I think this highlights a big limitation of LLMs. It doesn't "understand" the text. It's just piecing words together based on the probability. In this case, the

I think this highlights a big limitation of LLMs. It doesn't "understand" the text. It's just piecing words together based on the probability. In this case, the key word was "its", which it ignored - but makes a huge difference to the meaning of the sentence.