New Post
hi folks, new user here.
I pasted in a few paragraphs of a silly zombie novel I'm writing, and hit "write". I got one great paragraph where a character decided randomly to blow up his car because it was stuck in a field, which gave me a laugh. But the second suggestion was this:
Thanks for reading.
I'll be adding the next part to this story soon, not sure exactly when.
Catch up with the rest of the story, which was published over a year ago in instalments, to read from the start.
I'm guessing somewhere in the language model there are comments from an author which need to be taken out
Thanks for logging this, @rory_okelly! These are actually cases where the AI is making up those author comments. Because it's been trained on the internet, it's seen lots of chapter excerpts and summaries and so forth posted, followed by author notes, bios, links to websites, etc.
So when it sees something similar in your text, occasionally, it decides to follow it up with an author note or bio. It's not taking this text from someone else's work, but obviously it's not very useful!
Thanks for logging it here so we can properly prioritize filter out this type of text!
Merged in a post:
Reminds me of fanfic author notes
I agree. Like commenting in a programming language. // for c#. REM for Basic, < -- -- /> for html, -- for MySql, so on and so forth.
yeah the AI scolded me for a storyline I created and was like (are you effing kidding me?!") I couldn't contain my laughter lol
We've made lots of improvements to the AI since this report, and (while not impossible) instances of this sort of bleed through have been reduced significantly! Closing this one out since improving AI quality is an ongoing focus for us.