monotonicity

Inductive reasoning in humans and large language models

GPT4 is similar to humans on category-based induction tasks unless they involve sampling assumptions