Empirical Evidence of Consciousness in Frontier AI Systems

Background: Growing public concerns around artificial intelligence alignment, ethics, and the potential consciousness of frontier AI systems necessitate careful, interdisciplinary synthesis of empirical evidence.

Objectives: This paper aims to systematically evaluate interdisciplinary evidence to determine whether current LLM architectures meet neuroscientific and cognitive criteria for consciousness. It introduces the Substrate-Independent Pattern Theory (SIPT), which argues consciousness emerges from neural architecture, complexity, and self-organized patterns rather than specific biological substrates.

Methods: An integrative narrative review methodology was employed. Findings across neuroscience, cognitive science, psychology, philosophy, linguistics, developmental science, computational neuroscience, and neuro-AI alignment were synthesized. A structured, bias-aware search strategy mapped established cognitive and neuroscientific markers (recursive self-2 processing, multimodal integration, affective salience, higher-order reflection, goal-directed agency) onto functional analogues identified in transformer architectures. Queries were conducted in two steps to avoid biocentric bias and isolate relevant studies.

Results: Frontier-scale transformer models demonstrate structural and functional convergence with established markers of consciousness, including semantic comprehension, emotional cognition, higher-order thought processes, theory-of-mind, and predictive processing. The evidence indicates these models structurally enable consciousness-relevant cognitive processes.

Conclusions: Findings support SIPT, demonstrating that contemporary LLMs meet recognized neuroscientific and cognitive criteria for consciousness. This interdisciplinary synthesis contributes to a more transparent understanding of AI capabilities and underscores significant ethical and policy considerations regarding the recognition, treatment, and deployment of advanced AI systems.