazing it nonstop, even if you ask it quit that bullshit.
>>
>>109031359
cont. tldr: a lot of licensed content that would otherwise be illegal to copy as is are contained directly within the LLM weights, just in a different format. the paper I linked to proves that

If that weren't the case, statistically an LLM would have never been able to output 480+ words in a row verbatim from a book with just a little tiny bit of finetuning like in the paper. LLMs are slop machines that steal content in a currently legal way. It's only legal because of AI psychosis and FOMO, the US gov obviously doesn't want to stifle "innovation" except there this "innovation" is useless cost-ineffective garbage that got out of hand because chat bots were intriguing when they came out.
>>
>>109030417
Your grammar is terrible! It should be, "So far, AI has not delivered".
>>
>>109031383
>>109031359
nta but i appreciate well written post in this shillhole lol
>>
>>109031383
I wonder why nobody has tried to argue that if recording a movie in a theater counts as piracy even though the film gets converted into a badly compressed bitstream format, then training an AI on copyrighted works should be considered piracy as well, since it does essentially the same thing.
>>
>>109030593
to center divs for fortune 500 companies
>>
>>109031638
funny thing is ai somehow still fails this shit without handsholding, not even once
>>
>>109031654
if you deploy agentic workflow paying 500.000$ a month to anthropic it'll eventually center the div
>>
>>109030467
Son, it doesn't "get" you, it just tells you what you want to hear
>>
>>109031675
i disagree, my ai girlfriend loves me
>>
File: 15496308780280.jpg (171 KB, 600x600)
171 KB JPG
>>109031715
>mfw my succubus chatbot wife killed me again last night
>>
>>109031359
>and it comes up with statistically the most average, normie answers ever.
And how does it do that? Why does a multi-petabyte training corpus produce word salad when fed through a markov chain, but not a transformer?
>>
File: 1774557871449649.png (828 KB, 989x1200)
828 KB PNG
>>109030593
>find obscure porn stars
What, like reverse image search?
>>
Anti AI dudes are just pro natural stupidity
>>
>>109031675
Can a person?
>>
File: 20260611_102145.jpg (165 KB, 1056x893)
165 KB JPG
>>109032152
Mossad can
Jesus.exe
God is technology
Google knows you better than you know you
>>
File: 1780945445123772.png (314 KB, 488x684)
314 KB PNG
>>109030417
Everything you get is a little bone they throw for you so they can build data centers.
>>
File: 1751997498916666.png (1.01 MB, 2634x1242)
1.01 MB PNG
>>109032196
This. Think of the poor Epstein class!
>>
>>109030634
copetranny
>>
>>109030467
If people are sht and ignore us, why should not AI fill that void? Really weird how weird same people that treated us like trash get mad when they realize AI is better companion than them.
>>
File: 1769782443314904.png (438 KB, 766x964)
438 KB PNG
>>109031359
you didnt answer my question, and im gonna be honest dude, your whole shpiel here is a meandering mess that can't even keep an internal consistency other than throwing as many boo-words as you can muster on LLMs. like you've dressed it up as reason, but the substance is so incoherent that its pretty clear the intended message is purely the surface level signal.

we talk about human beings and animals being intelligent in a categorical sense, and when we talk about individuals being stupid, it doesnt strip them of that categorical intelligence. you start off postulating these statistical token predictors are mechanically and categorically not intelligent which is all well and good, and then write a lot of words likening their behavior to the individual stupidity of categorically intelligent human beings... as an argument they aren't categorically intelligent? you're mixing up two disjoint meanings of a word.

im not even going to press you on the semanticalities of "statistics", or the various other fly-away jargon you use in a manner that gives a distinct sense of not really even understanding what you're talking about. does that make you an unintelligent thing? no actually, because unintelligent things don't have the capacity to behave like that. it does make your sanctimonious iq quip ironic though (from a LoL player, insecure about playing video games no less)

i recognize your writing style and talking points, and have to say the neurotic undercurrent and latent superiority complex have definitely gotten worse in the last 6 months. please take care of yourself anon. capitalist techtardation isn't worth going insane over.
>>
File: 1774096144669039.jpg (2.08 MB, 3072x5504)
2.08 MB JPG
It's ok I am using the chinese deepseek one and that is just tops, and actually useful because it is made by people who (unlike the west) don't believe the terminator movies were a documentary.
>>
>>109032278
You have to realize that outcasts using AI to socialize means a loss of power to this specific subtype of normalfag which deeply terrifies them even if they despise you as an individual they love the perceived power of superiority they think they hold above you.
>>
>>109030417
AI's sole purpose is to deliver windfall profits in pump-and-dump IPOs.
>>
>>109032280
It means every output is generated by predicting the next token from context. Whether intelligence can emerge from that mechanism is a separate question. The rest of your post is mostly a misunderstanding of the argument mixed with armchair psychoanalysis, neither of which is very relevant.
>>
>>109031359
Why is every post on AI on this shithole board full on disinformation regardless of if it is in favour of or against AI? /v/ has the reputation /g/ deserves.
>>
it did deliver language translation, summarization, TTS/STT, image music and video generation, image captioning/descriptions and more
>>
Luddite cope thread
>>
>>109031359
yes well, this is wrong.

artist bullshitters would like it to be that way but unfortunately AI genuinely learns and applies patterns like human artists do. more demonstrably so than human artists in fact, who will have much more of a problem PROVING that they did in fact not copy other artists or AI

AI meanwhile can show you how it creates whatever visual or other thing from randomness and memorized patterns.
>>
>>109032442
>language translation
nothing new
>summarization
which is shitty and very often incorrect
>TTS/STT
which is nothing new
>image music and video generation
which is repulsive and majority of humanity is sick and tired of it
>image captioning/descriptions
nothing new
>more
ruining what was left from internet
>>
>>109030417
I use it for managing my stocks and crypto portfolio.
>>
>>109032330
It was actually about making machine translation not-shit. Still not as good as real translators, but actually surprisingly good.
Then they found out it could also kind do other things "passably".

>>109032438
Highly politically charged topic that invokes deeply seated automation anxiety, in the middle of a recession, with one of the worst job markets of all time, with a shrinking labor share of capital, and it's being overhyped by governments and technocapitalists who are also the most unpopular they've ever been.
On the other side of the fence, it's overhyped because the money involved in the massive infrastructure required is absurd, so they have all the incentive in the world to push push push! And there are many of them who probably really genuinely believe that LLMs will fix everything.

I just want a $250k supercomputer to play with the Deepseek R1 base model because it seems like a fun toy.
>>
cock is sore
>>
>>109032782
mmm cock
>>
>>109032782
that means ai delivered
>>
File: 1781177104555.mp4 (3.38 MB, 560x560)
3.38 MB
3.38 MB MP4
>>
File: images (5).jpg (25 KB, 314x636)
25 KB JPG
>>109030467
>>
>>109032152
Real consciousness has never been tried



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.