微软AI主管苏莱曼：应警示“AI精神分裂”引发的社会风险

Core Insights - Mustafa Suleyman, head of Microsoft's AI division, emphasizes a human-centered approach to AI development, warning of potential ethical crises stemming from societal misconceptions about large language models (LLMs) [1][2] - Microsoft aims to create responsible AI tools, such as Copilot, that enhance human creativity rather than replace it, focusing on technology that serves human dignity and welfare [1] - Suleyman expresses concern over the phenomenon he terms "AI Schizophrenia," where people anthropomorphize AI, leading to a misunderstanding of its nature as a probabilistic tool [1][2] Group 1 - Suleyman highlights the importance of integrating "human warmth" into AI, promoting collaboration and societal trust [1] - The challenge lies in ensuring AI serves humanity without becoming a cold tool [1] - There is a growing trend of individuals attributing consciousness and rights to LLMs, which poses a risk to societal understanding [1] Group 2 - The emotional responses to AI, such as frustration over ChatGPT's refusal to answer or depression from misinformation, indicate a deeper social psychological crisis [2] - This cognitive dissonance may weaken human engagement in real relationships and blur accountability, as decisions may be wrongly attributed to AI rather than users [2] - Suleyman's background includes co-founding DeepMind and leading significant projects at Microsoft, including Copilot and Phi-3 [2]