LLM-Based Data Science Agents: A Survey of Capabilities, Challenges, and Future Directions
York University(约克大学);Vector Institute for AI(人工智能矢量研究所);Dialpad Inc.(Dialpad公司);Nanyang Technological University(南洋理工大学);Salesforce AI Research(Salesforce人工智能研究)
Mizanur Rahman, Amran Bhuiyan, Mohammed Saidul Islam, Md Tahmid Rahman Laskar, Ridwan Mahbub, Ahmed Masry, Shafiq Joty, Enamul Hoque
备注:
Survey paper; 45 data science agents; under review
Recent advances in large language models (LLMs) have enabled a new class of AI agents that automate multiple stages of the data science workflow by integrating planning, tool use, and multimodal reasoning across text, code, tables, and visuals. This survey presents the first comprehensive, lifecycle-aligned taxonomy of data science agents, systematically analyzing and mapping forty-five systems onto the six stages of the end-to-end data science process: business understanding and data acquisition, exploratory analysis and visualization, feature engineering, model building and selection, interpretation and explanation, and deployment and monitoring. In addition to lifecycle coverage, we annotate each agent along five cross-cutting design dimensions: reasoning and planning style, modality integration, tool orchestration depth, learning and alignment methods, and trust, safety, and governance mechanisms. Beyond classification, we provide a critical synthesis of agent capabilities, highlight strengths and limitations at each stage, and review emerging benchmarks and evaluation practices. Our analysis identifies three key trends: most systems emphasize exploratory analysis, visualization, and modeling while neglecting business understanding, deployment, and monitoring; multimodal reasoning and tool orchestration remain unresolved challenges; and over 90% lack explicit trust and safety mechanisms. We conclude by outlining open challenges in alignment stability, explainability, governance, and robust evaluation frameworks, and propose future research directions to guide the development of robust, trustworthy, low-latency, transparent, and broadly accessible data science agents.
EarthSE: A Benchmark for Evaluating Earth Scientific Exploration Capability of LLMs
Shanghai Jiao Tong University(上海交通大学);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);The Hong Kong Polytechnic University(香港理工大学);The Chinese University of Hong Kong(香港中文大学)
Leveraging ChatGPT's Multimodal Vision Capabilities to Rank Satellite Images by Poverty Level: Advancing Tools for Social Science Research
Center for Applied Intelligent Systems Research (CAISR), Halmstad University, Sweden(应用智能系统研究中心(CAISR),哈马斯特德大学,瑞典);Department of Human Geography, Lund University, Sweden(人类地理系,吕勒奥大学,瑞典);Department of Earth and Environmental Sciences, Lund University, Sweden(地球与环境科学系,吕勒奥大学,瑞典)
CoWeaver: A Bi-directional, Learnable and Explainable Matching Engine for Mixed Human-Agent Science Collaboration
McGill University(麦吉尔大学);Mila – Québec AI Institute(Mila——魁北克人工智能研究所);University of Connecticut(康涅狄格大学);Sichuan University(四川大学);Gradient;Princeton University(普林斯顿大学)
The Alien Space of Science: Sampling Coherent but Cognitively Unavailable Research Directions
Max Planck Institute for Human Development(马克斯·普朗克人类发展研究所);Max Planck Institute for Intelligent Systems(马克斯·普朗克智能系统研究所);ELLIS Institute Tübingen(图宾根ELLIS研究所);Polytechnique Montreal(蒙特利尔理工学院);CIFAR AI Chair(CIFAR人工智能主席);Mila – Quebec AI Institute(魁北克人工智能研究所);Tiptree Systems(Tiptree系统)
OceanGPT: A Large Language Model for Ocean Science Tasks
College Computer Science and Technology, Zhejiang University(浙江大学计算机科学与技术学院);Donghai Laboratory(东海实验室);School of Software Technology, Zhejiang University(浙江大学软件学院);Huzhou University(湖州大学);Ocean College, Zhejiang University(浙江大学海洋学院);Zhoushan-Zhejiang University Ocean Research Center(舟山浙江大学海洋研究中心)
Bridging Technology and Humanities: Evaluating the Impact of Large Language Models on Social Sciences Research with DeepSeek-R1
School of Physics and Information Technology, Shaanxi Normal University(陕西师范大学物理与信息技术学院);School of Foreign Languages, Northwest University(西北大学外国语学院);School of Computing, University of Georgia(佐治亚大学计算学院)
SCICONVBENCH: Benchmarking LLMs on Multi-Turn Clarification for Task Formulation in Computational Science
Rensselaer Polytechnic Institute(拉特格斯理工学院);University of Texas at Arlington(德克萨斯大学阿灵顿分校);Pacific Northwest National Laboratory(太平洋西北国家实验室);National Renewable Energy Laboratory(国家可再生能源实验室)
Cognitive Science-Inspired Evaluation of Core Capabilities for Object Understanding in AI
University of Cambridge(剑桥大学);Universitat Politècnica de València(瓦伦西亚理工大学);Sigmund Freud University(西格蒙德·弗洛伊德大学);University of Primorska(普里莫尔斯卡大学);Helmholtz Institute for Human-Centered AI(亥姆霍兹以人为本人工智能研究所)
Promising and worth-to-try future directions for advancing state-of-the-art surrogates methods of agent-based models in social and health computational sciences