If you want to do something generic, like “make a text editor” or “make a todo app”, then this is where the AI is most successful. LLM’s work best when they have many examples of similar projects. But if there are many similar projects, why don’t you just use one of them instead. The more unique a project, the more it’s actually worth doing, but the less useful LLM’s are.
Sometimes writing a clear description of what you want done in english isn’t much easier than writing it in python. This is especially true if your thinking is in visualizations and algorithms not in english. The english language is a bit of a mess and was mostly designed before computers became a thing, it is far from the optimal language in which to instruct machines.
LLM generated code seems in practice to offer a large increase in code quantity in exchange for a modest decrease in code quality. This is often a bad tradeoff.
If you want to do something generic, like “make a text editor” or “make a todo app”, then this is where the AI is most successful. LLM’s work best when they have many examples of similar projects. But if there are many similar projects, why don’t you just use one of them instead. The more unique a project, the more it’s actually worth doing, but the less useful LLM’s are.
Sometimes writing a clear description of what you want done in english isn’t much easier than writing it in python. This is especially true if your thinking is in visualizations and algorithms not in english. The english language is a bit of a mess and was mostly designed before computers became a thing, it is far from the optimal language in which to instruct machines.
LLM generated code seems in practice to offer a large increase in code quantity in exchange for a modest decrease in code quality. This is often a bad tradeoff.