Unverified50% confidenceFactExact time
When reasoning chains grow longer and multi-level derivation is required, the accuracy of System 1 models collapses sharply, because a single forward pass cannot carry too many intermediate states.
1
Sources
50%
Confidence
Long-term
Relevance
10/4/2026
First Seen
Sources
Related Entities
Related Claims
Unverified让大模型一句话生成完整App往往产出结构混乱、难以修改的结果，被称为"一次性生成的脆弱性问题"（one-shot fragility），任务复杂度超阈值时全局一致性急剧下降78% similarUnverifiedNaïve post-training quantization causes individual weight errors to accumulate and amplify across multiple layers—a phenomenon known as compounding error—and at 2-bit or 1-bit precision the model structure collapses75% similarUnverified长程Agent任务的实现依赖ReAct框架等推理-行动循环，错误会在链条中级联放大，导致完成率远低于单步问答74% similarUnverified级联失败是指单次工具调用错误会污染后续上下文，导致模型在错误的中间状态上继续推理，最终使整个任务偏轨74% similarVerified多代理系统存在失控的递归调用、协调开销激增和错误逐层放大（级联失败）的隐患74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/973487API
curl https://kongchang.com/api/v1/knowledge/claims/973487MCP
get_claim(id=973487)