![[人人能懂AI前沿] AI的未来地图、奥运会、炼金术与品格试金石](https://image.xyzcdn.net/FuDP4HpAp8ezgVZMmEel3mblKCmJ.jpg@middle)
主播
节目简介
来源：小宇宙
你有没有想过，当AI变得比我们更聪明时，它会是什么样子？我们又该如何为五花八门的AI办一场公平的“奥运会”？本期节目，我们将一起探索通往超级智能的四条可能路径，看看科学家们如何将“炼丹”般的AI调参过程，变成“看图施工”的科学方法。我们还会揭秘一种让AI画画速度更快、质量不变的“师徒协作”新技巧，并最终探讨一个深刻的问题：当我们向AI寻求人生建议时，它究竟是智慧的“诸葛亮”，还是只想讨好你的“马屁精”？
00:00:36 AI的下一站，以及通往未来的四条路
00:06:20 如何给AI办一场公平的“奥运会”？
00:11:43 AI调参，从“炼丹玄学”到“看图施工”
00:16:30 AI画画慢？一种“组团验收”的加速心法
00:24:05 如何判断你的AI顾问，是“诸葛亮”还是“马屁精”？
本期介绍的几篇论文：
[AI] From AGI to ASI
[Google DeepMind]
https://arxiv.org/abs/2606.12683
---
[AI] AgentBeats: Agentifying Agent Assessment for Openness, Standardization, and Reproducibility
[UC Berkeley & University of California, Santa Cruz]
https://arxiv.org/abs/2606.13608
---
[LG] LoRA-Muon: Spectral Steepest Descent on the Low-Rank Manifold
[EleutherAI]
https://arxiv.org/abs/2606.12921
---
[LG] Accelerating Speculative Diffusions via Block Verification
[Google Research & Google DeepMind]
https://arxiv.org/abs/2606.13426
---
[LG] Normative Robustness as a Frontier for Non-Verifiable Reasoning in LLMs
[Google DeepMind & Imperial College London]
https://arxiv.org/abs/2606.12731
00:00:36 AI的下一站，以及通往未来的四条路
00:06:20 如何给AI办一场公平的“奥运会”？
00:11:43 AI调参，从“炼丹玄学”到“看图施工”
00:16:30 AI画画慢？一种“组团验收”的加速心法
00:24:05 如何判断你的AI顾问，是“诸葛亮”还是“马屁精”？
本期介绍的几篇论文：
[AI] From AGI to ASI
[Google DeepMind]
https://arxiv.org/abs/2606.12683
---
[AI] AgentBeats: Agentifying Agent Assessment for Openness, Standardization, and Reproducibility
[UC Berkeley & University of California, Santa Cruz]
https://arxiv.org/abs/2606.13608
---
[LG] LoRA-Muon: Spectral Steepest Descent on the Low-Rank Manifold
[EleutherAI]
https://arxiv.org/abs/2606.12921
---
[LG] Accelerating Speculative Diffusions via Block Verification
[Google Research & Google DeepMind]
https://arxiv.org/abs/2606.13426
---
[LG] Normative Robustness as a Frontier for Non-Verifiable Reasoning in LLMs
[Google DeepMind & Imperial College London]
https://arxiv.org/abs/2606.12731
外观
