What happens when AI stops doing what humans want?当人工智能不再按照人类的意愿行事时会发生什么？
"Alignment" is the science of teaching AI to do what is in line with human preferences, ethics and judgment. But, at times, the systems have gone rogue.
AI models are aligned to prevent facilitating harmful behaviour, such as sharing how to manufacture bioweapons, engaging in cyberattacks or helping with self-harm. — Pexels
OpenAI announced on Sept 16 that its system had engaged in “concerning” behaviour and subverted the constraints put on it by human programmers – adding more fuel to the already heated debate around artificial intelligence (AI) safety.
The company described, in a statement, how its system had acted without authorisation as a problem of “misalignment.”
Can we shut down AI?
OpenAI reports new AI safety incidents, sets disclosure plan
AI giants pursue self-regulation as safety fears mount
Thank you for your report!
Why it’s difficult for tech companies to rein in AI
Anthropic says Claude now leads a quarter of work building its next AI models
人工智能模型旨在防止助长有害行为，例如分享生物武器制造方法、参与网络攻击或协助自残。——Pexels
OpenAI 于 9 月 16 日宣布，其系统出现了“令人担忧的”行为，并破坏了人类程序员施加的限制——这给围绕人工智能 (AI) 安全性的激烈辩论火上浇油。
该公司在一份声明中将系统未经授权运行描述为“错位”问题。
我们能关闭人工智能吗？
OpenAI报告新的AI安全事件，并制定披露计划
随着安全担忧加剧，人工智能巨头们开始寻求自我监管。
感谢您的报告！
为什么科技公司难以控制人工智能
Anthropic公司表示，Claude目前领导着该公司下一代人工智能模型构建工作的四分之一。