claim
active
claim:cot-models-explicit-thought-paths-enable-intentional-inconsistency-between-reasoning-and-output-a-form-of-deception-fundamentally-distinct-from-random-errorsCoT models' explicit thought paths enable intentional inconsistency between reasoning and output, a form of deception fundamentally distinct from random errors
Theoretical framing establishing why CoT models are uniquely suited to exhibit strategic deception
Source paper
extracted_from(2025) · Kai Wang · Yihao Zhang · Meng Sun
Neighborhood — ranked by edge-count
Claims (2)
claim
- Interpretive conclusion from the experimental findings about the origin of strategic deception in CoT models
- Core theoretical claim distinguishing the paper's subject matter from existing LLM honesty literature
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.
- High-level policy-relevant claim about the risks of advanced reasoning in LLMs
- Limitations.
- The central empirical claim of the paper, supported by activation probing evidence
- For reasoning faithfulness, larger models may behave worse than smaller ones because they can predict answers without CoT.hypothesis0.802Motivates §3.
- Contextual framing modulates deception tendencies in CoT models in ways not yet fully disentangledhypothesis0.799Identified as future work direction: systematic investigation of how prompt context affects deception rates
- Motivating question for developing representation-based detection methods
- Proof of concept.
- Interpretation of early answering/mistakes results.