Super interesting. I wonder what other latent domain-specific intelligence GPT-3 picked up during training, that is parseable with text in and text out. Like a flash cards generator?
Polar (https://getpolarized.io/) has a GPT-3 based flash card generator from text highlights. It's available to premium subscribers.
Hmm, I like this direction - so maybe, as the user is navigating the incident, let them steer the model with questions and/or additional lines. Is that sort of what you'd envision?