We show that LLMs like GPT can mistrust beneficial algorithmic advice through 2 experiments with 5 real-world datasets.

Work in progress!