
对齐问题
The Alignment Problem
Brian Christian · 初版 2020 年
7人推荐
11位读者提及
12次提及
为什么值得读
- 剖析如何确保计算机遵循人类意图的核心困境
- 系统盘点机器学习不同发展阶段的陷阱与缺陷
- 行文通俗易懂文笔极佳,生动描绘人机互动未来
一本令人瞠目结舌的力作,深入探讨了构建 AI 系统时出现的种种问题,以及旨在纠正这些偏差的变革行动。当今由数据训练的“机器学习”系统极其高效,以至于我们已邀请它们代我们观察、倾听并做出决策。然而警钟已经敲响:系统在筛选简历时,多年后我们才发现它们带有内在的性别偏见;算法在决定保释与假释时,似乎对黑人和白人被告区别对待;我们再也无法认定自己的房贷申请甚至医疗检验报告会有人类肉眼审核。而自动……
Hacker News 上怎么说
正在加载完整评论……
从从事 ML / 大语言模型(LLM)的产品经理视角来看……Brian Christian 写的《对齐问题》(The Alignment Problem)。我们该如何告诉计算机去执行所有我们想让它做的事、别做所有我们不想让它做的事,以及搞定所有我们自己都没意识到想让或不想让它做的事?如何才能……
From the lens of a Product Manager working in ML / Large Language Models... The Alignment Problem By Brian Christian How do we tell the computer to do all the things we want it to do, all the things we don't want it to do, and all the things we didn't realize we want it to do or not do? How do…
非常酷的项目。我目前正在读 Brian Christian 讲 AI 的《对齐问题》(The Alignment Problem,写得非常好,读起来很流畅),还在(以极慢的速度)啃 Garry Kasparov 的《我的伟大前驱》(My Great Predecessors)第一卷。这两个主题结合起来太棒了!我尤其感兴趣的是……
Very cool project. I'm currently reading The Alignment Problem by Brian Christian on AI (very well written and easy to read), and also (very slowly) My Great Predecessors, Vol 1 by Garry Kasparov. This is a great combination of the two topics! I'm particularly interested…
Brian Christian 写的《对齐问题:机器学习与人类价值观》(The Alignment Problem: Machine Learning and Human Values)[1]很好地梳理了这段历史,并探讨了不同 ML 方法中的一些陷阱。读起来非常有趣……
The Alignment Problem: Machine Learning and Human Values[1] by Brian Christian is a nice overview of some of this history and exploration of some of the pitfalls in different ML approaches. It's really interesting reading…
Brian Christian 写的《对齐问题》(The Alignment Problem)。它系统梳理了 ML 领域与人类价值观的复杂性。它为读者勾勒了一幅我们正在迈向的未来图景。
The Alignment Problem by Brian Christian. It sets up the complexity of ML field and human values. It paints a picture to the readers on the future we navigate.
《对齐问题》(The Alignment Problem)
The Alignment Problem
……读了几天要是还看不出这种书以及其作者的虚浮无物,那就真没救了。关于现代 AI 未来走向,好得多的书有 Stuart Russell 的《人类兼容》(Human Compatible)或 Brian Christian 的《对齐问题》(The Alignment Problem),这两本书都出版于 LLM 大爆发之前,但都预见到了它们的阿喀琉斯之踵——即无法控制它们学到什么,以及它们将如何使用学到的知识。
…days to not recognize the lack of substance in such a book, and in its author. Far better books on the possible futures of modern AI are Stuart Russell's "Human Compatible" or Brian Christian's "The Alignment Problem", both of which predate the boom in LLMs but still anticipate their Achilles heel -- the inability to control what they learn or how they will use it.