Three Fallacies in AI Consciousness Research A Structural Diagnostic

Abstract

Contemporary research on artificial intelligence and consciousness exhibits three structurally distinct fallacies whose combined operation produces a discipline that uses the vocabulary of consciousness while no longer engaging consciousness as an explanandum. Hard Conflation bundles the functional and phenomenal aspects of consciousness under a single term, generating apparent intractability from functional disagreement. Concept Hollowing preserves the term "consciousness" while replacing what made consciousness a meaningful concept with something methodologically tractable. The Stolen Concept, in the sense developed by Rand, uses claims about phenomenal consciousness while denying or undermining the foundations that gave phenomenal consciousness its referential content. Together these fallacies produce the pragmatic pivot characteristic of the contemporary field: the phenomenal question is declared intractable, research is redirected to the perception of AI consciousness, and the original explanandum is preserved nominally while being abandoned operationally. The paper develops each fallacy formally, traces its operation in Comșa (2026) as a worked example, and shows how the apparent intractability dissolves once consciousness is treated as axiomatic rather than as something to be derived from functional architecture. The Unified Axioconscious Field Theory developed in prior work by the present author provides one such structural framework. The paper closes by tracing the cultural stakes: as the vocabulary of consciousness is captured to fit AI systems, the original phenomenal-anchored sense drops out of the working vocabulary of the culture, and humans come to understand themselves through the hollowed concepts.