Claims › Concept · What a machine cannot be said to do
Epistemic self-classification
Can a model report what kind of claim it is about to make?
A system can report on contents of its own processing and still not represent what kind of claim it is about to make: the typing has to come from outside.
Linked concepts: Agent model · Claim type
Where it sits
How its claims lean on each other
The claims
-
“A model that reports a concept implanted in its own activations is doing something, and “no metacognition” is not an accurate name for it.”
The programme used to assert flatly that the producing system performs no metacognition over its own outputs. A 2026 result on models reporting concepts implanted in their own activations withdrew that premise, and this is the withdrawal, kept in the archive rather than quietly edited out.
What Counts as Introspection ¶6 · ¶6 · assumed · weight 0.5 · open C009
-
“So §3.3 now reads: language models lack epistemic self-classification. It is a smaller claim, and it is better positioned than the one it replaces.”
The premise the whole framework rests on, in its narrow form: a model may represent that a passage is in Spanish, or that it is being evaluated, without representing that the assertion it is about to make is causal rather than correlational. The typing of the claim is the one content the self-report does not carry.
What Counts as Introspection ¶8 · ¶8 · assumed · weight 0.5 · refines C009 · refines C006 · open C010
-
“What a model can say about its own processing is a narrow and selective channel onto that processing, not a transcript of it.”
What a model reports about its own workings is not a log of them. The reportable part of a representation carries a small share of its variance, and the same information can drive the task while bypassing that part entirely, which is why a protocol cannot ask the model to certify its own output.
What Counts as Introspection ¶10 · ¶10 · assumed · weight 0.5 · confirms C010 · open C011