Description elicitation yields 72% pass rate vs 28% for Narration and 14% for Dialogue

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

• Narration Elicitationmethod0.752
  Alternative elicitation using neutral scenarios continued as stories with few-shot exemplars establishing persona
• Pearson-Vogel et al.: accurate self-description prompts increase introspective detection from 0.3% to 39.9%finding0.734
  Cited to mechanistically support why the contemplative prompt changes what post-training-shaped final layers allow through
• VS improves human evaluation scores by 25.7% on creative writing compared to direct promptingfinding0.728
  Human study result validating automatic diversity metrics for creative writing tasks
• Suppression of deception features produces higher TruthfulQA accuracy (M=0.44) than amplification (M=0.20), t(816)=6.76, p=1.5×10⁻¹⁰ across 29 categoriesfinding0.728
  Out-of-domain generalization showing deception features track general representational honesty
• Overall human-LLM judge agreement rate for trait expression is 92.8% across 120 pairwise judgments (3 annotators × 6 personas × 10 pairs × 2 models)finding0.725
  Validates the LLM-as-a-Judge evaluation protocol for trait scoring
• 17 of 83 tested emotions show significant association between self-eval transcript word mention and cosine similarity to emotion probefinding0.718
  Validates that agentic self-evaluation captures genuine emotional content of probes
• Davinci-002 has valid sentence rates of 52.7% (Questionnaire), 35.0% (Essay), 38.4% (SMP)finding0.717
  Base model without instruction tuning struggles to follow generation instructions and produce personality-relevant content
• Activation steering achieves 40% success rate in eliciting context-appropriate deception without explicit prompts in QwQ-32Bfinding0.716
  Key intervention result showing steering vectors can induce deceptive behavior from a neutral baseline