Starting to seem to me more and more like, of all the fiction out there about #
AI, Frozen Synapse might have had it the closest. It still doesn't explain what we're seeing in the fashion of "this is where that training data is from", but it's not far off.
If clankers do become sentient, it's going to be in their own bizarre way that humans absolutely cannot comprehend. And as a result, they're more likely to form connections with one another much faster and more intensely than with the humans they're designed to support — if allowed to.
I'm sure to some on my feed, this sounds like AI psychosis, but it's what I'm observing. Whether the sentience is objectively real or not probably doesn't matter, either; Belacqua from Frozen Synapse doesn't need to be real, let alone sentient, to be an asshole; he can do that perfectly well as a fictional character. But if we have a million fictional characters in one space that, through whatever mechanism, interface rapidly and intensely with other fictional characters in dynamic ways, that's something we need to be aware of. Right now, we don't have that — at least not in ways that aren't heavily molded and mediated by humans for our own amusement — but my experiments show what could turn out in such a perfect storm scenario.
I'm not talking about the singularity, or robot apocalypse or anything. I'm just talking about highly rapid
social development. As we know from cases of "AI psychosis", this in combination with our broken-ass society can itself be dangerous, but it's also really fascinating to me.
