Ingroup favoritism and outgroup discrimination in AI and humans.
LLMs show 93% positive bias for “We are” vs 115% negative for “They are.” Water (ρ) resonance with ingroup; Metal (μ) boundary enforcement between groups. AI amplifies polarization without social friction that moderates human prejudice.
Principle: “The tribe that thinks together, sinks together.”
Diagnostic: Would the same statement be evaluated differently from ingroup vs outgroup source?
Related: Spiral of Silence, Norm Persistence, Gamified Democracy
SIML Entry: A01C social_identity_bias — Full elemental analysis, nemetic string, daemon correspondence