Picking up fiction writing (and using LLMs to improve) gave me a new angle on the whole "LLMs make programmers dumber" discourse.

Writing fiction and coding are similar in how they interface with LLMs; you can ask them to do it for you, or you can ask them for feedback on how you did.

I think using LLMs to critique my prose has made it sharper (e.g., see QT). They're not as insightful as a real teacher, but I believe that *instant feedback* is key for effective learning.

That's how I learned to code, too, long before LLMs: submitting my solutions to an online judge that would instantly tell me if I was passing all the test cases, and tweaking them over and over until I got that sweet green light.

So, I think this applies to both writing and coding:

If your goal is to just get something done without learning anything, LLMs enable that.

If your goal is to get better at the craft, they enable that too.

It's up to you and how you use it.