Toward an Offline Generative AI Dialogue Model to Support the Automation of a Materials Synthesis Laboratory

Abstract

Materials synthesis laboratories face challenges due to repetitive and error-prone tasks, necessitating innovative solutions. This review explores the potential of large language models (LLMs) to develop an offline dialogue model based on generative artificial intelligence, designed to automate processes in materials synthesis laboratories through conversational interaction with robotic systems and human personnel. Through a systematic literature review from 2019 to 2025, 13 studies are analyzed across three categories: generative AI and language models, automation and robotics in laboratories, and automation and evaluation frameworks. The findings highlight improvements in efficiency (20–90%) and cost reduction (up to 84%) in robotic and scientific tasks but also point out limitations in accuracy (60–80%) and reproducibility (65%). Offline operation requires optimization for local hardware and data security. A hybrid model with human feedback is recommended to maximize impact in materials synthesis, addressing both technical and ethical challenges.