Empowering Generalist Material Intelligence with Large Language Models.

Area of Science:

• Materials Science
• Artificial Intelligence
• Computational Science

Background:

• Large language models (LLMs) are emerging as key drivers in materials science.
• Generalist Materials Intelligence (GMI) integrates reasoning, modeling, and validation.
• The agent-in-the-loop paradigm leverages LLM-based agents for autonomous discovery.

Purpose of the Study:

• To review the transformative impact of LLMs in materials science.
• To discuss challenges and propose strategies for LLM adoption in GMI.
• To outline a roadmap for implementing materials-aware LLMs.

Main Methods:

• Comprehensive literature review of LLM applications in materials science.
• Analysis of LLM roles in data extraction, property prediction, structure generation, synthesis planning, and self-driven labs.
• Discussion of challenges including knowledge misalignment, resource demands, and performance inconsistencies.

Main Results:

• LLMs are revolutionizing traditional materials science tasks and catalyzing the agent-in-the-loop paradigm.
• LLMs bridge the gap between ontology, concept, computation, and experiment.
• Significant challenges exist in scaling LLM adoption due to knowledge misalignment and resource intensity.

Conclusions:

• LLMs are pivotal for advancing GMI, enabling autonomous materials discovery.
• Addressing LLM limitations in adaptability, efficiency, and trustworthiness is crucial.
• A strategic roadmap is proposed for the equitable and democratic implementation of materials-aware LLMs.