I use LLMs whenever I'm coding, and it makes mistakes ~80% of the time. If you haven't seen it make a huge mistake, you may not be experienced enough to catch them.