A new study reveals that large language models (LLMs) exhibit myopic planning, relying more on shallow search strategies rather than deep reasoning, which contrasts with human planning methods. This insight could guide improvements in aligning LLM decision-making with human expertise in strategic tasks.