Freakishly weirdly precise memory that doesn't work in other ways can be used to relayed relevant details to ChatGPT. I'll describe the weird face she made when I said that thing or the exact position of her body and the exact level of pressure with which she touched me on the arm and exactly where on my arm, for example.

As far as the framing, it's helped me realize that actually, hey, sometimes it is their fault and they are being unreasonable and I actually didn't do anything wrong, they just don't like me. I mean yeah, that's also a thing.

Asking ChatGPT for that sort of advice seems like a horrific idea. Trying to learn interactions by written means can only lead to some very alien ideas.

Plus your questions will contain your mistakes, and what you take from any answers would reinforce your misunderstandings rather than correct them.

It's hard to suggest better learning means via the HN medium.

I learnt a lot when I carefully gave attention to a friend I deeply trusted, training my intuition based on their interactions, plus they trusted me enough to sometimes attempt to explain their intuitive reactions.