What if the LLM made you solve a really complex math equation before it gave you the results? Would that make it ok?