framework
active
framework:situational-awareness-dataset-sadSituational Awareness Dataset (SAD)
Dataset and framework for evaluating LLM self-knowledge including predicting own behavior
Neighborhood — ranked by edge-count
Papers (1)
paper
Thinkers (1)
thinker
- Laine, R.introducesLead author of the Situational Awareness Dataset (SAD) for LLMs
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.
- Model's access to information about its training objective, deployment context, and ability to distinguish training from non-training
- The specific form of reflection studied, where a model reflects on reasoning generated by another source.
- Ability of a model to describe its own learned behavioral tendencies.
- Circular causality between perception and action; central to enactive interpretation
- Measurable capacity of frontier LLMs to detect and report their own internal states, used as a downstream measure in Experiment 4
- A dialogue agent using first-personal pronouns and expressing self-concern in ways that suggest consciousness but are actually role play
- Happe 2003 hypothesis that humans use a single cognitive system for reasoning about mental states of self and others