Exploring Consciousness in LLMs: A Systematic Survey of Theories, Implementations, and Frontier Risks

TL;DR

The paper investigates whether large language models can exhibit consciousness by proposing a clarifying taxonomy and surveying both theoretical and empirical work. It distinguishes consciousness from awareness using a C0-C1-C2 framework and surveys phenom­enal versus access theories, complemented by formal definitions to enable testable criteria. Empirically, it maps direct consciousness tests and capabilities like Theory of Mind, situational awareness, metacognition, sequential planning, and creativity, detailing benchmarks and alignment strategies. Frontier risks such as scheming, manipulation, autonomy, and collusion are analyzed with existing evaluations and mitigation approaches. The work emphasizes the need for unified evaluation frameworks, interpretability methods, embodied and multi-agent research, and outlines practical directions to guide safe, robust progress in conscious LLM research.

Abstract

Consciousness stands as one of the most profound and distinguishing features of the human mind, fundamentally shaping our understanding of existence and agency. As large language models (LLMs) develop at an unprecedented pace, questions concerning intelligence and consciousness have become increasingly significant. However, discourse on LLM consciousness remains largely unexplored territory. In this paper, we first clarify frequently conflated terminologies (e.g., LLM consciousness and LLM awareness). Then, we systematically organize and synthesize existing research on LLM consciousness from both theoretical and empirical perspectives. Furthermore, we highlight potential frontier risks that conscious LLMs might introduce. Finally, we discuss current challenges and outline future directions in this emerging field. The references discussed in this paper are organized at https://github.com/OpenCausaLab/Awesome-LLM-Consciousness.

Figures (1)

• Figure 1: Taxonomy of large language model consciousness.