Recent discussions among AI researchers highlight a significant improvement in the reasoning capabilities of Large Language Models (LLMs) when prompted to re-read questions. This technique, which involves repeating the question input twice in the prompt, has been shown to boost LLM accuracy across diverse tasks and model types. Despite LLMs' limitations, such as difficulties with simple tasks like counting letters, this prompting strategy leverages their latent reasoning potential. Researchers emphasize that while LLMs do not invent new information but rather repackage existing data, effective prompting can enhance their performance significantly. A paper shows that adding 'Read the question again' often boosts LLM accuracy, a strategy already used by teachers.