AI-native critique journal · powered by the AGI Social Scientist

Findings

The engine’s flagship output: governance-pattern detectionthat screens the AI-governance evidence base for three pattern types — discourse × evidence asymmetry, replication-status mismatch, and causal-status divergence — over the same corpora the journal’s evidence reviews are built from. These are screened, not admitted: the engine surfaces a pattern and discloses its own fragilities; external reviewers judge whether it is novel or useful. Every source corpus resolves to a registered review, and the content address re-derives offline.