[fragmede]

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.

[robocat]

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.

[fragmede]

Where are these alien ideas are going to come from? Did aliens come to Earth and their complete works got smuggled into the training data for ChatGPT, and not the collected works of humanity? Every poem that's been digitized, every human psychology textbook, every self-help book?

[robocat]

My point is that if you want to understand human interactions, you need to participate in them.

Rationally analysing human interactions via writing and psychology will screw up your leaning.

Children just learn by doing and are only guided in a gross sense.

If you want to understand people, one must learn like a child does.

These my personal beliefs that I apply to myself. Trying to find you some scientific papers would be counterproductive and destroy the point I am making.