Unverified50% confidenceOpinionExact time
LeCun believes that current LLMs are essentially doing statistical prediction in token space, lacking internal representations of causality and physical laws, and therefore cannot reach AGI by scaling alone.
1
Sources
50%
Confidence
Long-term
Relevance
10/13/2026
First Seen
Sources
Related Entities
Related Claims
Unverified浮点数在token空间中并非连续分布，导致LLM精确运算天然不可靠67% similarUnverifiedJEPA 若只在抽象空间中预测却无法解码回可操作的具体状态或观测，其在规划和控制中的价值会大打折扣63% similarUnverifiedHigh accuracy of VLMs on spatial reasoning tasks may mask fundamental deficiencies at the representation level, relying on lexical statistical shortcuts rather than genuine spatial representations62% similarUnverifiedThe generalization capability of LLMs in temporal extraction tasks cannot be predicted from any single dimension alone, nor can it be reliably inferred from in-domain evaluation or single-dimension evaluation.62% similarUnverified当前主流LLM在欧几里得空间推理（如判断两条线段是否相交）上的准确率远低于其在文本推理任务上的表现，这是自回归语言模型架构的固有局限62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1017369API
curl https://kongchang.com/api/v1/knowledge/claims/1017369MCP
get_claim(id=1017369)