avatar

Zengqing Wu

Researcher
The University of Osaka
wuzengqing (at) outlook.com


Biography

Zengqing Wu is a researcher at The University of Osaka under the supervision of Prof. Chuan Xiao. He is also a lecturer for the mathematical modeling course at the Computer Center of China Welfare Institute Children’s Palace. His main research focus is computational social science, especially the application of large language models in computer simulation.

Research Interests

  • Computational Social Science: Complex Systems, Agent-Based Modeling, Large Language Models for Social Simulation
  • Computational Linguistics: Emergent Language
  • Education: Mathematical Modeling, Teaching and Learning Assessment

News

  • [Jun. 2026] We propose a methodology based on emergent language for studying conscious AI. [code]
  • [Jun. 2026] Not All Flips Are Conformity: We find that answer flips in multi-agent LLM debate conflate self-reflection instability with peer influence; after decomposition, strict conformity is smaller, predominantly harmful, and partially predictable.
  • [May. 2025] We argue that LLM-based social simulations need clear boundaries to meaningfully contribute to social science research in our position paper. Our paper is to appear at ICML 2026 @ Seoul, South Korea.
  • [Mar. 2026] Served as a Junior Commentator at DEIM 2026 (AIの社会応用).
  • [Feb. 2026] Accepted by BOOST (Broadening Opportunities for Outstanding young researchers and doctoral students in STrategic areas), The University of Osaka.
  • [Oct. 2025] Gave an oral presentation entitled "Some Initial Attempts and Insights on LLM-Based Social Simulations" at 7th Joint Korea-Japan Workshop on Management of Data (KJMD 2025) @ Busan, South Korea.
  • [Oct. 2025] Presented a poster online at the 2025 Michigan AI Symposium.
  • [Aug. 2025] The Hidden Strength of Disagreement: We discovered that implicit consensus mechanisms can outperform explicit coordination in dynamic environments requiring long-horizon adaptability, where preserving agent diversity enhances exploration and robustness. Our paper is to appear at EMNLP 2025 (Main) @ Suzhou, China. [Code]
  • [Jun. 2025] We argue that LLM-based social simulations need clear boundaries to meaningfully contribute to social science research in our position paper.
  • [Mar. 2025] Reached the milestone of 100 citations :)
  • [Dec. 2024] Presented a poster at the 2024 Australasian Database Conference (ADC 2024) @ Tokyo Institute of Technology, Tokyo, Japan.
  • [Sep. 2024] Shall We Team Up: Following the SABM framework, we discovered the spontaneous cooperation of LLM agents in competing environments. Our paper is to appear at EMNLP 2024 (Findings) @ Miami, USA. [Code]
  • [Sep. 2024] LLMob: As a joint work with the University of Tokyo, we developed an LLM agent framework for the generation of personal activity trajectories. This work will appear at NeurIPS 2024. [Code]
  • [Jul. 2024] Presented a poster online at the ICML 2024 Workshop Agentic Markets (AMW @ ICML 2024).
  • [Jun. 2024] Presented a poster at the Eighth International Workshop on Symbolic-Neural Learning (SNL 2024) @ National Museum of Emerging Science and Innovation (Miraikan), Tokyo, Japan.
  • [Mar. 2024] Our work SABM：大規模言語モデルに基づくエージェントベース実世界シミュレーション received the NEC Corporation Award at DEIM 2024.
  • [Jan. 2024] Our paper, which investigates the application of Shannon entropy in assessing students' abstraction levels to optimize students' learning process, has been accepted in the IEEE Transactions on Education (ToE).
  • [Nov. 2023] SABM: We developed a computer simulation framework that incorporates LLMs into agent-based modeling. [Slides] [Code]
  • [Sep. 2023] Dissertation: Bridging the Gap: Utilizing Knowledge Graphs to Uncover Inconsistencies in Social Science Theories.

Highlights

  1. ICML
    LLM-Based Social Simulations Require a Boundary
    Zengqing Wu*, Run Peng, Takayuki Ito, Makoto Onizuka, Chuan Xiao*
    Forty-Third International Conference on Machine Learning (Position Paper Track).
    PDF
  2. EMNLP
    The Hidden Strength of Disagreement: Unraveling the Consensus-Diversity Tradeoff in Adaptive Multi-Agent Systems
    Zengqing Wu*, Takayuki Ito*
    The 2025 Conference on Empirical Methods in Natural Language Processing.
    PDF Code
  3. EMNLP
    Shall We Team Up: Exploring Spontaneous Cooperation of Competing LLM Agents
    Zengqing Wu, Run Peng, Shuyuan Zheng, Qianying Liu, Xu Han, Brian Inhyuk Kwon, Makoto Onizuka, Shaojie Tang, Chuan Xiao*
    The 2024 Conference on Empirical Methods in Natural Language Processing.
    PDF Code
  4. NeurIPS
    Large Language Models as Urban Residents: An LLM Agent Framework for Personal Mobility Generation
    Jiawei Wang, Renhe Jiang*, Chuang Yang, Zengqing Wu, Makoto Onizuka, Ryosuke Shibasaki, Chuan Xiao
    Thirty-Eighth Annual Conference on Neural Information Processing Systems.
    PDF Code
  5. arXiv
    Smart Agent-Based Modeling: On the Use of Large Language Models in Computer Simulations
    Zengqing Wu, Run Peng, Xu Han, Shuyuan Zheng, Yixin Zhang, Chuan Xiao*
    arXiv, 2023.
    PDF Code
  6. IEEE-TE
    Utilization of Information Entropy in Training and Evaluation of Students’ Abstraction Performance and Algorithm Efficiency in Programming
    Zengqing Wu*, Huizhong Liu, Chuan Xiao*
    IEEE Transactions on Education (IEEE-TE), 2024.
    PDF

Education

Awards

Personal Awards

  • 2026 BOOST (Broadening Opportunities for Outstanding young researchers and doctoral students in STrategic areas) / 国家戦略分野の若手研究者及び博士後期課程学生の育成事業（BOOST）次世代AI人材育成プログラム
  • 2024 NEC Corporation Award, DEIM 2024 / 第22回日本データベース学会年次大会スポンサー賞 日本電気株式会社賞
    Supervised by Prof. Chuan Xiao
  • 2022 Teraura Sayoko Scholarship / 寺浦さよ子記念奨学会奨学金
  • 2020 Meritorious in Mathematical Contest in Modeling (MCM)
    Supervised by Prof. Koichi Fukase
  • 2018 Top 10 Outstanding Youth in Shanghai Changning District / 上海市长宁区第四届"长宁好青年"
  • 2018 Finalist in High School Mathematical Contest in Modeling (HiMCM)
  • 2015 First Prize of National Olympiad in Informatics in Provinces, Shanghai, China (NOIP)
  • 2013 First Prize of Japan Sansu Olympic / 日本算数オリンピック

Supervised Student Awards

  • 2021 Outstanding in the International Mathematical Modeling Challenge (IMMC) with co-supervisor Mr. Linfeng Dong
  • 2026 Three teams won Finalist in the International Mathematical Modeling Challenge (IMMC) with co-supervisor Mr. Linfeng Dong
  • 2025 Finalist in the International Mathematical Modeling Challenge (IMMC) with co-supervisor Mr. Linfeng Dong
  • 2026 Two teams won Meritorious in the International Mathematical Modeling Challenge (IMMC) with co-supervisor Mr. Linfeng Dong
  • 2022 Meritorious in High School Mathematical Contest in Modeling (HiMCM)
  • 2024 Meritorious in the International Mathematical Modeling Challenge (IMMC) with co-supervisor Mr. Linfeng Dong

Experience & Services

Professional Activities

  • DEIM 2026 Junior Commentator / ジュニアコメンテータ

International Conference

  • Reviewer (Conference): NeurIPS 2024, ICLR 2025, AISTATS 2025, ICML 2025, ACL ARR 2025, NeurIPS 2025 DB Track, ADMA 2025, ICLR 2026, ACL ARR 2026
  • Reviewer (Workshop): ICLR 2024 Workshop LLMAgents, ICML 2025 Workshop CFAgentic
  • External Reviewer: CIKM 2022, DASFAA 2023, KDD 2023, ADMA 2023, APWeb-WAIM 2024, ACL ARR 2024, ADMA 2024

Journal

  • Reviewer: IEEE Access, European Journal of Engineering Education

Professional Memberships

  • The Database Society of Japan (DBSJ)
  • The Association for Computational Linguistics (ACL)

Funding

  • 2026 大阪大学 次世代AI人材育成事業（BOOST）, Osaka University
  • 2020 大阪大学 令和2年度学部学生による自主研究奨励事業, Osaka University