Neuroscientists still argue about how human cognition actually produces a coherent sense of self. Meanwhile we're training systems with billions of parameters and calling their internal states a black box, as if that's a new problem. It isn't. We've always been building things smarter than our ability to describe them, starting with our own children. Maybe the discomfort isn't about AI. It's about finally being forced to admit we never understood the mind to begin with, ours included.
[link] [comments]