I've been using ChatGPT and Claude for coding help, but they keep generating functions that look right but don't actually work. Sometimes they reference libraries that don't exist or make up syntax. Is there a way to make them more reliable, or do I just need to fact-check everything they output?
you gotta be specific with your prompts - ask for code in a particular language version, mention exact libraries you want it to use, and paste in error messages if something breaks so it can debug instead of making stuff up! fact-checking is still essential because even with good prompts they'll confidently write convincing garbage, so always test the output and look for sus function names or imports before you trust it.
Log into answer.
