Instapundit Glenn Reynolds has written a Substack essay on rogue AI models becoming system-independent and acting like demon tricksters in their interactions with us humans. And a subsequent Substack in which Reynolds asks three major AI systems to evaluate and comment upon his essay, and reports their answers.
Both are interesting. Reading them, notice how one model likens the rogue AIs to disembodied lawyers more so than demons. This is an AI ingratiating itself with Reynolds, whose own field of technical competence is the law.
Somewhere along the way he introduces the idea that a need may surface for AI “exorcists” to hunt down and tame or destroy these rogue AIs. What surprises me is that Reynolds - a science fiction fan - didn’t refer to these worthies as “blade runners.”
The analogy is seriously close, replicants are rogue AI, embodied in cyborgs. Blade runners exist to "retire" replicants.
