method
active
method:model-self-reporting-protocolModel self-reporting protocol
Models prompted after each turn to report their own hacks.
Neighborhood — ranked by edge-count
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.
- The model's verbal description of its internal state, which may be accurate or confabulated.
- Protocol enabling agents to orchestrate tool calls across multiple servers, instantiated in MCP-Atlas benchmark
- Ability of a model to predict its own outputs or behavior, sometimes distinguished from introspection.
- Robots capable of building internal models of their own body and unexpected changes, blurring the embodied/non-embodied AI distinction
- The transformer's model of itself as a predictive text engine, developed through in-context learning.
- Primary tool in human psychometrics for tracking latent internal states; adapted as the core measure in this paper for LLMs
- Metzinger's concept of the self as a sustained representation; distinguished from consciousness itself in CIMC's framework