Discussion about this post

User's avatar
David Raitt's avatar

As I understand them, AI models tend to generate similar output to what they've been trained on. Which suggests that the improv information ChatGPT has "learned" is mostly about technique. Which further suggests that the people who create that information write mostly about technique.

Beginning improvisers must learn the basics to be able to get on their feet. But after a certain point, some continue hunting for techniques, optimizations, and hacks as a hedge against the uncertainty, to try to guarantee "successful" improv according to whatever their definition is. They gloss over advice like "be human" or "throw away the rules and just play." Give them a new pattern game to play and they're all over it.

This is an important reminder to keep the human part of our teaching intact. We have to work harder at finding new ways to express those ideas. Thanks, Lloydie!

I'm looking forward to those RHIIF workshops!

Expand full comment
2 more comments...

No posts