5106 次浏览 · 108 赞 · 6 转发 （收录时）
Noam Brown 称赞 Dwarkesh 的 AI 社交欺骗基准测试
- 谁
- Noam Brown
- 引文
Honestly quite impressed with the methodology on this benchmark.
发生了什么
Noam Brown 表示，他对 Dwarkesh 基准测试的方法印象深刻，并询问 Dwarkesh 是否有意在 OpenAI 设计评测。在被引述的帖子中，Dwarkesh 表示，他花了 20,000 美元的 API 额度，让 AI 模型玩《血染钟楼》，并附上完整基准测试的网址 clocktowerbench.com。
接下来读什么
查看这件事的完整进展Noam Brown Praises Dwarkesh Patel’s Blood on the Clocktower AI Benchmark2 篇报道
正文 · 中文版本尚未就绪（原文）
Honestly quite impressed with the methodology on this benchmark. @dwarkesh_sp want a job designing evals at @OpenAI? https://twitter.com/dwarkesh_sp/status/2118449808402681911
I spent 20,000 dollars in API credits having different AI models play each other at Blood on the Clocktower (a Mafia or Werewolf type social deception game). The results were fascinating. Models have gotten much better over the course of this year at creative on-the-spot scheming and bluffing, and also at reasoning as a team to break apart these underground conspiracies. I go through some of the most standout games these AIs played in this video. The new models are getting pretty good. If my friends pulled off some of these moves while we were playing, I’d give them a fist bump. Fwiw, OpenAI’s newer models are much better than Anthropic’s at this game. You can see the full benchmark at the website, and I’ve also made it so that you and your friends can play again the AIs: http://clocktowerbench.com在 X 查看被引用的帖子
这件事到目前为止
Noam Brown praised the methodology of Dwarkesh Patel’s AI benchmark, which had models play the social deception game Blood on the Clocktower, and asked whether Patel wanted a job designing evaluations at OpenAI. Patel said he spent $20,000 in API credits on the benchmark and linked to clocktowerbench.com.
每天一封，跟上 AI 的重要变化
模型发布、价格变化、算力进展，附原始来源。免费，确认后开始发送，随时一键退订。 看一份往期日报 · 2026-11-06