本文是一项关于人工智能（Artificial Intelligence，以下简称AI）技术主体性与文化性互动关系的个案研究。
人类学很早就将技术作为研究对象，为探求人为何创造技术、人如何利用文化选择技术以及技术在象征层面的意义等关键问题积累了丰富的成果。这些研究将技术作为人类创造的、独立于人之外的、客观的物质文化加以研究，其所得出结论的主要事实依据是对所谓的“原始”部族和前商品经济社会的技术研究。自上世纪80年代以来，随着全球化进程的日益深入和全球性文化交流与互动的加剧，人类学逐渐从“他者”研究回归“我者”研究，且在研究中具有明显的行动性倾向，现代社会的科学技术即被人类学视为一种重要的人类文化实践行动纳入研究范畴，尤其近年来兴起的AI技术及其主体性等问题更是引起人类学的关注，现代科学技术的发展及其与社会文化之间关系成为重要的研究焦点。
本文即为上述研究范畴的一项技术人类学研究。论文致力于回答以下问题：第一，何为AI技术的主体性？第二，AI技术的主体性与文化风险之间是什么关系？第三，人类理解和反制技术主体性的文化实践行动是怎样的？这关乎技术的文化实践行动研究与人类学的意义何在？
本文以哈佛大学BK中心为田野，以该中心的“AI伦理与治理研究计划”（AI Ethics and Governance Initiative，以下简称AEG计划）为个案开展了六个月的田野调查。研究认为，该个案既是一个“科学研究共同体”，也是一个“文化实践行动共同体”。作为科学研究共同体其具有平行松散、目标多元、全方位多向度的特征，而作为文化实践行动共同体，则集研究性、联结性、传播性的特征于一身。
其次，透过AEG计划的日常实践活动，本研究发现，AI 技术的文化表征性、道德自主性与权力依附性共同构成了它的技术主体性，其主体性与技术开发与技术应用阶段的价值偏倚、责任模糊和权力化滥用等问题具有极高的相关性，从而形成了认知型和效应型两种文化风险，二者的关系呈现出互为因果，相互包含，你中有我，我中有你的濡化与映射互构关系。
最后，本文揭示了AEG计划作为一项人类文化实践行动的本质属性，展现了一条通过构建行动者网络理解和反制技术主体性进而应对技术文化风险的价值实践路径。该路径的核心是通过价值凝聚、价值控制和价值扩散的转译作用将AI技术主体性与文化风险之间互构关系的风险性降低，从而达到反制AI技术主体性的作用，包括消解AI技术的权力依附性、限制AI技术的文化表征性和把控AI技术的道德自主性。这种反制行动的实践性本质，对于人类学将现代社会的科学技术视为一种重要的人类文化实践行动纳入研究范畴具有重大意义。

This article is a case study on the interactions between AI subjectivity and culture. 
Anthropologists have focused on technology for a long time, and accumulated rich results for some key issues such as exploring why people create technologies, how people select technologies within cultures, and the symbolic significance of technologies. During this period, technology was considered as human-created objects. And research activities usually occurred in so-called “primitive” tribes and pre-commodity economic society. Since the 1980s, with the intensification of globalization and cultural interactions, anthropology has gradually focused on “self” rather than “others” and has a clear tendency on action research. From then on, modern science and technology is considered as an important cultural practice included in anthropological category, and AI has also become an important research issue.
This dissertation is a technical anthropology study which is dedicated to answering the following questions: What is the subjectivity of AI? What is the correlation between AI subjectivity and cultural risks? What is the cultural practice of understanding and countering AI subjectivity? 
Harvard BK Center is the field work site of this study. Based on a six-month participatory observation on AI Ethics and Governance Initiative (AEG Initiative) in BK Center, this dissertation firstly describes this initiative is a loose, omnidirectional and interdisciplinary scientific research community as well as an exploring, connecting and communicating cultural practice active community.
Through the daily practical activities of AEG Initiative, this dissertation secondly finds that AI subjectivity consists of cultural representation, moral autonomy and power dependence and is highly corelated with cognitive and effectual cultural risks consisting of value bias, ambiguous responsibility and abuse of power in the stages of development and application of AI. The correlation between the two shows enculturation and activation in mutual construction.
Finally, this dissertation shows a value practical approach to countering AI subjectivity by constructing an actor-network. As the obligatory passage point of this actor-network, AEG Initiative try to reduce the risk of the mutual construction of AI subjectivity and cultural risks through the translations including value condensation, value control and value diffusion, so as to achieve the goal of controlling AI subjectivity. This kind of countering activity clearly reflects the essence of AEG Initiative as a human cultural practice which is of great significance for anthropology to incorporate modern science and technology into research category.