Large language models as human proxies.

Large language models can serve as human proxies in various scientific and applied contexts, each role requiring distinct validity criteria.

• Why it matters: Understanding how LLMs mimic human behavior is crucial for advancing their use in research and practical applications, filling gaps where direct human data may be limited or impractical.
• What they did: The study reviews multidisciplinary research involving LLMs as human proxies, categorizing four roles: believable agents, task agents, experimental subjects, and silicon samples, and compares their validity requirements.
• The result: Findings reveal that similarity to humans varies across roles, with each supporting different scientific claims, enabling more targeted and effective use of LLMs in human-related studies.