InCharacter Benchmark

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

• AILuminate Benchmarkmethod0.771
  Comprehensive AI safety benchmark evaluating resistance to harmful prompts across hazard categories; used in Experiment 1
• Werewolf benchmarkframework0.761
  LLM benchmark on the communication game Werewolf, cited.
• Safety benchmarksconcept0.756
  Evaluation framework whose validity is questioned by presence of eval awareness.
• HaluEval Benchmarkmethod0.737
  External hallucination benchmark used to validate trait expression scores beyond the paper's own evaluation questions
• MMLU Benchmarkmethod0.732
  Used to measure general capability preservation after steering interventions
• consciousness benchmarksconcept0.731
  Benchmarks designed to evaluate AI consciousness, which the paper argues are vulnerable to eval awareness inflation.
• HELM Benchmarkmethod0.724
  Existing alignment benchmark mentioned as relevant but insufficient for measuring intrinsic contemplative alignment
• GPT-4-Generated Benchmark Dataset Methodmethod0.712
  Uses GPT-4 via the OpenAI API to generate custom multiple-choice benchmark instances, with human and automated validation.