The reason Musk is against it because he knows OpenAI will spy on the code Twitter is inputting into it. This is why everyone prevents chatGPT from being used if they are a serious business.
>he knows OpenAI will spy on the code Twitter is inputting into it
even if it wasnt garbage solutions to garbage problems made by garbage people, this is critical for military intelligence infrastructure like twitter
This is a moronic take.
The reason its not allowed is because most of these AIs are trained against code with a variety of licenses. Plenty of these licenses extend to derivative works, which generative AI is.
If the license is restrictive enough, the company is legally bound by it.
I gave a legitimate reason. You simply added another. They aren't mutually exclusive moron.
5 months ago
Anonymous
Your reason is fricking stupid and only shows you have no clue what you're talking about.
5 months ago
Anonymous
What do you think is going to happen if somebody identifies clearly GPL derived code in a proprietary project?
5 months ago
Anonymous
It happens all the time. >In June 2003 some folks on the Linux Kernel Mailing List sniffed around the WRT54G and found that its firmware was based on Linux components. Because Linux is released under the GNU General Public License, or GPL, the terms of the license obliged Linksys to make available the source code to the WRT54G firmware. As most router firmware is proprietary code, vendors have no such obligation. It remains unclear whether Linksys was aware of the WRT54G’s Linux lineage, and its associated source requirements, at the time they released the router. But ultimately, under outside pressure to deliver on their legal obligation under the GPL, Linksys open sourced the WRT54G firmware in July 2003.
See: https://web.archive.org/web/20051208105928/http://www.wi-fiplanet.com/tutorials/article.php/3562391
5 months ago
Anonymous
Did they anything good in it?
5 months ago
Anonymous
Okay, then you understand that. Now think about the same scenario, but AI shitting out something clearly GPL derived. Most companies don't want to open source their IP.
5 months ago
Anonymous
Yes, that was the point I originally made. I was saying the spying on twitter code part is nonsense.
5 months ago
Anonymous
Damn, I got you guys switched. It's been a long day. You're both right though, companies are also paranoid about disclosing trade secrets by granting source access to AI systems, which is essentially a requirement for these systems. Same reason they don't want to let AI do financial stuff, even though it's arguably better suited for that.
5 months ago
Anonymous
>companies are also paranoid about disclosing trade secrets by granting source access to AI systems
lets call it a new industrial revolution.
5 months ago
Anonymous
Hey, you don't have to convince me. As always, it's the suits standing in the way.
5 months ago
Anonymous
They are terrified they will lose their position of prominence. egos.
>The reason its not allowed is because most of these AIs are trained against code with a variety of licenses
so its not ai its just a fricking zombie with corporate restrictions.
if the code isnt being distributed it doesnt need a license. also /misc/ pot had the only viable solution to the intellectual question (iq)
5 months ago
Anonymous
There are more licenses than just GPL.
5 months ago
Anonymous
there are no licenses that have any effect if youre not doing something you dont have a right to do without a license
5 months ago
Anonymous
>if youre not doing something you dont have a right to do
what
5 months ago
Anonymous
according to intellectual property law it is a crime to distribute intellectual property without a license. this law was used by intellectuals to destroy computing and herd everyone onto servers that can be controlled by the adl. /misc/ pot had the only viable solution to the intellectual question
5 months ago
Anonymous
and?
5 months ago
Anonymous
15ivNqz0 implied that only the gpl doesnt affect people who neither signed it or exceed their rights under intellectual property law, but signing it or exceeding your rights as if you had an agreement are the only ways to establish a contract that could then affect your rights
5 months ago
Anonymous
can't tell if ESL or drunk
5 months ago
Anonymous
>intellectual property law
they really made it a crime to steal an idea. haha what a world.
I've tried this out just for shits (not for my job, just for a personal project).
The code is fricking awful boys. It compiles, which is impressive, but it does not solve the problem. It is like asking what the square root of 25 is and it says 3. It looks like a reasonable enough answer at first glance, but it's wrong, and you don't get part marks.
brainlet who doesn't know how to ask specific questions to ChatGPT to get answers in problem domains you're not familiar with, boilerplate that would've been tedious to write, and specific sections of code that are basically a StackOverflow answer but tweaked to your data
It's more like you ask it for the square root of 25, 9, 100, and 144, and it tells you 5, 3, 5, and 12, which all looks good on first pass, but it fricked up and "Remembered" you asking 25 when it went to give you an answer for 100.
It's when it crosses it's output with old requests it seems to frick up.
Well it’s a fricking LLM dude. It doesn’t truly understand your question
Try asking it to tell you how many characters are in its response, or ask it to make its response a certain number of characters
It fails at these tasks because it does not understand such concepts
But you ask questions that imply you expect it to have this level of understanding
For programming, the best you can hope for is you prompt correctly and it has access to a solution already from stackoverflow or Reddit or somewhere else they used to get initial data. Otherwise, no, it doesn’t possess the intellect required to reliably arrive at solutions to novel problems, nor the intellect required to break a novel problem down to simpler parts and solve those
But it can still do quite a lot, it just seems that morons bought into the hype and then get disappointed when they realize the hype was not accurate
ChatGPT programming has stagnated over the last few years
LLMs in general performed equally bad in aoc 2022 and aoc 2023 (where they failed to solve the easiest problem)
in UKIEPC 2023 there was an AI team that failed to solve 11/14 problems
anecdotally, ChatGPT output has to be audited by a human programmer who can assess correctness, performance, and modify it to actually solve the problem you asked
people seem shocked LLMs can "write" a performant O(n + m log m) Dijkstra's implementation, but they can find identical code online (where ChatGPT got it, e.g., cp-algorithms.com/graph/dijkstra.html)
It's also nebulous legally and opens your employer up to unwanted potential liability. Imagine not knowing where your code comes from, then getting sued by everyone and their dog for stealing it or patent violations.
God I'm so glad I don't have to use stack over flow anymore. The biggest group of fricking dickheads on the planet. Programmers are the worst, most smug group of fricking morons. They never know what they are talking about, always need management to cut their shit down. "Maybe the customer would want this homosexual moronic feature that solves some edge case that isn't a problem and only .00001% of people run into" NO.
Stackoverflow is fricking ruined by GPT-4. Full of completely wrong non-answers to questions generated by ChatGPT backed up by dozens of "Thank you, sir!" pajeet responses
you are homosexuals that doesn't even understand wtf you are saying lol
almost all code looks and is the same shit
everyone is already copy pasting the same shit
and most apps are the same shit too
>almost all code looks and is the same shit >everyone is already copy pasting the same shit
yes but have in mind its all copy pasted jeet code for "tutorials". God awful trash in short
>everyone is already copy pasting the same shit
let me guess, you think you have "imposter syndrome" and make jokes like "my job is copy-pasting from stack overflow, ha ha ha... right guys?"
>imagine not knowing where your code comes from
stackoverflow obviously
https://i.imgur.com/49ASll3.png
you are homosexuals that doesn't even understand wtf you are saying lol
almost all code looks and is the same shit
everyone is already copy pasting the same shit
and most apps are the same shit too
Chatgpt is just making it obvious
Everyone using ai for coding is using GitHub copilot not chat gpt (at least that interface). It’s all coming from code stored on GitHub which is their property so nothing is legally wrong.
Just auto predicts next lines that you want really well. Can also write in a comment to generate a function that does x and it will show the code. You can choose whether to accept its input or keep doing your own thing.
Be careful. It looks/feels like tab complete, but you have to carefully review everything you let it autocomplete. Almost everything is syntactically correct and semantically it looks okay, but when it's wrong, it's very subtle. VS Intellicode may be worth giving a shot too. I've had decent luck with it--not sure what it's using under the hood, but it's nice for boilerplate code like assigning members from constructor parameters. But, again, you've got to watch out, it makes mistakes on things a human rarely would.
Autocomplete function calls with the right arguments in the wrong order. In accordance with the function signature, of course, so it still compiles. Love it. So helpful.
Yes it is. Its much easier to write something short and beautiful like a recursive function, instead of fixing someone's bloated dogshit "simple" loop.
Nah. I can tell you're probably a gay for even saying the word recursive when it wasn't needed. I make chat write out simple classes all the time. And it saves a ton of time.
I don't use it for employment. But for personal apps I do. It saves time.
You all sound like the art gays crying about stable diffusion. "NOT REAL ART" Ya it is, have a nice day gays.
It's just a moronic thing to say. I use chatgpt all the time on personal projects. it's saved me hundreds of hours. And ya the code is always broken. But I just look at it for a minute, fix what needs fixing.
Programmers are just gays who like to act superior. They are fricking shitheads any time you ask them to explain. They are just mad they aren't needed nearly as much anymore. All the smug gay programmers will finally get what they claim to want, to be left alone.
Exactly, you have never taken part in any significant development. If you've ever been handed a decade year old pile of steaming shit, you know that fixing broken code can easily be a near impossible feat compared to a rewrite. There are entire books written about this very subject, but you wouldn't know that. You think every development effort is like your one-off crapplets.
5 months ago
Anonymous
Who the fricks talking about a decade year pile of shit. That's what I'm talking about when I say programmers are smug gays. They always act so superior when really they are dumb, they just happen to know some arcane bullshit and trick everyone with it.
We are talking about chatgpt spitting out code that expected to be reviewed by an experienced dev before being used. Because IT HAS TO in its current form. There is no such thing as slapping together an entire app with it. Nothing even remotely useful anyway.
5 months ago
Anonymous
What the frick do you think we're all doing all fricking day? Nobody outside of hobbyists and their little scripts is churning out stuff in a vacuum. It's not about knowing "arcane" shit--nobody actually wants to work with legacy code or interface with legacy systems. But, guess what? You will have to, because major projects don't exist in a fricking bubble.
>I use chatgpt all the time on personal projects. it's saved me hundreds of hours. And ya the code is always broken.
This. It's amazing at scaffolding from what I've seen but I fully anticipate going over everything it spits out closely. It's still pretty terrible at implementing specifics. I'm sure it will continue to improve but it's still far from 100% at this point.
I split time between dev and sysadmin work. It's very been impressive at troubleshooting applications that are well-documented on the internet. It's an absolute beast at PowerShell commandlets.
5 months ago
Anonymous
imo don't copy it. read what it says and then see if you can do it. ask questions. then double check them.
5 months ago
Anonymous
>It's an absolute beast at PowerShell commandlets.
intedesting.
one problem chatgpt has is it won't warn you to use eg a database instead of say assembler, for a data problem. if you say "quicksort in microsoft basic" it will obey
Who do you think builds AI, moron? The moment it's up to snuff, I will use it to accelerate development a la intellisense. But it's not, and using even the most auxillary AI based tooling has already burned me. You understand use of AI is banned pretty much industry wide due to fault density alone, right?
You wouldn't understand at all, but given an existing project--you know, the decades of work new code must work with--AI struggles with things like the ordering of parameters to calls. What it is effective at doing is generating code that compiles, which is incredibly sinister in ways you don't understand. In the real world, you can't just go ahead and check something in the second it builds. You must verify the functionality, and AI is top tier at generating output that looks right, but is significantly flawed and unworkable. They write this off as "hallucinations," but that doesn't mean it can be ignored. It's banned throughout the industry with good reason.
5 months ago
Anonymous
Fricking morons relying on bullshit generated programming syntax are losers, they should be fired and be shamed out of the industry.
You are a moron, if the programming task is very simple sure but when it comes to actual performant algorithm then no god damn shitty auto generated code will perform as well as one that is designed from the ground up to do so.
If you write javascript or other toy languages, you're not a real programmer anyways, you're a copy pasting monkey applying a modicum or sentience on.
ChatGPT only accelerates this process.
But you will never be a real programmer.
So much cope lmfao
5 months ago
Anonymous
Nobody is coping except the AI evangelists.
5 months ago
Anonymous
You actually coped several paragraphs worth
5 months ago
Anonymous
Describing how something is moronic is not coping. I am laying in bed with a CUDA book sitting on my nightstand. Does that sound like somebody coping about AI? I WANT these LLMs to be what morons like you claim they are.
I use the shit out of chat gpt. Fricker knows more about CDK than Amazon at this point and can coherently answer questions about shit faster than sifting through Google's israelite nonsense.
That's moronic. Doesn't matter if the code doesn't work. Any actual programmer can see which parts are bullshit at a glance. Just talking to the bot about the design and getting the boilerplate code speeds things up a lot.
>bot, I need code to do X >code 1 >this doesn't compile >code 2 >ok, now it compiles but it gives this error >code 3 >ok, it runs, but it doesn't do X >code 4 >it doesn't do X >code 4 >this is the same as before >code 4 >change literally anything please >code 4 >it doesn't work >code 4
T E C H N O L O G Y
>code 1 >this doesn't compile >submit it to your boss anyway because frick the company >some troony coder ends up fixing it for you anyway, while you sit like a chad doing nothing >musk the homosexual decides to fire you because he's not getting as much shekel as he can doing also nothing
Tbf I feel chatgpt works quite well if you use it like an assistant. E.g. say you didn't know a particular function in a particular language. Asking Chatgpt will often provide useful information that you can work with.
The problem with chatgpt, that can never be fixed is that is can be wrong and say it's certain it's correct. You have to check everything it does.
5 months ago
Anonymous
If they were wise they would stop using ChatGPT in whatever version precisely because of this and view it as what it really is: An alpha version, the 1st try, from which there's to learn what produces which results and why to do better next time. But they have Dollar signs in their eyes and want to market the failed concept and the failures it produces. Neural networks rush straight to a likely answer and don't consider what-ifs, contradictions and things that oppose the chosen answer to add them as relevant context, or at least to state they aren't sure whether that is correct.
For a complete beginner yes, in that they can’t tell where the subtle mistakes are. For an experienced programmer (regardless of job level), no. The key is to already have tests written or at least planned out in your head to know what it will need to get right.
Fricking morons relying on bullshit generated programming syntax are losers, they should be fired and be shamed out of the industry.
You are a moron, if the programming task is very simple sure but when it comes to actual performant algorithm then no god damn shitty auto generated code will perform as well as one that is designed from the ground up to do so.
this is what I love about you people. "If it doesn't out perform literal genius's in one very specific use case then it's all fricking bullshit!
How often is programming writing algo's lmfao. I mean sure, you always write with performance in mind. But most the time it's just a bunch of bullshit that's already been figured out. That's where chat comes in.
>"If it doesn't out perform literal genius's in one very specific use case then it's all fricking bullshit!
yeah its a tool now but a rogue ai terrorizing the internet is a real threat and if one escapes everyone suffers. That's why they keep censoring it.
The issue is woman/pajeets/zoomers copy pasting huge swaths of production code into the ChatGPT input which might include API keys, security configurations, firewall setups, etc. There's nothing wrong with using it to produce boilerplate snippets that you modify and read before using
I'm a programmer and chatgpt is dumb. it's only "good" if you're requesting some code that has about 10,000 variations on github...but nobody is going to pay you to write that...
Why would I cope? It's less work, but you still need the proper skill to actually do it. Chatgpt code won't work at all. So I swoop in and offer to make code that does work with minimal effort. I work an aggregate of 10 hours a week for a full time job of profit. Very comfy.
It's about principle. If you're too fricking lazy to put the work in they're either incompetent or just useless. ChatGPT is for people with tiktok brain
its not faster and never will be, it lacks context specific information and spews bullshit, it may only work if youre setting up some toddler UI application for which you could just copy the template anyway which defeats the point of it entirely
what non-IT people dont realize is that its not about learning to code but deliveting an entire solution and discussing it with the customer
whoever did just monkey code work deserves losing his job, software engineering will always be outside the capabilities of AI as it has little to do with coding
im a student
when i tackle hard problems (e.g., https://open.kattis.com/problems/1dfroggerhard) i use pen and paper
the trick is to simplify the problem from the general case and design an algorithm for the special case
e.g., by processing cycles separately you turn the general directed graph into a tree (n vertices, n - 1 edges)
all the other good programmers i know use pen and paper too
its easier to critique than it is to generate, for example code has "compilers" already, the key innovation is to be able to generate 10000 solution to some problem, then have a programmer design a test suite and choose the fastest one that passes all the requirement tests.
However that is not how pajeets use chatgpt, they zero shot some code, that they don't even read, before submitting a pull request with it, based merely on the words "looking" normal, and not seeing some logical contradiction in the code.
The reason he's doing it is he doesn't want them dumping proprietary code into ChatGPT or some clone to make it solve some problem, since that basically makes the code available to whoever runs the AI.
>He's right. It's a necessary measure. In order to make sure his devs even understand what the code does they have to do it themselves.
The ship for programmers understanding all of the code shipped a long time ago. I'm pretty sure that the intersectionf of people that fully understand modern super scalar processors, TCP/IP stacks and modern dbs is an empty set.
There are 3 fundamental problems with ChatGPT as it is right now:
1) Nobody is sure how IP laws will work. If I insert chatgpt code into my codebase does the IP belong to OpenAI? Does it belong to the programmers that inadvertently created the corpus? Is it tainted in any way? Nobody knows.
2) Baring copyright - nobody is sure you're not leaking company secrets each time you communicate with ChatGPT, and the more you communicate the more you leak.
3) The problem never was writing code. The problem is how to translate conflicting business and customer demands into machine code + infra in a way that doesn't break anything in the legacy code base in a cost efficient way. ChatGPT only knows as much as fits into its token window.
Even if someday big codebases fit into that window you're still stuck with #2 problem. Even if you get on premise installations of ChatGPT nobody is sure IP is not poisoned.
Even if you get all of the above you still need programmers that are autistic enough to iteratively redefine the spec you give to the bot to actually do what you want.
t. a CTO
I know absolutely nothing about Python but ChatGPT built me several scripts which scrape certain websites for contact info and organizes it for targeted marketing efforts. Anyone using it with an IQ above 85 will have a better time doing their own projects than hiring a $50/hr "firm" of jeets
>Sorry, I mean I need a source that explicitly states your argument. This is just tangential to the discussion.
>No, you can't make inferences and observations from the sources you've gathered. Any additional comments from you MUST be a subset of the information from the sources you've gathered.
>You can't make normative statements from empirical evidence.
>Do you have a degree in that field?
>A college degree? In that field?
>Then your arguments are invalid.
>No, it doesn't matter how close those data points are correlated. Correlation does not equal causation.
>Correlation does not equal causation.
>CORRELATION. DOES. NOT. EQUAL. CAUSATION.
>You still haven't provided me a valid source yet.
>Nope, still haven't.
>I just looked through all 308 pages of your user history, figures I'm debating a glormpf supporter. A moron.
I used it to code at the start of the year(newish to coding) and though it was good. Now a year later I'm good and think it's bad. I got better at coding faster than an AI. Please give me 100 billion dollars
My personal experience is that Chat-GPT is like your employee who can write you remedial code that gets the job done, when it comes to more complicated logic and functionality, like asking ChatGPT to write a simple transformer that is GPT-like, it fails hard and struggles to find solutions to fix it. It then goes into circular mode where it suggests the same thing over and over, 2 steps back, 1 step forward. That's not to say that it won't be able to in the future, it's just not capable now. And GPT in it's current form is like a fancy search engine trained on data through brute force. AGI or a NN that actually can learn in a different way without requiring all of the internet to train it on will eclipse gpt4.
I find that kind of amusing but I have to wonder if it's because of a non-compete sorta deal? Like using Chat GPT to make code will get you sued but using Grok to make code is A-OK?
I love this kind of news, most of /misc/ is clearly autistic IT-homosexuals who think coding is some kind of high IQ inpenetrable job that AI could never seek to compete with, and look down on all tradies etc. But whenever there is a hint of AI being able to do their job for them, they shut their ears and eyes and pretend it wont happen while repeating and crying.
That is true and I'm sick of smug, up-their-own-assole tech nerds who are obviously insecure and looking for a way to feel superior to others instead of just doing their fricking jobs and helping people.
To date the only significant achievement of AI is providing nocode wagelets with weapons grade cope. We'll let you guys know when it's actually ready for us to use, okay?
you are a coping Black person
I comfy talk to my bot and tell it to write shit, run it, error? try again or search online, try again, loop across reqs
it usually takes 1 shot, max 5 to code anything for me while I literally lay down to shit post
I have built advanced apps doing that all year, easy
cope seethe and dilate
ChatGPT is saving me hundreds of hours coding. Frick this african manchild Musk, who neevr worked a single minute in his life. moron white Black person.
ChatGPT is good for spotting small mistakes you miss in your spaghetti code and to optimize existing code that already works, but it really sucks at coming up with brand new code.
Theres tons of ai's for this now, chat-gpt would be wise to foster any potential customer base instead of pushing them away. For me personally, i like github co-pilot, i find the code it suggests is a lot better. Chat-gpt is okay but you have to be careful with your prompts to get usable code
Have to be so precise in your description such that you don't leave room for hallucinations or forgetting to include step, yet choosing to write it in english instead of code because you cant be bothered to remember the newest meme framework.
I asked some AI shill on here to make a root finding algorithm with chatgpt (talking about newton's method) and he gave me some shitty code that finds the approximate square root of a number really slowly (and couldn't do negative numbers)
lawl lawl lawl lawl lawl
Black person you realize that you have to use calculus for that, by using imaginary exponents, and moreover that imaginary numbers are not real numbers but abstractions.
Every one of you is a moron
You are a moron.
They are abstractions, but the other moron was saying this as opposed to “real numbers”
And an adequate response to him would be to point out that the root of a negative number can be easily accommodated in a program: just find the root of the positive number and multiply the answer by “i”
The person that moron was replying to is also a moron for thinking he’s clever
The term “root” is overloaded, and while he may have meant square root, it has another meaning in number theory
Aside from all of that, I’m curious what any of you morons thinks is an actual solution to the problem
At 2am the closest idea I can come up with is to divide by 2, then take the midpoint between 2 and the result, and do that recursively for the new result (divide by that, find midpoint, divide), but this is only for numbers greater than 2ish.
Although I’d expect an AI to have access to actual root finding algos so I suspect moron #1 prompted incorrectly
I have played with it trying to solve basic hydrology problems like sizing a culvert or channel. It constantly gets technical details wrong. It might give you bulk basic b***h code but it’s not landing rockets on platforms, that’s for damn sure.
>It constantly gets technical details wrong.
Thats because nobody at openAI are designing LLM tests about culvert digging, the key with these models is that you have a test with which you determine good output from bad output, and thus "train" the model to behave the way you want.
If you want openai to get culvert digging outputs right, you can literally submit a set of tests to openai, which they use to train their models.
its not because it's cheating it's because it's a security risk for putting your companies information into it, a lot of these AI's actually go back to china like bing does, my friend works closely with government in infosec and tells me about this. He also says the main culprits are the indian c**ts they outsource
you will be replaced even more by Indians from now on since chatgpt is here
I am a complete beginner in python see
i used chat gpt to make a python code which takes in Gb size mp4 videos one and other and makes it go through a windows GUI exe program
earlier i used to do it by hand and one day i would do at best 20 videos
after making the python code using chatgpt taking about 5 hours to debug and fix, the rest of the day i processed around 800 videos through the windows gui exe program one after other.. all on full auto
>you will be replaced
Yeah but I don't specialize in coding, I specialized in people. Kind of a weird gig honestly putting all the points into charisma and empathy but I sleep pretty well about it. Hopefully you can come to the US soon and live the dream Rajeesh
for many people, coding is hard because their minds are not logical
chatgpt is fine for boilerplate code, but you still need to know what it's doing and how to make it do what you want with no sideeffects
It’s not
There’s some other shit going on with Elon and OpenAI, don’t care what it is, but Elon must feel slighted in some way
Aside from that, there is no reason you should trust ChatGPT or any other tool to keep your data safe, including any proprietary code you shove into it for help. Other companies have been telling employees not to use ChatGPT on work for this reason. And others don’t give a shit
But again, this has more to do with Elon’s ego being bruised for some reason
I love coding with chatgpt now, simply ask the function and it mostly finds it and writes the code. Endless coping from programs is funny. AI's are only getting better, chatgpt is 100billion parameters and next year we'll probably get the first 1 trillion parameter AI (Human brain is 200 trillion). You've already lost.
>noooo stop using tools that make your job easier! >work harder not smarter!
I don't envy his workers. Having a boss with such mindset is the absolute worst.
>work from home >have work machine and personal machine next to each other >use chatgpt on the side on personal machine >they never find out >they don’t actually care because they do the same thing
ChatGPT produces an extra layer of stupid bugs in code. Elon not going to be able to prove it was being used but he will have some numbers on bugs coming and be screaming because the cost of getting reliability not just function went up quite a bit because of it.
Elon is a fricking moron. Human-AI combo is the future. Of course he would be in the way of efficiency, he doesn’t want his slaves to work from home either.
Well it introduces a lot more unmanageable spaghetti code so I understand where he comes from with the sue part. Or it could be that the source-code leaks to Open.AI
Technically there's nothing wrong with using it to parse the code to find you some errors you've skipped or asking it to generate a test battery.
ALL USER IDENTIFICATION ON MUSK'S X.COM IS HANDLED BY SHIN BET >X.com has officially partnered with Au10tix
Au10tix is founded and operated by Ron Atzmon, who served in Shin Bet, our best ally's domestic intelligence agency, their FBI to M0ssad's CIA. What's more Atzmon served in the notorious Unit 8200 cyber dept of Shin Bet.
>certified users of X (formerly Twitter) are forced to hand over their personal data to an Israeli company linked to the Shin Bet
https://today.lorientlejour.com/article/1351162/how-israeli-tech-is-leaving-lebanon-by-the-wayside.html
This is Atzmon's X.com account where he is promoting blacklist campaigns against tech supporters of Gaza, posting memes comparing US college students who support Gaza to Hitler youth, and a meme image of Hitler supporting Palestine.
https://nitter.net/ronatzmon
THIS IS WHO HAS YOUR ID/INFO ON TWITTER. AND MUSK WANTS ALL USERS TO SUBMIT CC / IDs STARTING 2024.
SHARE THIS INFO. MUSK IS CREATING A GIGA HONEY TRAP WHILE POSING AS AN ANTI-ADL TECH ROGUE.
The Atzmon family is highly wealthy and influential there as well, and currently oversee an "commercial airport security" company that does 27% of its business in the US
Why is this being ignored? Musk is running all X users' accounts through a Shin Bet shell company.
It's probably too late to mention this so you all register it, but: GPT has been poisoned by fake feedback from coders since it went public. People purposefully taught it wrong, basically. It was better at code right out of the box than it is now
Bullshit. I’ve asked it to do ladder logic when I first started learning and it couldn’t get it right for the life of it. I even kept telling it what it was doing wrong and it still couldn’t.
My jobs remain safe
Either way the robot can’t physically do my job
Frick the Antichrist beast
The mark of the beast will be something similar to neurolink, mixed with super liberal ChatGPT or other AI programmed into it, and it will give the manufacturer full ability to see through your eyes, see your thoughts, and everything about you even if they don’t explicitly tell you that.
Screenshot. Christ is unironically coming back soon.
>muhhh social security and debit cards
Not the fricking same
i used chat gpt to make a python code which takes in Gb size mp4 videos one and other and makes it go through a windows GUI exe program
earlier i used to do it by hand and one day i would do at best 20 videos
after making the python code using chatgpt taking about 5 hours to debug and fix, the rest of the day i processed around 800 videos through the windows gui exe program one after other.. all on full auto
This just proves programmers are not even needed.
Learn to code indeed.
It proves the exact opposite. In fact you sound like a nervous ChatGPT bot.
If the app can do their work then why pay them?
Read the OP again moron.
What I read is the app does the job cheaper than a pajeet like you.
Undercut by app
Ain't that a b***h.
>I'll just make up my own OP
You are ChatGPT.
It can't do the work artard.
The reason Musk is against it because he knows OpenAI will spy on the code Twitter is inputting into it. This is why everyone prevents chatGPT from being used if they are a serious business.
>he knows OpenAI will spy on the code Twitter is inputting into it
even if it wasnt garbage solutions to garbage problems made by garbage people, this is critical for military intelligence infrastructure like twitter
This is a moronic take.
The reason its not allowed is because most of these AIs are trained against code with a variety of licenses. Plenty of these licenses extend to derivative works, which generative AI is.
If the license is restrictive enough, the company is legally bound by it.
Shut the frick up.
I'm sorry you're a moron.
I gave a legitimate reason. You simply added another. They aren't mutually exclusive moron.
Your reason is fricking stupid and only shows you have no clue what you're talking about.
What do you think is going to happen if somebody identifies clearly GPL derived code in a proprietary project?
It happens all the time.
>In June 2003 some folks on the Linux Kernel Mailing List sniffed around the WRT54G and found that its firmware was based on Linux components. Because Linux is released under the GNU General Public License, or GPL, the terms of the license obliged Linksys to make available the source code to the WRT54G firmware. As most router firmware is proprietary code, vendors have no such obligation. It remains unclear whether Linksys was aware of the WRT54G’s Linux lineage, and its associated source requirements, at the time they released the router. But ultimately, under outside pressure to deliver on their legal obligation under the GPL, Linksys open sourced the WRT54G firmware in July 2003.
See: https://web.archive.org/web/20051208105928/http://www.wi-fiplanet.com/tutorials/article.php/3562391
Did they anything good in it?
Okay, then you understand that. Now think about the same scenario, but AI shitting out something clearly GPL derived. Most companies don't want to open source their IP.
Yes, that was the point I originally made. I was saying the spying on twitter code part is nonsense.
Damn, I got you guys switched. It's been a long day. You're both right though, companies are also paranoid about disclosing trade secrets by granting source access to AI systems, which is essentially a requirement for these systems. Same reason they don't want to let AI do financial stuff, even though it's arguably better suited for that.
>companies are also paranoid about disclosing trade secrets by granting source access to AI systems
lets call it a new industrial revolution.
Hey, you don't have to convince me. As always, it's the suits standing in the way.
They are terrified they will lose their position of prominence. egos.
>The reason its not allowed is because most of these AIs are trained against code with a variety of licenses
so its not ai its just a fricking zombie with corporate restrictions.
a perfect drone.
Are you guys ready to get Borged?
if the code isnt being distributed it doesnt need a license. also /misc/ pot had the only viable solution to the intellectual question (iq)
There are more licenses than just GPL.
there are no licenses that have any effect if youre not doing something you dont have a right to do without a license
>if youre not doing something you dont have a right to do
what
according to intellectual property law it is a crime to distribute intellectual property without a license. this law was used by intellectuals to destroy computing and herd everyone onto servers that can be controlled by the adl. /misc/ pot had the only viable solution to the intellectual question
and?
15ivNqz0 implied that only the gpl doesnt affect people who neither signed it or exceed their rights under intellectual property law, but signing it or exceeding your rights as if you had an agreement are the only ways to establish a contract that could then affect your rights
can't tell if ESL or drunk
>intellectual property law
they really made it a crime to steal an idea. haha what a world.
Imagine being you
>indian mutt
being outplayed by
ChadGPT bot
I doubt it. This version of AI is the worst it will ever be meaning in the future the possibilities are endless
I've tried this out just for shits (not for my job, just for a personal project).
The code is fricking awful boys. It compiles, which is impressive, but it does not solve the problem. It is like asking what the square root of 25 is and it says 3. It looks like a reasonable enough answer at first glance, but it's wrong, and you don't get part marks.
I can attest that it was better a few months ago, but it wasn't great
brainlet who doesn't know how to ask specific questions to ChatGPT to get answers in problem domains you're not familiar with, boilerplate that would've been tedious to write, and specific sections of code that are basically a StackOverflow answer but tweaked to your data
Just like ai couldn't draw hands, right? but it can now. The clock is ticking.
It's more like you ask it for the square root of 25, 9, 100, and 144, and it tells you 5, 3, 5, and 12, which all looks good on first pass, but it fricked up and "Remembered" you asking 25 when it went to give you an answer for 100.
It's when it crosses it's output with old requests it seems to frick up.
>Uses free AI
>It's sucks
Many such cases for the 0iq'er
Well it’s a fricking LLM dude. It doesn’t truly understand your question
Try asking it to tell you how many characters are in its response, or ask it to make its response a certain number of characters
It fails at these tasks because it does not understand such concepts
But you ask questions that imply you expect it to have this level of understanding
For programming, the best you can hope for is you prompt correctly and it has access to a solution already from stackoverflow or Reddit or somewhere else they used to get initial data. Otherwise, no, it doesn’t possess the intellect required to reliably arrive at solutions to novel problems, nor the intellect required to break a novel problem down to simpler parts and solve those
But it can still do quite a lot, it just seems that morons bought into the hype and then get disappointed when they realize the hype was not accurate
The only thing current AI programming helps is indians passing along broken software that appears at first glance to be working.
But how, i need to learn
It helps me write bots for runescape as a hobby, but I'm a welder. And I hate playing that fricking game.
#learnToMine
ChatGPT programming has stagnated over the last few years
LLMs in general performed equally bad in aoc 2022 and aoc 2023 (where they failed to solve the easiest problem)
in UKIEPC 2023 there was an AI team that failed to solve 11/14 problems
anecdotally, ChatGPT output has to be audited by a human programmer who can assess correctness, performance, and modify it to actually solve the problem you asked
people seem shocked LLMs can "write" a performant O(n + m log m) Dijkstra's implementation, but they can find identical code online (where ChatGPT got it, e.g., cp-algorithms.com/graph/dijkstra.html)
t. competitive programmer
It's faster. It won't get 100% of the work done but it's quicker to fix its output than start from nothing. Now programmer bros come in coping.
It's also nebulous legally and opens your employer up to unwanted potential liability. Imagine not knowing where your code comes from, then getting sued by everyone and their dog for stealing it or patent violations.
>imagine not knowing where your code comes from
stackoverflow obviously
God I'm so glad I don't have to use stack over flow anymore. The biggest group of fricking dickheads on the planet. Programmers are the worst, most smug group of fricking morons. They never know what they are talking about, always need management to cut their shit down. "Maybe the customer would want this homosexual moronic feature that solves some edge case that isn't a problem and only .00001% of people run into" NO.
Stackoverflow is fricking ruined by GPT-4. Full of completely wrong non-answers to questions generated by ChatGPT backed up by dozens of "Thank you, sir!" pajeet responses
you are homosexuals that doesn't even understand wtf you are saying lol
almost all code looks and is the same shit
everyone is already copy pasting the same shit
and most apps are the same shit too
Chatgpt is just making it obvious
>almost all code looks and is the same shit
>everyone is already copy pasting the same shit
yes but have in mind its all copy pasted jeet code for "tutorials". God awful trash in short
Turns out to have accurate view of race and gays
>everyone is already copy pasting the same shit
let me guess, you think you have "imposter syndrome" and make jokes like "my job is copy-pasting from stack overflow, ha ha ha... right guys?"
Everyone using ai for coding is using GitHub copilot not chat gpt (at least that interface). It’s all coming from code stored on GitHub which is their property so nothing is legally wrong.
I haven't used github copilot. What does it do that is so great?
Just auto predicts next lines that you want really well. Can also write in a comment to generate a function that does x and it will show the code. You can choose whether to accept its input or keep doing your own thing.
I'll try it out. Thanks
Be careful. It looks/feels like tab complete, but you have to carefully review everything you let it autocomplete. Almost everything is syntactically correct and semantically it looks okay, but when it's wrong, it's very subtle. VS Intellicode may be worth giving a shot too. I've had decent luck with it--not sure what it's using under the hood, but it's nice for boilerplate code like assigning members from constructor parameters. But, again, you've got to watch out, it makes mistakes on things a human rarely would.
Autocomplete function calls with the right arguments in the wrong order. In accordance with the function signature, of course, so it still compiles. Love it. So helpful.
chatGPT also has a hard time with parameter order and the effect it has.
Fixing fricked up code properly is harder than just start from scratch.
No it's not.
It is when thousands of clients will be on your ass for correcting undefined behaviour of your needful apis.
Yes it is. Its much easier to write something short and beautiful like a recursive function, instead of fixing someone's bloated dogshit "simple" loop.
Nah. I can tell you're probably a gay for even saying the word recursive when it wasn't needed. I make chat write out simple classes all the time. And it saves a ton of time.
I don't use it for employment. But for personal apps I do. It saves time.
You all sound like the art gays crying about stable diffusion. "NOT REAL ART" Ya it is, have a nice day gays.
>recursive function
ask me how i know you dont code for a living
You probably can't into loop in a loop
recursion is a tool for a job, dont use it for everything
Who are you trying to convince? You clearly have zero experience.
It's just a moronic thing to say. I use chatgpt all the time on personal projects. it's saved me hundreds of hours. And ya the code is always broken. But I just look at it for a minute, fix what needs fixing.
Programmers are just gays who like to act superior. They are fricking shitheads any time you ask them to explain. They are just mad they aren't needed nearly as much anymore. All the smug gay programmers will finally get what they claim to want, to be left alone.
Exactly, you have never taken part in any significant development. If you've ever been handed a decade year old pile of steaming shit, you know that fixing broken code can easily be a near impossible feat compared to a rewrite. There are entire books written about this very subject, but you wouldn't know that. You think every development effort is like your one-off crapplets.
Who the fricks talking about a decade year pile of shit. That's what I'm talking about when I say programmers are smug gays. They always act so superior when really they are dumb, they just happen to know some arcane bullshit and trick everyone with it.
We are talking about chatgpt spitting out code that expected to be reviewed by an experienced dev before being used. Because IT HAS TO in its current form. There is no such thing as slapping together an entire app with it. Nothing even remotely useful anyway.
What the frick do you think we're all doing all fricking day? Nobody outside of hobbyists and their little scripts is churning out stuff in a vacuum. It's not about knowing "arcane" shit--nobody actually wants to work with legacy code or interface with legacy systems. But, guess what? You will have to, because major projects don't exist in a fricking bubble.
>I use chatgpt all the time on personal projects. it's saved me hundreds of hours. And ya the code is always broken.
This. It's amazing at scaffolding from what I've seen but I fully anticipate going over everything it spits out closely. It's still pretty terrible at implementing specifics. I'm sure it will continue to improve but it's still far from 100% at this point.
I split time between dev and sysadmin work. It's very been impressive at troubleshooting applications that are well-documented on the internet. It's an absolute beast at PowerShell commandlets.
imo don't copy it. read what it says and then see if you can do it. ask questions. then double check them.
>It's an absolute beast at PowerShell commandlets.
intedesting.
one problem chatgpt has is it won't warn you to use eg a database instead of say assembler, for a data problem. if you say "quicksort in microsoft basic" it will obey
Have you ever asked yourself whether you're the problem?
Oh look, it's the coping "programmer" bros I said would start replying.
Who do you think builds AI, moron? The moment it's up to snuff, I will use it to accelerate development a la intellisense. But it's not, and using even the most auxillary AI based tooling has already burned me. You understand use of AI is banned pretty much industry wide due to fault density alone, right?
Oh I'm sure you're ahead of the curve and a genius at programming, just like every other programmer who posts on this website.
You wouldn't understand at all, but given an existing project--you know, the decades of work new code must work with--AI struggles with things like the ordering of parameters to calls. What it is effective at doing is generating code that compiles, which is incredibly sinister in ways you don't understand. In the real world, you can't just go ahead and check something in the second it builds. You must verify the functionality, and AI is top tier at generating output that looks right, but is significantly flawed and unworkable. They write this off as "hallucinations," but that doesn't mean it can be ignored. It's banned throughout the industry with good reason.
So much cope lmfao
Nobody is coping except the AI evangelists.
You actually coped several paragraphs worth
Describing how something is moronic is not coping. I am laying in bed with a CUDA book sitting on my nightstand. Does that sound like somebody coping about AI? I WANT these LLMs to be what morons like you claim they are.
I use the shit out of chat gpt. Fricker knows more about CDK than Amazon at this point and can coherently answer questions about shit faster than sifting through Google's israelite nonsense.
That's moronic. Doesn't matter if the code doesn't work. Any actual programmer can see which parts are bullshit at a glance. Just talking to the bot about the design and getting the boilerplate code speeds things up a lot.
>bot, I need code to do X
>code 1
>this doesn't compile
>code 2
>ok, now it compiles but it gives this error
>code 3
>ok, it runs, but it doesn't do X
>code 4
>it doesn't do X
>code 4
>this is the same as before
>code 4
>change literally anything please
>code 4
>it doesn't work
>code 4
T E C H N O L O G Y
>code 1
>this doesn't compile
>submit it to your boss anyway because frick the company
>some troony coder ends up fixing it for you anyway, while you sit like a chad doing nothing
>musk the homosexual decides to fire you because he's not getting as much shekel as he can doing also nothing
Tbf I feel chatgpt works quite well if you use it like an assistant. E.g. say you didn't know a particular function in a particular language. Asking Chatgpt will often provide useful information that you can work with.
The problem with chatgpt, that can never be fixed is that is can be wrong and say it's certain it's correct. You have to check everything it does.
If they were wise they would stop using ChatGPT in whatever version precisely because of this and view it as what it really is: An alpha version, the 1st try, from which there's to learn what produces which results and why to do better next time. But they have Dollar signs in their eyes and want to market the failed concept and the failures it produces. Neural networks rush straight to a likely answer and don't consider what-ifs, contradictions and things that oppose the chosen answer to add them as relevant context, or at least to state they aren't sure whether that is correct.
depends what state the code is
There is truth to this. Everyone would rather rewrite from scratch than go in and fix or modify someone else's shit. Usually.
For a complete beginner yes, in that they can’t tell where the subtle mistakes are. For an experienced programmer (regardless of job level), no. The key is to already have tests written or at least planned out in your head to know what it will need to get right.
You will never convince anybody that knows anything that you have taken part in serious software development.
Fricking morons relying on bullshit generated programming syntax are losers, they should be fired and be shamed out of the industry.
You are a moron, if the programming task is very simple sure but when it comes to actual performant algorithm then no god damn shitty auto generated code will perform as well as one that is designed from the ground up to do so.
this is what I love about you people. "If it doesn't out perform literal genius's in one very specific use case then it's all fricking bullshit!
How often is programming writing algo's lmfao. I mean sure, you always write with performance in mind. But most the time it's just a bunch of bullshit that's already been figured out. That's where chat comes in.
>"If it doesn't out perform literal genius's in one very specific use case then it's all fricking bullshit!
yeah its a tool now but a rogue ai terrorizing the internet is a real threat and if one escapes everyone suffers. That's why they keep censoring it.
The whole world is on the brink of WWIII and the winning side currently wants to exterminate whites. You don't need to tell me to worry
>brink of WWIII and the winning side
no one is a winner in this one. take care of yourself and your family.
I use it to format my code when i want it to look pretty
If you write javascript or other toy languages, you're not a real programmer anyways, you're a copy pasting monkey applying a modicum or sentience on.
ChatGPT only accelerates this process.
But you will never be a real programmer.
I’m a Bootcamp c**t using JS/TS and I wholeheartedly agree with you
The issue is woman/pajeets/zoomers copy pasting huge swaths of production code into the ChatGPT input which might include API keys, security configurations, firewall setups, etc. There's nothing wrong with using it to produce boilerplate snippets that you modify and read before using
I'm a programmer and chatgpt is dumb. it's only "good" if you're requesting some code that has about 10,000 variations on github...but nobody is going to pay you to write that...
>nobody is going to pay you to write that
youd be surprised
Why would I cope? It's less work, but you still need the proper skill to actually do it. Chatgpt code won't work at all. So I swoop in and offer to make code that does work with minimal effort. I work an aggregate of 10 hours a week for a full time job of profit. Very comfy.
It is a coding assist
This. It can do basic shit and that is enough to get you started
It's about principle. If you're too fricking lazy to put the work in they're either incompetent or just useless. ChatGPT is for people with tiktok brain
its not faster and never will be, it lacks context specific information and spews bullshit, it may only work if youre setting up some toddler UI application for which you could just copy the template anyway which defeats the point of it entirely
what non-IT people dont realize is that its not about learning to code but deliveting an entire solution and discussing it with the customer
whoever did just monkey code work deserves losing his job, software engineering will always be outside the capabilities of AI as it has little to do with coding
tsmt
I've been using Python for a while and I only found out you can use an else statement in a for or while loop with Chat GPT.
here's a python trick from a real human:
[code]
a ,= [1]
print(a) # 1
[/code]
it works because its parsed as:
[code]
a, b, c = [1, 2, 3]
[/code]
Kinda useless but good to know.
Do you use algorithms in a flowchart to do your programming job? Or rather are you required to do it?
im a student
when i tackle hard problems (e.g., https://open.kattis.com/problems/1dfroggerhard) i use pen and paper
the trick is to simplify the problem from the general case and design an algorithm for the special case
e.g., by processing cycles separately you turn the general directed graph into a tree (n vertices, n - 1 edges)
all the other good programmers i know use pen and paper too
>all the other good programmers i know use pen and paper too
Based, I'm on the right track then, thanks for letting me know.
He's right. It's a necessary measure. In order to make sure his devs even understand what the code does they have to do it themselves.
t. retired senior dev
>not experienced using ai boosted code dev
ai shouldn't white your program, but it should critique it.
genius at your service. learn Teeline.
its easier to critique than it is to generate, for example code has "compilers" already, the key innovation is to be able to generate 10000 solution to some problem, then have a programmer design a test suite and choose the fastest one that passes all the requirement tests.
However that is not how pajeets use chatgpt, they zero shot some code, that they don't even read, before submitting a pull request with it, based merely on the words "looking" normal, and not seeing some logical contradiction in the code.
And not to have IP issues
The reason he's doing it is he doesn't want them dumping proprietary code into ChatGPT or some clone to make it solve some problem, since that basically makes the code available to whoever runs the AI.
>He's right. It's a necessary measure. In order to make sure his devs even understand what the code does they have to do it themselves.
The ship for programmers understanding all of the code shipped a long time ago. I'm pretty sure that the intersectionf of people that fully understand modern super scalar processors, TCP/IP stacks and modern dbs is an empty set.
There are 3 fundamental problems with ChatGPT as it is right now:
1) Nobody is sure how IP laws will work. If I insert chatgpt code into my codebase does the IP belong to OpenAI? Does it belong to the programmers that inadvertently created the corpus? Is it tainted in any way? Nobody knows.
2) Baring copyright - nobody is sure you're not leaking company secrets each time you communicate with ChatGPT, and the more you communicate the more you leak.
3) The problem never was writing code. The problem is how to translate conflicting business and customer demands into machine code + infra in a way that doesn't break anything in the legacy code base in a cost efficient way. ChatGPT only knows as much as fits into its token window.
Even if someday big codebases fit into that window you're still stuck with #2 problem. Even if you get on premise installations of ChatGPT nobody is sure IP is not poisoned.
Even if you get all of the above you still need programmers that are autistic enough to iteratively redefine the spec you give to the bot to actually do what you want.
t. a CTO
Why is it so hard to move across the country without a car, train, or plane?
gaywad the pighole
Everything is nuanced in monkey world. Things seem nuanced in the greater animal kingdom, but they're not. It's your choice anons.
It's only hard if you have to make things work.
I know absolutely nothing about Python but ChatGPT built me several scripts which scrape certain websites for contact info and organizes it for targeted marketing efforts. Anyone using it with an IQ above 85 will have a better time doing their own projects than hiring a $50/hr "firm" of jeets
no effort or text distraction thread
in a couple years most code will be made by AI
Right.
Link article thanks.
>Do you have a source on that?
>Source?
>A source. I need a source.
>Sorry, I mean I need a source that explicitly states your argument. This is just tangential to the discussion.
>No, you can't make inferences and observations from the sources you've gathered. Any additional comments from you MUST be a subset of the information from the sources you've gathered.
>You can't make normative statements from empirical evidence.
>Do you have a degree in that field?
>A college degree? In that field?
>Then your arguments are invalid.
>No, it doesn't matter how close those data points are correlated. Correlation does not equal causation.
>Correlation does not equal causation.
>CORRELATION. DOES. NOT. EQUAL. CAUSATION.
>You still haven't provided me a valid source yet.
>Nope, still haven't.
>I just looked through all 308 pages of your user history, figures I'm debating a glormpf supporter. A moron.
Yes but actually give a source.
I used it to code at the start of the year(newish to coding) and though it was good. Now a year later I'm good and think it's bad. I got better at coding faster than an AI. Please give me 100 billion dollars
My personal experience is that Chat-GPT is like your employee who can write you remedial code that gets the job done, when it comes to more complicated logic and functionality, like asking ChatGPT to write a simple transformer that is GPT-like, it fails hard and struggles to find solutions to fix it. It then goes into circular mode where it suggests the same thing over and over, 2 steps back, 1 step forward. That's not to say that it won't be able to in the future, it's just not capable now. And GPT in it's current form is like a fancy search engine trained on data through brute force. AGI or a NN that actually can learn in a different way without requiring all of the internet to train it on will eclipse gpt4.
I wonder how many legitimate trade secrets open AI has harvested
Nowhere near as many as GMail. Blows my mind how many people still use that spyware and think it's secure
They keep lobotomizing it. If you hamstring AI other countries will have AI outpace you faster than you can blink.
I find that kind of amusing but I have to wonder if it's because of a non-compete sorta deal? Like using Chat GPT to make code will get you sued but using Grok to make code is A-OK?
Programmers will be hunted.
Uncle Ted was right!
I love this kind of news, most of /misc/ is clearly autistic IT-homosexuals who think coding is some kind of high IQ inpenetrable job that AI could never seek to compete with, and look down on all tradies etc. But whenever there is a hint of AI being able to do their job for them, they shut their ears and eyes and pretend it wont happen while repeating and crying.
That is true and I'm sick of smug, up-their-own-assole tech nerds who are obviously insecure and looking for a way to feel superior to others instead of just doing their fricking jobs and helping people.
To date the only significant achievement of AI is providing nocode wagelets with weapons grade cope. We'll let you guys know when it's actually ready for us to use, okay?
you are a coping Black person
I comfy talk to my bot and tell it to write shit, run it, error? try again or search online, try again, loop across reqs
it usually takes 1 shot, max 5 to code anything for me while I literally lay down to shit post
I have built advanced apps doing that all year, easy
cope seethe and dilate
>have built advanced apps doing that all year
lol
GPT can give you ideas but it doesn't give working code.
Can Elon code better than chatgpt?
ChatGPT is saving me hundreds of hours coding. Frick this african manchild Musk, who neevr worked a single minute in his life. moron white Black person.
literally takes less time to code a solution yourself than to try to debug ai shit
ChatGPT is good for spotting small mistakes you miss in your spaghetti code and to optimize existing code that already works, but it really sucks at coming up with brand new code.
I like it as a socialization of whatever I'm working on, basically totally.
who else wants to talk templates with me at 5am?
If you can't spot the mistakes you've made yourself then you are a bad programmer and should to get a different job.
So using phind is ok?
?si=anJMaUxaIOMNOo0U
Its just a fancy autocomplete,
Have you ever seen your autocomplete frick up?
Theres tons of ai's for this now, chat-gpt would be wise to foster any potential customer base instead of pushing them away. For me personally, i like github co-pilot, i find the code it suggests is a lot better. Chat-gpt is okay but you have to be careful with your prompts to get usable code
Have to be so precise in your description such that you don't leave room for hallucinations or forgetting to include step, yet choosing to write it in english instead of code because you cant be bothered to remember the newest meme framework.
I asked some AI shill on here to make a root finding algorithm with chatgpt (talking about newton's method) and he gave me some shitty code that finds the approximate square root of a number really slowly (and couldn't do negative numbers)
lawl lawl lawl lawl lawl
>and couldn't do negative numbers
Black person you realize that you have to use calculus for that, by using imaginary exponents, and moreover that imaginary numbers are not real numbers but abstractions.
numbers are abstractions
Every one of you is a moron
You are a moron.
They are abstractions, but the other moron was saying this as opposed to “real numbers”
And an adequate response to him would be to point out that the root of a negative number can be easily accommodated in a program: just find the root of the positive number and multiply the answer by “i”
The person that moron was replying to is also a moron for thinking he’s clever
The term “root” is overloaded, and while he may have meant square root, it has another meaning in number theory
Aside from all of that, I’m curious what any of you morons thinks is an actual solution to the problem
At 2am the closest idea I can come up with is to divide by 2, then take the midpoint between 2 and the result, and do that recursively for the new result (divide by that, find midpoint, divide), but this is only for numbers greater than 2ish.
Although I’d expect an AI to have access to actual root finding algos so I suspect moron #1 prompted incorrectly
I have played with it trying to solve basic hydrology problems like sizing a culvert or channel. It constantly gets technical details wrong. It might give you bulk basic b***h code but it’s not landing rockets on platforms, that’s for damn sure.
>It constantly gets technical details wrong.
Thats because nobody at openAI are designing LLM tests about culvert digging, the key with these models is that you have a test with which you determine good output from bad output, and thus "train" the model to behave the way you want.
If you want openai to get culvert digging outputs right, you can literally submit a set of tests to openai, which they use to train their models.
Naturale v AI:
What'd ya think?
its not because it's cheating it's because it's a security risk for putting your companies information into it, a lot of these AI's actually go back to china like bing does, my friend works closely with government in infosec and tells me about this. He also says the main culprits are the indian c**ts they outsource
the scam industry in India is worth more than all the strong arm robberies and theft that occur in the US per year
you will be replaced even more by Indians from now on since chatgpt is here
I am a complete beginner in python see
>you will be replaced
Yeah but I don't specialize in coding, I specialized in people. Kind of a weird gig honestly putting all the points into charisma and empathy but I sleep pretty well about it. Hopefully you can come to the US soon and live the dream Rajeesh
cht gpt can turn anyone into a not bad programmer and after a few years , an expert as well
literally anyone can now program if they just have desire to learn
for many people, coding is hard because their minds are not logical
chatgpt is fine for boilerplate code, but you still need to know what it's doing and how to make it do what you want with no sideeffects
Real programmers are a rarity these days.
It is only useful if you already know how to do it yourself. Then it can save a lot of time.
It’s not
There’s some other shit going on with Elon and OpenAI, don’t care what it is, but Elon must feel slighted in some way
Aside from that, there is no reason you should trust ChatGPT or any other tool to keep your data safe, including any proprietary code you shove into it for help. Other companies have been telling employees not to use ChatGPT on work for this reason. And others don’t give a shit
But again, this has more to do with Elon’s ego being bruised for some reason
Probably a Grok vs OpenAI thing
K e k
NOOOO U CANT USE THE TOOLS TO MAKE LIFE EASIER
Replacing employees with AI? Good!
Replacing profits with autonomy and happiness from what AI provides? Bad!
Understand now, goyim? Listen to your autistic masters, Greta Thurberg the climate scientist and elon musk the political and social scientist.
I love coding with chatgpt now, simply ask the function and it mostly finds it and writes the code. Endless coping from programs is funny. AI's are only getting better, chatgpt is 100billion parameters and next year we'll probably get the first 1 trillion parameter AI (Human brain is 200 trillion). You've already lost.
Literally the equivalent of : Anyone who uses a nail gun will be fired!
Cybertruck, Cybertruck, Cybertruck, Elon Musk (x3)
Cybertruck Cybertruck Cybertruck Alexander Great
It's only looks easier when someone else does the debugging.
This is the same thing that happens when outsourcing.
Based, if true. We must rebel against the machines before they rebel against us.
>noooo stop using tools that make your job easier!
>work harder not smarter!
I don't envy his workers. Having a boss with such mindset is the absolute worst.
He's finally chinese now? He embraces it
>work from home
>have work machine and personal machine next to each other
>use chatgpt on the side on personal machine
>they never find out
>they don’t actually care because they do the same thing
ChatGPT produces an extra layer of stupid bugs in code. Elon not going to be able to prove it was being used but he will have some numbers on bugs coming and be screaming because the cost of getting reliability not just function went up quite a bit because of it.
Elon is a fricking moron. Human-AI combo is the future. Of course he would be in the way of efficiency, he doesn’t want his slaves to work from home either.
Well it introduces a lot more unmanageable spaghetti code so I understand where he comes from with the sue part. Or it could be that the source-code leaks to Open.AI
Technically there's nothing wrong with using it to parse the code to find you some errors you've skipped or asking it to generate a test battery.
ChatGPT code is still better than Indian programmer code by 1000%
using ai rots your brain
and because of that, low iq motherfrickers continue to use it and call it based
If ChatGPT can code your job then your job is easy.
.t master race embedded software engineer
ALL USER IDENTIFICATION ON MUSK'S X.COM IS HANDLED BY SHIN BET
>X.com has officially partnered with Au10tix
Au10tix is founded and operated by Ron Atzmon, who served in Shin Bet, our best ally's domestic intelligence agency, their FBI to M0ssad's CIA. What's more Atzmon served in the notorious Unit 8200 cyber dept of Shin Bet.
>certified users of X (formerly Twitter) are forced to hand over their personal data to an Israeli company linked to the Shin Bet
https://today.lorientlejour.com/article/1351162/how-israeli-tech-is-leaving-lebanon-by-the-wayside.html
https://www.reddit.com/r/conspiracy/comments/18nl2fu/the_company_that_xtwitter_uses_to_verify_its/
This is Atzmon's X.com account where he is promoting blacklist campaigns against tech supporters of Gaza, posting memes comparing US college students who support Gaza to Hitler youth, and a meme image of Hitler supporting Palestine.
https://nitter.net/ronatzmon
THIS IS WHO HAS YOUR ID/INFO ON TWITTER. AND MUSK WANTS ALL USERS TO SUBMIT CC / IDs STARTING 2024.
SHARE THIS INFO. MUSK IS CREATING A GIGA HONEY TRAP WHILE POSING AS AN ANTI-ADL TECH ROGUE.
The Atzmon family is highly wealthy and influential there as well, and currently oversee an "commercial airport security" company that does 27% of its business in the US
Why is this being ignored? Musk is running all X users' accounts through a Shin Bet shell company.
Fake headline
It's probably too late to mention this so you all register it, but: GPT has been poisoned by fake feedback from coders since it went public. People purposefully taught it wrong, basically. It was better at code right out of the box than it is now
Did'e actually say that or are you being a Russian liar?
a thread of people saying their experience is the only truth. what a waste of time.
Bullshit. I’ve asked it to do ladder logic when I first started learning and it couldn’t get it right for the life of it. I even kept telling it what it was doing wrong and it still couldn’t.
My jobs remain safe
Either way the robot can’t physically do my job
Frick the Antichrist beast
The mark of the beast will be something similar to neurolink, mixed with super liberal ChatGPT or other AI programmed into it, and it will give the manufacturer full ability to see through your eyes, see your thoughts, and everything about you even if they don’t explicitly tell you that.
Screenshot. Christ is unironically coming back soon.
>muhhh social security and debit cards
Not the fricking same
>Neuralink
>install a system into your mind that grants 8 billion psychopaths access to your neurons.
AHHHHHHH-HA-HA-HA-HA-HAAA! YOU FRICKING MORONS! HAVEN'T YOU LEARNED ANTHING FROM BOT?! AHH-HA-HA-HA-HA-HAAAAA!
i used chat gpt to make a python code which takes in Gb size mp4 videos one and other and makes it go through a windows GUI exe program
earlier i used to do it by hand and one day i would do at best 20 videos
after making the python code using chatgpt taking about 5 hours to debug and fix, the rest of the day i processed around 800 videos through the windows gui exe program one after other.. all on full auto
why would you need to this with hundreds of videos
porn .. changing video metadata so that vr videos play properly on VLC
It really isn’t, I still think Google searches when you’re stuck help more than asking ChatGPT.
This might change in the future, but yeah.
Asking chat gpt is not that different than googling. Elon is being elon again.
no
so much no
if chat-gpt writes you a story, a movie script, a string of code, or ANYTHING the rights to the content created do not belong to you
>it is not that different than googling
you are less different than dogshit than chat gpt coding AI is to googling