AI Agents, Data Power, and the Ideological Horizon of AGI

Abstract

The global technology industry has converged on a narrative framing AI agents—systems capable of autonomous, multi-step task execution—as progress toward artificial general intelligence (AGI). This article argues that this narrative primarily serves an ideological function. First, the inference from agent capabilities to general intelligence is philosophically unwarranted: autonomy, persistence, and planning do not address the understanding deficit limiting current large language models. Second, agent deployment enables consolidation of data power through data monopolization, infrastructural control, and labor restructuring consistent with what scholars term technofeudalism. Third, the AGI narrative obscures this transformation by naturalizing corporate dominance as inevitable and providing moral cover that weakens democratic oversight. Recognizing this ideological function—and what is genuinely new in agents, viz., that they intermediate action rather than only information—reframes agent deployment as a political question about control and delegation rather than a technical trajectory toward inevitable intelligence.