Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A coder considers the waning days of the craft

newyorker.com

Post reply on HN
Print view View original

141–150 of 1001 posts

Re: A coder considers the waning days of the craft

#141

Post by silenced_trope »

hiddencost wrote: post #130

> At one point, we wanted a command that would print a hundred random lines from a dictionary file. I thought about the problem for a few minutes, and, when thinking failed, tried Googling. I made some false starts using what I could gather, and while I did my thing—programming—Ben told GPT-4 what he wanted and got code that ran perfectly. I mean ... IDK, if you can't write that on your own without research, I don't…

I'm the opposite. I feel like it is true. I very much will NO LONGER THINK about "simple" problems.

My most recent thing is for a game I'm working on. I got a bunch of free sound FX from freesound.org, but they were AIFF format. I googled and found an online converter, then asked ChatGPT to use ffmpeg to write a script that goes through the directory converting the AIFFs to MP3s.

In the past if I would have written such a script it would have required me to "look into how to do it" and then I would have saved the script somewhere on my machine or online. Now I just deleted it right away. If I ever need to do it again or something similar I'll just ask ChatGPT and get it from scratch again.

Re: A coder considers the waning days of the craft

#142

Post by mastazi »

If you want to become an architect, you don't need previous experience as a bricklayer.

Many other fields besides construction are similar: most positions don't require previous experience as "manual labour".

Software development was an exception to the rule, maybe it no longer is.

Re: A coder considers the waning days of the craft

#143

Post by josephg »

fisherjeff wrote: post #41
cglan wrote: post #6

Maybe I’m in the minority. I’m definitely extremely impressed with GPT4, but coding to me was never really the point of software development. While GPT4 is incredible, it fails OFTEN. And it fails in ways that aren’t very clear. And it fails harder when there’s clearly not enough training resources on the subject matter. But even hypothetically if it was 20x better, wouldn’t that be a good thing? There’s so much of t…

Also, I think we are quite a ways out from a tool being able to devise a solution to a complex high-level problem without online precedent, which is where I find the most satisfaction anyway. LLMs in particular can be a very fast, surprisingly decent (but, as you mention, very fallible) replacement for Stack Overflow, and, as such, a very good complement to a programmer's skills – seems to me like a net positive at l…

We’ll see - but given the gap between chatgpt 3 and 4, I think AIs will be competitive with mid level programmers by the end of the decade. I’d be surprised if they aren’t.

The training systems we use for LLMs are still so crude. ChatGPT has never interacted with a compiler. Imagine learning to write code by only reading (quite small!) snippets on GitHub. That’s the state llms are in now. It’s only a matter of time before someone figures out how to put a compiler in a reinforcement learning loop while training an LLM. I think the outcome of that will be something that can program orders of magnitude better. I’ll do it eventually if nobody else does it first. We also need to solve the “context” problem - but that seems tractable to me too.

For all the computational resources they use to do training and inference, our LLMs are still incredibly simple. The fact they can already code so well is a very strong hint for what is to come.

Re: A coder considers the waning days of the craft

#144

Post by w4ffl35 »

cglan wrote: post #6

Maybe I’m in the minority. I’m definitely extremely impressed with GPT4, but coding to me was never really the point of software development. While GPT4 is incredible, it fails OFTEN. And it fails in ways that aren’t very clear. And it fails harder when there’s clearly not enough training resources on the subject matter. But even hypothetically if it was 20x better, wouldn’t that be a good thing? There’s so much of t…

I've been saying the same thing. Coding is the worst part of the process. I've been doing it for 20 years professionally and another 10 or more on top of that as a hobby. Don't care about code, just want to make things. Code sucks.

Re: A coder considers the waning days of the craft

#145

Post by mathieuh »

FigurativeVoid wrote: post #23

I may be in the minority, but after the initial “wow” period, I have been underwhelmed with co-pilot. Don’t get me wrong, there are few times it has really really impressed me. And there are a few things it really shines at, but most of the time I find it getting in the way. I’m not so concerned about the craft of programming. But those that make a good living automating trivial tasks should be more worried.

Completely the same, sometimes it does something cool but a lot of the time I start writing a comment to get it to generate even something simple like e.g. some boilerplate test code and it can’t manage it. It’s happened often enough that I don’t bother most of the time now.

Plus it really annoyingly doesn’t seem to handle quotes and brackets, if I’m halfway through writing a line of code and it suggests a reasonable completion I’ll press tab and it’ll fuck up all the brackets and quote marks, I don’t know what I’m doing wrong here but it really annoys me.

We had a guy from Github do a demonstration for us and the most complicated thing he got it to do was write some getters and setters for a class representing a linked list iirc. And even then he spent more time having to cycle through suggestions to find one that he wanted than he would have spent just writing it out of his head.

Yet to be convinced personally.

Re: A coder considers the waning days of the craft

#146

Post by tingletech »

gymbeaux wrote: post #134
hiddencost wrote: post #130

> At one point, we wanted a command that would print a hundred random lines from a dictionary file. I thought about the problem for a few minutes, and, when thinking failed, tried Googling. I made some false starts using what I could gather, and while I did my thing—programming—Ben told GPT-4 what he wanted and got code that ran perfectly. I mean ... IDK, if you can't write that on your own without research, I don't…

Yeah it’s not a challenging problem, but the specific code example isn’t the point. Replace his example with “at one point, we wanted a command that would convert a PNG to a JPG….” most of us couldn’t do that without Google.

why not use `shuf` from GNU core utils for the random and `convert` from ImageMagic for the png to jpg?

Re: A coder considers the waning days of the craft

#147

Post by MattGaiser »

hiddencost wrote: post #130

> At one point, we wanted a command that would print a hundred random lines from a dictionary file. I thought about the problem for a few minutes, and, when thinking failed, tried Googling. I made some false starts using what I could gather, and while I did my thing—programming—Ben told GPT-4 what he wanted and got code that ran perfectly. I mean ... IDK, if you can't write that on your own without research, I don't…

Especially since that particular task has been done thousands upon thousands of times. I have done it to win arguments on Hacker News[0]. It wouldn't surprise me if the exact code to do it (except for changing file names and number of lines) existed on Stack Overflow.

[0] https://news.ycombinator.com/item?id=28587016

Re: A coder considers the waning days of the craft

#149

Post by kenjackson »

josephg wrote: post #143
fisherjeff wrote: post #41

Earlier quoted context omitted.

Also, I think we are quite a ways out from a tool being able to devise a solution to a complex high-level problem without online precedent, which is where I find the most satisfaction anyway. LLMs in particular can be a very fast, surprisingly decent (but, as you mention, very fallible) replacement for Stack Overflow, and, as such, a very good complement to a programmer's skills – seems to me like a net positive at l…

We’ll see - but given the gap between chatgpt 3 and 4, I think AIs will be competitive with mid level programmers by the end of the decade. I’d be surprised if they aren’t. The training systems we use for LLMs are still so crude. ChatGPT has never interacted with a compiler. Imagine learning to write code by only reading (quite small!) snippets on GitHub. That’s the state llms are in now. It’s only a matter of time b…

Why do you say the snippets are small? They don’t get trained on the full source files?

Re: A coder considers the waning days of the craft

#150

Post by enjoylife »

hiddencost wrote: post #130

> At one point, we wanted a command that would print a hundred random lines from a dictionary file. I thought about the problem for a few minutes, and, when thinking failed, tried Googling. I made some false starts using what I could gather, and while I did my thing—programming—Ben told GPT-4 what he wanted and got code that ran perfectly. I mean ... IDK, if you can't write that on your own without research, I don't…

Yeah, this author sounds more of an up-and-coming software developer. The cited example is something that someone should be easily able to do in their day to day language.

The ability for chatGPT to essentially translate/expand your knowledge out to other languages (eg “how do I read in a file to utf8 again” ). It’s all just more leverage and power to solve the mundane, faster. The key point is you have to know the rough solution space already.

Post reply on HN