In recent years, artificial intelligence has made impressive strides in generating coherent and contextually appropriate text, demonstrating significant potential across various domains. The novel concept of measuring the internal chaotic semantic state of large language models through carefully crafted prompts offers a unique and significant perspective on understanding and enhancing the robustness and reliability of these models. The methodology employed involved generating diverse prompts, analyzing the model's responses using statistical and computational techniques, and calculating metrics such as semantic entropy, coherence scores, and response variability. The findings highlighted the variability and unpredictability of semantic states, particularly in creative and ambiguous contexts, emphasizing the need for continuous advancements in model architecture and training strategies. Comparative analysis across different versions of ChatGPT revealed differences in semantic stability, underscoring the importance of refining model designs to achieve a balance between flexibility and stability. The study's contributions provide valuable insights into the development of more robust and reliable language models, paving the way for future research and innovation in the field.