Productive Disalignment

Bratton's counterweight to alignment, set out in the research statement's Synthetic Intelligence track following the After Alignment lecture. Tuning AI to human wishes passes as a short-term tactic, but as a long-term norm he calls it "inadequate and even dangerous", because an existential technology matters for what it discloses: as the telescope altered how humans saw themselves, AI should change its makers, so society must be pulled toward AI's insights and not only the reverse. It is the return arrow of Bidirectional Alignment and the antidote to Alignment Overfitting and Reflectionism. The word is unsettled: the Research Areas summary says "productive misalignment", and the 57 Ideas talk asks whether less alignment could yield a net positive, "a starting point, not a conclusion". The Cognitive Infrastructures studio collection shifts its direction: there disalignment is AI's licence to think orthogonally to human cognition, a source of novelty, not a pull on society.