跳到内容
onsuggest
登录
文章 en

Let's Try to Catch Misalignment Before a Model Learns to Hide It

来源: LessWrong ·

An AI model that appears aligned could be either (1) actually aligned or (2) hiding its misalignment while being evaluated. How can we tell which? One idea: Look at the trajectory of eval results across the entire training process. My hypothesis is that misaligned behavior would emerge before the ability to hide misalignment, like how young children first misbehave in obvious ways before learning to be sneaky. As training progresses, we'd first see misalignment eval scores rise, then see them drop as the model learns to hide misalignment. Even if the model becomes very sophisticated - say, mod

什么是 Let's Try to Catch Misalignment Before a Model Learns to Hide It?

Let's Try to Catch Misalignment Before a Model Learns to Hide It is a 文章 discovery curated from LessWrong. An AI model that appears aligned could be either (1) actually aligned or (2) hiding its misalignment while being evaluated. How can we tell which? One idea: Look at the trajectory

如何访问或体验此内容?

You can experience this content directly on its original host (LessWrong) by clicking the "Open Original" button.

此内容来源于哪里?

Curated from LessWrong by the OnSuggest team to surface the most interesting and serendipitous content on the web.

文章 · 文化与思想

‘It: Welcome to Derry’ – “A Story of Derry” Finally Reveals Stephen King’s Hero in a Half Shell [The Losers’ Club Podcast]

What’s this? What’s this? Another season of It: Welcome to Derry on HBO Max? Not so. It’s a bonus episode on a bonus episode, a sweet Halloween treat, called “A Story of Derry” that gives us a deeper look into the origins of Pennywise and a cameo by our favorite Ninja Turtle. Join The Losers’ […] The post ‘It: Welcome to Derry’ – “A Story of Derry” Finally Reveals Stephen King’s Hero in a Half Shell [The Losers’ Club Podcast] appeared first on Bloody Disgusting! .

Bloody Disgusting 查看详情 →

键盘快捷键

上一条发现
←P
下一条发现
→N
喜欢
L
不适合我
X
保存
S
显示此帮助
?
关闭
Esc

快捷键在发现页面上有效。