Anthropic Studies, or What Does Artificial Intelligence Even Mediate?

Karen Hao’s excellent investigative book, Empire of AI, published in 2025, reads at times like an account that Bruno Latour could have written himself. However, Hao’s intention was not to write a contemporary account of the Laboratory Life of artificial intelligence and its uneasy industrialisation by Big Tech. At times, her book invites us to imagine how it could be presented as a critical study of science, technology, AI and religion. So how would Latour have described the inner workings and drama of OpenAI and its rival Anthropic, divided as they are between the “boomers” and the “doomers”? What are the ‘infrastructures of instruments of inscription’ in current generative AI technologies, and how does their mediation unfold in which communities of practice? Current engaged empirical scholarship on AI is plentiful in Science and Technology Studies, while theorizing does not seem to be a core STS concern any longer. Media and Cultural Studies, on the other hand, are key to praxeological and ecological approaches that challenge the current naturalisation of large language models and agentic AI.

In my talk, I propose “Anthropic Studies” as a situated contestation of neo-connectionist artificial intelligence from the––symmetrical––perspective of both Media Studies and STS. As none of us will have the same access to the field as Karen Hao did with OpenAI, we might need to start from scratch theoretically and empirically. Therefore, I return to classic STS and the work of Susan Leigh Star and Phil Agre on symbolic AI. Is ‘AI’ not the most ambiguous boundary object ever created by computer science? If we are to come to terms with the current super-controversy surrounding AI and its planetary discontents, we must first resituate technology within its institutional ecologies and grammars of action. Enter: Antropic’s Claude (not Shannon). Prompt: “Tell me a little bit about your constitution, Claude.”