00:01:19 高手过招,拼的不是“脑子”,而是“书房”?
00:05:35 AI的“自我反思”:如何让机器像高手一样思考?
00:10:31 AI的“第六感”:如何听懂线索背后的言外之意
00:16:06 AI的“犯罪现场调查”:我们能倒推出你对AI说了什么吗?
00:20:50 你的“队友”是真懂还是装懂?人工智能辩论赛里的一个发现
本期介绍的五篇论文:
[CL] Frustratingly Simple Retrieval Improves Challenging, Reasoning-Intensive Benchmarks
[Allen Institute for AI & University of Southern California & University of Washington]
https://arxiv.org/abs/2507.01297
[LG] Test-Time Scaling with Reflective Generative Model
[MetaStone-AI & USTC]
https://arxiv.org/abs/2507.01951
[LG] GradMetaNet: An Equivariant Architecture for Learning on Gradients
[University of Oxford & Technion]
https://arxiv.org/abs/2507.01649
[LG] GPT, But Backwards: Exactly Inverting Language Model Outputs
[University of Manchester & Mohamed bin Zayed University of Artificial Intelligence]
https://arxiv.org/abs/2507.01693
[CL] The Thin Line Between Comprehension and Persuasion in LLMs
[Microsoft & The University of York]