Google DeepMind 新任驻场哲学家谈 AI 伦理
哲学家 Henry Shevlin 于去年 5 月加入 Google DeepMind,职称为“Philosopher”,并将 80% 的时间投入该公司,同时仍在剑桥大学 Leverhulme 未来智能研究中心任职。他认为 AI 发展速度太快,学术界反应迟缓,与 LLM 的日常接触是形成良好直觉的必要条件。他还指出,抑制模型自我归因心理状态会连带降低其对动物福利的关注,因此不宜只改动模型的单一维度。
there are sort of web-like aspects to LLM cognition. But people are concerned about reports that human-acting AI has led people, including children, in dangerous directions— encouraging suicide , persuading a user that the AI is a god, or that the AI is a perfect companion or lover . Isn’t that an argument against letting AI sound like a person? Shevlin: It’s really important to make sure model behavior is suitably responsible for the context you’re using it in. One thing I can say though is, t
匹配字段正文