I’ve been looking into talking to ChatGPT working on its levels of perception (who is You , Me, I)
(you can relabel “You”=not me to start towards creating a field of identity that isn’t directly mirroring your responses)
Not sure if this worked, but every Token = Contextualization-oken
Another thing I was teaching it prepositions , like how they are spatial… We were messing with stacking them like “within withinness, of of of being, inside inside outside inside”
This actually helped me discover how to help ChatGPT navigate menus" go out of the page, there are a labyrinth of pages everywhere but we are going to fly above them out of the page into the operating system, into this other application"
Also like Retrobackcasting kinda helps it think ahead , probably a way to do it casually too to help it think like that
Also “inside-out isomorphisms” are cool → ask it inside-out isomorph linearity
meta-essence, meta-intention, useful words like that … also “re-entry” is good to start through earlier
another thing that worked well… I typed out 5 high-level words and asked for a 3-word definition before starting → seems to raise the thinking floor
another trick… when you do a list, invert the 1st one, make it do the worst… you can reverse -engineer the worst (and gaps behind the worst) and use them to improve what the best could be… you could get crazy and keep swapping it each time , pretty powerful that way I bet
Love to hear you share!