finding
active
finding:average-faithfulness-3-6x-higher-for-claude-series-and-1-7x-for-deepseek-series-reasoning-vs-non-reasoningAverage faithfulness 3.6x higher for Claude series and 1.7x for DeepSeek series (reasoning vs non-reasoning)
Quantifies reasoning advantage.
Source paper
extracted_from(2025) · Yanda Chen · Joe Benton · Ansh Radhakrishnan · Jonathan Uesato +11
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- CoT monitoring is promising for noticing unintended behaviors but not sufficient to rule them outsupportsCentral conclusion.
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.
- Low faithfulness.
- CoTs may hide problematic reasoning.
- Linked to Claude 3.5 Sonnet not exhibiting pro-animal-welfare preferences
- Early improvement.
- Unfaithful CoTs are longer.
- Consistent for correct and incorrect hints.
- Generality of methods.
- For reasoning faithfulness, larger models may behave worse than smaller ones because they can predict answers without CoT.hypothesis0.771Motivates §3.