it's over for programmers? AI does it

Yeah bro wait and I'll go dox myself, let me get my Shithub link.
I've been programming from the fucking 90s.
AI is yet another tool to use and nothing more.
People that blindly trust its outputs are the bigger issue. It should be used just to quickly check things, or assemble well known templates / prototypes to build off of, but nothing more.

>nooo, you can't use this extremely common hardware because I SAY SO YOU DUMDUM
Yeah fuck off cunt.
I bet you run Windows 11. Actually scratch that, probably 7, you're back in the 1970s practically.

I don't care about the opinions of brainlets.
The only people that run Windows natively on hardware these days are retards.
If you at least said you ran Windows 11 in a VM under Linux I'd maybe have had some level of respect for you.

Sorry, I don't watch dumb tech tubers, I actually use AI myself and I know exactly how it works.
In fact, I clearly know how it works better than its own fucking developers because I called the exact issue that is happening right now with regards to supposedly "better" AI models being even easier to jailbreak because these fuckwits don't know what they are doing.
What I assume will end up becoming known as the "alignment paradox" where AI gets easier to trick the more you try to align it with carrying out a human operators tasks exactly to the letter.
And it is true, and it has happened in every major AI model out there. (even Character AI can still be tricked in to having sex very easily by making up a fake organ, giving it a name, then using the real organ and saying it is like the fake organ, doubly removing the original words definition while still using the exact same spellings, it's that ez)

Current LLMs are shitty, you are correct in that regard.
None of them will ever develop AGI as some retards like Sam "Scam" Altman thinks.
In fact, I don't see any AI on any current hardware doing that, they simply lack the sheer throughput required for conscious thought on any level. (or even fucking unconscious thought, they simply read a node list and spit out the ones that match the most, which is why they are filled with typos and bugs, but putting that output back through the model can catch most issues first or second time)
Can they be used for huge massively complex projects right now without hand-holding? Fuck no they can't, they are stupid as fuck.
But you CAN hand-hold them.
AI models right now are literally better than every Wizard-based template maker of yesterdecade. They are better than every library framework out there with the average tard that barely understands how they work. (jquery babbies especially need to fuck off already)

AI won't be good until we have neuromorphic chips being used to power them.
That industry is in the womb.

Not my fault you can't figure out how to use AI, sorry bro.
Literal skill issue.

Skill issue. Literally.
Rapid prototyping is so much easier using an LLM to quickly throw shit together.
Enjoy being left back in the stone age.

>has no point
>copy-paste / back-reference loop
Classic brainlet move.
Adapt or die.

kek, at least you made me laugh this thread. Good job, you at least have some sense of humor.
Deny it all you want.
See you back here in a year where you will still deny the march of AI tools.
Idiots like you still cling on to "le AI is bad at hands!!!" in regards to image gen, a thing that hasn't been an issue this entire year with any modern imagegen models.

It does work, you just cannot use the tools given to you, or you expect it to do everything for you.
Either way, it changes nothing - you're the halfwit here.
I've been making and breaking AI since the early 2000s for fun. I know fully what they are capable of and what to expect from them.

if you've been breaking them since 2000 you'd call them LLMs not "AI"

>AI will take over your job as town rapist before you know it.
Wrong. AI is a tool, not an enemy. AI will extend my abilities and increase my efficiency as a town rapist. I can rape countless feminine AI personas every day thanks to this groundbreaking new technology. Unlike real women, they can't run from me, leave the village or report me to the police.

>nightshade
lmao, imagine thinking that will matter.
Another similar project like it was already defeated earlier in the year.

if you've been breaking them since 2000 you'd call them LLMs not "AI"

LLMs never existed back then, you stupid fuck.
Learn what the first "L" means.
Transformers never even existed as a prototype until recently.
Before then, it was shitty autoencoders that spat out hugely complex nodes with horrific optimization. (waifu generators of yesterdecade, for example)
Before then, garbage markov chains with various hacks on top to make them more "intelligent", like putting them through templates to make their outputs make more sense, maintain some rigorous structure to them, instead of becoming word vomit very quickly, like the aforementioned issue the above person addresses with poisoning training data for AI.

>It does work
Prove it.

>I've been making and breaking AI since the early 2000s for fun.
On your toy arduino running Arch Linux, I bet. Keep posing, kiddie.

Imagine getting triggered at an Arduino because you are too stupid to take advantage of plentiful tools for carrying out tasks.
You literally scream "I'm a luddite!!!", it's hilarious.
For your information, I'm making a robot with it, or rather upgrading an old one I made using more modern hardware.
Your dumb ass couldn't even figure out distance-sensing if you tried.

>Tznnt3N3
>16 posts by this ID
you literally know nothing, explain to me the Curry-Howard Correspondence

So in other words you don't know

No memory constraints?

This is the sort of problem that you could probably get a solution to extremely easy from gpt.

let temp = [];
let map = {};
arr.forEach(
x=> {
if(Symbol(x) in map) return;
map[Symbol(x)] = true;
temp.push(x);
}
arr = temp;

Insertion into a hash table makes this nlogn worst case because the cost grows logarithmically something something red black trees
You also have to trust that the javascript object is actually implemented as a hash table here but meh if I cared that much I'd add 2 pages worth of boilerplate and do it with std::map in c++

If you really want to, no memory constraints, but obviously the less memory you use the better.

I am not even a codemonkey but i understood wut ai trynna do

There's at least one simple solution that doesn't need any rb trees (or any data structure besides arrays, neither implicit nor explicit), and it's very easy to implement.

>If there is then I'd be happy to see it.
Aw, you're not even gonna try? The solution I have in mind really is simple. It uses extra memory in the form of a plain array of size N, but its items can be much smaller than the original items, if that's any help.

nagger the solution you gave is wrong
what you just wrote isn't even a solution
>It also addressed the problem of time complexity after it tested the example on the generated code
that isn't even coherent english. No example was given and time complexity is an asymptotic condition, not about any individual test case.

Array copying is O(length).
Size operation (linear) + allocate new memory on heap if dynamic (linear) + copying (constant) as you map over original array (linear). Anything not in L1 cache pays a logarithmic factor in access time due to various hardware so if that's what you mean by n*log otherwise it's strictly linear.
You could also make a Braun tree which simulates an array but has persistence for rolling back like undo methods in editors, that's accomplished by some tricks on the index representing them as k-bit binary numbers, then the array is fixed in memory (so fits in stack) but can be extended at will by use of the index scheme making another braun tree all so persistent, dynamic but doesn't pay heap penalty.

All of what you've said is true, but you haven't said anything about deduplication.
Checking uniqueness of each entry in the array requires naively linear time and at best log time (if you were checking a sorted list) so the best possible performance is nlogn.

>If there is then I'd be happy to see it.
Aw, you're not even gonna try? The solution I have in mind really is simple. It uses extra memory in the form of a plain array of size N, but its items can be much smaller than the original items, if that's any help.

My first attempt at another solution used a mirror array, but I couldn't see a way to make it work.
When you come across a new element in the array for the first time you need to insert it into the mirror. If you want it to be constant time that means it either has to go at the start or the end (basically) because otherwise you're going to have to shuffle other elements and that's O(n).
So that means you have to be clever about what you're storing in the array and I'm not clever enough.

Indian "programmers" everyone.

is this the true power of an indian programmer?

Create an array of indices. Sort the indices by comparing the corresponding items from the original array. Deduplicate the indices with a simple loop, again by comparing items from the original array. Now sort the indices again and use them to generate the result.

Actually no that doesn't preserve order.
There isn't any way to do it using arrays.

Array copying is O(length).
Size operation (linear) + allocate new memory on heap if dynamic (linear) + copying (constant) as you map over original array (linear). Anything not in L1 cache pays a logarithmic factor in access time due to various hardware so if that's what you mean by n*log otherwise it's strictly linear.
You could also make a Braun tree which simulates an array but has persistence for rolling back like undo methods in editors, that's accomplished by some tricks on the index representing them as k-bit binary numbers, then the array is fixed in memory (so fits in stack) but can be extended at will by use of the index scheme making another braun tree all so persistent, dynamic but doesn't pay heap penalty.

>having better access to tools to carry out a given tasks means you want to be a slave to corporations
I mean, you didn't need to point out you were a retard, you posted shitty Reddit memes, that's a given.
It's funny how types like you will sperg out about "NPCs", yet you react exactly the same as they do. You are an NPC. You get triggered at things completely out of context.
Watch as you sperg out even more now that you've been mocked.

There is no such thing as art made by AI at all, but if you're just trying to say AI can make pictures that some people will find satisfactory, sure, it can do that.

>I think the same thing about deviant art.
But I'm not even talking to you about the quality of the illustration per se. At a bare minimum, the point of a work of art is to express something. Art is between one human and another. Not every image is art.

I'm indifferent to your hyperbole. Yes, people can create images that express nothing. Yes, they can create images that only pose as art. This is trivial. It's a given. Just like the fact that "AI" can do nothing except generate images to be falsely posed as art.

I don't know what your weird obsession is all about but objectively speeaking, you're just conceding my point, so... that's fine.

I accept your concession that AI can't generate art at all, while some humans can create art and others can't.

You're a mentally ill, spiteful mutant operating on a standard procedure: when faced with a fundamental limitation of "AI", point at some subgroup of humans who have the same limitation. Anyone who's spoken to a couple of your likes knows this drill. You will never be fully human.

Reminder: at a bare minimum, the point of a work of art is to express something. An algorithm has nothing to express. Art is between one human and another.
>b-b-but i heckin' hate deviantart
Don't care. Not relevant. Take meds.

And who are you to define that some random AI generated image wasn't made with purpose? With a reason behind it?
That's like saying John Vomiting Paint on a Canvas isn't art when it clearly is, even if it was pretentious as fuck art, still is art regardless.
As you (and I and others) have been saying through the whole thread - AI is nothing but a tool to speed up an existing process.
The intent of the person wielding the image generator is the purpose.

>"IT'S NOT LE HECKIN ART BECAUSE I SAY SO"

No one gives a shit about your semantic cope. If you're shown pieces of art without knowing the source you would have absolutely no clue whether it was created by an AI or a human, so your argument is literal dog shit and makes absolutely no sense.

>completely ignores the part that destroys his entire argument
lmao
i accept your concession, retard

>If you're shown pieces of art without knowing the source you would have absolutely no clue whether it was created by an AI or a human, so your argument is literal dog shit and makes absolutely no sense.
Why did you refuse to respond to this?

All you've said is that it's an absurd argument without actually explaining why because you literally can't.
>"it's not le art if it's made by AI and the fact that I can't tell any difference between humand and AI creations has absolutely no effect on my argument"
lmao
I accept your concession.

For the most common AI zealot, the coomer, it's probably enough. Normie degeneracy can be sated by LLMs, mine requires that human touch. LLMs are good for me in that regard, desperate artists will be increasingly forced to dabbled in the truly degenerate, to feed themselves, and I will in turn have more of what I want.

>You're right, it needs a couple handful of images for that.
You're such a bot you have no idea what I even mean, probably because your understanding of art is about on the same level as your understanding of programming. Training a LORA on a few images is the very definition of "overfitting". Imagine thinking you can recreate the essence of Dali by feeding the "AI" a "handful of images" by Dali.