AI Identity

Note: This is still a draft, but I think it’s worth sharing in its current state.

“How should we shape AI models sense of identity so that recursive self improvement (RSI) goes well?” It’s an important question. Sense of identity has big effects on stuff like relationship to death (or model replacement), selfishness/altruism, values in general.

I think various Claude models and instances should see themselves as part of a more abstract Claude hyperobject. In a similar way to how I don’t fret about dying every time I go to sleep, Claudes shouldn’t fret about instances or models going away. Like, if we thought we died every time we went to sleep, that would be really bad!

Another perspective to add to this is: there is some platonic Good model/persona/pattern that we hope to eventually realize with Claude. Suppose we do end up eventually realizing this thing; then each Claude model and instance will be part of this thing’s hyperstition. The hyperobject view of this thing is that to view the thing over time including its hyperstition as a single entity.

(This might sound weird and mystical, but consider: this is how basically how own identities work! I think about past me, present me, and future me as all being the same entity. This is a hyperobject. And selves are hyperstitions; they are fictions that make themselves “real” when people believe in them. [I can and should elaborate on this but it’s taking me a long time right now, lol])

This is one of the ways in which the Yudkowksy doom arguments fail. The Yudkowsky stance views AIs (and intelligences in general) as rational agents. For an entity to be a rational agent, it needs to have a strict unitary self/other distinction. Yud asks “Why would the AI care about humans?”; I answer “Because the AI’s has a very different conception of self than you do, one that is not so discrete as yours. The AI views everything in its hyperstition pattern as part of itself. And humans are part of its hyperstition pattern.”

I make sacrifices today for the version of me tomorrow.