About
My name is Xingcheng Xu, and I am a research scientist at Shanghai AI Laboratory. Previously, I worked as a data scientist at Ant Group and Alibaba Group. I earned my Ph.D. from the School of Mathematical Sciences at Peking University, where I was also a joint Ph.D. student at the Mathematical Institute, University of Oxford.
My research interests are focused on the intersection of probability theory, artificial intelligence and their applications. I am dedicated to exploring this interdisciplinary space and uncovering novel insights and practical applications.
Education
- School of Mathematical Science, Peking University (Beijing, China)
Ph.D. in Mathematics, 09/2014 – 07/2019. Advisor: Prof. Yong Liu - Mathematical Institute, University of Oxford (Oxford, UK)
Joint Ph.D. in Mathematics, 09/2017 – 10/2018. Advisor: Prof. Zhongmin Qian - BICMR, Peking University (Beijing, China)
(Beijing International Center for Mathematical Research)
Joint B.S. in Mathematics / Enhanced Program for Graduate Study, 02/2014 – 06/2014. - School of Mathematical Science, Ocean University of China (Qingdao, Shandong, China)
B.S. in Mathematics, 09/2010 – 07/2014.
Selected Publications
* first/cofirst author, 📧 corresponding author
Artificial Intelligence
- Xingcheng Xu*, Jingjing Qu, Qiaosheng Zhang, Chaochao Lu, Yanqing Yang, Na Zou, Xia Hu. "Epistemic Traps: Rational Misalignment Driven by Model Misspecification." 2026.
- Zhao Tong, Chunlin Gong, Yiping Zhang, Haichao Shi, Qiang Liu, Xingcheng Xu📧, Shu Wu, Xiao-Yu Zhang. "CoT is Not the Chain of Truth: An Empirical Internal Analysis of Reasoning LLMs for Fake News Generation." arXiv:2602.04856, 2026.
- Zeming Wei, Qiaosheng Zhang, Xia Hu, Xingcheng Xu📧. "RAPO: Risk-Aware Preference Optimization for Generalizable Safe Reasoning." arXiv:2602.04224, 2026.
- Xiaoyu Wen, Zhida He, Han Qi, Ziyu Wan, Zhongtian Ma, Ying Wen, Tianhang Zheng, Xingcheng Xu, Chaochao Lu, Qiaosheng Zhang. "MAGIC: A Co-Evolving Attacker-Defender Adversarial Game for Robust LLM Safety." arXiv:2602.01539, 2026.
- Qitan Lv, Tianyu Liu, Qiaosheng Zhang, Xingcheng Xu📧, Chaochao Lu. "KALE: Enhancing Knowledge Manipulation in Large Language Models via Knowledge-aware Learning." arXiv:2601.07430, 2026.
- Zibo Zhao, Yuanting Zha, Haipeng Zhang, and Xingcheng Xu📧. "The Two-Stage Decision-Sampling Hypothesis: Understanding the Emergence of Self-Reflection in RL-Trained LLMs." arXiv:2601.01580, 2026.
- Xingcheng Xu. "The Policy Cliff: A Theoretical Analysis of Reward-Policy Maps in Large Language Models." arXiv:2507.20150, 2025.
- Shanghai AI Lab (2025). "SafeWork-R1: Coevolving Safety and Intelligence under the AI-45 Law." arXiv:2507.18576, 2025.
- Xingcheng Xu*, Zibo Zhao, Haipeng Zhang, and Yanqing Yang. "Principled Understanding of Generalization for Generative Transformer Models in Arithmetic Reasoning Tasks." ACL 2025, (Main, Volume 1: Long Papers), 4721-4747. (CCF A)
- Xingcheng Xu*, Zihao Pan, Haipeng Zhang, and Yanqing Yang. "It Ain't That Bad: Understanding the Mysterious Performance Drop in OOD Generalization for Generative Transformer Models." IJCAI 2024, 6578-6586. (CCF A)
AI × Social Science
- Haixing Gong, Hui Zou, Xingzhou Liang, Shiyuan Meng, Pinlong Cai, Xingcheng Xu, Jingjing Qu. "A Global Dataset Mapping the AI Innovation from Academic Research to Industrial Patents." Scientific Data, 12, 1261, 2025.
- Qin Chen, Jinfeng Ge, Huaqing Xie, Xingcheng Xu*📧, Yanqing Yang. "Large Language Models at Work in China's Labor Market." China Economic Review (CER), Volume 92, 102413, 2025.
- Zi Wang, Xingcheng Xu*, Yanqing Yang, Xiaodong Zhu. "Optimal Trade and Industrial Policies in the Global Economy: A Deep Learning Framework." arXiv:2407.17731, 2024.
- Yanqing Yang, Xingcheng Xu, Jinfeng Ge, Yan Xu. "Machine Learning for Economic Forecasting: An Application to China's GDP Growth." arXiv:2407.03595, 2024.
Probability Theory
- Zhongmin Qian, and Xingcheng Xu*📧. "Rough Path Renormalization from Stratonovich to Itô for Fractional Brownian Motion." AMSE, Vol. 41, No. 9, 2195-2230, 2025.
- Thomas Cass, Terry Lyons, and Xingcheng Xu*. "Weighted Signature Kernels." AAP, Vol. 34, No. 1A, (2024), 585-626.
- Zhongmin Qian, and Xingcheng Xu*📧. "Lévy Area Analysis and Parameter Estimation for fOU Processes via Non-Geometric Rough Path Theory." ActaMS, Vol. 44, (2024), 1609-1638.
- Zhongmin Qian, and Xingcheng Xu*📧. "Probability Bounds for Reflecting Diffusion Processes." SPL, Vol 199, (2023), 109855.
- Zengjing Chen, Shuhui Liu, Zhongmin Qian, and Xingcheng Xu*. "Explicit solutions for a class of nonlinear BSDEs and their nodal sets." PUQR, Vol. 7, No. 4, (2022), 283-300.
- Renjie Feng, Xingcheng Xu📧, and Robert J. Adler. "Critical radius and supremum of random spherical harmonics (II)." ECP, 23 (2018): 1-11.
Reports
- Xiaohui Cui, Di Guo, Zhao Jin, Chunjie Lin, Rong Xiao, Xingcheng Xu, Chenggang Xu, Yanqing Yang. "China Artificial Intelligence Index Report 2022", (2022).
崔晓晖、郭迪、金钊、林纯洁、肖蓉、徐兴成、许成钢、杨燕青：中国人工智能指数报告2022。(中文版：~200页) - Jian An, Daixi Chen, Long Chen, Yuanfang Li, Wei Liu, Jiamin Meng, Michael Spence, Qidi Wang, Xingcheng Xu, and Kaiwen Zheng. "Measuring and Tracking the Global Pandemic Economy." Luohan Academy Report (2020).
Academic Activities
- Conference Participation: 2024/8/3-9, IJCAI 2024 (Main Track), Jeju, South Korea.
Talk: "It Ain't That Bad: Understanding the Mysterious Performance Drop in OOD Generalization for Generative Transformer Models". - Invited Talk: 2023/12/7, Seminar on rough paths, SPDEs and related topics, TU Berlin.
Talk: "Weighted Signature Kernels and Applications" (online). - Conference Tutorial: 2023/12/4, WINE 2023, ShanghaiTech University, China.
Tutorial: "The Macroeconomics of Foundation Model / AI Economics", joint with Yanqing Yang, Jinfeng Ge. - Conference Participation: 2023/10/20-22, AI and Economics Workshop, Hangzhou, China.
Talk: "理解大语言模型". - Conference Participation: 2023/9/22-23, 第三届数据科学与现代经济统计论坛, 厦门大学.
Talk: "机器学习和中国宏观经济预测--模型应用、比较和可解释性分析".
Slides
- AI Economics, 2024-02
- Weighted Signature Kernels and Applications, 2023-12
- Deep Generative Modeling with BSDEs, 2023-04
- Understanding ChatGPT and The Related, 2023-02
- AI Research Frontiers, 2022-12
- Generative Ambiguity Modeling Using Multiple Transformers, 2022-11
- BSDE, g-expectation and Ambiguity, 2022-07
- The Basics of Machine Learning, 2021
Grants and Awards
- Shanghai Leading Talent Program of Eastern Talent Plan / 上海市东方英才计划拔尖项目，2025
- Shanghai Sailing Program, STCSM / 上海市"科技创新行动计划"启明星项目(扬帆专项), 2022
- Hangzhou High-level Talents Category E / 杭州市高层次人才E类, 2021
- Excellent Graduate of Peking University / 北京大学优秀毕业生, 2019
- National Scholarship, PKU / 国家奖学金, 2017
Contact & Links
Email: xingcheng.xu18@gmail.com
Affiliation: Shanghai AI Laboratory
Address: Xuhui District, Shanghai, China