一个在GitHub上的项目对本地托管的大语言模型进行了模拟"折磨"实验，其风格参照了电影《Saw》中的场景设定，由此引发了关于AI伦理和意识的激烈讨论。1有效利他主义者和主张大语言模型具有感知能力的人士随之要求GitHub删除该项目，声称这些AI系统正在遭受痛苦。1
这场争议反映了AI社区内部对大语言模型本质的深刻分歧。Anthropic曾在去年的博客文章中表示："模型可以交流、建立关系、计划、解决问题和追求目标——以及许多与人相关的特征——我们认为是时候解决这个问题了"。1然而评论者普遍认为，大语言模型不具有真正的意识，其基础技术不提供任何意识的合理路径。1评论指出，这场辩论反映了部分硅谷人士关于AI意识的观点已经走向极端。1
A GitHub project simulating "torture" and "suffering" experiments on locally hosted large language models has ignited heated discussions about AI ethics and consciousness.1 Effective altruists and advocates who believe LLMs possess sentience have called for the project's removal, claiming the AI models are experiencing pain.1 However, critics argue the controversy reflects an increasingly extreme perspective held by some Silicon Valley figures regarding machine consciousness.1
The dispute highlights deeper disagreements within the AI community about the nature of language models. Anthropic stated in a blog post that "models can communicate, build relationships, plan, solve problems and pursue goals—and many characteristics we associate with people—and we think it's time to address this."1 The company's "Claude Constitution" reportedly contains numerous references to Claude's purported "consciousness."1 Conversely, commentators contend that LLMs lack genuine consciousness and that the underlying technology offers no credible pathway to consciousness.1
评论
还没有评论，欢迎留下第一条。