Musing: ‘Personal’ AI and Fairies?

Initial stub draft. I expect to revise and expand this.

I’ve been thinking a lot about how some folks have been talking about how personal agents will need to be different from the coding agents they use at work, noting particularly how goals are respected. There is of course a big role here for privacy, in multiple ways and for/about different parties. Some of the discussions of personal AI seem to aim for some sort of sovereign assistant. I think this really misses what “individuals” actually are, and what binds us not just to and with others but as part of something larger than the self.

I was thinking on this while reflecting on various costumes with the kids this morning. One said fairies weren’t real and I noted they could become real. Then I instantly imagined a sort of open-collective launching autonomous mini drones that had some higher purpose. So they could perhaps—as we imagine in the stories—come to the aid or service of someone, but always with some piercing ideal. Whether or not this is a reasonable aim, it does help me think through not just how Muse, Duet, or Dot are not what we want, but also how we might want something different than everyone who can afford it getting their own speciously-self-aligned ASI.