I'm in the same boat. In a lot of ways, it's a de-compression algorithm that takes a succinct idea as input and appropriately inflates it with the verbosity exp

I'm in the same boat. In a lot of ways, it's a de-compression algorithm that takes a succinct idea as input and appropriately inflates it with the verbosity expected for a specified social context. The ChatGPT examples are very impressive, but after seeing dozens of them I find they highlight how much normal communication is boilerplate and pablum. Years ago, https://www.reddit.com/r/SubredditSimulator/ https://www.reddit.com/r/SubredditSimulator/ was in the uncanny valley using markov chains to get some moments of lucidity that were funny or plausibly real. But the joke when it succeeded wasn't that markov chains were amazing, it was that subreddit discourse is so predictable and repetitive. ChatGPT could probably generate threads with thousands of reddit comments that are indistinguishable from the real thing. That becomes more of an uncomfortable look in the mirror than anything else.