The forum's official policy is that "AI" such as ChatGPT should not be used to create posts, other than when used as a machine translation tool.
Don't post AI-generated code. This includes code you want help with. Do not use AI to generate script code and then ask people to fix it for you when it inevitably doesn't work.
AIs are still very bad at writing code, especially for relatively niche APIs that don't have a lot of example code for them to copy. They frequently invent object and method names that do not even exist, and generate code that could never work and which is of no use. And even if it does appear to work, with no one who actually understands or knows what it does, it could conceivably do anything.
It's better to simply ask if anyone has or is willing to write a script, or help you write one, than to post code that you don't understand and which doesn't work. People are often happy to help.
Do not use AI to generate excessively long and detailed feature requests and then expect people to read them.
A concise description of your idea is better than one which is hundreds of pages long but contains no extra information.
If you can't be bothered writing it yourself, why should anyone else bother to read it?
Some more information as to why AI-generated code and posts are not welcome on this forum.
LLMs are built on industrial-scale intellectual property theft. All that text that you think AI is "writing" for you is just regurgitated text that other people have written.
It regularly makes mistakes when generating code for Opus. It invents commands and APIs that don't even exist (and then people ask us to debug "their" code and tell them why it doesn't work).
Even when it has more examples to copy from, it rarely produces well-designed code that's easy to understand and that anyone will want to fix or maintain in the future. (This also makes it harder to review the code, and hides problems.)
It does not learn from its mistakes. It is never held accountable for its mistakes.
It is completely unsafe. It can be easily subverted by input data and has no concept of trust of adversaries. There's no way to fix that; it is inherent to how it works. (This shows how much the large tech companies really understood or cared about security.)
It is not deterministic. A prompt which generated working code one day may not the next. That's not something to build on top of.
It makes people who use it less competent over time.
It is terrible for the environment. (Shows how much big tech companies really cared about that.)
It has made it impossible to buy new computer hardware for reasonable money. (Remember when hardware prices went down over time?)
It hands too much control to corporations who have demonstrated they are downright psychotic.
Machine learning and LLMs have their uses, but Gen AI for anything creative, like code and art, is all downsides and no upsides. It's just a shortcut for people who don't want to learn how to do things themselves.
It's just a minor stand in the scheme of things that we are making here, but we're making it nonetheless. Don't post AI slop to this forum.