Is ai going to end the field of programming?

yes.

Attached: 15_37_15.png (1652x760, 150.07K)

You can get it to explain what code does but can you get it to write code.

Yes, ask it to write a hangman game or something and it will do it.

But does it actually write it or does it just mash a bunch of other people's code for hangman together, like the image "AI"s do?

this question also applies to human programmers

You are aware even best in class programmers are also just a mix of their experiences and things they've done in the past?
The biggest downside to AI currently when it comes to programming isn't the code it writes, it's the lack of contextual testing and the centralization of major corporations due to how costly it is to run. Once both are solved there's very little reason for huge categories of programmers, who do very repetitive things in their day to day work.

yeah, feels a lot like people who say this isn't happening are huffing some serious copium. then again we still have accountants when taxes have been a solved problem for at least a decade, so who knows.

>stackoverflow kiddies

I should have known. Fuck off retards. This will amaze you but at some point, someone had to write some original code. Try not to let that keep you awake at night.

It's crazy how far people will go to act as if their field won't be impacted by AI. It also doesn't mean one has to panic as if they'll be evicted next week.
We're already seeing benefits, despite AI we're using right now still being quite basic. We get to write better code in less time and can focus on the tougher issues. Eventually these issues will become easier, and AI will learn from them as well. Yet our onramp to get people from knowing nothing to being able to produce great work is still very lengthy.
The only actual obstacle is regulation. And I don't see programmers banding together the same way the tax industry has to prevent easy automatic taxes.

This might shock you, but you don't need to go down to the nearest well to fetch water yourself anymore. Don't think about that too long, I know how short fused you are.

The difference between you and me is that I can go to the well if I want to, you don't even know where it is.

there's only so much any one human can know, though. i dont know how to build a car engine, but i can change the oil and the spark plugs and even figure out when something's wrong. sure, i know less than someone who can build an engine, but who cares? me being unable to do that doesnt have any impact on what i ultimately want to do: drive my car and keep it working.

>Not being a webdev shitter
>The wheel is reinvented every 2 years
>AI will never be able to catch up

Attached: drink_pp_classic.png (500x500, 240.29K)

>the ""AI"" needs someone to write 100 hangman games and then needs an array of computers working for hours to train itself before it can generate 1 hangman game
yeah, I'll pass

i wish it would

Any Forums tards will really read a verge article and start thinking they know enough to be making these ridiculous claims, let alone have any sort of grasp on any sort of basic machine learning or other ai theory. go ahead and substantiate any of what you're saying with something meaningful. was there a particular paper or model that has you thinking your speculative claims are true or meaningful?

i know very little about the underlying technology here, i wont pretend otherwise. but i suppose when i tell the computer "write me a hangman terminal application", it spits out code that i can compile and run with zero extra work, i start to think that can be extrapolated to many other programs - if not now, then certainly in the span of a decade or so. if you have knowledge otherwise, im happy to learn and rectify my ignorance

wasn't AlphaCode able to bet 50% of leetcode users some months ago?

>i start to think that can be extrapolated to many other programs -
hangman is a pretty deterministic problem, and intelligent problem solving algorithms have been a thing since forever. yea sure you can extrapolate those algorithms to more complex deterministic problems such as producing a program that can produce said game, but when you start involving real world errata into problems and other systems that are much more complex than a hangman game, it's not so simple.
>if not now, then certainly in the span of a decade or so.
yea and they've been saying that for decades
>if you have knowledge otherwise, im happy to learn and rectify my ignorance
not really sure what to try to say here that would be meaningful, but you should really start looking into machine learning basics. machine learning is finding a best fit plane through high dimensional cartesian space.

Yeah... It doesn't understand the code you just wrote. The code is very close to code in a sample that was in its training set, and you've just prompted it to regurgitate the explanatory text that appeared below the example.

Throw an inc in there; does it now say that the program prints "2"?

let me know when it can fix bugs. until then, i ain't worried