Towards Intrinsic Interpretability of Large Language Models:A Survey of Design Principles and Architectures | ScienceToStartup