[dgellow]

Tell it what is wrong and why, it is surprisingly good at fixing itself with a bit of help. I was able to guide ChatGPT well enough to make it write an old school racing game using pygame. Start general, see what it gets correct and what should be changed, and give it better indications.

[Xelynega]

Chatgpt isn't a person though, it didn't fix anything. You gave it a prompt, it gave a result that was close. You add a line to the prompt and it gives you something closer to what you were looking for.

It looks more like refining a search pattern(like one might do with an LDAP query) on the part of the operator than it does an algorithm "fixing", "changing", and "being guided". It's interesting how we anthropomorphism the output of this algorithm compared to other APIs, even though the algorithm is closer to oher APIs than it is a human as far as we understand.

[dgellow]

I don’t think anyone here is confusing ChatGPT for an actual person.

What is really interesting with ChatGPT compared to other interactive software is that you can give instructions the way you would do it with a human. You can literally copy paste a compilation error, with no more context, and it will fix the previous program it generated. Even just pointing vaguely to something like “that does not look correct, you forgot some edge cases” will result in an improved version.