finding
active
finding:generic-person-1-person-2-dialogues-yield-probes-similar-to-human-assistant-onesGeneric Person 1/Person 2 dialogues yield probes similar to Human/Assistant ones
Representations not tied to Assistant.
Source paper
extracted_from(2026) · Nicholas Sofroniew · Isaac Kauvar · William Saunders · Runjin Chen +12
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Not Assistant-specific.
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.
- Empirical characterization of conversation domains that are safe for model persona stability
- Similar number and nature of exclusive features across models results from oversampling of Human/Assistant chat transcriptshypothesis0.755Motivated by Bricken et al; tested and not supported.
- Distinct representations for the operative emotion of the current versus the other speaker.
- Shows the key divide is passive vs. active framing, not the specific wording of instructions.
- The motivating question the paper sets out to answer by proposing role play and simulation metaphors
- Additional evidence that core representations are persona-relative, supporting Claim about persona-relative representations
- Conditional prediction about how a well-informed dialogue agent would handle questions of personal identity
- Addresses skeptical alternative that reports reflect only conversational content