AI for Science

Follow this topic Ask about this topic

AI for Science studies how computational methods can help researchers analyze data, build and test models, design experiments, and discover new explanations or materials. Approaches include scientific foundation models, retrieval and multi-agent systems, causal and symbolic methods, Bayesian optimization, and tools that connect AI to simulators, code, instruments, and high-performance computing. Benchmarks increasingly test scientific coding, image interpretation, simulator use, experimental design, and multi-step research workflows. AI often helps with well-specified tasks, but reliable open-ended discovery remains difficult: systems can make plausible errors, struggle with mechanistic reasoning, and need stronger verification, provenance, and evaluation against expert work.

Latest papers

161 papers10 papers in the last four weeks, down 23% on the four weeks before. 0.1% of all new papers.

AI-for-Science Low-code Platform with Bayesian Adversarial Multi-Agent Framework
Mar 3, 2026Zihang Zeng, Jiaquan Zhang, Pengze Li +2Scientific Code Generation AI for Science
Artificial Intelligence Innovation and Incubation Institute, Fudan University · Shanghai Innovation Institute · Shanghai Academy of AI for Science
OrchMAS: Orchestrated Reasoning with Multi Collaborative Heterogeneous Scientific Expert Structured Agents
Mar 3, 2026Yichao Feng, Haoran Luo, Zhenghong Lin +4Multi-Agent LLMs Multi-Agent Orchestration
Magellan Technology Research Institute (MTRI) · Nanyang Technological University, Singapore
The Alien Space of Science: Sampling Coherent but Cognitively Unavailable Research Directions
Mar 1, 2026Alejandro H. Artiles, Martin Weiss, Levin Brinkmann +6Research Idea Generation Large Language Model-Guided Scientific Discovery
Tiptree Systems · Max Planck Institute for Human Development, Berlin · Max Planck Institute for Intelligent Systems, Tübingen +4
Mapping the Landscape of Artificial Intelligence in Life Cycle Assessment Using Large Language Models
Feb 26, 2026Anastasija Mensikova, Donna M. Rizzo, Kathryn HinkelmanLife Cycle Assessment AI for Science
University of Vermont Burlington, VT, USA
PreScience: A Dataset and Benchmark for Scientific Forecasting
Feb 24, 2026Anirudh Ajith, Amanpreet Singh, Jay DeYoung +7AI for Science Forecasting Benchmarks
Allen Institute for Artificial Intelligence, Seattle, WA, USA · Knowledge Lab, University of Chicago, Chicago, IL, USA · School of Computer Science and Engineering, Hebrew University of Jerusalem, Israel +1
Transforming Behavioral Neuroscience Discovery with In-Context Learning and AI-Enhanced Tensor Methods
Feb 19, 2026Paimon Goulart, Jordan Steinhauser, Dawon Ahn +4AI for Science Tensor Decomposition
UC Riverside Computer Science & Engineering Riverside, CA, USA · UC Riverside Psychology Riverside, CA, USA · UC Riverside Electrical & Computer Engineering Riverside, CA, USA
Criteria-first, semantics-later: reproducible structure discovery in image-based sciences
Feb 17, 2026Jan BumbergerScientific Reproducibility AI for Science
Helmholtz Centre for Environmental Research – UFZ, Research Data Management - RDM, Permoserstraße 15, Leipzig, 04318, Germany · Helmholtz Centre for Environmental Research – UFZ, Department Monitoring and Exploration Technologies, Permoserstraße 15, Leipzig, 04318, Germany · German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Puschstraße 4, Leipzig, 04103, Germany
Anagent For Enhancing Scientific Table & Figure Analysis
Feb 10, 2026Xuehang Guo, Zhiyong Lu, Tom Hope +1Scientific Figure Understanding AI for Science
College of William & Mary · NIH - National Library of Medicine · The Allen Institute for AI (AI2)
Scientific feedback reveals productive regimes beyond training conventions
Dec 12, 2025Yoon Pyo Lee, Samrendra Roy, Souvik Chakraborty +1AI for Science RL for Generative Models
Department of Nuclear Engineering, Hanyang University, 222, Wangsimni-ro, Seongdong-gu, Seoul, 04763, Republic of Korea.
CLIMATEAGENT: Multi-Agent Orchestration for Complex Climate Data Science Workflows
Nov 25, 2025Chenyue Li, Hyeonjae Kim, Wen Deng +4AI for Science Multi-Agent Orchestration
1The Hong Kong University of Science and Technology
Agentic Artificial Intelligence for Multistage Physics Experiments at a Large-Scale User Facility Particle Accelerator
Sep 21, 2025Thorsten Hellert, Drew Bertwistle, Simon C. Leemann +2AI for Science Agentic Workflows
Lawrence Berkeley National Laboratory, Berkeley, California 94720, USA

Mar 3, 2026J/K move · Enter open · S save

AI-for-Science Low-code Platform with Bayesian Adversarial Multi-Agent Framework

Zihang Zeng, Jiaquan Zhang, Pengze Li, Yuan Qi+1

Artificial Intelligence Innovation and Incubation Institute, Fudan University · Shanghai Innovation Institute · Shanghai Academy of AI for Science

Scientific Code Generation AI for Science Multi-Agent Coding LLM Agent Reliability

Open paper AskPDF arXiv

Large Language Models (LLMs) demonstrate potentials for automating scientific code generation but face challenges in reliability, error propagation in multi-agent workflows, and evaluation in domains with ill-defined success metrics. We present a Bayesian adversarial multi-agent framework specifically designed for AI for Science (AI4S) tasks in the form of a Low-code Platform (LCP). Three LLM-based agents are coordinated under the Bayesian framework: a Task Manager that structures user inputs into actionable plans and adaptive test cases, a Code Generator that produces candidate solutions, and an Evaluator providing comprehensive feedback. The framework employs an adversarial loop where the Task Manager iteratively refines test cases to challenge the Code Generator, while prompt distributions are dynamically updated using Bayesian principles by integrating code quality metrics: functional correctness, structural alignment, and static analysis. This co-optimization of tests and code reduces dependence on LLM reliability and addresses evaluation uncertainty inherent to scientific tasks. LCP also streamlines human-AI collaboration by translating non-expert prompts into domain-specific requirements, bypassing the need for manual prompt engineering by practitioners without coding backgrounds. Benchmark evaluations demonstrate LCP's effectiveness in generating robust code while minimizing error propagation. The proposed platform is also tested on an Earth Science cross-disciplinary task and demonstrates strong reliability, outperforming competing models.