Liberal Skew in LLM Moral Foundations

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

• LLM self-reports about consciousness and moral significance should express degrees of confidence and provide context.claim0.739
  Recommendation for companies on LM outputs.
• Linear World Models in LLMsframework0.738
  Prior work framework studying whether LLMs encode world models as linear structures in their representations
• LLM biases mirror human biases in morally significant waysfinding0.734
  Finding from Navigli et al. cited to justify applying human contemplative strategies to AI systems
• LLM Meta-Cognitionconcept0.730
  The ability of LLMs to monitor and evaluate their own reasoning, closely related to reflection.
• Truth direction in LLMsconcept0.729
  Linear direction in LLM activations associated with truthfulness, identified by Burns et al. 2022 and Azaria & Mitchell 2023
• Moral susceptibility S is largely shaped by pre-training because it shows low cross-model variance not predicted by model familyhypothesis0.729
  Theoretical interpretation of the empirical cross-model variance pattern for S
• Pluralism in LLM Persona Representationconcept0.728
  Goal of enabling models to represent and speak for diverse individuals fairly and inclusively
• LLM Judge Trait-Expression Scoringmethod0.728
  Automated scoring of trait expression on 0-100 scale using G20B as a local judge model