LLMs / 2025-12-03

Note: An LLM was not used in writing this microblog entry.

This entry is part of a microblog series called LLMs

---

My experience with LLMs since my last update has been both positive and negative.

On the one hand, I’ve seen a quality and comprehension drop in some programmers. “I vibed it” is becoming a familiar explanation for both. I’m increasingly concerned.

On the other hand, I have found a workflow with LLMs that works for me. For those things that I could write myself and practically see in my head, this flow makes me happy:

1. Prompt a tool like Cursor or Claude Code with a paragraph description of what to do and where, with the permission to edit files.
2. Close it and return to Emacs.
3. Browse through the diff in magit.
4. Edit almost everything, rewrite some things, leave others that are good.
5. Continue myself by hand as normal.

This has been quite effective for me as a means to speed up tasks which are doing what I’ve done elsewhere with small differences. I’m using it as glorified autocomplete.