I would be surprised if adversarial examples for LLMs were also adversarial examples for humans
a Schelling point for those who seek one
