Event ·Multi-source

Yuchen Jin 谈新 AI 实验室：多数会失败，挑战 scaling laws 者才有机会

Yuchen Jin 在 X 上回应外界关于他批评新 AI 实验室的提问。他认为多数新实验室会失败属正常现象，但质疑照搬 OpenAI 和 Anthropic 打法、融资 10 亿美元再把 50% 花在 GPU 上的路线撑不起百亿估值。他更看好挑战 scaling laws、挑战 Transformer 架构和下一词预测范式的逆向者。他同时指出，优秀研究者未必是好创始人，研究到产品的转化是普遍短板。

Current read

SWE-bench 上表现很强的 LMArena 团队创始人聊了点行业实话：抄 OpenAI 作业的新实验室大多活不下去，他押注的是那些挑战 Transformer 和下一词预测范式的团队，还点破了研究员创业的通病。

11 sources56° AI heatLast updated 2026/10/25 17:04:49

Evidence

11 sources

Primary sourceYuchen Jin

2026/10/25 17:04:49View source

Related source 1rohanpaul_ai

2026/10/23 17:28:46View source

Related source 2Ethan Mollick

2026/10/23 17:19:55View source

Related source 3kimmonismus

2026/10/23 19:54:11View source

Related source 4Gergely Orosz

2026/10/23 21:14:49View source

Related source 5IT之家

2026/10/24 00:59:37View source

Related source 6Andrew Curran

2026/10/24 02:39:19View source

Related source 7Thariq

2026/10/24 04:04:29View source

Related source 8Blognone

2026/10/25 06:24:37View source

Related source 9Nathan Lambert

2026/10/25 11:19:36View source

Related source 10Bindu Reddy

2026/10/25 17:49:17View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story