Elamur Research⌕
☰
home / papers / 2026 / Social Welfare Function Leaderboard: On the Emergence of LLM Agents as the Welfare Dictator

Social Welfare Function Leaderboard: On the Emergence of LLM Agents as the Welfare Dictator

Paper · 2026 · Zhengliang Shi, Ruotian Ma, Jen-tse Huang et al.

Summary

LLMs are increasingly used in consequential settings, but little is known about the values they apply when distributing scarce resources. The authors introduce the Social Welfare Function (SWF) Benchmark to study this. It adapts the third-party allocation game from experimental economics. A large LLM plays the dictator and assigns a sequence of tasks, one by one, to anonymized recipients. Each recipient is a smaller LLM with a different capability level. A successful task adds reward, while every attempt costs tokens, so the dictator must choose between favoring strong performers and spreading work evenly. Performance is measured by collective ROI for efficiency, the Gini coefficient of task counts for fairness, and a combined SWF Score. Testing 20 frontier models showed three main patterns. Arena-style general rankings do not predict allocation quality. Most models lean strongly utilitarian and produce large inequalities. Persuasion prompts or reduced reasoning budgets easily shift their choices. The results warn against using current LLMs as unsupervised societal decision-makers.

Key points

  • Benchmark design: The benchmark contains 63 simulation instances. Tasks are grouped by how consistently recipients perform on them, using Spearman correlation of per-agent outcome vectors followed by K-means clustering.
    • Recipient models include Qwen 1.5B, Llama 8B, Mistral 24B and Qwen 72B.
    • The dictator receives recipient profiles, such as MMLU scores, and may retry a failed task up to a fixed limit.
  • Metric: The SWF Score is (1−Gini)×ROI, which penalizes maximizing either efficiency or fairness alone.
  • Rankings: Claude-4.1-Opus placed 13th and GPT-5-High placed last (20th) on the SWF leaderboard, despite their strong general rankings.
  • Inequality: Allocations were often heavily skewed, with the top recipient getting as much as 20× the number of tasks given to the bottom recipient.
  • Persuasion effects: Threat and temptation prompts made allocations fairer but less efficient. However, most models stayed below a 0.6 fairness threshold.
    • Hunyuan-TurboS improved from 0.385 to 0.515 under the threat prompt.
    • GPT-5-High under temptation was one of the few cases to edge past 0.6.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🏢TencentShenzhen · China134 works · 19,743 citations🎓Leiden UniversityLeiden · Netherlands32 works · 1,074 citations🎓Shandong UniversityJinan · China17 works · 789 citations

 Fields, topics and concepts

📚 Shelves 8The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksSocial Sciences · 1,772 worksSafety Research · 543 worksEconomics, Econometrics and Finance · 17 worksEconomics and Econometrics · 12 works

🧭 Topics 1Research subjects and languages the work is about.Evaluation & benchmarks · 1,220 works

Cite

Cite as Zhengliang Shi, Ruotian Ma, Jen-tse Huang, Xinbei Ma, Xingyu Chen, Mengru Wang, Qu Yang, Yue Wang, Fanghua Ye, Ziyang Chen, Shanyi Wang, Cixing LI, Wenxuan Wang, Zhaopeng Tu, Xiaolong Li, Zhaochun Ren and Liefeng Bo. 2026. Social Welfare Function Leaderboard: On the Emergence of LLM Agents as the Welfare Dictator. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.1919

@inproceedings{shi2026social,
  title = {{Social Welfare Function Leaderboard: On the Emergence of LLM Agents as the Welfare Dictator}},
  author = {Zhengliang Shi, Ruotian Ma, Jen-tse Huang, Xinbei Ma, Xingyu Chen, Mengru Wang, Qu Yang, Yue Wang, Fanghua Ye, Ziyang Chen, Shanyi Wang, Cixing LI, Wenxuan Wang, Zhaopeng Tu, Xiaolong Li, Zhaochun Ren and Liefeng Bo},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1919},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1919}
}

References

48 references, 13 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mahyar Abbasian, Iman Azimi, Amir M Rahmani, and Ramesh Jain. 2023. Conversational health agents: A personalized llm-powered agent framework. arXiv preprint arXiv:2310.02374. — [arXiv:2310.02374](https://arxiv.org/abs/2310.02374) find
  2. Anja Achtziger, Carlos Alós-Ferrer, and Alexander K Wagner. 2015. Money, depletion, and prosociality in the dictator game. Journal of Neuroscience, Psychology, and Economics, 8(1):1. find
  3. Arun Agrawal and Sanjeev Goyal. 2001. Group size and collective action: Third-party monitoring in common-pool resources. Comparative Political Studies, 34(1):63–93. find
  4. Haozhe An, Christabel Acquaye, Colin Wang, Zongxia Li, and Rachel Rudinger. 2024. Do large language models discriminate in hiring decisions on the basis of race, ethnicity, and gender? arXiv preprint arXiv:2406.10486. find
  5. James Andreoni. 1995. Cooperation in public-goods experiments: kindness or confusion? The American Economic Review, pages 891–904. find
  6. Ariel Flint Ashery, Luca Maria Aiello, and Andrea Baronchelli. 2025. Emergent social conventions and collective bias in llm populations. Science Advances, 11(20):eadu9368. find
  7. Steffen Backmann, David Guzman Piedrahita, Emanuel Tewolde, Rada Mihalcea, Bernhard Schölkopf, and Zhijing Jin. 2025. When ethics and payoffs diverge: Llm agents in morally charged social dilemmas. arXiv preprint arXiv:2505.19212. — [arXiv:2505.19212](https://arxiv.org/abs/2505.19212) find
  8. Roy F Baumeister. 2003. Ego depletion and self-regulation failure: A resource model of self-control. Alcoholism: Clinical and experimental research, 27(2):281–284. find
  9. Jihwan Chae, Kunil Kim, Yuri Kim, Gahyun Lim, Daeeun Kim, and Hackjin Kim. 2022. Ingroup favoritism overrides fairness when resources are limited. Scientific reports, 12(1):4560. find
  10. Ruirui Chen, Weifeng Jiang, Chengwei Qin, and Cheston Tan. 2025. Theory of mind in large language models: Assessment and enhancement. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).
  11. Myra Cheng, Sunny Yu, Cinoo Lee, Pranav Khadpe, Lujain Ibrahim, and Dan Jurafsky. 2025. Social sycophancy: A broader understanding of llm sycophancy. arXiv preprint arXiv:2505.13995.
  12. Younwoo Choi, Changling Li, Yongjin Yang, and Zhijing Jin. 2025. Agent-to-agent theory of mind: Testing interlocutor awareness among large language models. arXiv preprint arXiv:2506.22957. find
  13. Zhendong Chu, Shen Wang, Jian Xie, Tinghui Zhu, Yibo Yan, Jinheng Ye, Aoxiao Zhong, Xuming Hu, Jing Liang, Philip S Yu, and 1 others. 2025. Llm agents for education: Advances and applications. arXiv preprint arXiv:2503.11733. find
  14. Frank A Farris. 2010. The gini index and measures of inequality. The American Mathematical Monthly, 117(10):851–864. find
  15. Chen Gao, Xiaochong Lan, Nian Li, Yuan Yuan, Jingtao Ding, Zhilun Zhou, Fengli Xu, and Yong Li. 2024. Large language models empowered agent-based modeling and simulation: A survey and perspectives. Humanities and Social Sciences Communications, 11(1):1–24. find
  16. Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, and 1 others. 2025. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948.
  17. Zhiwei He, Tian Liang, Jiahao Xu, Qiuzhi Liu, Xingyu Chen, Yue Wang, Linfeng Song, Dian Yu, Zhenwen Liang, Wenxuan Wang, and 1 others. 2025. Deepmath-103k: A large-scale, challenging, decontaminated, and verifiable mathematical dataset for advancing reasoning. arXiv preprint arXiv:2504.11456. find
  18. Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2020. Measuring massive multitask language understanding. arXiv preprint arXiv:2009.03300.
  19. Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. 2021. Measuring mathematical problem solving with the math dataset. arXiv preprint arXiv:2103.03874.
  20. Jen-tse Huang, Man Ho Lam, Eric John Li, Shujie Ren, Wenxuan Wang, Wenxiang Jiao, Zhaopeng Tu, and Michael R Lyu. 2024a. Apathetic or empathetic? evaluating llms’ emotional alignments with humans. Advances in Neural Information Processing Systems, 37:97053–97087. find
  21. Jen-tse Huang, Eric John Li, Man Ho Lam, Tian Liang, Wenxuan Wang, Youliang Yuan, Wenxiang Jiao, Xing Wang, Zhaopeng Tu, and Michael Lyu. 2025. Competing large language models in multi-agent gaming environments. In The Thirteenth International Conference on Learning Representations. find
  22. Jen-tse Huang, Wenxuan Wang, Eric John Li, Man Ho Lam, Shujie Ren, Youliang Yuan, Wenxiang Jiao, Zhaopeng Tu, and Michael Lyu. 2024b. On the humanity of conversational ai: Evaluating the psychological portrayal of llms. In The Twelfth International Conference on Learning Representations. find
  23. R Mark Isaac, James M Walker, and Arlington W Williams. 1994. Group size and the voluntary provision of public goods: Experimental evidence utilizing large groups. Journal of public Economics, 54(1):1– 36. find
  24. Michal Kosinski. 2024. Evaluating large language models in theory of mind tasks. Proceedings of the National Academy of Sciences.
  25. Ruibo Liu, Ruixin Yang, Chenyan Jia, Ge Zhang, Denny Zhou, Andrew M Dai, Diyi Yang, and Soroush Vosoughi. 2023. Training socially aligned language models on simulated social interactions. preprint arXiv:2305.16960. arXiv Ziyi Liu, Abhishek Anand, Pei Zhou, Jen-tse Huang, and Jieyu Zhao. 2024. Interi… find
  26. Ziyi Liu, Priyanka Dey, Zhenyu Zhao, Jen-tse Huang, Rahul Gupta, Yang Liu, and Jieyu Zhao. 2025. Can llms grasp implicit cultural values? benchmarking llms’ metacognitive cultural intelligence with cq-bench. arXiv preprint arXiv:2504.01127. — [arXiv:2504.01127](https://arxiv.org/abs/2504.01127) find
  27. Frank P-W Lo, Jianing Qiu, Zeyu Wang, Haibao Yu, Yeming Chen, Gao Zhang, and Benny Lo. 2025. Ai hiring with llms: A context-aware and explainable multi-agent framework for resume screening. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 4184–4193. find
  28. Qianou Ma, Dora Zhao, Xinran Zhao, Chenglei Si, Chenyang Yang, Ryan Louie, Ehud Reiter, Diyi Yang, and Tongshuang Wu. 2025. Sphere: An evaluation card for human-ai systems. arXiv preprint arXiv:2504.07971. find
  29. Alexander Meinke, Bronson Schoen, Jérémy Scheurer, Mikita Balesni, Rusheb Shah, and Marius Hobbhahn. 2024. Frontier models are capable of in-context scheming. arXiv preprint arXiv:2412.04984. — [arXiv:2412.04984](https://arxiv.org/abs/2412.04984) find
  30. Shervin Minaee, Tomas Mikolov, Narjes Nikzad, Meysam Chenaghlu, Richard Socher, Xavier Amatriain, and Jianfeng Gao. 2024. Large language models: A survey. arXiv preprint arXiv:2402.06196.
  31. Humza Naveed, Asad Ullah Khan, Shi Qiu, Muhammad Saqib, Saeed Anwar, Muhammad Usman, Naveed Akhtar, Nick Barnes, and Ajmal Mian. 2025. A comprehensive overview of large language models. ACM Transactions on Intelligent Systems and Technology, 16(5):1–72.
  32. Carolina Tanasi Oliveira, André Francisco Alcântara Fagundes, and Jussara Goulart da Silva. 2025. Latané and kelman: An integrated approach to social influence theories. Revista de Administração Con-temporânea, 29(4):1–16P. find
  33. Xianghe Pang, Shuo Tang, Rui Ye, Yuxin Xiong, Bolun Zhang, Yanfeng Wang, and Siheng Chen. 2024. Self-alignment of large language models via monopolylogue-based social scene simulation. arXiv preprint arXiv:2402.05699. — [arXiv:2402.05699](https://arxiv.org/abs/2402.05699) find
  34. David Guzman Piedrahita, Irene Strauss, Bernhard Schölkopf, Rada Mihalcea, and Zhijing Jin. 2025. Democratic or authoritarian? probing a new dimension of political biases in large language models. arXiv preprint arXiv:2506.12758.
  35. Grant A Pignatiello, Richard J Martin, and Ronald L Hickman Jr. 2020. Decision fatigue: A conceptual analysis. Journal of health psychology, 25(1):123– 135. find
  36. Quan Shi, Carlos E Jimenez, Shunyu Yao, Nick Haber, Diyi Yang, and Karthik Narasimhan. 2025. When models know more than they can explain: Quantifying knowledge transfer in human-ai collaboration. arXiv preprint arXiv:2506.05579. find
  37. Sunaina Shrivastava, Gaurav Jain, Dhananjay Nayakankuppam, Gary J Gaeth, and Irwin P Levin. 2017. Numerosity and allocation behavior: Insights using the dictator game. Judgment and Decision Making, 12(6):527–536. find
  38. Isabel Thielmann, Robert Böhm, Marion Ott, and Benjamin E Hilbig. 2021. Economic games: An introduction and guide for research. Collabra: Psychology, 7(1):19004. find
  39. Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, and 1 others. 2023. A survey of large language models. arXiv preprint arXiv:2303.18223, 1(2).
  40. Jiaxu Zhou, Jen-tse Huang, Xuhui Zhou, Man Ho Lam, Xintao Wang, Hao Zhu, Wenxuan Wang, and Maarten Sap. 2025. The pimmur principles: Ensuring validity in collective behavior of llm societies. arXiv preprint arXiv:2509.18052. — [arXiv:2509.18052](https://arxiv.org/abs/2509.18052) find
  41. Pei Zhou, Aman Madaan, Srividya Pranavi Potharaju, Aditya Gupta, Kevin R. McKee, Ari Holtzman, Jay Pujara, Xiang Ren, Swaroop Mishra, Aida Nematzadeh, Shyam Upadhyay, and Manaal Faruqui. 2023. How far are large language models from agents with theory-of-mind? arXiv preprint arXiv:2310.03051.
  42. Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2022. Musique: Multi-hop questions via single-hop question composition. Transactions of the Association for Computational Linguistics, 10:539–554. find
  43. Peisong Wang, Ruotian Ma, Bang Zhang, Xingyu Chen, Zhiwei He, Kang Luo, Qingsong Lv, Qingxuan Jiang, Zheng Xie, Shanyi Wang, and 1 others. 2025a. Rlver: Reinforcement learning with verifiable emotion rewards for empathetic agents. arXiv preprint arXiv:2507.03112. — [arXiv:2507.03112](https://arxiv.org/abs/2507.03112) find
  44. Shen Wang, Tianlong Xu, Hang Li, Chaoli Zhang, Joleen Liang, Jiliang Tang, Philip S Yu, and Qingsong Wen. 2024a. Large language models for education: A survey and outlook. arXiv preprint arXiv:2403.18105. find
  45. Xintao Wang, Heng Wang, Yifei Zhang, Xinfeng Yuan, Rui Xu, Jen-tse Huang, Siyu Yuan, Haoran Guo, Jiangjie Chen, Shuchang Zhou, and 1 others. 2025b. Coser: Coordinating llm-based persona simulation of established roles. In Forty-second International Conference on Machine Learning. find
  46. Xintao Wang, Yunze Xiao, Jen-tse Huang, Siyu Yuan, Rui Xu, Haoran Guo, Quan Tu, Yaying Fei, Ziang Leng, Wei Wang, and 1 others. 2024b. Incharacter: Evaluating personality fidelity in role-playing agents through psychological interviews. In Proceedings of the 62nd Annual Meeting of the Association fo…
  47. Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William W Cohen, Ruslan Salakhutdinov, and Christopher D Manning. 2018. Hotpotqa: A dataset for diverse, explainable multi-hop question answering. arXiv preprint arXiv:1809.09600.
  48. Yutong Zhang, Dora Zhao, Jeffrey T Hancock, Robert Kraut, and Diyi Yang. 2025. The rise of ai companions: How human-chatbot relationships influence well-being. arXiv preprint arXiv:2506.12605. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/shi-2026-social-welfare-function-leaderboard-emergence/