Hi HN - looks like someone beat me to posting this here! This was my most recent weekend project, and I was curious to see how GPT would do when asked for output that a computer had to run, versus output for humans to read. The area of AI results that are wrong but wrong in plausible ways fascinated me.

Let me know if you have any questions!

This seems pretty interesting. But with shell auto-fill and spell-checker almost ubiquitous at this point, its pretty unnecessary.

(But now I'm thinking of that one game where you have to interact with a super-intelligent AI on a spaceship. I can't remember the name.)