From Linguistic Form to Computational Function: Rethinking AI Consciousness

Abstract

Rapid technological evolution has reignited a foundational debate that has simmered since the dawn of AI namely: can a machine, by virtue of its programming and architecture alone, possess genuine understanding, semantic content, or even consciousness? Or does it merely execute an elaborate simulation like a “stochastic parrot” that skillfully recombines statistical patterns without a shred of comprehension? This paper argues that the latter interpretation is correct, and that John Searle’s Chinese Room Argument (CRA) provides the essential philosophical lens for diagnosing the condition of modern LLMs. The study methodology integrates classical philosophy of mind with contemporary theoretical frameworks. Clear conceptual distinctions within consciousness studies are established to isolate the specific target of the distinction between phenomenal consciousness and intrinsic intentionality. Thereafter, a tripartite analytical framework with the Computational Theory of Mind (CTM), the Connectionist and Neural Network Theory, and the Integrated Information Theory (IIT) is constructed to explore the study from multiple convergent angles. Further, ChatGPT’s own introspective responses when prompted on topics of consciousness, meaning, and its own operations are presented and analysed. The outputs are sophisticated linguistic performances that, upon analysis, confessed the very absence they are asked to describe. The study reveals that the ChatGPT system is best characterized as an episodic, largely feed-forward transformer designed to exploit statistical regularities in language. It lacks the recurrent, integrative, and self-referential dynamics that contemporary theories of consciousness, especially IIT, associate with phenomenology.