The AI Command Pyramid: How Linguistic Control over AI Produces a New Social Hierarchy

This working paper proposes the concept of the AI Command Pyramid to explain an emerging form of inequality in human–AI interaction. Rather than framing AI inequality in terms of access, skills, or displacement, the paper argues that the primary axis of stratification lies in differential capacity to command, stabilize, and correct AI systems through language. Drawing on interaction-level analysis and a state-based framework, the paper conceptualizes linguistic control as a form of power that redistributes uncertainty downward while concentrating epistemic authority upward. The paper is theoretical and analytical in scope, focusing on interactional dynamics rather than system architecture, policy compliance, or performance benchmarking. Supplementary Note This record is supplemented by a conceptual note outlining the broadertheoretical framing of the work, including the notion of “AI command literacy”and its implications for human–AI supervision. The supplementary note is descriptive and exploratory in nature, intendedto provide contextual background rather than additional empirical claims.