Recently, LeCun questioned the reasoning capabilities of large language models on Twitter, noting that their emergent abilities are primarily the result of contextual learning. Research experiments have shown that large language models perform poorly in complex planning tasks, indicating limited reasoning abilities. The paper also points out issues in the study of large language models' planning tasks, including confusion between planning knowledge and the execution of plans. Improving the planning capabilities of large language models remains challenging. This discussion provides important insights into understanding the true capabilities of large language models in the fields of reasoning and planning.