The discussion centers on the challenges of using Large Language Models (LLMs) for engineering design, specifically in fields such as rocket engineering. The commentators express skepticism about the ability of current LLMs to handle complex engineering tasks, particularly those requiring the understanding of diagrams and drawings. They argue that while software engineering benefits from the tokenization compatible with LLMs, traditional engineering fields presented significant hurdles due to their reliance on visual languages and precise logical structures. One commentator notes that current models, like O3, struggle even with basic tasks such as interpreting electronic schematics, pointing out that the training data available for LLMs in engineering is limited compared to what exists for software development.