This is like writing in general. It's easy to edit crappy text you've written into something better.
It's completely impossible to do it to a text you didn't write at all.
LLM models are pretty good in doing the crappy first version. It might use abandoned packages or old APIs but the skeleton is there. It's not that hard to add some meat on the bones when the structure exists.
Recently I had to parse a pretty crappy XML format (planned by committee) with Go. I just fed the XML to GPT4 and asked it to parse specific values from it. It got like 95% there. I just had to do a few fixes and polish it a bit. Saved me a lot of headache and poking around in documentation.
This is like writing in general. It's easy to edit crappy text you've written into something better.
It's completely impossible to do it to a text you didn't write at all.
LLM models are pretty good in doing the crappy first version. It might use abandoned packages or old APIs but the skeleton is there. It's not that hard to add some meat on the bones when the structure exists.
Recently I had to parse a pretty crappy XML format (planned by committee) with Go. I just fed the XML to GPT4 and asked it to parse specific values from it. It got like 95% there. I just had to do a few fixes and polish it a bit. Saved me a lot of headache and poking around in documentation.