Large language models do a great job of emulating meaning. That's their biggest strength and greatest weakness. We got a finally good chatbot, at the cost of making people project so much more then a chatbot on it. LLMs can't mean anything. They can only produce surrogate meaning.
When you compress vast amounts of text into a model and query the model you can only get back information that is already in the training data set. It's inherently backwards looking, it cannot produce lowered entropy forward in time. You can't get more knowledge out of it, only less then you inputted in the first place.