The Algorithmic Statistician: Can Large Language Models Transform Policy Analysis?
DOI:
https://doi.org/10.5281/zenodo.17307063Keywords:
Large language models; public policy analysis; agent-based modeling.Abstract
Large Language Models (LLMs) are emerging as transformative tools for governance, but their empirical value in public policy analysis remains under-examined. Through a systematic review of 141 publications (Google Scholar, arXiv), this paper analyzes the capacity of LLMs to simulate heterogeneous agents, extract legal and political text, and support decision-making. By synthesizing advances in multi-agent modeling, neurosymbolic extraction, and cooperative AI platforms, the article demonstrates that: a) the heterogeneity of LLMs approximates demographic diversity more faithfully than representative models; b) hybrid systems achieve state-of-the-art accuracy in document analysis; and c) open frameworks, while fostering transparency and participation, entail risks of bias and opacity. An agenda is proposed to improve multimodal inference and ethical governance. In conclusion, LLMs are powerful—not self-sufficient—complements to expert judgment, capable of improving policy design and evaluation if integrated into responsible sociotechnical ecosystems.
Downloads
References
Hao, Y., & Xie, D. (2025, February 24). A multi-LLM-agent-based framework for economic and public policy analysis (arXiv:2502.16879). arXiv. https://arxiv.org/abs/2502.16879
Pan, S., Nagar, S., & Parkes, D. C. (2025). Can a society of generative agents simulate human behavior and inform public health policy? A case study on vaccine hesitancy (Version 3; arXiv:2503.09639). arXiv. https://arxiv.org/abs/2503.09639
Salekpay, F., van den Bergh, J., & Savin, I. (2024). Comparing advice on climate policy between academic experts and ChatGPT. Ecological Economics, 226, Article 108352. https://doi.org/10.1016/j.ecolecon.2024.108352
Lewington, A., et al. (2024). Creating a cooperative AI policymaking platform through open source collaboration. arXiv. https://doi.org/10.48550/arXiv.2412.06936
Buster, G., Pinchuk, P., Barrons, J., McKeever, R., Levine, A., & Lopez, A. (2024). Supporting energy policy research with large language models. arXiv. https://doi.org/10.48550/arXiv.2403.12924
Alkhaza’leh, Y. M. S. (2025). Artificial intelligence and its impact on social policy-making. Journal of Ecohumanism, 4(1), 2320–2337. https://doi.org/10.62754/joe.v4i1.6053
Shapira, E., Madmon, O., Reinman, I., Amouyal, S. J., Reichart, R., & Tennenholtz, M. (2024). GLEE: A unified framework and benchmark for language-based economic environments (arXiv:2410.05254). arXiv. https://arxiv.org/abs/2410.05254
Vatamanu, A. F., & Tofan, M. (2025). Integrating artificial intelligence into public administration: Challenges and vulnerabilities. Administrative Sciences, 15(4), Article 149. https://doi.org/10.3390/admsci15040149
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Encuentros. Revista de Ciencias Humanas, Teoría Social y Pensamiento Crítico
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
