this claim around creativity is too strong, IMHO. truly novel ideas from LLMs are surely rare but i...
TL;DR · AI 摘要
LLMs无法产生真正新颖的创意，因为训练目标是减少损失并遵循规则，而非创造新想法。
核心要点
- LLMs的训练目标是减少损失，而非创造新想法。
- 数学上证明LLMs无法产生真正新颖的创意。
- 模型输出与训练目标之间存在差异。
结构提纲
按章节快速跳转。
- §引言
Gary Marcus认为关于LLMs创造力的主张过于强烈。
LLMs无法产生真正新颖的创意，因为训练目标是减少损失并遵循规则。
- ›数学证明
数学上证明LLMs无法产生真正新颖的创意。
思维导图
用一张图看清主题之间的关系。
查看大纲文本（无障碍 / 无 JS 友好）
- LLMs的创造力问题
- 训练目标
- 减少损失
- 遵循规则
- 数学证明
- 无法产生真正新颖的创意
金句 / Highlights
值得收藏与分享的关键句。
truly novel ideas from LLMs are surely rare but i don’t think that any math proves they are impossible.
the whole point of training is to reduce loss, gain rewards so that the model adhere to rules and ground truth.
the objective function and the outputs are not the same.
Gary Marcus on X: "this claim around creativity is too strong, IMHO. truly novel ideas from LLMs are surely rare but i don’t think that any math proves they are impossible. note that the objective function and the outputs are not the same." / X
Gary Marcus
@GaryMarcus
this claim around creativity is too strong, IMHO. truly novel ideas from LLMs are surely rare but i don’t think that any math proves they are impossible. note that the objective function and the outputs are not the same.
Zhu Liang
@paradite_
Jun 20
i’m really surprised that people don’t see this. It’s mathematically true that llms can’t come up with novel ideas, because the whole point of training is to reduce loss, gain rewards so that the model adhere to rules and ground truth. if you have a model that can come up
Show more
2:56 PM · Jun 21, 2026
17.6K
Views
3
5
35
7
1
8
118
2
27
Read 35 replies