Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2501.08579 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918224990306304 |
|---|---|
| author | Wang, Qian Wu, Jiaying Jiang, Zichen Tang, Zhenheng Luo, Bingqiao Chen, Nuo Chen, Wei He, Bingsheng |
| author_facet | Wang, Qian Wu, Jiaying Jiang, Zichen Tang, Zhenheng Luo, Bingqiao Chen, Nuo Chen, Wei He, Bingsheng |
| contents | Large Language Models (LLMs) are increasingly employed for simulating human behaviors across diverse domains. However, our position is that current LLM-based human simulations remain insufficiently reliable, as evidenced by significant discrepancies between their outcomes and authentic human actions. Our investigation begins with a systematic review of LLM-based human simulations in social, economic, policy, and psychological contexts, identifying their common frameworks, recent advances, and persistent limitations. This review reveals that such discrepancies primarily stem from inherent limitations of LLMs and flaws in simulation design, both of which are examined in detail. Building on these insights, we propose a systematic solution framework that emphasizes enriching data foundations, advancing LLM capabilities, and ensuring robust simulation design to enhance reliability. Finally, we introduce a structured algorithm that operationalizes the proposed framework, aiming to guide credible and human-aligned LLM-based simulations. To facilitate further research, we provide a curated list of related literature and resources at https://github.com/Persdre/awesome-llm-human-simulation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2501_08579 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | LLM-based Human Simulations Have Not Yet Been Reliable Wang, Qian Wu, Jiaying Jiang, Zichen Tang, Zhenheng Luo, Bingqiao Chen, Nuo Chen, Wei He, Bingsheng Computation and Language Large Language Models (LLMs) are increasingly employed for simulating human behaviors across diverse domains. However, our position is that current LLM-based human simulations remain insufficiently reliable, as evidenced by significant discrepancies between their outcomes and authentic human actions. Our investigation begins with a systematic review of LLM-based human simulations in social, economic, policy, and psychological contexts, identifying their common frameworks, recent advances, and persistent limitations. This review reveals that such discrepancies primarily stem from inherent limitations of LLMs and flaws in simulation design, both of which are examined in detail. Building on these insights, we propose a systematic solution framework that emphasizes enriching data foundations, advancing LLM capabilities, and ensuring robust simulation design to enhance reliability. Finally, we introduce a structured algorithm that operationalizes the proposed framework, aiming to guide credible and human-aligned LLM-based simulations. To facilitate further research, we provide a curated list of related literature and resources at https://github.com/Persdre/awesome-llm-human-simulation. |
| title | LLM-based Human Simulations Have Not Yet Been Reliable |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2501.08579 |