Scientific LLMs: A Data-Centric Survey and Roadmap

This August 2025 paper offers an extensive overview of the evolution and application of Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs) within scientific research, primarily focusing on the period from 2018 to 2025. It details how these AI models have progressed through various paradigm shifts, from initial transfer learning to sophisticated scientific agents capable of autonomous research. The document thoroughly examines the diverse data modalities—including visual spectra, microscopy images, molecular encodings, and time-series data—across six key scientific domains: Chemistry, Materials Science, Physics, Life Sciences, Astronomy, and Earth Science. Furthermore, it addresses critical issues surrounding data quality, traceability, timeliness, privacy, and bias within scientific datasets, while also highlighting the importance of robust evaluation benchmarks and tool integration for advancing scientific AI.Source:https://arxiv.org/pdf/2508.21148

There are no comments yet.

Please log in to write the first comment.