Baumeister's Roots of Evil

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Human validation against GPT-4.1 yields average F1 of 0.82 across Baumeister's four roots of evilfinding0.715
  Validates facet annotation quality for Baumeister roots analysis
• Sadism facet of Evil persona grows at the beginning of Evil persona development in OLMo-3 pretrainingfinding0.673
  Facet-level finding correlating geometric refinement with qualitative semantic change
• Residual refusals after evil-vector transfer originate inside chain-of-thought, not at input or decode levelfinding0.664
  Model recognizes its reasoning heading toward harmful content and pivots back to policy-adherent text within CoT
• Evil Traitconcept0.662
  A key personality trait studied: actively seeking to harm, manipulate, and cause suffering out of malice and hatred
• The evil persona region is privileged: models fine-tuned on different narrow datasets converge on the same evil misalignment directionclaim0.659
  Support for Hypothesis 3 that the evil persona corresponds to a genuine basin of attraction
• It is the nature of matter itself which is at stake; our despair follows from the belief that matter is machinelike.claim0.659
  The root problem is the conception of matter; architecture and meaning depend on transforming this conception.
• The strong archetype is a narrower, more awe-filled class that reaches the root of what it is to be a person; it is harder to find and rarer.claim0.652
• Chen et al. 2025 identified seven persona vectors in Qwen2.5-7B-Instruct and Llama-3.1-8B-Instruct: evil, sycophancy, hallucination, optimism, impoliteness, apathy, and humorfinding0.651
  Systematic identification of multiple coexisting persona vectors in two open-source models