概念

Yudkowsky-style argument

Q: Yudkowsky-style argument 最近有什么新动态？

traeai 已收录 1 篇与 Yudkowsky-style argument 相关的内容。最新一篇是「What it's really like to run AGI safety at Google DeepMind (and where I disagree with 'doomers') | Rohin Shah」，由 80,000 Hours Podcast 发布。

基于误泛化和对抗样本的 AGI 风险理论，由 Eliezer Yudkowsky 提出。

已跟踪 1 条高相关材料

TraeAI 观察

如果只读 3 篇

What it's really like to run AGI safety at Google DeepMind (and where I disagree with 'doomers') | Rohin Shah

80,000 Hours Podcast · 9 分

Rohin Shah 认为，尽管 AGI 安全风险值得重视，但灾难性对齐失败并非不可避免，常规对齐技术有望成功防止最坏情况，且当前主流担忧（如欺骗性行为）在实际训练中并不构成默认路径。

Rohin Shah on What It's Really Like to Run AGI Safety at Google DeepMind (and Where I Disagree with 'Doomers')

80,000 Hours Podcast6月2日27820 字 (约 112 分钟)

Rohin Shah argues that while AGI safety risks deserve attention, catastrophic misalignment is not inevitable, and prosaic alignment techniques are likely sufficient to prevent worst-case outcomes, especially since current concerns like deception are not default behaviors in real training.

入选理由：Rohin Shah 认为灾难性 AGI 对齐失败不是默认结果，缺乏足够强的论证支持其必然发生。

FeaturedPodcast#AGI#AI Safety#DeepMind#Alignment#Rohin Shah英文

跨材料问答 · Yudkowsky-style argument

回答基于：Yudkowsky-style argument 相关 1 条材料