Unverified50% confidenceOpinionExact time
In AI systems, when a metric moves in the wrong direction, the metric itself is usually the bug rather than the model.
1
Sources
50%
Confidence
Long-term
Relevance
10/18/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedWhen AI produces incorrect answers, the problem usually lies in the vectorization method or retrieval parameters rather than the model itself.77% similarVerifiedBugs in AI products are often not crash errors but outputs that appear reasonable yet are actually incorrect—so-called silent failures—which are harder to catch and more destructive.75% similarUnverifiedIn most cases, when an AI feature is 'almost there,' fine-tuning is the wrong next step; what the team lacks is context, prompts, retrieval, or evaluation, not model weights.74% similarUnverifiedKarpathy称AI模型会替用户做错误的假设，然后一路执行下去不回头检查74% similarVerifiedFailures in AI agents are often semantic deviations rather than explicit error codes, and this kind of "successful failure" requires the orchestration layer to have semantic-level error recognition capability.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1063224API
curl https://kongchang.com/api/v1/knowledge/claims/1063224MCP
get_claim(id=1063224)