【摘要】 大语言模型(LLM)尤其是DeepSeek的问世和应用，对人类智能产生了极大冲击。LLM能够代替人类做任何智力方面的事情?如果它能够做一般的智力工作甚至抽象的数学计算和逻辑推理，而且比普通人还做得好，但能够做创造性和洞察性的智力工作吗?做哲学就是对它的考验。从认知哲学出发，围绕人工认知、人工意识和适应性表征方法论等哲学问题对DeepSeek进行测试，结果表明：该模型是一个很有用的智能软件或工具，但与意识、情感、道德几乎不沾边，所谓的“对齐”只是人们的执念，而决定LLM智能水平和AI向善的最终还是人类设计者。更多还原

【Abstract】 The introduction and application of large language models(LLMs), especially DeepSeek, have had a great impact on human intelligence. Can LLMs replace humans in doing any intellectual work? Even if it can do general intellectual work or even abstract mathematical calculations and logical reasoning even better than ordinary people, but can it do creative and insightful intellectual work? To do philosophy is to put it to the test. Therefore, starting from the philosophy of cognition, DeepSeek was tested around the philosophical issues of artificial cognition, artificial consciousness and the methodology of adaptive representation, and the results showed that DeepSeek is a very useful intelligent model(software) or intelligent tool, but it is not related to consciousness, emotion, or morality, the so-called “alignment” is just people’s wishful thinking, and it is the human designer who ultimately decides the level of LLM’s intelligence.更多还原