Let me clarify a logic that many people might not understand. Current AI does not generate content through complete comprehension, but rather through data model matching. Most of the time, the results produced by comprehension-based content generation and data model matching-based content generation are similar. However, non-comprehension-based content generation means that there are unreliable situations, and the reasoning produced by it is also unreliable. This is why, despite AI having a vast knowledge base and being able to reason, it still cannot be relied upon to discover many of the knowledge gaps in human understanding. Only when AI achieves comprehension-based content generation can it ensure the accuracy of answers in complex and ever-changing environments, and the reasoning produced will also be accurate.