AI实验室正在大量招募哲学家，因为这门学科对智能、心智与道德的提问，正好对应大型语言模型与AI代理人的新风险。DeepMind与Anthropic都建立了内部哲学团队；WIRED估计DeepMind至少有10名、Anthropic有4名哲学家。这波招聘也反过来影响学界，许多大学已开设AI伦理课程或计算机科学与哲学的联合项目。

这些哲学家多半处理的是具体问题，而非抽象的意识神秘学：公平性、错误资讯、恶意滥用、失控代理人，以及价值对齐。Iason Gabriel在DeepMind的研究，从早年的算法偏差转向如何让技术「主动向善」；Julia Haas则在Nature发表框架，测试LLM是否具备道德能力。Anthropic的Amanda Askell更直接参与模型开发，为Claude撰写宪章，规定模型该如何行事、应遵守哪些价值。

学界对此看法分裂。支持者认为，在公司控制基础技术的情况下，房间里有哲学家总比没有好；批评者担心这会变成伦理洗白、为行销和 hype 服务。作者指出，这些公司终究对投资者与股东负责，不太可能因哲学意见而改变竞争路线。即便如此，哲学家仍在被聘用：2026年4月，DeepMind又增聘一名剑桥大学高级研究助理，研究机器意识与超智能，职称干脆就是「哲学家」。

AI labs are hiring philosophers because questions about intelligence, mind, and morality now map onto the risks created by large language models and AI agents. DeepMind and Anthropic have both built internal philosophy teams; WIRED counts at least 10 philosophers at DeepMind and 4 at Anthropic. The trend is also reshaping academia, where many universities now offer AI ethics courses or joint computer science-philosophy programs.

Most of this work is practical rather than mystical: fairness, misinformation, malicious misuse, errant agents, and value alignment. Iason Gabriel’s work at DeepMind has moved from earlier concerns about algorithmic bias toward how to make technology “actively good,” while Julia Haas has published a Nature framework for testing whether LLMs show moral competence. At Anthropic, Amanda Askell is directly involved in model development and wrote Claude’s constitution, setting out how the model should behave and what values it should uphold.

Academics are divided. Supporters say that if corporations control foundational technology, it is better to have a philosopher in the room; critics warn of ethics-washing and hype. The article argues that these companies remain answerable to investors and shareholders, so philosophy is unlikely to override competition. Even so, hiring continues: in April 2026, DeepMind added a senior research associate from Cambridge to work on machine consciousness and superintelligence, with the job title simply “philosopher.”