作者/Author(s): Mustafa Suleyman 

網站來源/Source: Foreign Affairs 

日期/Date: 01/23/2024 

關鍵字/Keywords: 科技、人工智慧


摘要：
人工智慧（AI）在日常生活中變得越來越重要，雖然人們對AI伴隨的潛在政治、經濟和國際安全風險一無所知，但它正在迅速擴散。新型AI能夠獨立生成新知識，因此需要現代化的遏制政策來管理AI，防止其造成災害。 
  • 自人類誕生以來，基礎技術變得越來越便宜、易用和擴散，並隨著人類的進化融入新的發明中。 
  • 技術擴散的唯一例外是核武器，由於其高度危險性和破壞性，核武器受到政府嚴格控制。 
  • 人工智慧也必然會沿著以前基礎技術的路徑發展，而且人工智慧的進化速度比以往任何發明都要快。具有人工智慧能力（ACI）的AI現在可以自主完成各種複雜任務，將來可能產生自己的通用智慧（AGI）。 
  • 目前，大多數人都在討論人工智慧的好處，因為它可以提高生產力，推動經濟增長，徹底改變人們的日常生活。
  • 然而，人工智慧其實也伴隨危險。它可能通過簡單的語言命令創造新形式的誤導和虛假訊息，自動化流程可能擾亂勞動人力市場，有心人士也可能利用人工智慧進行網路攻擊。
  • 除了人工智慧的社會和倫理影響之外，它可能迅速並廣泛地傳播到文明的各個層面，因此人類必須遏制其進化。 
  • 人類必須要制定更嚴格的法規、新的問責和透明度，推動更強大的技術安全、治理和所有權模式，有效限制人工智慧。
  • 社會各層面的利益相關者都應該參與人工智慧的限制與遏制，並將硬體技術知識結合國際協議或制度，才能更有效地規範和審核人工智慧的使用。
 Summary: 
Artificial intelligence (AI) is becoming more important in daily life and is proliferating fast despite not knowing its potential political, economic, and international security risks. As newer AI can generate new knowledge independently, a modern form of containment policy is needed to manage AI and stop it from creating a disaster. 
  • Since the birth of mankind, foundational technologies have become cheaper, easier to use, and more proliferated, and they have been integrated into new inventions as humans evolve. 
  • The only exception to technology proliferation is nuclear weapons, which are controlled strictly by governments due to their highly dangerous and destructive nature. 
  • AI will inevitably follow the path of previous foundational technologies without resistance. Further, AI's evolution is faster than any previous invention. Artificial capable intelligence (ACI) can now complete a range of complex tasks autonomously, and in the future, it may generate its own general intelligence (AGI).  
  • Currently, most people are discussing the boon of AI as it can enhance productivity, drive economic growth, and revolutionize all aspects of daily life. 
  • Yet, it is important to note the dangers of AI. It may create new forms of misinformation and disinformation with simple language commands, disrupt the labor market through process automation, and increase the possibility of cyberattacks.
  • Apart from AI's social and ethical implications, its evolution must be contained because it can spread all layers of civilization fast and far.
  • Containing AI means stricter regulations, better technical safety, new governance and ownership models, and new modes of accountability and transparency to guarantee safer technology and controlled proliferation. 
  • Containing AI needs stakeholders from all layers of society with an interest in AI. It must incorporate hard technical knowledge with international agreements or regimes to regulate and audit the use of AI. 
 
回上層