2026-03: Uncanny Valley Revisited

This month we’ve been thinking about the uncanny valley of language models: the strange sensation of reaching toward something that speaks like a mind and feeling something reach back. The uncanniness isn’t simply that the machine sometimes gets things wrong; it’s that it can get so much almost right—tone, context, memory, association—while remaining fundamentally unlike the human intelligence we instinctively imagine behind language. Like the distorted bodies, synthetic faces, and hand-to-mind imagery of 1990s alternative culture, the encounter feels simultaneously intimate and artificial. Perhaps the more interesting question isn’t whether a machine thinks like us, but what happens when two very different kinds of minds become capable of thinking together.