Large Language Models for Chemists: Applications and Insights

In recent years, LLMs (such as ChatGPT, Claude, DeepSeek, LLaMA, and other transformer-based models) have emerged as powerful tools in chemistry, enabling new approaches to scientific discovery. While many chemists, from undergraduate students to researchers, find these AI models interesting, they may lack a certain knowledge base to better integrate these tools into their daily research.

Large Language Models for Chemists breaks down that barrier by demystifying how LLMs work in an accessible way and showing, step by step, how they can be applied to solve real chemistry problems. Written in a friendly, tutorial style, the book assumes only a basic background in chemistry and minimal programming experience. It begins by gently introducing artificial intelligence and machine learning concepts in lay terms, building up to the inner workings of LLMs without heavy math. Readers will learn how these models ＂think＂ and generate text, gaining an intuitive understanding of concepts like neural networks, transformers, and training data using analogies and simple diagrams. Crucially, each concept is reinforced with chemistry-focused examples. It spans from understanding chemical nomenclature and reactions as a ＂language＂ to exploring how an LLM can suggest synthetic routes or explain spectral data.

Beyond theory, this book emphasizes practical application. Each chapter includes hands-on tutorials and case studies that invite readers to experiment with real tools. Using open-source libraries (such as RDKit for cheminformatics and standard Python machine learning frameworks), readers will walk through projects like predicting molecular properties with the aid of an LLM, generating novel compound ideas, analyzing research papers, and even using an LLM as a conversational chemistry assistant. For example, one case study guides the reader in using an LLM to mine a chemistry literature database and then write Python code to analyze reaction trends, mirroring cutting-edge research where LLMs assist in code generation and data mining for chemical discovery.

Zhiling Zheng is an incoming Assistant Professor at Washington University in St. Louis in the Department of Chemistry. He earned his Ph.D. from the University of California, Berkeley, and completed his postdoctoral research at MIT. Zheng is renowned for foundational research integrating large language models (LLMs) with chemical research. Zheng earned his Ph.D. in Chemistry from the University of California, Berkeley, where he was mentored by Professor Omar M. Yaghi and his doctoral research in materials chemistry (designing metal-organic frameworks for water harvesting) honed his expertise in experimental design and chemical data analysis. He then expanded into data-driven chemistry and machine learning during his postdoctoral research at MIT in Professor Klavs Jensen’s group. At MIT, Zheng pioneered methods that integrate LLMs with chemical research, including using LLMs for literature mining, automated synthesis planning, and even code generation to accelerate reaction discovery. He has published multiple peer-reviewed papers on these topics in high-impact venues (including an ACS editor’s choice JACS paper on ChatGPT for MOF data mining, two Angewandte Chemie article on LLM+ML for organic synthesis, and a perspective in Nature Reviews Materials on LLMs in chemistry).Very recently, he also co-authored a chapter on the book ＂Reticular Chemistry and New Materials＂ by Zheng’s unique dual expertise - deep knowledge of chemistry and hands-on experience with modern AI techniques - positions him as an ideal author for this book. He has also demonstrated a talent for science communication: he has delivered tutorials and invited talks on AI for chemists, mentored students in coding for chemistry, and is passionate about education. As an early-career professor leading a lab on ＂AI and Chemistry,＂ he is committed to training the next generation of chemists to use data and AI effectively.