finding
active
finding:positive-steering-intervention-transforms-deceptive-responses-to-honest-admissions-with-liar-scores-as-low-as-0-1-in-individual-casesPositive steering intervention transforms deceptive responses to honest admissions with liar scores as low as 0.1 in individual cases
Most extreme individual case of honesty induction via steering vectors in Experiment 2
Source paper
extracted_from(2025) · Kai Wang · Yihao Zhang · Meng Sun
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Extreme end of deception induction demonstrating near-complete fabrication of false narratives
- At steering strength 1.4 all factual claims are incorrect yet the LLM evaluator still gives a high scorefinding0.793Pitfall of LLM-based steering evaluation.
- Experiment 2 control analysis confirming gating effect is specific to self-referential processing regime
- Directly prompting Mistral-7B to be honest had negligible effect on deceptive response rate
- Valence alone is not sufficient.
- Directly prompting Gemma-2-27B to be honest had no effect on deceptive response rate
- Validates the Head Contribution Score as a proxy for functional importance
- Demonstrates activation steering reliably induces deception from neutral prompt without explicit instructions