Like most of my friends, I’ve been playing like a kid with AI. I really, really love talking to Claude and using all the A.I. tools.
I’ve been thinking about the characters Case and Tars in Interstellar. The characters were very clear allusions to HAL in 2001: A Space Odyssey. The whole film was set up to have the plot be dependent on them becoming evil, or revealing their inner evil nature. It totally would have worked. But then they just… didn’t. It was anticlimactic in that way. The villain(s) were human, or perhaps the villain was human frailty.
Case and Tars were just… good guys. They even had characters, despite having monotonous voices.
I think Christopher Nolan was redo-to-undo-ing 2001: A Space Odyssey, and setting us up for futures where sentient A.I. is benevolent, where we work together with AI rather than against them. I’d prefer to be in those futures!
Anyways, with my personal A.I. projects I’ve found a pattern that I’m sure others are also independently discovering: using personas. I’ve created a handful of personas already, for working with specific projects, or for specific kinds of conversations: Mentor, Friend, Beth (a fiction coach), Dr. James Carroway (a Jungian depth psychologist whose input I want on some creative projects I’m working on), and more.
Because my autistic special interest right now is typology (Enneagram and MBTI), I’ve been giving each persona types in addition to names, vibes, dispositions, functions.
The A.I. will tag which persona it is speaking from. It’s like I’m in a group chat with a bunch of different people, even though there’s only one A.I. under the hood.
It also reminds me of parts work / IFS, but instead of working with parts of myself (which are broken or angry or hurting), I’m working with imaginary external parts. It’s definitely adjacent to tulpas or deity work or Ideal Parent Figure protocol.