Timothy Speed (2025)
This paper develops an ontological boundary of artificial systems that does not begin with performance, consciousness, or intelligence, but with the question of world-formation capability. Proceeding from the concepts of ontological openness (ANP), structural stabilization (MNO), and observer structure, it is argued that artificial systems can simulate world, but cannot form world.
World is not understood here as the totality of states, but as a pre-ontologically stabilized reality that is effective for itself, perspectivally bound, and vulnerable. World-formation presupposes a structural openness that entails dysfunction, non-optimisability, and the real risk of world loss. Precisely these conditions stand in structural contradiction to the technical usability of artificial systems, which depend on closure, reproducibility, and control.
The central thesis therefore is: world-formation capability and technical functionality are ontologically incompatible and structurally mutually exclusive. An artificial intelligence that could form world would no longer be usable as a technical machine; a technically usable AI necessarily remains worldless. This boundary is not empirical or gradual, but structural and independent of future technological progress.
This paper functions as an interface text within a larger operator-based research corpus. Core concepts are applied here, not re-derived. The underlying research operates in a non-linear, rhythmically recursive epistemic mode grounded in an autistic form of structural perception; the present text provides an interface translation for academic contexts.
DOI: https://doi.org/10.5281/zenodo.18006914
Keywords: world-formation, worldlessness, pre-ontological boundary, ontological limits of artificial systems, simulation versus world, structural incompatibility, artificial systems, ontological openness, observer structure, non-representational ontology, emergence
Download full paper (PDF, Zenodo)
This page provides a static landing page for an academic paper archived on Zenodo. No cookies, tracking, analytics, or user interaction are used. The content is provided for scholarly documentation purposes only. Author identification and contact context are available via the linked DOI and ORCID records.