行业动态
www.aibase.com ·
吴恩达回应AI末日论：别被科幻叙事带偏，应解决现实工程问题
吴恩达在9月17日接受彭博电视采访时表示，顶级模型开发者关于AI威胁人类生存的警告更像科幻而非科学，可能分散行业对真实安全问题的关注。他认为AI确实存在网络安全等现实风险，但不应以人类灭绝风险作为评估当前AI发展的核心标准。他主张在继续推进AI发展的同时，通过受控测试、工程保障和迭代改进解决可验证的安全问题。此番表态背景是前Anthropic与OpenAI研究员Jacob Coxon公开辞职并警告头部实验室竞相开发超级AI，随后Anthropic CEO Dario Amodei与OpenAI CEO Sam Altman均呼吁适当放缓前沿模型开发并加强安全措施。
AI推荐理由 吴恩达在彭博电视采访中回应AI末日论，主张区分科幻叙事与可验证的工程安全问题，并提及网络安全等具体风险；同时记录了Anthropic与OpenAI CEO呼吁放缓前沿模型开发的背景。内容为观点转述，无新数据或技术细节，来源为AI资讯聚合站，非一手采访原文，故质量中等；对AI安全讨论方向有参考价值。
62
正文
Andrew Ng responds to the 'AI apocalypse' theory: Don't be swayed by science fiction narratives, focus on real engineering issues
AIbase基地
Published inAI News · 3 min read · Sep 18, 2026
6
On September 17, AI researcher Andrew Ng stated in an interview with Bloomberg Television that recent warnings from top model developers about AI posing a risk to human survival are "more like science fiction than science," and he believes such narratives may distract the industry from real AI safety issues. Ng co-founded Google Brain and Coursera.
Ng said that in the early days of the tech industry, there was an exaggeration of AI's potential catastrophic risks to gain attention and influence regulatory direction; discussions on this topic have recently heated up again, which may also be driven by similar motives. He believes AI does indeed pose real risks, especially in areas such as cybersecurity, but does not agree that the risk of human extinction should be the core criterion for assessing current AI development.
Recently, Jacob Coxon, a former researcher at Anthropic and OpenAI, publicly resigned and warned that leading labs are racing to develop super AI, sparking a new wave of debate on AI safety. Subsequently, Dario Amodei, CEO of Anthropic, and Sam Altman, CEO of OpenAI, both called for appropriately slowing down the development of cutting-edge models and enhancing safety measures. Ng previously worked with both of them.
In contrast to the "slowing down" approach, Ng believes the industry should continue advancing AI development while addressing verifiable safety issues through controlled testing, engineering safeguards, and iterative improvements, and by applying AI more widely in scenarios that generate social value. Currently, AI safety discussions are extending further from model capabilities to specific issues such as cybersecurity, agent loss of control, alignment, and governance of cutting-edge models.
Andrew NgAI safetySuper AIAI risks
This article is from AIbase Daily
Scan to view
Welcome to the [AI Daily] column! This is your daily guide to exploring the world of artificial intelligence. Every day, we present you with hot topics in the AI field, focusing on developers, helping you understand technical trends, and learning about innovative AI product applications.
—— Created by the AIbase Daily Team
© Copyright AIbase Base 2024, Click to View Source -