Text this: The influence of mental state attributions on trust in large language models