That’s certainly one possibility. But another possibility is that the people praise LLMs are not very good at judging whether the code it generates is of good quality or not…