Mastodon Feed
Boosted by soatok@furry.engineer ("Soatok Dreamseeker"):
patcharcana@furry.engineer ("Patch Arcana") wrote:
We fantasize about the LLM having tells because it's easier to think about than the actual truths:
- The average sum of human writing is pretty fucking dismal, and;
- The only way to tell an LLM output from a human output is to see if it's internally consistent and has a point.
And that latter is a problem: it means you actually do have to read the slop. Worse, you have to be able to read critically, which isn't exactly something you can assume the average reader is capable of.