Exploring the nature of reasoning in AI models, questioning if making LLMs express their thoughts out loud limits their potential.