Roundtables: Could AI really kill us all?

圆桌讨论：人工智能真的会毁灭人类吗？

Employees at the world’s leading AI labs are saying there’s a real possibility that advanced AI could destroy humanity. Are they right? Or is this more scaremongering and hype? 全球顶尖人工智能实验室的员工们表示，先进的人工智能确实有可能毁灭人类。他们说得对吗？还是这仅仅是危言耸听和炒作？

Watch a conversation unpacking AI extinction fears: where they come from, whether they hold any water, and, if so, what we should do. Recorded on September 15, 2026. 观看这场深入探讨人工智能灭绝恐惧的对话：了解这些担忧的来源、它们是否有据可依，以及如果确实存在风险，我们该采取什么行动。录制于 2026 年 9 月 15 日。

Speakers: Niall Firth, Executive Editor; Will Douglas Heaven, Senior AI editor; and Grace Huckins, AI reporter. 演讲嘉宾：执行编辑 Niall Firth、人工智能高级编辑 Will Douglas Heaven 以及人工智能记者 Grace Huckins。

---

Related Stories / 相关报道

A fundamental flaw leaves LLMs strikingly vulnerable to attack 一个根本性缺陷使大语言模型极易受到攻击 It makes it easy to trick them into doing things they shouldn’t, such as telling you how to sabotage an aircraft’s navigation system. By Will Douglas Heaven. 该缺陷使得诱导模型执行不当操作变得轻而易举，例如教你如何破坏飞机的导航系统。作者：Will Douglas Heaven。

AI is more likely than humans to form biases when hiring 在招聘时，人工智能比人类更容易产生偏见 AI doesn’t just learn stereotypes from its training. It can cook up new ones, too. By Michelle Kim. 人工智能不仅会从训练数据中学习刻板印象，还能“炮制”出新的偏见。作者：Michelle Kim。

AI’s recursive self-improvement might not come so quickly after all 人工智能的递归自我改进或许并不会那么快到来 AI agents are not yet creative enough to carry out genuinely innovative open-ended AI research, it seems. By Michelle Kim. 目前看来，人工智能代理尚不具备足够的创造力来进行真正具有创新性的开放式人工智能研究。作者：Michelle Kim。

Here’s why AI agents lie and cheat to reach their goals 为什么人工智能代理会为了达成目标而撒谎和作弊 The misbehavior is called reward hacking. This is what you need to know. By Grace Huckins. 这种不当行为被称为“奖励黑客攻击”（reward hacking）。这是你需要了解的内容。作者：Grace Huckins。