[goodside]

As someone's who's played with GPT-3 prompting a lot, I can tell you almost no curation would be needed to get this result. This output is pretty mundane — it's even written in the perfunctory "so there you have it" tone that GPT-3 uses by default, where it speaks like the task is mutually understood to be silly and it's doing the the bare minimum needed to prove it understands. GPT-3 will easily write bare-bones stories like this for just about any subject. It's much harder getting it to write long stories with coherent structure.