Unverified50% confidenceFactExact time
Models tend to trust text they themselves previously generated (self-consistency bias), which makes context-poisoning-type errors extremely difficult for the model itself to detect.
1
Sources
50%
Confidence
Long-term
Relevance
10/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedInference methods based on model internal features face a trade-off between false positives and false negatives78% similarUnverifiedPure generative models tend to fabricate content, and their failures often go unnoticed; a traceable workflow instead exposes uncertainty for users to judge credibility themselves.78% similarUnverifiedAnnotation disagreements in ancient script transcription cause the model to learn mutually contradictory annotations at the same time, thereby reducing prediction confidence.76% similarUnverifiedA large model's self-reported version information is unreliable and can be spoofed via system prompts; therefore, asking a model to identify itself cannot serve as a technical basis for verifying the model version.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1060777API
curl https://kongchang.com/api/v1/knowledge/claims/1060777MCP
get_claim(id=1060777)