Towards Trustable Language Models: Investigating Information Quality of Large Language Models