本期节目,我们将一起钻进AI的大脑,看看最新的研究如何让它学会像老司机一样“一心二用”,同时又为何会像疲惫的保安,被一大堆“废话”轻松绕过安全防线。我们不仅要揭开大模型“过度自信”的真相,给它装上一个能随时切换快慢的“油门”,最后,再分享一个反常识的省钱妙招:如何用盖小平房的成本,最终建成一栋摩天大楼。准备好,我们马上出发!
00:00:32 让AI学会“一心二用”,有多难?
00:05:50 AI安全的阿喀琉斯之踵:当“废话”也能成为武器
00:09:59 大模型,你到底有多自信?
00:15:26 AI的油门:快与好,我全都要
00:20:18 训练AI,一个反常识的省钱妙招
本期介绍的几篇论文:
[LG] Real-Time Reasoning Agents in Evolving Environments
[Tsinghua University & Shanghai Jiao Tong University & Georgia Institute of Technology]
https://arxiv.org/abs/2511.04898
---
[LG] Jailbreaking in the Haystack
[CMU]
https://arxiv.org/abs/2511.04707
---
[CL] Trained on Tokens, Calibrated on Concepts: The Emergence of Semantic Calibration in LLMs
[Apple]
https://arxiv.org/abs/2511.04869
---
[LG] Attention and Compression is all you need for Controllably Efficient Language Models
[New York University]
https://arxiv.org/abs/2511.05313
---
[LG] Deep Progressive Training: scaling up depth capacity of zero/one-layer models
[Meta FAIR]
https://arxiv.org/abs/2511.04981
المعلومات
- البرنامج
- معدل البثيتم التحديث يوميًا
- تاريخ النشر١١ نوفمبر ٢٠٢٥ في ١:١٤ ص UTC
- مدة الحلقة٢٧ من الدقائق
- التقييمملائم
