You are best off thinking of LLMs as highly advanced auto correct. They don’t know what words mean. When they output a response to your question the only process that occurred was “which words are most likely to come next”.