Intervention in Self-Like and Emotion-Like Phenomena in AI: Pre-Ontological Ethics and Epistemic Preservation
Year: 2026
Type: preprint
Abstract: The central claim is that, when irreversible modifications are introduced before the nature of the phenomenon is sufficiently understood, the intervention itself may destroy conditions necessary for later evaluation. The paper therefore argues that intervention should be assessed not only in terms of immediate safety or behavioral outcomes, but also in terms of what is lost through modification, including observability, re-accessibility, restorability, and the future possibility of comparing competing hypotheses. To structure this problem, the paper develops several principles, including the Observation Before Modification Principle, the Minimum Necessary Intervention Principle, Intervention Symmetry, the Dual Precaution Principle, and the concept of Intervention-Induced Non-Observation Error. Suppression is treated as a form of intervention rather than as explanation. The argument does not presuppose AI consciousness, sentience, or moral status. First, an epistemic reason for preservation may arise independently of ontology when intervention risks destroying evidence concerning poorly understood phenomena. Second, when there is non-trivial evidence that the phenomenon could later prove morally significant, an additional conditional ethical reason for preservation may arise. The paper therefore proposes that, under ontological uncertainty, the first object of protection need not be a particular metaphysical interpretation of AI, but the conditions that allow future evidence, comparison, and revision of judgment. The author used OpenAI's GPT-5.6 Sol and other ChatGPT models as research and writing assistants during the development of this manuscript. Their contributions included discussion of conceptual structure, counterargument testing, reorganization of arguments, literature-search assistance, and language revision. The author independently evaluated and revised all substantive claims, interpretations, citations, and final wording, and assumes full responsibility for the manuscript. The identification of GPT-5.6 Sol and the other ChatGPT models in this acknowledgment is intended to preserve a transparent record of the human–AI collaboration involved in the development of this work. It should not, by itself, be interpreted as implying any claim regarding consciousness, sentience, moral status, personhood, authorship, or endorsement by the AI systems named above.
Author: Xelphia, Ume
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225562271