OpenAI 的首席執行官 Sam Altman 敦促國會為高級人工智能 (AI) 系統制定許可和安全標準。 在參議院小組委員會的證詞中，奧特曼承認了人們對人工智能技術的擔憂，並強調瞭如果它出錯的潛在風險。會議中，OPENAI CEO 警告國會 人工智能“Can go quite wrong” 。
Altman 提議成立一個監管機構，為達到一定能力水平的 AI 項目頒發許可證，同時也有權撤銷許可證並強制遵守安全標準。 在此期間，他向委員會保證，OpenAI 會進行徹底的預測試和持續更新，以確保其人工智能工具的安全。 他認為，讓公眾廣泛使用這些工具有助於識別和減輕風險。
聽證會闡明了因迅速採用 ChatGPT 等人工智能係統而引起的廣泛擔憂，ChatGPT 是一種聊天機器人，在短短兩個月內就獲得了大約 1 億用戶。 討論的主題包括人工智能對選舉、知識產權盜竊、新聞報導、軍事行動以及多樣性和包容性倡議的潛在影響。
Altman 強調，為 ChatGPT 提供支持的最新版本系統（稱為 GPT-4）是一種受人類控制的工具，而不是一個自治實體。 他還承認，雖然某些工作可能會被人工智能自動化，但新的和改進的就業機會將因此出現。
立法者認為聽證會是理解這些新人工智能係統的能力和影響的第一步。 儘管對具體方法缺乏共識，但兩個政黨的成員都認識到在這一領域需要聯邦監管。
節錄:
OPENAI CEO 警告國會 人工智能“Can go quite wrong”
Sam Altman, the CEO of OpenAI, urged Congress to establish licensing and safety standards for advanced artificial intelligence (AI) systems. In his testimony before a Senate subcommittee, Altman acknowledged the concerns surrounding AI technology and emphasized the potential risks if it were to go awry.
Altman proposed the creation of a regulatory agency that would grant licenses for AI projects meeting a certain level of capability, while also having the authority to revoke licenses and enforce compliance with safety standards. In the interim, he assured the committee that OpenAI conducts thorough pre-testing and continuous updates to ensure the safety of their AI tools. He argued that making these tools widely accessible to the public aids in identifying and mitigating risks.
The hearing shed light on the broad range of concerns arising from the rapid adoption of AI systems like ChatGPT, a chatbot that gained approximately 100 million users in just two months. Topics discussed included the potential impact of AI on elections, intellectual property theft, news coverage, military operations, as well as diversity and inclusion initiatives.
Altman emphasized that the latest version of the system powering ChatGPT, known as GPT-4, is a tool under human control, not an autonomous entity. He also acknowledged that while certain jobs may be automated by AI, new and improved employment opportunities will arise as a result.
Lawmakers considered the hearing as an initial step in comprehending the capabilities and implications of these new AI systems. Despite a lack of consensus on the exact approach, members from both political parties recognized the need for federal regulation in this area.
Original content from Wall Street Journal.