Having spent some small time in the information theory and signal processing world, it infuriates me how often people champion LLMs for writing things like data dictionaries and documentation.
Information is measured in information theory as “the difference between what you expected and what you got”, ergo, any documentation generated automatically by an LLM is by definition free of Information. If you want something explained to you in English then it can be generated just as easily as and when you want it, rather than stored as the authoritative record.
Ah yes, comments and commits written by LLMs, who wouldn’t want that.
Having spent some small time in the information theory and signal processing world, it infuriates me how often people champion LLMs for writing things like data dictionaries and documentation.
Information is measured in information theory as “the difference between what you expected and what you got”, ergo, any documentation generated automatically by an LLM is by definition free of Information. If you want something explained to you in English then it can be generated just as easily as and when you want it, rather than stored as the authoritative record.
You’re not looking at a bigger picture. That also raises the question of your alleged qualifications.
It’s much better than commit messages like “WIP”, “Quick fix”, “Refactoring” and other mental diarrhea every developer is guilty of.