Semantic Fidelity Project: Meaning Preservation Across AI and Representational Systems

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This collection contains the core papers of the Semantic Fidelity Project, part of the broader Reality Drift framework. The project examines what happens to meaning as information is retrieved, compressed, summarized, generated, mediated, and recursively reused across artificial intelligence and other representational systems. Semantic Fidelity is defined as the degree to which the relationships that make a representation meaningful survive transformation, including context, hierarchy, causality, uncertainty, intent, tone, and reference. The series distinguishes semantic fidelity from factual accuracy, source support, coherence, and textual similarity. It examines how meaning can degrade before visible factual failure occurs, how recursive compression and generative reconstruction can conceal structural loss, how weakened corrective constraint allows fidelity decay to persist, and how language itself functions as a compressed trace of cognition within human-AI systems. The collection concludes with a working lexicon for diagnosing representational integrity, semantic drift, recursive compression, constraint collapse, verification failure, provenance failure, optimization dynamics, and related mechanisms within the Reality Drift framework. The six papers are intended as a public reference collection for researchers, builders, evaluators, educators, and others working on meaning preservation, AI-mediated reasoning, and representational systems.