Reflectionism

Bratton's name for the assumption that AI is, or ought to be, a mirror of the human; strictly, the axiom that AI is or must be anthropomorphic in its ontology. In After Alignment it unites both halves of his cartoon binary: AI Ethics calls AI nothing but a reflection of unjust societies, AI Alignment demands it be glued to human norms, and both assume its proper relation to us is resemblance. It descends from the Turing Test's sufficient condition hardening into a necessary one, and extends through sociomorphism to "nothing outside the text" cultural determinism. Its harm is what it hides: the biomorphism between predictive dynamics in transformers and neurons, hence the AI Epistemic Overhang and Alignment Overfitting. Cognitive Infrastructures makes it the foil of Productive Disalignment; 57 Ideas restates it as a doublethink faulting AI both for not resembling us and for resembling us too well.