I’m quite happy with this one. I liked where it led me, and where it concluded.
I’m enfolding GPT-3 into my writing process, meaning I’m doing a fair amount of revising sentences in completions. This feels a lot like revising my human-only work, boiling ten-word sentences down to four-word sentences, looking for redundancies and inefficiencies in the language while retaining the jist.
The most interesting part of this way of working is adjusting my subsequent prompts to match the voice of the output. I don’t want the seams to be visible. While GPT-3 strives to match my tone, I strive to match its tone, until we meet somewhere in the middle.