Leveraging Foundational AI Models for Polymer Design

Abstract

Recently, there have been immense advances in the development of large-scale artificial intelligence (AI) models across numerous information modalities including text, video, and images. These foundation models, pre-trained on large corpuses of data, can be tuned for specific applications in materials and chemistry—often requiring significantly lower volumes of data. Consequently, there are many reports on leveraging large language models or similar systems in chemistry domains.[1-3] Despite these successes, there remains a significant gap between advances in foundation models, their practical use in routine research activities, and their overall utility for complex polymeric materials. In this talk, we will discuss our approaches towards solving a range of issues encountered during the development, training, deployment, and use of foundation models for polymeric materials, including representation and organization of historical experimental data.[4] We will also discuss how integration of foundation models and large language models into a single interface can facilitate a seamless user experience for utilization of AI for routine research tasks in polymer chemistry. Finally, specific uses of these technologies in combination with generative AI for polycondensation polymerization, organocatalysis, and polymer recycling will be highlighted in the presentation. 1. Jablonka, K. M.; Schwaller, P.; Ortega-Guerrero, A.; Smit, B. Is GPT-3 All You Need for Low-Data Discovery in Chemistry? ChemRxiv (2023). https://doi.org/10.26434/chemrxiv-2023-fw8n4. 2. Zheng, Z.; Zhang, O.; Borgs, C.; Chayes, J. T.; Yaghi, O. M. ChatGPT Chemistry Assistant for Text Mining and Prediction of MOF Synthesis. arXiv (2023). https://doi.org/10.48550/arXiv.2306.11296. 3. Castro Nascimento, C. M.; Pimentel, A. S. Do Large Language Models Understand Chemistry? A Conversation with ChatGPT. J. Chem. Inf. Model. 63, 1649–1655 (2023). https://doi.org/10.1021/acs.jcim.3c00285. 4. Park, N. H.; Manica, M.; Born, J.; Hedrick, J. L.; Erdmann, T.; Zubarev, D. Y.; Adell-Mill, N.; Arrechea, P. L. Artificial Intelligence Driven Design of Catalysts and Materials for Ring Opening Polymerization Using a Domain-Specific Language. Nat. Commun., 14, 3686, (2023). https://doi.org/10.1038/s41467-023-39396-3.