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

Using ChatGPT to generate a GPT project end-to-end

github.com

Post reply on HN
Print view View original

31–40 of 225 posts

Re: Using ChatGPT to generate a GPT project end-to-end

#31

Post by AnimalMuppet »

haunter wrote: post #7
lionkor wrote: post #2

What value does a developer deliver when their entire process is done by an LLM? Is there no desire for a creative process?

People had the same thoughts about cake mixes in the 40s. Oh you can just buy a cake mix? That's not cooking anymore444 https://www.youtube.com/watch?v=r6wKaLQ66r8

Fun fact: People didn't like cake mixes when they first came out, precisely because it wasn't "really cooking". Then someone (Betty Crocker?) changed the mix (and the instructions) so that the person had to add an egg, not just water. Then the humans felt like they were actually cooking, and cake mixes were more accepted.

A really smart AI would leave enough for the humans to do that they feel like they're still in charge.

Re: Using ChatGPT to generate a GPT project end-to-end

#32

Post by ixaxaar »

nklrta wrote: post #23
dhoe wrote: post #15

Earlier quoted context omitted.

It's a spot on - moving up a level in the abstraction hierarchy.

Human: ChatGPT, please devise an algorithm that solves the traveling salesman problem in polynomial time. A: Certainly, as an AI language model I'm happy to oblige. Here is the algorithm. Human: I used a higher abstraction level! I have solved P == NP!

I wish it were that easy.

If you see toward the end where it generates Einstein's field equations, I had to start from real numbers (dedekind cuts) to real manifolds to pseudo-Reimannian manifolds to the curvature metric to the final thing.

Re: Using ChatGPT to generate a GPT project end-to-end

#33

Post by freehorse »

jvanderbot wrote: post #19
andai wrote: post #17

Earlier quoted context omitted.

I asked GPT to write a program which displays the skeleton of a project, i.e. folders, files, functions, classes and methods. I put that at the top of the prompt. I had a spooky experience with a project that was written almost entirely by GPT. I gave it the skeleton and one method and asked it for a modification. It gave it and also said "don't forget to update this other method", and showed me the updated code for…

It's very good at guessing from function names and context. This is very impressive the first few times, and very frustrating when working with existing codebases, because it assumes the existence of functions based on naming schemes. When those don't exist, you can go ask it to write them, but this is a rabbit hole that leads to more distractions than necessary (oh now we need that function, which needs this one). I…

It only works well with codebases that are probably quite well represented in its training data ime. For more obscure ones one is better off just doing it by themselves. Finetuning may be a way to overcome this, though.

Re: Using ChatGPT to generate a GPT project end-to-end

#34

Post by jiggywiggy »

amelius wrote: post #24

Is anyone letting an LLM code and run its code by itself, then iteratively fix any bugs in it without human intervention until it e.g. passes some black box tests? Would it be possible to significantly improve an LLM using such unsupervised sessions?

Yeah at some point gpt4 loses track and just consistently is wrong.

Lately I can't feed it too much info, the longer the context the more issues.

With your suggestion, it doesn't know which part of the iteration is correct at the moment. For us iteration is logically but for chatgpt I think it's just more variables that make the chance of being wrong larger. So you need to build that in somehow that it can iteratively filter and prompt

Re: Using ChatGPT to generate a GPT project end-to-end

#36

Post by eurekin »

jvanderbot wrote: post #18
zerop wrote: post #16

It's good at writing new code, with sufficient prompting. But the big open question as of now for engineering orgs is - can it edit existing code like developers, just by instructions. Is there any hands on experience anyone has on copilot-x?

Copilot without -x does this relatively well. But it's a hands-on process. I can't just give it some source files and say "go". But it can easily make you 10x. I often spend more time tab-completing than writing.

Any publically available demos?

I've only played around with faux pilot and it seemed that it was mostly limited by my prompting skill.

Also, is copilot aware of anything outside of currently edited file?

Re: Using ChatGPT to generate a GPT project end-to-end

#37

Post by smokel »

amelius wrote: post #24

Is anyone letting an LLM code and run its code by itself, then iteratively fix any bugs in it without human intervention until it e.g. passes some black box tests? Would it be possible to significantly improve an LLM using such unsupervised sessions?

You might be interested in Auto-GPT: https://agpt.co/, an attempt at an autonomous AI based on GPT.

It does not feed back information into GPT, so the LLM is not improving. Such a system would require both guts (insanity?) _and_ money to pull off.

Re: Using ChatGPT to generate a GPT project end-to-end

#38

Post by eurekin »

MacsHeadroom wrote: post #21
zerop wrote: post #16

It's good at writing new code, with sufficient prompting. But the big open question as of now for engineering orgs is - can it edit existing code like developers, just by instructions. Is there any hands on experience anyone has on copilot-x?

GPT-4 32k can fit a few thousand lines of code into a single prompt and output diff's, patches, and/or the fully modified code. For best results, pasting any relevant documentation into the prompt can help.

Can you demo it, please?

Re: Using ChatGPT to generate a GPT project end-to-end

#39

Post by hklwrand »

braindead_in wrote: post #25

Combine TDD and Self debugging into a workflow and you almost have a new paradigm of software development where entire applications can be developed with a series of prompts. Software programmers have finally programmed themselves out of jobs! It's kind of poetic justice that LLMs trained on open source code is replacing us. We should have never listened to Richard Stallman. /s

I'm not writing new open source. First, web developers have profited from the work of others while locking down their own work on servers.

They became rich and started looking down on those who wrote real complex code.

Now "Open"-AI launders the output while keeping their IP safe. I wonder though if the whole thing is a scheme by Microsoft to kill open source.

Re: Using ChatGPT to generate a GPT project end-to-end

#40

Post by gwoolhurme »

braindead_in wrote: post #25

Combine TDD and Self debugging into a workflow and you almost have a new paradigm of software development where entire applications can be developed with a series of prompts. Software programmers have finally programmed themselves out of jobs! It's kind of poetic justice that LLMs trained on open source code is replacing us. We should have never listened to Richard Stallman. /s

If you really do feel that way. Honest non troll question. Why not quit? There still seems plenty of software work even if there was a feedback loop in the workflow. However if you feel we’ve wrote our selves out why not do something more physical. Not to sound snarky.
Post reply on HN