Axiomatics by Prompts with Prompted Prompts: Introducing Reality Pictures to Replace Language-Modelled with Existence-Guided Narrative Patterns
Year: 2026
Type: preprint
Abstract: This paper continues an earlier experiment in prompt-based political reasoning by askingwhether narrative-driven discussion can be partially replaced by an axiomatic sequence ofprompts. The point of departure is a deliberately strong initial prompt concerning Russia’s forcedfederalization and demilitarization, treated not as a conclusion to be debated but as a workingaxiom from which further prompts can be generated.The paper argues that conventional political discussion, whether human or AI-assisted, tends tobecome trapped in normative reformulations, objections, and rhetorical loops that obscure ratherthan clarify reality. In contrast, the proposed method seeks prompts that are increasinglydescriptive rather than normative, so that they function as compact pictures of reality rather thanas wish-expressive political narratives.Using the war in Ukraine and the contemporary Russian imperial system as a test case, the papershows how an initial prompt can generate a sequence of further prompts describing militarydeadlock, imperial self-destruction, social degradation, and the internal human cost of theRussian system. Particular emphasis is placed on the claim that the first victims of an imperialorder may be the people living within it, and that this perspective remains largely invisible instandard Western political discourse.The paper does not present a finished formal theory. Instead, it offers a first methodological steptoward a prompt-based axiomatics in which political reasoning proceeds by making assumptionsexplicit, refining them through further prompts, and replacing language-modelled narrativepatterns with existence-guided pictures of reality.
Author: Zeps, Dainis
Institution: University of Latvia
Language: English
Subfield: Linguistics and Language
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7167596070