Consciousness is a defining feature of the human mind, and as large language models (LLMs) advance, questions about their potential for consciousness become pressing. This paper clarifies commonly confused terms like LLM consciousness and awareness, then systematically reviews existing theoretical and empirical research on the topic. It also highlights potential frontier risks that conscious LLMs might pose, discusses current challenges, and outlines future directions for this emerging field.
Large language models show early signs of representing aspects of self-consciousness within their internal mechanisms, but these representations are difficult to alter through direct manipulation and can instead be strengthened by fine-tuning on core concepts. The study defines self-consciousness for language models using causal structural games, refines ten core concepts, and tests ten leading models across four stages: quantification, visualization, manipulation, and acquisition. Results suggest that while models have not achieved full self-consciousness, certain concepts are discernibly encoded, and targeted fine-tuning can enhance these representations.