The article explores the relationship between values and AI, distinguishing between two concepts of AI: computational artifacts and sociotechnical systems. It argues that values cannot be 'embedded' in computational artifacts because they are ontologically distinct from these artifacts. However, when AI is understood as a sociotechnical system, values and AI share the same ontological category—both are social. The relationship between values and AI is best understood as a dimension of the relationship between technology and society, which can be theorized in multiple ways. The literature on co-production shows that technology and society are co-productive, and values are generated through this process. The article critiques the idea of embedding values in AI, emphasizing that values are fluid and context-dependent, and their attribution to AI is influenced by social relations and meanings. Examples of AI systems, such as an ad distribution system, Tay the Twitter bot, and the Iron Dome, illustrate how values are attributed to AI based on its outcomes and the social context in which it operates. The article concludes that the relationship between values and AI is best understood through the concept of AI as a sociotechnical system, recognizing the interplay between computational artifacts and social relations.The article explores the relationship between values and AI, distinguishing between two concepts of AI: computational artifacts and sociotechnical systems. It argues that values cannot be 'embedded' in computational artifacts because they are ontologically distinct from these artifacts. However, when AI is understood as a sociotechnical system, values and AI share the same ontological category—both are social. The relationship between values and AI is best understood as a dimension of the relationship between technology and society, which can be theorized in multiple ways. The literature on co-production shows that technology and society are co-productive, and values are generated through this process. The article critiques the idea of embedding values in AI, emphasizing that values are fluid and context-dependent, and their attribution to AI is influenced by social relations and meanings. Examples of AI systems, such as an ad distribution system, Tay the Twitter bot, and the Iron Dome, illustrate how values are attributed to AI based on its outcomes and the social context in which it operates. The article concludes that the relationship between values and AI is best understood through the concept of AI as a sociotechnical system, recognizing the interplay between computational artifacts and social relations.