前OpenAI安全总监批评公司忽视人工智能安全，呼吁加强下一代模型准备和安全性
原OpenAI超级对齐团队负责人Jan Leike在离职后首次公开批评了公司忽视人工智能安全的问题。Leike指出，在过去几个月里，其团队在开展安全研究时遇到了计算资源不足的难题，这直接影响了安全研究的进展。他认为，OpenAI作为一家肩负全人类责任的科技公司，应该将更多的精力投入到下一代模型的安全准备、监控和防患于未然的工作中。
Leike强调，OpenAI需要明确如何管理和控制比人类更聪明的AI系统，并提出了包括安全性、对抗性鲁棒性、超级对齐、保密性以及社会影响等方面的关注点。他批评公司在过去几年中，安全文化和流程被华而不实的产品所取代，这可能会对整个社会造成长远的影响。
Leike的这一表态引起了业界的广泛关注，人们对于人工智能的安全性和伦理问题再次进行了深入的讨论。业界专家普遍认为，随着AI技术的快速发展，确保AI系统的安全性、可控性和透明度已经成为全球科技公司必须面对的挑战。
此消息一出，OpenAI方面尚未作出回应。但可以预见的是，Leike的批评将促使OpenAI重新审视其安全策略，并可能在未来的产品开发中更加注重安全性和伦理考量。
英语如下：
Title: Former OpenAI Security Chief Warns Against Pursuing Glamorous Products at the Expense of AI Safety
Keywords: Criticism, Security, Responsibility
News Content:
Former OpenAI Security Director Criticizes Company for Neglecting AI Safety, Urges Enhanced Preparedness and Security for Next-Generation Models
Jan Leike, the former head of the OpenAI Super Alignment team, made his first public criticism of the company’s neglect of AI safety issues since his departure. Leike pointed out that his team faced challenges with insufficient computational resources while conducting safety research in the past few months, directly impacting the progress of their safety studies. He believes that as a tech company bearing the responsibility for humanity, OpenAI should allocate more resources to prepare for the safety of the next generation of models, monitoring, and taking preventive measures.
Leike emphasized that OpenAI needs to clarify how to manage and control AI systems smarter than humans, and raised concerns about areas such as safety, adversarial robustness, super alignment, confidentiality, and social impact. He criticized the company for replacing its safety culture and processes with superficial product-centric efforts over the past few years, which could have long-term implications for society.
Leike’s statement has attracted widespread attention in the industry, prompting a deep discussion on the safety and ethical issues of artificial intelligence. Industry experts普遍认为普遍认为, as AI technology continues to develop rapidly, ensuring the safety, controllability, and transparency of AI systems have become challenges that global tech companies must face.
As of the publication of this news, OpenAI has not made a response. However, it is expected that Leike’s criticism will prompt OpenAI to reassess its safety strategy and possibly place greater emphasis on safety and ethical considerations in future product development.
【来源】https://mp.weixin.qq.com/s/nERwuSXyXa7Blrsw-uW0EA
Views: 2