I suppose it's a gradual slope here, with spell checkers on the one side, a grammar checker a bit further ahead, then Google Doc's autocomplete, and then close to the very end of the slope you have this system that just needs a few sentences and completes the entire essay for you.
One question then is - at which point on the slope does the comment stop being a direct product of your mind? Another question is - in what ways does this distinction matter?
I think that one of the neat new narrative art forms that GPT-3 like algorithms will make possible is a prose cinema where the author is more like a director, prompting agents to react to one another. Like lining up billiards shots. I think a bunch of people are doing stuff kinda like that now at least in regard to poetry generation.
At what point does the billiard ball stop being an object of your will?
