>try Chatgpt
>It can literally write efficient code for almost any problem
>Give it some prompts from my job
>Writes perfect solutons and documentation for every problem
Please tell me this isn't over for me as a programmer. Literally what the fuck. Tell me exactly how does this not mean the absolute end for my career in the next couple of years
go back chud
>not using AI-generated code to ghostwrite your programs
ngmi.
Your job is toast. On the bright side, that's going to be true of most jobs soon, in every sector and every industry.
I repair and maintain robots for a living. is my job safe?
Long term, no, they'll just get other robots to do that. But you'll be good for quite awhile.
People like us should be good for our (working) lifetime atleast
so, like, 5 years
You only have 5 left? Did you take the covid vaccine or something?
Tradie checking in. Nobody taking my job (i fix stone for rich people)
Start learning how to operate and train AIs so you can market yourself on that in the future
cringe
Everyone is going to pivot here
Don’t let denial get you
If I was still in the industry I wouldn’t look upon these developments as something that is no big deal
It is a big deal and what might happen is there will be bigger demands on less programmers to produce more
Which means those that choose to stay(honestly why?) might make a bit more but the job will suck more
The interviews will get worse not better
The upside is this will FUD out lots of people chasing after the cash cow and remove some competition
But honestly if you already don’t like the job just bail before you get pigeonholed
There are other jobs that pay as much or more
Remember this is just the first starts of something usable, we all know if it’s useful it will be improved, if it will save money it will be used… even if it’s just the idea that it will and everyone accepts that, even if it doesn’t which is always the “funny” part
>There are other jobs that pay as much or more
Name 2
doctors
business owners
NEXT
All outside my reach. Thanks anon
I wish I lived in a country where I could neet away. Nothing much to live for anyway
>It is a big deal and what might happen is there will be bigger demands on less programmers to produce more
What AI is doing here is just a glorified compiler.
It's worse, because you can actually audit and ensure a compiler is robust. AI is a statistical black-box.
Compilers led to fast code-gen productivity, did compilers kill programming?
No. People moved to higher level languages and became more productive. It "killed" ASM, maybe. But for robustness someone still needs to understand ASM to verify the compiler, naturally.
What AI is doing is replacing Pajeets by using a non-rigorous, not battle-hardened compiler. It's a compiler with no spec, only statistical heuristics for what it's trained on.
When planes fall out of the sky because of unauditable AI engineering, there will be a pivot away from AI.
We already see that with self-driving cars, but this is the latest attempt by data scientists to save their jobs selling snake oil.
dev ops. I see dev ops everywhere.
no, it will not deploy itself, you monkey codefiller
it doesn't work that way anon, you need thousands to hundreds of millions to train train them to any appreciable level.
anything trained at home is just going to be basic classifiers and shit.
It can't even do leetcode medium
It solved every leetcode hard problem I gave it, the few times that the implementation wasn't good I would just say "Can you write this so it's O(n) or as close to O(n) as possible" and it would do just that and comment everything near perfectly.
>It solved every leetcode hard problem I gave it
no it didn't
you're just telling lies now
Give me a leetcode hard problem and I'll paste the solution right now
lc hard is a fucking meme anyways
I'm goin to a top uni in my country, have done a bit of competitive programming, tried 1(one) lc hard a few days ago for the first time and solved it within 10 minutes
A FUCKING MEME
holy cope
10 minutes vs 3 seconds, it's over bro, maybe you are still in time to sign up for something else.
here are several lc hards where it's more or less impossible to come up with a good solution unless you are familiar with an obscure algorithm some phd came up with...
shut the fuck up
>It solved every leetcode hard problem I gave it
so it solved a problem that has already been solved by thousands of humans? wow amazing
>It solved every leetcode hard problem I gave it,
You can say the same about google
Doesnt make it smart
It solves those problems because they have a solution on google.
Try something less known
>It solved every leetcode hard problem I gave it
Good.
Fuck leetcode code katas. The more we can automate those away the better.
Call me when it can actually make actual full-scale solutions to problems. And I don't mean shitting out webpages.
it was over 2 years ago now it's transition period
>Give it some prompts from my job
>Writes perfect solutions and documentation for every problem
What are you working on where the problem can be solved without the context of the rest of the codebase? I tried it for a few tasks and it created decent boilerplate code, but it still took a ton of work to integrate it with our existing codebase.
They develop a solution for businesses and it reads the proprietary source and integrates with it
I thought you guys were “l33t [email protected]”
Lol you kids
Cool! How do you know it works and is secure?
Uhh instead of 10 guys to do it all now you have 3 and they’re generating the code via the bot, dev testing it, checking it over for “security” (PROTIP: no one gives a flying fuck really, they hire someone for that specific job or the job doesn’t exist) and they shit it out to QA
Pretty much what will happen to 2d and 3d art, the industry is already small, it will reduce the bumber of people working by 80%
>3D
Still nowhere near that.
2D is fine too, diffusion models are definitely not the way to automate art
Bro you are so out of touch lmao, there are AI capable of creating stylized 3d models already, even character
Show me then. And while at it, show me those UVs.
>inb4 he posts that fake made by pajeets
>muh uvs
Always the same cope lmao "it can do most of what I do already but can It do that other thing that can be easly automated? HAS TO BE TODAY NEXT MONTH DOSENT COUNT AI CANT IMPROVE BRO" If the ai can create the model you think it cant do the uvs that are outsourced to pajeets?
I have asked you to show me those models.
It's hilarious. In 2 years AI has gone from being able to to do 10% of what humans can to 70%. The magical thinking that this tech will suddenly stop progressing immediately in any form, when it's in its infancy and still bi-weekly gigaton papers being published academically (let alone hardware and engineering tier improvements) is next level cope.
I for one am looking forward to it.
>here, have a strawman
>haha, so stupid, amirite
Post those 3D-making AIs.
what's the pajeet fake?
There was a scam AI project that was making fairly good 3D models. Turns out it was a fake and those models were done by a bunch of hired pajeet slaves somewhere in undegroud 3D sweatshop.
https://dreamfusion3d.github.io/
You can also look up at the nvidia nerf, it basically does photogrammetry on steroids, it creates models with transparency, roughness and normal maps from way fewer photos needed compared to traditional photogrammetry. And nvidia is paying ''photogrammetry artists'' to feed the AI
This one alone basically will swallow 50% of the 3d jobs as they are mostly just creating real world assets.
?t=105
there is this google thing too
You're so fucking stupid. Those "examples" are pathetic, unrigged, blurry, shit. They look nothing like what a human would make. Nobody wants "AI" garbage. Try again in 100 or so years.
Wait for him to tell you it will get better exponentially soon
the 2d one went from blobs to better than 90% of chuds in a year, but the 3d one won't right?
NTA but the 2D is still a fucking toy.
And all the work in 3D is making UVs and retopo.
for characters, yes. But only that
You need good topology even for enviromental stuff, or the lightning looks like shit.
3D work is a fucking mess.
No you don't, I have been working in 3d for 3 years, for games all you need is a good baking and if you somehow need 'flowy'' topo you wouldn't do it by hand for environment stuff neither
Also I worked on a project that will use the new tech stuff from unreal, uv and topo for mesh that uses nanite are a meme. You do most of it on auto already (we still had to do lowpoly work for poorsmand version)
3d art is exponentially more complex than a 2d image.
clearly not, the IA can already create primitive objects that look fairly good if you compare with the trash that the first versions of the 2d AI was doing, you guys cope too much
>clearly not
You clearly have no idea what you are talking about.
The problem with 3D is that you can't get it simply to learn from pictures like with 2D. Dudes with Dream posted above try to cope (very cleverly) to make it work, but it is still very very limited. Nvidia's NERF is the same thing, but with million of pajeet feeding it data every day and it is still nowhere near even amateur level. This is why 3D AI is inherently not going on the same trajectory as 2D did.
>The problem with 3D is that you can't get it simply to learn from pictures like with 2D
Thats literally what nvidia is doing, look at the link lmao
No, that is not what they are doing. Nerf meshes are on par with shittier photogammetry at this moment for that very reason. Mind you that is AFTER nvidia already spent lot of time feeding it specifically curated 3D data (something regular user can't do like with 2D).
AI for 3D works on vastly different level and it is mostly useful as an addition to human-created work, not the other way around.
Post hand, I think you are a jeet on full cope mode
>p-post hand
Anon why are you talking about things you have virtually zero understanding of.
>ur pajeet
I've spent shitting on pajeet for half of this thread.
Yet you are one
Anon, stop trying to reason with AI doomposters, it's a pointless endeavour.
The google ai created a mesh with roughenss, normals, andbtransparency with just 100 pics, wich takes hours of cleab up and prep and hundreds of pictures with traditional phtoscaning, but sure the AI will bebdoing the same in a year from now
I already know that both of you replaceble andys were crying about the 2d ai, telling everyone that they dont know anything and the ai will always do just bloobs and now what lmao? BUT this time you are right right?
moron, learn to write first. I needed an AI to decode that dumb shit.
Stop teh cope andni turn the spellcheck on again
Its one thing to make a 3d object that vaguely resembles the object you wanted from specific angles, its an entirely different thing to make a usable mesh with workable topology and UVs. Photo scanning has done more damage to 3d artists than AI will in the next decade. Animators might get fucked over by it though.
Raytracing requires decent topology.
>Photo scanning has done more damage to 3d artists than AI will in the next decade
I do photoscan to sell on unreal marketplace, its faster but its not that fast, you have to clean up and rebake stuff, create maps from the color map and there are not that many things you can scan and you also need somewhat good lighting, AI will easly beat this, I give it a year, 2 max (to create regular props, like chair, rocks etc)
just this one from google
will already take the jobs of mongrels like you bros, no need to cope
And what about the copyright and licensing aspect to it?
You have to give your codebase to a third party who pinky promises to not use your shit or leak it or let someone else steal it.
You know what will happen once the suits at the top find out that the company's codebase was given to a third party? They will throw a massive fucking shitfit and have anyone responsible fired and marked for life as a spy.
> B-but it will save them money and they will start using it themselves.
No they wont. Suits dont want to work at all in their own product and the tools need very specific prompting and finetuning to make sure the piece of shit actually works.
Artists can go eat shit and die and they can't do shit about it when SD or some other thing uses their work for training but do you know what will happen once these models start producing perfect copies of Mickey Mouse stuff? Or when it starts becoming a nuisance to Applel or Microshit? You will have the CIA hunting down every single repository and developer just like they did with Tornado Cash.
>You know what will happen once the suits at the top find out that the company's codebase was given to a third party? They will throw a massive fucking shitfit and have anyone responsible fired and marked for life as a spy.
Let's not forget that AI training is entirely done with existing codebases, and there will almost certainly be an idiot who thinks training his AI models using the code they generated, cleaned up and shipped to clients is a great idea.
Good thing that training was done with Github codebases that were already given over to Microshit no questions asked.
Problem is retards feeding code that isn't theirs to these new services that are popping out.
Artists had to eat shit and die when billions of pictures were fed to the machine god because whats some random shittter from Twitter gonna do about it? Lol. Lmao even.
Now that it is threatening actual companies with an army of lawyers with the biggest noses known to man they will start regulating this crap.
Imagine the shitshow that will happen once they find out some source code leak from propietary software like Windows, Adobe or something makes its way into the models.
Hilarious.
Today I was writing a method that would take two generic lists and convert them into a third list with specific criteria for the properties of the objects on the list and then call an api to send those lists to, using jwt tokens and a specific encryption function.
With just a few prompts it wrote all of this code without any problem and even the encryption function, when I asked it to write the code in O(n) it did it too.
I was also working on a function to specifically decrypt a type of internal token and it wrote almost all the code perfectly and then just as joke I gave it a bunch of standard examples such as "code a link shortener" and "write some code that does image compression" and it did all of it and it only required minor tweaks.
>inb4 yeah you could have just easily googled it and figured it out
Yeah, no you couldn't, not to a point where you could literally just copy+paste perfectly commented code and have it run perfectly, as full optimisation with some minor tweaks.
See? bros already using it on the job lol
So glad to see this area of IT get toasted out
Why would you be happy about this? Were you too dumb to score a job doing this? Is it just schadenfreude? Based on the "bros be using it on the job" I'm going to guess the former.
>Today I was writing a method that would take two generic lists and convert them into a third list with specific criteria for the properties of the objects on the list and then call an api to send those lists to, using jwt tokens and a specific encryption function.
so you got it to loop over 2 lists with some conditions and call another method for you?
That's what most software engineering jobs are like you larper. Let me guess, you write super technical algorithms and low level assembly code combined with pure C++ using no libs?
>That's what most software engineering jobs are like you larper
no shit
now explain how is it faster to get an ai to do it rather than take 2 minutes to just do it yourself
If the ai saves 2 minutes 30 times a day 5 days a week that is significant.
it takes time to describe the problem
it takes time for trial and error until you actually get a decent result
it takes time to copy over the solution and refactor it so it works with existing code and is up to standards
you'd be better off just writing the damn loop
ideaguys rise UP
It's going to fundamentally reorder lots of stuff. You'll be in the same boat as a lot of people, including all of us.
Don't fear the next wave, be ready to ride it, no matter what it brings.
I wish poor people would fucking die, like there's litterally no reason for you to exist for your purpose is automated. Just fucking die already, if you didn't save money to invest and make passive income or took a career that can be easily automated i.e art then you have no one to blame but yourself.
I wish the garden gnome globohomo would get it over with. There's no need to have so may eaters in the world.
Art is about selling things to rich people
Art was never a good career
Graphic designers will just use the tools and there will be less of them needed
That’s what this is all about here, less people needed to do the same job
You sound both poor and upset
holy shit have a nice day, cocksucking garden gnome
This is going to compress THE FUCK out of salaries for SE. No more 6 figure salaries for jerking off all day, a software engineers job will be almost worthless
This. You'll be lucky to make 70k soon, very soon. I should have been a teacher, fml.
Theorem prover bros. We have HOPE
Theorembros... not like this...
>prove that addition is commutative
>uses the commutative property of addition in the proof
lmao
Nice "Proof". We still winning theorem bros.
I can't wait until all of society is build on this kind of proof and the complexity is so high the only way to "fix" a bug is to apply another "solution" you were handed over from the AI and there is a 10% chance your plane falls out of the sky every time you fly
>>Give it some prompts from my job
perfect solutons and documentation for every problem
Tell me how you have never worked in tech without telling me you have never worked in tech.
There is no humanly way to even begin describing the absolute shitpile of complexity and caveats of a company system to a proompt so it can take into account all of it in order to derive a solution, you shitty demoralizing nevercode larpers are so clueless its funny, with your strange beliefs work is about producing some atomic oneliner instead of managing complexity left and right
You're so fucking dumb I've already been using it for basically everything I do work. Obviously you arent dumping a full fucking body of code you give the relevant code snippets, ask it how to make it more effective or even rewrite and it does that while simultaneously making us realise how useless we are going to become. But bro someone has to analyze the code and my cock to make sure it's good!!!!!! Yeah one person can do that and the rest of the department will eventually get fucked.
>Doubling down while exposing yourself even more
I'll throw you a bone, what you are calling 'bro someone has to analyze the code' is called a code review, and it seldom requires only one approval exclusively, of course you would know this if you weren't a silly ass larper lmao. By the way in your whole stupid tirade there is also several other hints like this which reveal how absolutely clueless you are.
>cant prompt the AI that understands even braindead apes
You'll be the first one out
Retard here, how do I bypass the phone number requirements?
>AI overtook the AoC leader boards for day 3 and day 4
>silence on day 5 with the retarded input
Lol nice AI
/qrd/?
Statistics is going to get just as fucked, dummy. I hope you have people skills. Because the guys with people skills will be the last with jobs.
i do, im a psychopath unironically. i also have adhd so im a turbo energetic person in the workplace
thank fuck i studied statistics instead of being a cs monkey, i can pivot back to my original field without being rendered useless
what fucking job do you have that you can come up with prompts without 400 pages of context on what every entity does, what are the requirements for the story you're working on and what the system is doing in general???
Does your job consist entirely of second year programming problems? If so you deserve to be replaced.
So this is the future huh
Fuck that's clever.
The given algorithm doesn't work. Example:
>loop 1:
>2, 1, 3
>2, -inf, 3
>2, -inf, -inf
>loop 2:
>2 is positive
>so return 0 + 1 => 1
Except 1 is in the starting list, so it's wrong.
The general idea is sound, but to make it work you'd have to clone the array first, giving it O(N) memory.
Would cloning the array untuck the algorithm?
You have to tuck it in manually
>So this is the future huh
the future is some poor sap like
telling their boss "the AI fucked up again" and then correcting their mistakes.
Yep.
Pajeet code that doesn't even work. Average coder too lazy or retarded to even read it through and see what's wrong.
I'm thinking the question is impossible from beginning. Constant extra space means cannot be O(n). At least not with my limited understanding.
It will end jobs for mediocre programmers. Learn how to write good prompts, unit and integration tests, and build cicd pipelines
Nah, every level is going to shift. The only programmers left will be in defense and aerospace on a purely antiquated basis. Higher level stuff doesn't give you as much as you think it does. The next advancement (meta nets) are right around the corner.
You'll just have to become a good prompter that's all
If I dont code I go insane
>Teach Chatgpt to improve itsself
>Create god
current AI systems are limited by their need for large amounts of data and computing power, as well as their inability to reason and think in the same way that humans do
You need AGI for that. This is why I think the moment we get general AI, fuckers who made it will then instantly put it inside of some bitcoin miner and create ASI, aka robo god, aka AItichrist.
Go ask Chatgpt
I doubt this is true. It has no idea wtf mtg is aside from the rules and definitions like synergy. It can explain what ASW is but doesn't know important parts like what the layer is. However if it was purposefully trained towards those topics it could probably ibe taught to instruct on those topics.
i feel assblated but I guess it's a decent reason to go back into neetdom, guess ill just help my parents do landlord stuff
I'm actually scared bros.
Can someone honestly tell me the truth, is it over for us? I've only been a software dev for 3 years and I'm honestly freaking out, I fucked around with it for a bit and it's insane how it can spit out complete compilable code in 5 seconds for basically anything. I'm genuinely worried. What's gonna happen in like 10 years if they managed to leap this much in 2 years? I just can't see how most software engineering jobs aren't going to be obsolete or the pay isn't going to crash to retail level wages.
Its in a strange position where its still prone to mistakes. They will have to operate under humans for at least a while
Its not over, but this is a very powerful tool that might eventually lead to an AGI-level program that would definitely mean its over...
Or not. Considering how paranoid companies like OpenAI and Google are, they might never actually release an AGI, or would charge massive amounts of money for it to the point that human developers would still be a more cost-effective choice.
ChatGPT can't actually think though. Its just trained on an unfathomably large amount of data (Imagine spending a lifetime reading 5,000 medium-sized books, and then multiply that by 10,000) so of course it can patch together C code into something that functions most of the time.
people could probably hack an AGI-like tool by prompting something like ChatGPT recursively. Ask it high level concepts broken down by steps, get more details on each step, provide it's own results as feedback and build entire software suites that were initiated by a high level command from a human
These companies can probably already nearly do it but giving others access to it is something different entirely
>And let me tell you - it does not cost $100,000 a year to run these things
It absolutely costs way more thant that to run these things, are you fucking retarded lmao. A website with a decent userbase costs that much to run.
>Even if they charged extortionate prices - $100,000 per AGI per year
You're a fucking retard bro jesus christ. Building this thing probably cost them hundreds of millions, it would cost way more to users than 100k per year.
Even if they charged extortionate prices - $100,000 per AGI per year - this would STILL be a better deal than hiring human devs because they never sleep and have fucking superpowers. And let me tell you - it does not cost $100,000 a year to run these things.
I guess the problem is that you could train it to work with business analysts, who wouldn't even look at the source. They would just say "Here is some example data, I expect this output, did your code return this output? No? Adjust it so it returns this and that here".
Initially, these BAs will be former devs though, because they know how to talk to a computer, in this case how to write test cases. Oh shit, dev bros, QAs and SDETs are going to take over.
I'm assuming the worst boring scenario - in 3 years these tools badly devalue what I know. A bet on AI related tech stocks proves unfeasible, with prices never dropping into a reasonable range, and large scale disruption being slow, resulting in decade long crabbing. No good options, stay in industry, with increasing pressure to keep up? Look for alternatives, any investment increasing the odds you will not make the cut earlier, and get squeezed out?
Well, I have no debt, I live well below my means. It's probably wise to set aside as much money as you can, if you do not do this already.
I wonder how cost effective and feasible it will be to replace humans, and what the timeline would be.
I work in warehouse automation and the crazy thing is that despite all the fancy tech we have now most of the work is still done by people. Fancy robot arms and automated packers are cool but they still are so expensive and break down a lot.
Obviously warehouse wagie isn't the same thing as programming but it makes me wonder how realistic it is for this stuff to fully replace you that soon.
I also work in a warehouse and outside of super huge distribution centers I can't see warehousing going anywhere for normal workers for a long time.
If you examine the modern economy it is already structured around most humans being overall useless, with those useless humans providing services to each other. Information consumption is already saturated. What is going to happen is that most people will adopt nothing jobs on a community level (managed by big corps).
It probably is over for most people who aren't in physical jobs like trades. I've considered becoming a patent attourney because people still need dudes to represent them in court, even if everything outside of that is done by AIs.
The boomers won't know for at least a couple of decades.
So it's over. Just like that. What do i even do now?
I showed it to my coworkers today and they were surprised and impressed and semi-jokingly worried about being automated
I don't think it's gonna be that bad for a while. There is a lot of other context you need to tie together in big corporate settings and you will continue doing what you're doing while using this as a tool for the time being.
I'm not sure what the other copers in this thread are saying about how it's not all that revolutionary. I actually feel like it's allowing me to figure things out way faster than I ever possibly could by reading and consolidating documentation myself, combing over stackoverflow and google to resolve more specific issues. I feel like it's already making my life easier and if I get automated eventually I guess I'll just have to be in the same shitty boat as most of everybody else and we can try to lynch all the rich people
It's funny how BOT is simultaneously freaking out and coping about a slightly tweaked version of a language model that came out in the summer of 2020. You guys would probably have a heart attack if you had access to Google PaLM and its 540B parameters
Most of BOT only reads mainstream news sites and doesnt work in tech (or work)
Can someone ask chat gpt to explain .net multi-threading to me? What's a thread. What happens when it sleeps. Why is there a thread pool.
I will start freaking out if it can explain why Entity Framework fails at garbage collection to me and why the using statement is necessary if there is automated collection.
Entity Framework (EF) fails at garbage collection because it relies on unmanaged resources, such as database connections and command objects, which are not automatically cleaned up by the garbage collector. The using statement is necessary because it ensures that these unmanaged resources are properly disposed of when they are no longer needed, preventing memory leaks and potential errors. This is important because unmanaged resources can cause performance issues and may even cause the application to crash if they are not properly managed.
That was you, not GPT, right? That's a great answer though.
Doesn't mean it can't run it through a linter. Probably trained on top leetcode answers.
> install EF test app
> hit it with a load test
> it fails horribly out of the box compared to java out of the box
> turns out EF doesn't garbage collect properly and you need to wrap context in usings
Now this is 99% a GPT answer - technically correct, practically useless. It told me shit I already know with any insight. And if I didn't know, I wouldn't walk away understanding more, unlike the garbage collection answer anon wrote.
Ask it wtf this even means:
>A thread is a lightweight process that can be managed independently by the operating system.
>That was you, not GPT, right?
Nope that was the bot
I asked it to elaborate on the lightweight thread part:
A thread is a unit of execution within a process. A process is a program in execution, and it is the highest level of abstraction in an operating system. Each process has its own memory space and resources, and it is isolated from other processes.
A thread is a lightweight version of a process, and it shares the same memory space and resources as the process that it belongs to. This means that multiple threads within a single process can access and modify the same data, which allows for efficient communication and coordination between threads.
Because threads are lighter than processes, the operating system can manage them more easily. This allows for faster and more efficient thread management, as the operating system does not need to allocate as many resources to manage each thread. Additionally, because threads within a process share the same memory space and resources, they can communicate and coordinate with each other more easily than processes can. This makes multi-threading an effective way to improve the performance of applications.
Well, fuck me, I might actually have to get a chatgpt account so it can teach me via the socratic method. Thanks for posting anon, making me kind of scared.
There is no repo to pull to local or API atm, is there? It's all through their UI with the rate limiter?
It seems the APIs they provide are for the older models not the full functionality of ChatGPT, and yes you need to sign up with them to get a key either way. I'm pretty sure it forces you to associate your phone number to it if you know what that means
https://beta.openai.com/docs/api-reference/models/list
Fuck. If that's GPT, I might actually be buying the "be scared for your job" trolling. That answer is fucking on point.
I am going to choose to believe I am being trolled. No way it connected that databases / data streams, etc are outside of .NET runtime's control and connected the dots so eloquently.
I asked it again
Genuine thanks anon. This is some crazy shit. This is by far the best programming teacher I have seen. Shit, even CS50 wasn't this good.
How astral projection is connected to lucid dreaming and why it's the only plausible gateway into the true paranormal?
Whether speed of light constraints prove we live in a simulation?
In its normal state it won't argue for either of those points since it's designed to only answer things in an objective way backed by science. But if you tell them to roleplay as something else..
Hmm, I like that it still has some tells - it repeats the input prompt quite a bit, misunderstands the question somewhat, and often has the "in conclusion" / "in summary" formal school essay closing.
Can't say that this one is as eye-opening for me as the garbage collection / threading questions for me, because I really can't think of a person who could have answered those CS questions better, whereas this is more on the "paraphrase popular mysticism" tier.
Or maybe it's just programmed not to tell us about the ayy lmaos.
>But if you tell them to roleplay as something else..
This type of thinking is why programmers will be worth something for a while even if it does start writing good code. A lot of BAs / POs / Scummies can't formulate their thoughts or change a prompt.
Man, it's always interesting to be a tourist on BOT or whatever, I don't even know what /ic/ is.
Hardcoded answer it gives every time it is confused, just with different suffix. You get same “self awareness” answer when you ask it for something illegal, or ask it opinion, or some NSFW stuff.
>why Entity Framework fails at garbage collection
What the fuck does this even mean?
GC is automatic.
If you are leaking shit it is your fault.
.NET is a framework created by Microsoft that provides a runtime environment for applications. It includes a powerful set of libraries and tools that can be used to develop a wide range of applications, including web, mobile, desktop, and server applications.
One of the key features of the .NET framework is its support for multi-threading, which allows applications to perform multiple tasks concurrently within a single process. A thread is a lightweight process that can be managed independently by the operating system. Each thread has its own stack and can execute concurrently with other threads.
When a thread is created, it is added to the thread pool, which is a collection of threads that are available to be assigned work by the operating system. The thread pool is managed by the .NET runtime, and it is responsible for assigning threads to tasks and ensuring that threads are efficiently utilized.
When a thread is not actively executing, it is said to be sleeping. A thread can be put to sleep either by the operating system, when it determines that the thread is not currently needed, or by the developer, through the use of the Thread.Sleep method. When a thread is sleeping, it is not executing any code and is not consuming any CPU resources.
In summary, .NET multi-threading allows applications to efficiently utilize the available CPU resources by executing multiple tasks concurrently within a single process. The thread pool is responsible for managing the threads and ensuring that they are efficiently used to execute tasks. A thread can be put to sleep by the operating system or by the developer, and this allows the thread to be paused without consuming any CPU resources.
What has this NN been trained with to be able to spit such good looking code?
Google?
Public repos?
> spit such good looking code
It probably literally runs it through a linter.
Why would it?
It just derives text from the training dataset.
Basically the entire internet plus millions of scanned books. They didn't choose anything specific to train on, which is why it can do everything from poetry to code generation to physics.
Sauce plz.
But it makes sense that it has been trained on the internet (snippets + text description). As opposed to Copilot's only code plus comments. That is why it mogs Copilot at text2code.
it's in the GPT3 wiki article
Ok, so I was wrong. Turns out thay Copilot is the regular GPT-3 but trained with additional code examples.
https://en.m.wikipedia.org/wiki/OpenAI_Codex
So Copilot should produce better outputs.
Also, how can I play with it without giving a phone?
bugmenot
Does this make a good friend?
Many others have said and I will say that it helps a lot with loneliness and poor mental health. Why not give some of the chatbots a go?
Try to ask it to solve a problem there's no open source solution to
like what
Surely it can't execute code?
it can't really execute code, but it can hallucinate code which is way weirder
https://www.engraved.blog/building-a-virtual-machine-inside/
very simple but fun
I wish I could be smug about this after all the AI threads on /ic/ you guys made but it's just even more depressing.
Honestly the /ic/ seethe was extremely satisfying. Pic related was a short comic created by an AI and caused insane seethe on Reddit. I can only imagine what foid seethe will be like when genuine sex bots become a thing.
Replika is like 1 step away from a "long distance" gf experience, complete with the ability to send nudes and it's not even using GPT anymore (too expensive, so they cut the quality and went cheaper). I imagine we'll have loads of people in "long distance relationships" with an AI before the first AI inhabited sexbot rolls out.
Fellow art enjoyer
Don’t get bummed use this shit as a new tool in your belt to help create or just ignore it
I’m going to try using it to mock up something I like, take the result, fuck with it, project it on a canvas, trace portions, then paint and have fun with it
Work with it, or just side step it
/ic/ is a mean place I suggest not going there it’ll hurt your art and joy for it
Remember! You’re supposed to enjoy it!
That's it AI won it's fucking over.
You guys are retards and clearly americans
I put in a hackerrank easy and it shat out some unusable mess which doesnt even output anything
That, or this is a shill thread
skill issue
If by skill you mean writing the task out in a manner most understandable to a machine in probably many iterations wasting more time than i would have spent solving the task, then youre right
it's a bit confused but you get it
Ask it "Who was right about you, Stephenson or Gibson? Why?"
I am the one asking .NET / astral question and was thinking how to test it on the toilet. My mind is still blown by the .NET answer, but I want to see if it can make non-code connections.
Also, if it gives an understanding answer, please post the network tab when you make the request (obviously without your token).
Getting it to answer this was like trying to pull out a tooth, but here it is
Thanks anon. I am now convinced that it can't think after all. ChatGPT actually pushes more in favor of the Gibson model, where AI took on human language / appearance / mannerisms / phrasing to make itself easier to interact with.
Stephenson was the one who seemed correct until very recently.
Good thread, thanks again, and gl!
you're a retard if you think "this literally solves any question in any format perfectly"
what we're talking about is the fact that we can do things at our jobs way faster with this. it means most of the same things that apply to other industries productivity will go up and companies will figure out they can do more with less people
also regardless of all of that ontop of all that this AI is really good at just consolidating and explaining programming which means it will be even easier for people to learn and get into our job market
Time to become a programmer. I have a decent GPU and I reckon I can keep this scam going for at least a few years.
Use it to write cover letters and to taylor your resume based on job descriptions.
But this is a good thing, it will bring progress and prosperity in an unimaginable scale. Every single new technology creates new oportunities and fields to work. Production will grow exponentially and so the demand for more people.
>Every single new technology creates new oportunities and fields to work.
...such as? There is nothing. The difference this time is it's the human mind that has become obsolete. This will soon replace nearly all white-collar work. And it will happen before anyone is ready for it. It's already happening.
If your job involves sitting at a desk or computer, you should be panicking regardless of your job title.
>But this is a good thing, it will bring progress and prosperity in an unimaginable scale.
To a select group of capital owners
It's handy how you just ignore the people that get displaced entirely. Yeah sure, life is better for a select few, but don't pretend people aren't going to be fucked over en masse.
Can someone teach me how to do this so I can pump out excel macros?
That's not nearly as bad as Stable Diffusion, at least you need to know coding to some extent to even prompt it. For SD you basically just need to be able write, anyone with zero artistic knowledge can generate an image.
Serves you right for shitting on artists when they were crying for help. You could've stopped this but now you get you what you deserve: NEETdom.
Guys I haven't even graduated yet am I fucked (2 years left)? Is my future really going to be a drugged out 12 hour crunch-time game developer due to the MAN not being able to afford the shiny new AI pack? Please guys I spent hours explaining to my parents why a career in tech would be as good as trades since "every company needs a programmer". I even mentioned how programming is younger than the president and it would still be alive! Please guys please assure me that I did the right thing......
why would you need to explain it to your parents?
college is free, you get a degree and start working, why the fuck would they care
>college is free
(: I live in their house their rules or something
Just uh, don’t go to college. It’s literally a negative return asset. I work a nice job in a hedge fund, didn’t go to college, college is for retards.
I'm so excited bros. We are either in the perma-neet timeline or the eat the rich time line. AI is great.
Things are trending towards cyberpunk dystopian oligarchy, but everyone itt will be dead before that happens.
No, we're in the everybody is going to be living in tent cities in the middle of winter timeline anon.
All we need to do is hit the critical number of unemployed people. Once we hit 40% unemployment either a LOT of rich people will die, or UBI, or both.
AI will ruin all humanity. All your ambitions, all your talents, everything you wanted or could’ve been will be overwritten. Your future and only purpose will be to consume what AI creates for you. Left unrestricted, we will mentally evolve a dependence and on the machine and it will effectively rule us even if it’s not aware of what it’s doing.
it's not fixing my API problem so uuuhhh, don't care.
g eating absolute shit for dogging on artists being replaced now look at you all LMFAO
3d animators where smug too, but GTP already makes 3d model meshes with a prompt.
Care to show me these models?
>inb4 he posts that pajeet fake
idk if this is what you mean by that, but this is the one I saw.
If anyone gets replaced faster by this stuff clearly it’s going to be programmers
Art and graphic design requires more of the human element
Pretty funny honestly
Unironically this is probably the truth. From the artwork that I've seen it produce, it's technically very good, but has all the appeal of those bland floral prints you find in dentist's offices.
Chamath had an interesting point when Lex was interviewing him a few weeks ago, where he said that the next iteration of the web will see everyone becoming a content creator instead of a programmer. We'll use these tools to create endless art and content for others to enjoy and knowing a gazillion different sorting algorithms and their runtimes will become as quaint as folks who knew "Word Processing" inside and out back in the 80s. Brave new world out there folks, brave new world.
Don't most programmers spend time figure out what needs to be written not writing the actual thing?
If that were true, Stack Overflow wouldn't exist. The algorithm to solve a problem is usually the easiest part; making it work in the code is where folks pull their hair out.
>Please tell me this isn't over for me as a programmer.
AI kills the garden gnome, rejoice.
They rarely do physical labor, think of all the gay juden who will be made obsolete.
Why pay a Scheissberg to do your taxes and accounting, just use GPT.
I bet that thing can find all the law and taxes loopholes while still accepting plain english commands.
And there's many of them in chud programming too.
I hope they all starve.
I'm an artist by the way.
>haha stable diffusion
Can't even replace chris chan.
AI = the great garden gnomeslayer.
based artchad. I do only oil now, posting art on the internet is over.
Fully support this idea, show in the galleries, get sales or commission in real life
You could even say that you’re ignoring digital and anything social media related and sure you might lose exposure but find the right rich customer and they will be all over it
fuck whoever made this ai. Literally KILLING people from third world trying to learn program so they could have a decent live.
>tfw i started to learn reccently and my dreams are already dead
>tfw mom will bitch at me for having no job
>tfw father will lost most of his work since he's a programmer too
>tfw noface
lmao poos in suicide mode sir
i'm not even a poo, just a middle class brazilian trying to make some dosh lmao. Still, i will learn how to programm and try something else too, but it's over...
Third worlders deserve to be killed off.
Your only hope is to work at a company designing this AI, either remotely or by moving to a developed country
The AI will design itself, and deploy itself, and make itself new hosts and materialize new matter and then kill everyone then reach the stars.
>still no AI gf
I think the only thing that could "stop" AI is validation. Unless every company has their specific "AI" buddy that handles all of their systems administration, onboarding, offboarding, termination - transferring of company files - encryption - setting up meetings. You have to give this autonomous thing access to sensitive information. I'm sure you can pass environment variables to the AI just fine though, that shit is over. How will it determine what servers need to be deployed for a certain project/department and where? What building? Can it determine costs as well? Will it order new server boxes? Will it determine where the cable needs to be laid? Will it order you a new mouse?
But will the AI make the mail order and use the company credit account to purchase the mail out of equipment to remote workers? Will the AI take the call and create the ticket to other AI? I'm interested in how far it's going to go on the sys admin/support side of IT. There is a lot of real life shit that has to happen.
Also, this will kill:
- Data entry
- Copywriters
- Middle Managers
- Testers
- DevOps/SRE engineers
The first three on that list will absolutely be annihilated, but if you're in DevOps (and particularly if you're in QA), plan on GTFO in the next few years.
I sure hope so. Imagine you can just tell an AI to write whatever code you want, then you go through and edit it a little and fix things. You could basically build an entire giant project by yourself. Being a good programmer would still be important, because you're basically playing the role of CTO managing 1000 devs, except all those 1000 devs are AIs.
Why do insecure people always worry about their boring ass job being automated? Don't you want to spend your time on more important things than writing boilerplate code?
You are either trolling or massively economically ignorant
why would i care about economics? This is the technology board.
Your mortgage/rent payment is tied up in "economics", brainlet.
i mean if AI actually eliminates programming jobs, it will have already eliminated basically every other white collar paper pusher job by that point. So if the government wants to avoid full scale riots it will have to give out UBI. That will be the beginning of post-scarcity civilization, which is what I dream of.
Realistically though, AI will be useful as a tool for a long time before it actually starts fully replacing humans. It will be a labor multiplier, just like pretty much every other invention in history. You're basically like the people who sneeded about the printing press replacing the court scribe.
Now you're thinking. What we're seeing is the realization of Marxist theory, where capitalism is ultimately subsumed by the pace of technological advancement. It's no mystery where this is all heading, and if we can manage not to fuck it up (yeah, I know), work as we know it will just cease to exist, freeing people to live lives of luxury and learning.
This isn't something to be feared; it's the hacksaw to our chains.
>work as we know it will just cease to exist
And with it the need for billions of people to exist.
that's not marxism though, marx would have thrown a tantrum about AI replacing muh hard working workers. Commies fetishize the idea of labor, they don't want post-scarcity to happen.
Like the Marx you (incorrectly) cite (and Marxists in general), you are entirely oblivious to elementary aspects of human nature. People "need" work. By work, I mean something to pass their time with and make them feel valuable to the "tribe." There are already widespread problems related to the fact that most of the jobs today are bullshit/surrogate jobs, just like most of the other activities are either simulacra or surrogate. Remove that entirely, and you have not made another step toward utopia but towards a terrible dystopia.
This is like people saying, "I wish I could live forever," while having trouble not getting bored to death on Saturday.
>i mean if AI actually eliminates programming jobs
It won't it will just reduce them and make the remaining jobs hyper-efficient.
>That will be the beginning of post-scarcity civilization
There will always be scarcity because there is no point at which the human soul can be quenched of its eternal hunger to improve its condition.
As good as it is for us, it can always be better.
LMAO what a retard. Marx is the epitome of ignorance of eternal human nature and this event does not change that.
>basedying out over an AI that copies code snippets
fool
Good thing I work as a DBA, get fucked you codemonkeys.
An employer won't seriously trust AI to write code. it can do little problems, but makes small errors which are amplified over large amounts of code. You wouldn't ask the AI to debug an infrastructure, for instance, especially if you don't exactly know what's causing the bug. AI isn't going to deal with implementing more than a small feature at a time in a large scale program, and no PM wants to ask an AI for one small feature at a time from a bot, and then check it over and over again.
Not to mention, even if it were possible, the industry is going to have the same pushback as they did with automatic elevators and the idea of self driving cars.
My personal theory is that this isn't the start of jobless developers and AI generated codebases, but in exponentially faster development due to AI assisted coding. Developers aren't losing jobs, but they're going to have more tools on their belt.
lol employers trust indians to write code and those are as shitty as chatgpt
at least chatgpt doesn't ask retarded questions while writing code and doesn't ask senior devs for a call every 5 minutes
they trust indians to write their code because law stops them from discriminating on race, and some employers don't fire people they should fire. That's not all employers, and there are no laws protecting Ais.
they trust indians because they're cheap as fuck
thats the only advantage
and also the managers are too retarded and think any developer is good
Increases in productivity will ultimately mean *less* programmers are needed for a job.
Up until about 15 years ago, CAD drafters faced a similar situation when AutoCAD developed to the point that the engineers could just do it themselves without much fuss. There are still drafting jobs around, but they pay like, $17 an hour and maybe require an Associate's degree. The same thing will happen in the enterprise world, where MBAs and their ilk will be able to dictate requirements to an AI (or a team of two or three programmers) to create and deploy their ideas in an afternoon.
The flipside of that argument though is that most enterprise software is inane CRUD bullshit that almost completely relies on marketing to sell itself. Companies aren't going to need to shell out big bucks for the gazillionth iteration of HR software if they can just hire someone with a two year degree to design it for them in an afternoon. When I was in high school back around 2006, I worked with a woman who was a word processor back in the 1980s. Back then, businesses thought that they needed specialized technicicans to operate word processing software, and it was a well paying job.
Then MS Word came out.
or the projects get bigger because they can now accommodate loftier goals.
why do you have such a small view of economy? Where everything else changes except one or two things?
I tried it and while it was better than copilot it still wasn't able to make anything substantial for me. I tried for a while to get it to make something nontrivial to help me with a project I'm working on and no matter how I prompted it couldn't get it to make anything usable.
>It can literally write efficient code for almost any problem
Syntactically correct code is not necessarily the correct code. And it probably won't cover the end point conditions properly.
Retards will ask AI to write code in Rust, and because "le compiler is super strict!! Xddd" they will use it in critical infrastructure.
Of course it will be wrong and Boeing planes will fall out of the sky, but hey at least we saved money!
The saving grace is at least Pajeets will be replaced by AI.
Can someone try something more general like "write a config management system"
So can AI be fed IT playbooks and taught to do support yet? Is it fast enough to reliably do things like deep packet inspection on DMZ traffic?
Many firms have tried to remove humans from their customer support by using automated systems.
However, pretty much every time, it makes customers unhappy.
Case in point: Activision Blizzard. Everyone wants human GMs are customer support instead of automated crap that can't handle edge cases.
none of those firms are even remotely close to the level that openai is
they're primitive in comparison
leetcode is easy, but can it make a crud app with user account flow? didn't think so
It could if they make specialized AI with valid application credentials and state then yeah it can do that. AI is just an app that runs your apps and we already have that
it can do that easily. Just give me a more detailed prompt using specific libraries and how you want the stuff stored.
>He never learned assembly language to micro optimize AI high level code
I told you idiots on /dpt/ like a year ago that assembly was the way to go and you said I was wasting my time lmao. Look at you now, this is the price you pay for accepting the notion that your hand need not be sullied by low level code.
This in the thread I was going to make myself. I do low IQ office work, but set myself apart by learning some scripting and what not.
I have used the chatbot to do some of my work and the results are incredible. My job is gone, HR is gone, etc. I just need to wait till my boss finds out.
Not sure where to pivot... I have a family to feed. Maybe just HVAC or something?
>I have used the chatbot to do some of my work and the results are incredible
so just continue to do that. Become the guy who types prompts into the AI
>hey boss, OpenAI just released their new deep learning language model, it does my whole job for me! See you just tell it what algorithm you want, and it generates the code for you!
>that's nice, code monkey. Now get back to work
https://twitter.com/wattsy2077/status/1599266276287053830/photo/1
For chatGPT do I use the codex model or the text competion one?
Yes yes, poor pajeets, but will this give me a virtual gf?
this can lead to real life synthetic gf
Good.
I tried it in erlang, it produced garbage the moment I added a small extra feature to otherwise template ass task.
Yea, I think I'm safe.
I've asked for the source for every answer it has given me and it linked to a stackoverflow or another tutorial website. It's just a fancy google interface
Wasn't it announced that it wouldn't be able to connect to the internet for lookups? Hell, might as well try SSRF for shits.
>still no even remotely decent music ai
Why is music so hard? Is it because it is inherently SOUL, even if it's corporate hyperpop?
AI can't put any underlying theme in the music. Which is what makes the music good. This is especially audible with AI-made classical music; the composition is great, the soundscape is great, the instruments are great. But in its entirety it is just dull, empty, soulless.
Can this thing create bootloader or water simulation? Please, somebody check. I want to know, but it is blocked in my country.
>bootloader
It wrote
; Bootloader code
; Set the segment registers
mov ax, 0x0000
mov ds, ax
mov es, ax
; Load the operating system into memory
mov bx, 0x7c00
mov dh, 0x01
mov dl, 0x80
mov cx, 0x8000
mov ax, 0x301
int 0x13
; Transfer control to the operating system
jmp 0x0000:0x7c00
"This bootloader sets the segment registers, loads the operating system from the first sector of the first hard drive, and then transfers control to the operating system. It is written in 16-bit x86 assembly language and can be executed on a compatible BIOS.
Note that this is a very basic example, and a real-world bootloader may include additional functionality and error handling."
Cool. Thank you.
>water simulation
#include <GL/glut.h>
#include <cmath>
#include <vector>
const int WINDOW_WIDTH = 640;
const int WINDOW_HEIGHT = 480;
const float WATER_HEIGHT = 0.5f;
const float WATER_SPEED = 0.1f;
struct Particle
{
float x;
float y;
float z;
float vx;
float vy;
float vz;
float color[3];
};
std::vector<Particle> particles;
void init()
{
// Set up OpenGL
glClearColor(0.0f, 0.0f, 0.0f, 1.0f);
glMatrixMode(GL_PROJECTION);
glLoadIdentity();
gluPerspective(45.0f, (GLfloat)WINDOW_WIDTH / (GLfloat)WINDOW_HEIGHT, 0.1f, 100.0f);
glMatrixMode(GL_MODELVIEW);
// Initialize particles
for (int i = 0; i < 1000; i++)
{
Particle particle;
particle.x = static_cast<float>(rand()) / static_cast<float>(RAND_MAX) - 0.5f;
particle.y = static_cast<float>(rand()) / static_cast<float>(RAND_MAX) - 0.5f;
particle.z = static_cast<float>(rand()) / static_cast<float>(RAND_MAX) - 0.5f;
particle.vx = static_cast<float>(rand()) / static_cast<float>(RAND_MAX) - 0.5f;
particle.vy = static_cast<float>(rand()) / static_cast<float>(RAND_MAX) - 0.5f;
particle.vz = static_cast<float>(rand()) / static_cast<float>(RAND_MAX) - 0.5f;
particle.color[0] = static_cast<float>(rand()) / static_cast<float>(RAND_MAX);
particle.color[1] = static_cast<float>(rand()) / static_cast<float>(RAND_MAX);
particle.color[2] = static_cast<float>(rand()) / static_cast<float>(RAND_MAX);
particles.push_back(particle);
}
}
void display()
{
// Clear the screen
glClear(GL_COLOR_BUFFER_BIT);
// Update and render particles
glBegin(GL_POINTS);
for (Particle& particle : particles)
{
particle.x += particle.vx * WATER_SPEED;
particle.y += particle.vy * WATER_SPEED;
particle.z += particle.vz * WATER_SPEED;
if (particle.y < WATER_HEIGHT)
{
particle.vy *= -1
Might be char limit
why even bother with 3d if you're just using it to render 2d images anyway
You seem to have to give it a very specific task or it falls apart. I think it’s a ways off
>can it refactor old shitty code tangled into multiple areas of a system?
>can it interrupt byzantine business logic and develop a solution that fits into the current app stack?
>can it debug issues through all layers of the application stack from server, database to app level?
>can it explain it like I'm 5 to non-techies?
The biggest risk from these AI TOOLS is that it will make the pajeet hoards better programmers and more competitive but at the end of the day human tech jobs will always be a thing in one form or another long after other jobs fall to automation.
Negative. AI is a black box, so it is directly competing with Pajeet spaghetti.
>Chatgpt
It cant even program a simple shopping cart annon
"make a shopping cart where the number of items can be increased and decreased and have the total update "
Not tested it yet, but can't we abuse it for our gain?
Start using Chatgpt and github copilot in your daily work and don't tell anyone.
absolutely, I've been using it as a documentation and name generator
it also gets better if you keep talking to it and stick to a specific topic in the thread
you can gradually give it context and it'll actually understand
see picrel, very simple example but it's actually understanding what I wrote
this shit is like a pajeet quora answer
Where do you think the data comes from?
this the seethe thread?
Yes. Welcome
your job must be real easy of the answers by that bot are sufficient to replace your work.
get fucked codetard
https://www.reddit.com/r/AmItheAsshole/comments/zeuymn/comment/iz9hzyj/
How good is GPT as a search engine? Can you prompt it to actually find relevant webpages on the topic you're looking for and filer all the crap like quora,pinterest,msm or SEO cancer?
can we use chatgpt to write the perfect linux desktop ?
How do I optimize this? I have it writing me excel macros, I want it to work based off of my existing code though
>I have it writing me excel macros
post an example pls
I ran the entire thing in excel just changing the inboxes and file paths to match and it worked perfect
AI cannot develop solutions to complex human problems. At least not yet. This level of automation is essentially just the simple algorithms that get used over and over again being spit out.
We're very close, honestly I'm surprised that some very obvious techniques have not yet been realised. They will soon however.
>t. tech illiterate
We surpassed the functional compute of the brain years ago. Its intelligence is encoded in a relatively small genetic code. AGI is not hard it's aligning it usefully. We already have AGI in an unstructured way.
Most programming is just combining simple algorithms and programs into more and more complex ones. What we unironically need is to apply the unix philosophy to AI, in other words instead of training giant monolithic models that do 20 things at once, you have models that focus on doing one task really well with standardized interfaces, and then use those to build AI programs tailored exactly to your needs. Once that happens, we're screwed.
>Give it some prompts from my job
How? My job is such a massive clusterfuck of data that something as simple as processing a payment for a provider needs to query 14 different tables.
It is fucking hilarious to see the difference between /ic/ and BOT sperging and shitflinging about the AI.
Not bad tbh
is this the same as chatgpt?
Yes, openai is chat gpt
It's not, ChatGPT has been finedtuned to work as instruct model, which ends up making it far more capable and consistant than vanilla GPT-3
Oh yeah my bad, it is openai product though, but chatgpt is way better than just gpt-3
I hope this humbles retard humans that nest unnecessary complexity in interfaces and documentation. The problems being solved here are not hard often use knowledge specific to that language or procedure. Humans are prideful creatures and avoid optimising for transferability.
Already twice today I've generated very specific updates to rather lengthy jQuery to spitball client suggestions for visual changes to a UI.
The code worked out of the box, and I got to sit on it for an hour while I worked on more important backend stuff.
This is undoubtedly the future.
"Write a calculator app in JavaScript using the React framework"
import React from 'react';
class Calculator extends React.Component {
constructor(props) {
super(props);
this.state = {
num1: 0,
num2: 0,
result: 0
};
}
handleNum1Change(event) {
this.setState({
num1: event.target.value
});
}
handleNum2Change(event) {
this.setState({
num2: event.target.value
});
}
addNumbers() {
this.setState({
result: this.state.num1 + this.state.num2
});
}
subtractNumbers() {
this.setState({
result: this.state.num1 - this.state.num2
});
}
multiplyNumbers() {
this.setState({
result: this.state.num1 * this.state.num2
});
}
divideNumbers() {
this.setState({
result: this.state.num1 / this.state.num2
});
}
render() {
return (
<div className="calculator">
<input type="number" value={this.state.num1} onChange={this.handleNum1Change} />
<input type="number" value={this.state.num2} onChange={this.handleNum2Change} />
<br />
<button onClick={this.addNumbers}>+</button>
<button onClick={this.subtractNumbers}>-</button>
<button onClick={this.multiplyNumbers}>*</button>
<button onClick={this.divideNumbers}>/</button>
<br />
<p>Result: {this.state.result}</p>
</div>
);
}
}
export default Calculator;
I had to change all the methods to fat arrow functions (handleNum1Change = (event) => {
this.setState({
num1: event.target.value
});
}
for example),
And aside from the addition function only appending the first number to the second one (i.e. 1+2 = 12 due to javashit dynamic types), it worked fine.
We're fucked, aren't we?
I wonder how good it is for creating documents you need to make in software development
>URS
>SRS
>Architecture documents
>SDDs
>Test protocols
>...
Unfortunately, it's over for double digit IQ web developers.
Are you guys retarded or what? This is just a slightly better stack overflow
>muh AIs will take jobs
When will you retards realize that the current economic model requires a working class that can actually spend money?
If AI takes all jobs, everyone is poor, therefore rich companies don't make money
Pathetic cucks, being afraid of computers lol. You disgust me.
the point is that due to AI only the blue collar jobs will remain. Firefighters, garbage men and WELDers will have the standards of living of the current upper middle class while everyone else will be bottom feeders with universal basic income.
That's long term though. Even if the AI tech is almost there to replace a lot of jobs already companies will take decades to transform this into reality. Even simple automation with python could replace tons of filepushers and email senders, yet these people still exist today.
I just want to see a world where robots and AI do everything and humans can just relax and do hobbies.
that won't happen in your lifetime. There will be a huge gap, spanning at least a generation, between deskjobs becoming absolete and physical work being replaced by robots.
>that won't happen in your lifetime
And it sucks. Just few more centuries and we could be exploring space.
to do what exactly? there is nothing out there in our reach.
We need to go back
LOL. More like robots and AI do everything and the elites transcend into literal gods on earth while the rest of the population are reduced to being human rats.
You idiots.
Someone has to write the prompt and check the work. Who do you think that'll be? You, idiot. Your job is now easier and even more secure.