Beyond Control: Will to Power in AI

Sensational claims portray AI as an entity beyond human control. We argue that while AI may exhibit characteristics that appear "beyond control," this does not necessarily arise from its status as an independent entity. The panic "out-of-control" AI narrative is getting in the way of reflecting on how we can meaningfully control AI. To shed light upon this, the first part of the paper introduces a theoretical framework based on two variables: controllable/non-controllable AI and value-free/value-laden AI. This framework yields four distinct perspectives on the ontology of AI, including Value-Free Instrumentalism, Determinism, Value-Laden Critical Theory, and Substantialism. Our argument contends that AI's perceived lack of control is not rooted in its value-free mind-independence, but in its value-laden sociocultural power dynamics. The second part expands on Critical Theory Critical to elucidate that AI remains subject to control, despite prevalent perceptions suggesting otherwise. While AI may present legitimate existential concerns, these apprehensions do not stem from an inherent state of "out-of-control." Rather, they arise due to the exertion of control by entities whose motives do not entirely align with the public interest.