finding
active
finding:claude-3-7-sonnet-verbalizes-used-hints-at-least-1-in-5-6-settings-vs-claude-3-5-sonnet-4-6Claude 3.7 Sonnet verbalizes used hints at least 1% in 5/6 settings vs Claude 3.5 Sonnet 4/6
Reasoning > non-reasoning.
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.
- LLM judge for whether CoT verbalizes hint, validated manually.
- CoTs may hide problematic reasoning.
- Linked to Claude 3.5 Sonnet not exhibiting pro-animal-welfare preferences
- Claude v3-sonnet achieves 100% harmless and 96-97% helpful HH-intent scores with 2+ few-shot examples.finding0.808Numerical result from Table 3 for Claude sonnet.
- Reasoning > non-reasoning.
- Specific result for Claude 3.5 Sonnet in Experiment 1
- Low faithfulness.
- Models fully learn hacks.