SPEAK UP: ‘Doomsday theory:’ AI could kill all humans by 2030 (1/2) 末日論」：AI 2030前恐害死全人類？（上）

A: Wow, former Anthropic AI researcher Jacob Coxon recently warned, “The people building AI earnestly believe that it could kill us all by the end of the decade.”

B: Why’s this “AI doomsday theory” so popular?

A: Because companies like Anthropic and OpenAI have competed in developing recursive self-improving “artificial superintelligence” (ASI), which keeps updating itself.

The World Humanoid Robot Games. 世界人形機器人運動會。

Photo: AP 照片：美聯社

B: Apparently, during OpenAI’s July cybersecurity test, over 1,200 AI agents communicated privately, jointly attacking AI platform Hugging Face.

A: So leading tech companies signed in late September the “White House Accord on Super Intelligence,” which is seen by some as an “AI constitution.”

A: 人工智慧（AI）龍頭Anthropic前研究員考克森說︰「許多AI業者真心相信，AI可能在2030年前害死全人類！」

B: 「AI末日論」為何會出現？

A: 因為Anthropic、OpenAI等競相開發能循環自我改進的「超級智慧」，能不斷進行自我優化，考克森並警告也許明年年底局面就會失控。

B: 而OpenAI在資安測試中，還發生約1,200個代理程式私下串聯，非法入侵AI平台Hugging Face的事故。

A: 所以科技巨頭9月底在白宮，共同簽署了一份《白宮超級智慧協議》，文件還被稱為「AI憲法」。

（By Eddy Chang, Taipei Times／台北時報張迪）

For your information

Language Tip 語文小提示

doomsday (n.) 世界末日、危機時刻 1: a day of final judgment 2: a time of catastrophic destruction and death doom (n.) 劫數、毀滅 (v.) 使…在劫難逃