Comparing Humans and GPT-4 on Abstraction Tasks - arxiv.org

Clear