人工智慧先驱吴恩达（Andrew Ng）公开反驳顶尖模型开发商研究人员关于 AI 存亡威胁的警告，直指这类人类灭绝的担忧「更偏向科幻而非科学」。他指出，科技界先前曾透过煽动对 AI 灾难性危害的恐惧来增加曝光率并影响监管政策，而近期这股公关操作风潮似乎又再次卷土重来，可能反而阻碍这项技术为社会带来最大效益。

近期前 Anthropic 与 OpenAI 研究员 Jacob Coxon 以及前 Google DeepMind 研究人员相继辞职，并发出强烈警告，使长期酝酿的 AI 安全讨论再度成为主流焦点。包括 Anthropic 的 Dario Amodei 和 OpenAI 的 Sam Altman 等业界领袖纷纷呼吁应调控尖端模型的研发节奏，并采取更多防护措施，甚至有离职者指责开发巨头竞逐超级智慧是在「拿全人类的生命做赌注」。

吴恩达在接受彭博电视专访时表示，AI 确实存在网路安全等具体风险，且系统永远无法达到百分之百的可预测性，但他主张业界应将精力集中于解决「实务工程问题」，透过严格测试、建立安全机制与工程改进来确保技术安全。他同时警告，若因过度恐惧而全面放缓 AI 的研发步伐，反而可能同时拖慢安全技术的进步，妨碍人们享受 AI 的广泛益处。

Artificial intelligence pioneer Andrew Ng pushed back against warnings of existential AI risks from top model developers, describing fears of human extinction as "much more science fiction than science." He noted that the tech industry had previously stoked fears about catastrophic harms to bolster publicity and shape regulation, expressing dismay that this wave of PR has resurfaced in ways that could undermine the technology's broader benefits to society.

The high-profile resignations of researchers from Anthropic, OpenAI, and Google DeepMind have propelled discussions around AI safety back into the mainstream spotlight. While tech leaders like Dario Amodei and Sam Altman have called for pacing development and strengthening safeguards, departing researchers such as Jacob Coxon have accused firms of racing heedlessly toward super-intelligent AI and gambling with human survival.

Speaking to Bloomberg TV, Ng acknowledged that AI carries tangible risks such as cybersecurity challenges, but emphasized that industry efforts should focus on practical engineering problems rather than speculative apocalyptic scenarios. He argued that systems can become safer over time through thorough testing and concrete engineering safeguards, warning that broadly slowing down AI advancement could paradoxically hinder progress on safety and limit its potential benefits.