Live data from Hacker News

GPT-4 is phenomenal at Code

github.com

91–100 of 221 posts

Re: GPT-4 is phenomenal at Code

#91

That's cool and all, but can it do "in this existing code, please amend this feature"? Also useful, "please cover this code with meaningful unit tests"?

>Also useful, "please cover this code with meaningful unit tests"?

ChatGPT can already do that, assuming you provide reasonably well-formed specifications as input.

Re: GPT-4 is phenomenal at Code

#92
post #14

Remember the comments about artists and creatives when DALL-E came out? I remember.

I think my barber and my plumber have safer jobs from automation than average coders.

So far, I haven't seen GPT being used as a tool for programming by non-programmers.

Re: GPT-4 is phenomenal at Code

#93
post #65

Earlier quoted context omitted.

Let’s play this out. A company fires all the SE’s and GPT-4 takes over their place. Who will be there to ask the right questions? Middle management? Who can validate it spit out what you actually want? Project managers? Who will determine what questions to ask GPT-4? Product owners? Who will decide what type of tech (language etc…) to ask GPT-4 to write? College grads turned into corporate GPT-4 pushers? Who will be…

You don't need to fire all software engineers. You could let's say replace 75% of them with GPT.

Those software engineers are going to be very overloaded.

Will they be troubleshooting 100% of the code with only 25% of the staff?

GPT might become a tool in our arsenal but not a replacement. It needs to be able to say I don’t know before it spits out nonsense.

Re: GPT-4 is phenomenal at Code

#94

Earlier quoted context omitted.

email us at sualeh@anysphere.co :) We would love for you try it out.

Thanks! Appreciate it. I will reach out to try it for fun, but don't think my employer will be down with using it on our actual code.

We are designing the IDE for enterprise users actually. I.e. we can do an on prem deployment.

Re: GPT-4 is phenomenal at Code

#95
post #84

It managed to write, in one shot, a working λ-calculus parser, using a very specific programming style I asked it in JavaScript, and then translated it all to Python, including sarcastic, rhyming GLaDOS comments. https://twitter.com/VictorTaelin/status/1635726202231988225 It also seems to be extremely competent at writing Agda types and proofs. We need some tool that highly integrates it with entire codebases, allowi…

I don't think this is a good test. Every CS graduate student has to write a lambda-calculus parser. There must be thousands of implementations on the web. It really is not strange that GPT-4 can reproduce this.

Let's invent a better test then. What could be an example task that isn't widely available on the internet already? I'm having a hard time coming up with anything that isn't reproduced in many places.

Re: GPT-4 is phenomenal at Code

#96
post #66
post #14

Remember the comments about artists and creatives when DALL-E came out? I remember.

I remember it being vastly overhyped about how artists were now obsolete, but it really hasn't happened. People have quickly became very attuned to the "AI art style"--humans are excellent at pattern recognition, after all--and if anything, it's made people appreciate how important an artists' individual style and attention to detail is.

I'm sorry but am I the only one experiencing time dilation? You can't say ' X was supposed to be a big deal but was not' when the number of months since X began is single-digit.

Come back in ten years and tell me that there was no threat.

Also, for anyone living under a rock, the 'AI art style' --- by which you probably mean MidJourney 3 style --- has been steadily disappearing across MJ 4 and 5.

So, yes, actually, artists may now be 'obsolete', at least in the sense of 'viable form of capitalism for an adult North American to pursue as a profession'.

Circumstances for artists were already the worst they'd ever been, in real terms, even before this. Going forward, 'graphic designer', like 'music engineer', is going to be five people with real talent and 20,000 cool kids with trust funds. Everyone else fights for a job at Wal-Mart.

Best queue for your shift, the pinnies that fit well always get grabbed first

Re: GPT-4 is phenomenal at Code

#97
post #14

Remember the comments about artists and creatives when DALL-E came out? I remember.

I think my barber and my plumber have safer jobs from automation than average coders.

The problem is and always been how do you describe what you want the computer to do to the computer. This is where this will see some traction as a tool that helps developers do this faster, but my bet is that these sorts of systems will still work a lot better using an artificial language to state the requirements and these advances will be integrated into our languages and compilers.

Re: GPT-4 is phenomenal at Code

#98

Earlier quoted context omitted.

I made the case that anyone that solves problems with objectively correct solutions (programmers) would always be more at risk of AI taking their jobs than people that solve problems with subjectively correct solutions (artists, some designers).

Programmers' solutions aren't all objective, or at least shouldn't be. Simple tasks are. In some cases people are using overly low-level tooling, creating a lot of repetitive work that the AI would do better, and they might be at risk.

Objectively correct in the sense that we need 2+2 to equal 4 and we don't care how you got there kind of objectively correct.

Whereas an artist might need 2+2 to equal apple, and we can only subjectively solve for why or how.

Re: GPT-4 is phenomenal at Code

#99
post #53

Earlier quoted context omitted.

> Also useful, "please cover this code with meaningful unit tests"? GPT-3.5 is already really good at that. If you aren't using ChatGPT for your unit tests I'd suspect you're on a Performance Improvement Plan by now

What portion of the code do you feed it as an input?

You feed it design specifications as input, the same way that a human SDET would do it. The challenge, as always, is to have well-written specification docs.

Re: GPT-4 is phenomenal at Code

#100
post #65

Earlier quoted context omitted.

Let’s play this out. A company fires all the SE’s and GPT-4 takes over their place. Who will be there to ask the right questions? Middle management? Who can validate it spit out what you actually want? Project managers? Who will determine what questions to ask GPT-4? Product owners? Who will decide what type of tech (language etc…) to ask GPT-4 to write? College grads turned into corporate GPT-4 pushers? Who will be…

You don't need to fire all software engineers. You could let's say replace 75% of them with GPT.

You could replace 75% of them with a rubber duck. Soon you might replace 22.5% of them with GPT.
Post reply on HN