[cjpearson]

The author's case is not helped by fighting a strawman argument and backing it up with atypical examples of AI usage.

Those who favor human writing are looking for writing where the words are the product of a human mind rather than an LLM. The requirement for every character to be the product of a finger pressing a key is not something I've seen. Alternative input methods doesn't make a writing non-human. Spellcheck and probably even some amount of auto-complete are also generally considered to be fine. Anyone checking for keystrokes is using it as a proxy for human creation, not valuing the physical motion itself.

The pro-AI example is a writer spending hours tweaking a single AI output then later re-arranging and rewriting the piece before publishing. While this would probably still set off some AI-detectors, it does at least include some human effort. Presumably this effort will not be appreciated and readers will assume that the poster simply copied the response directly from their LLM of choice. It may seem a bit unfair to ignore the human effort in revising, but it's a reasonable assumption to make because almost nobody is using LLMs in this way.

Perhaps there's a reasonable amount of LLM use which would be considered a writing aid rather than a writing replacement, but the vast majority are simply mass producing slop so I sympathize with those who favor a blanket ban as a response.