Agustin V. Startari (Universidad de la República; Universidad de Palermo; Universidad de la Empresa) has posted The Grammar of Asymmetric Visibility: AI, Zionism, and the Reallocation of Political Agency on SSRN. Here is the abstract:
This article introduces asymmetric visibility as an AI-mediated condition in which subordinated political actors remain visible while their grammatical agency is weakened. Building on the concept of responsibility loss developed in the first article of the series, the paper argues that AI-generated and AI-mediated conflict discourse does not merely describe political violence, occupation, sanctions, resistance, or humanitarian suffering. It redistributes the grammatical conditions under which actors appear as agents, victims, risks, threats, humanitarian populations, or moderation objects.
The article focuses on Palestine, Zionism, anti-Zionism, Iran, United States foreign-policy discourse, Israel-linked institutional discourse, and platform moderation as test cases for analyzing agency asymmetry. Its central claim is not that AI systems intentionally favor one political position, nor that visibility itself disappears. Rather, the problem is formal: dominant-power actors may remain more frequently represented as institutional, defensive, diplomatic, or security-oriented subjects, while subordinated actors are more likely to appear through humanitarian, securitized, or moderation-oriented categories. A population may therefore be repeatedly mentioned, described, classified, and sympathized with while still being denied the grammatical status of a political subject.
Methodologically, the article proposes an Agency-Preservation Rate and an Asymmetric Visibility Index to measure the disparity between the agency preserved for dominant-power actors and the agency preserved for subordinated actors across AI-generated or AI-mediated conflict discourse. These indicators examine subject position, verbal agency, passive attribution, agent deletion, risk-object conversion, humanitarian framing, security framing, moderation framing, and political-subjecthood retention. The paper therefore shifts the evaluation of AI discourse from representation alone toward the measurable preservation or weakening of political agency.
The article concludes that visibility is not equality. In conflict discourse, the decisive question is not only whether AI systems mention Palestine, Iran, occupied populations, sanctioned societies, or anti-imperial speech, but how they grammatically position them. AI ethics and platform governance should therefore move beyond bias detection, hate speech classification, and misinformation control, and ask whether machine-generated discourse preserves the grammar through which political agency remains recognizable. Asymmetric visibility begins where recognition ends: when the dominated are seen, but not grammatically allowed to act.
To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum