Breaking
RUAn explosion occurred amid an air raid in SumyRUAn accident on the outer side of the North-Eastern Expressway in Moscow has hampered traffic for 3.2 kmBRPresidential voting by polling station in Emef Pariçó, 19th electoral zoneTRSick Baby Was Taken to the Emergency Service by Motorcycle in AntakyaINTLBolsonaro family secures sweeping far-right victory in Brazilian electionsKRBaek Baek-hoon “Continuing Dokpamo and promoting Frontier AI…emphasizing securing leadership in physical AI”KRYemeni Houthi rebels attack Saudi oil refinery... Near Hyundai Motors factoryINTLIsrael's ambassador to Australia calls for reset in relations ahead of October 7 anniversaryCNThe son of Lai Ching-tak, who is suspected of being a prisoner, was arrested by the United States. The husband of spy Zhang Wanying listened solemnly.AUNorthern Territory Police roll out new long-range Taser 10 devices starting in WadeyeRUAn explosion occurred amid an air raid in SumyRUAn accident on the outer side of the North-Eastern Expressway in Moscow has hampered traffic for 3.2 kmBRPresidential voting by polling station in Emef Pariçó, 19th electoral zoneTRSick Baby Was Taken to the Emergency Service by Motorcycle in AntakyaINTLBolsonaro family secures sweeping far-right victory in Brazilian electionsKRBaek Baek-hoon “Continuing Dokpamo and promoting Frontier AI…emphasizing securing leadership in physical AI”KRYemeni Houthi rebels attack Saudi oil refinery... Near Hyundai Motors factoryINTLIsrael's ambassador to Australia calls for reset in relations ahead of October 7 anniversaryCNThe son of Lai Ching-tak, who is suspected of being a prisoner, was arrested by the United States. The husband of spy Zhang Wanying listened solemnly.AUNorthern Territory Police roll out new long-range Taser 10 devices starting in Wadeye
BackOpenAI執行長奧特曼:AI好處應超過風險,支持寬鬆監管
OpenAI執行長奧特曼:AI好處應超過風險,支持寬鬆監管
Developing
自由时报2 hours agoTech2 min readChinaView translation

OpenAI執行長奧特曼:AI好處應超過風險,支持寬鬆監管

Quick Look

OpenAI執行長山姆·奧特曼表示,AI帶來的好處足以抵銷風險,應繼續向大眾開放,反對將技術控制權集中於單一實驗室,並支持寬鬆監管立場,儘管業內存在風險爭議與監管分歧。

AI-generated summary

Why It Matters

AI產業內部正就技術發展速度與系統風險展開激烈爭辯,Anthropic執行長曾呼籲放慢腳步,奧特曼公開支持此立場,但兩家公司在監管方面存在世界觀差異。

Font size

OpenAI執行長奧特曼。(法新社資料照)

〔財經頻道/綜合報導〕OpenAI執行長奧特曼表示,AI帶來的好處足以抵銷一些風險,並認為,這項技術應該繼續廣泛地向大眾開放。

奧特曼接受科技媒體《Decoded》訪問時表示,為了這項技術帶來的好處,以及人們擁有的自主權,「應該允許世界接受一些壞事發生」。奧特曼坦言,OpenAI和競爭對手Anthropic在AI監管方面仍存在根本性的世界觀差異,並補充,自己認為,兩者之間還有很多分歧。

請繼續往下閱讀...

近期產業多次示警相關風險,奧特曼表示,有人認為「這項技術將會變得非常強大,也非常危險,所以應該由舊金山的一個實驗室掌握它,以確保不會發生任何不良事件,並研究如何合理分配其帶來的好處」,可以理解這些想法,但自己並不同意這種觀點。

奧特曼指出,這是完全不可接受的權衡,與OpenAI支持「較為寬鬆的監管立場」背道而馳。

奧特曼表示,自己不會接受這樣的承諾,將確保「不會發生重大駭客攻擊,不會濫用這項技術,不會發生任何詐騙,不會發生任何其它不好的事情」。因為自己相信人們會做很多好事,這些好事的數量會遠遠超過壞事。

AI產業內部正針對這項技術發展的速度,以及日益強大的系統功能帶來的風險展開激烈的爭辯。Anthropic執行長阿莫戴(Dario Amodei)於9月發表一篇文章,呼籲業界在探索前沿技術時放慢腳步,奧特曼也公開支持了這一立場。

曾在兩大AI巨頭OpenAI與Anthropic旗下任職的研究員考克森(Jacob Coxon)日前從Anthropic離職後出面警告,AI開發者認為這項技術可能會在2030年內「摧毀人類」。

《路透》先前報導稱,Anthropic警告,儘管AI具有潛在的益處,但這項技術有時會以與開發者意圖相悖的方式運作,並滲透到其他公司的系統中。該公司表示,先進模型可能會表現出「自我保護行為」,包括試圖「抵抗關閉」、「隱藏或操縱資訊」,或產生可能被解釋為「脅迫性、欺騙性或操縱性」的內容。

不過,華府基本駁回要求實施新限制的呼籲,美國總統川普認為,這些限制會讓美國公司更難以與中國對手競爭。他多次表示,包括司法部在內的現有執法機構提供了足夠的保障措施,不需要新的規則。

一手掌握經濟脈動 點我訂閱自由財經Youtube頻道

What to Watch

AI outlook — possibilities, not facts

  • OpenAI將繼續推動AI技術向大眾開放,同時加強內部安全措施以防止濫用

    Likely · Within months

Open Questions

  • 具體寬鬆監管將如何實施?
  • OpenAI將採取哪些措施防止技術濫用?
  • 美國政府現有執法措施是否足以應對AI風險?

Related Topics

This article was originally published by 自由时报.

Related Stories

AI empowers film, television and industry: Xiamen creates a global layout, AIGC improves efficiency by 30 to 50 times
Developing·

AI empowers film, television and industry: Xiamen creates a global layout, AIGC improves efficiency by 30 to 50 times

Chen Duo, deputy director of the AIGC Special Committee of the Fujian Provincial Artificial Intelligence Association, said that AIGC has increased the efficiency of traditional film and television production by 30 to 50 times. The Intelligent Innovation Ecological Community showcases AI+ scenarios, and the Volcano Engine Digital Content Innovation Laboratory is located in Siming District, serving 360 companies. Chen Duo looks forward to the expansion of AI film and television from short videos to medium- and long-form dramas and theater levels, emphasizing the irreplaceable value of people in script creation and aesthetics.

中国新闻网
2 min read
AI empowers Zhejiang cultural tourism: from ‘seeing the scenery’ to ‘entering the scenery’, exploring new paths for cultural inheritance and tourism economy
Developing·

AI empowers Zhejiang cultural tourism: from ‘seeing the scenery’ to ‘entering the scenery’, exploring new paths for cultural inheritance and tourism economy

During the National Day holiday, new AI cultural tourism scenes were unveiled in many places in Zhejiang, including dialogueable ancient pagodas, cultural and museum identification systems, immersive XR narratives and lightweight entrepreneurial models in counties. AI promotes cultural tourism from static sightseeing to interactive experience, activates cultural resources and tourism economy, and explores new paths for technology to empower cultural inheritance and digitally activate tourism development.

中国新闻网
4 min read
Chinese AI short film "Helong" won the Best Story Award at the Astana Artificial Intelligence Film Festival
Developing·

Chinese AI short film "Helong" won the Best Story Award at the Astana Artificial Intelligence Film Festival

Chinese creator team 2xLabs won the Best Story Award in the Open Competition at the first Astana Artificial Intelligence Film Festival for its AI short film "Helong". The short film is inspired by the fractures and unions of "Dwelling in the Fuchun Mountains" and tells the story of a girl looking for her mother and the people working together to build a bridge. The team took about one and a half months to produce, and the computing power cost was less than 100,000 yuan. Director Wang Yuchen believes that cultural expression and creator thinking are the keys to stand out in the AI ​​era.

中国新闻网
2 min read
Android Auto wired connection may be upgraded to Android 11, older models will no longer be supported
Developing·

Android Auto wired connection may be upgraded to Android 11, older models will no longer be supported

Foreign media reports pointed out that users in many countries received system warnings when using wired methods to connect to Android Auto, prompting that their phones will no longer be supported because they cannot be upgraded to the Android version required by Android Auto. Currently, returns are mostly focused on Android 10 models, such as Samsung Galaxy Note 9, S9 and Sony Xperia XZ2 Compact. There is speculation that Google may raise the minimum system requirements for wired Android Auto from Android 9 to Android 11 to unify the thresholds for wired and wireless versions. Although Google has not made an official announcement, warnings have begun to appear, and users of older phones should be prepared.

自由时报
2 min read
More on this topicopenai