The debate about AI taking over jobs versus just being a fancy autocomplete has been ongoing, with mixed feelings towards LLMs.
The author had a realization that LLMs could be used to improve old, messy code instead of just generating new code.
By feeding old code into LLMs with prompts for refactoring and improvements, the author found the process surprisingly effective in beautifying the code.
The author concludes that LLMs can enhance developers' ability to work on less glamorous tasks like refactoring and improving existing codebases, making old projects presentable and useful again.