Large language model for knowledge synthesis and AI-enhanced biomanufacturing
Wenyu Li, Zhitao Mao, Zhengyang Xiao, Xiao‐Ping Liao, Mattheos Koffas, Yixin Chen, Hongwu Ma, Yinjie Tang
Abstract
Large language models (LLMs) are transforming synthetic biology (SynBio) education and research. In this review we cover the advancements and potential impacts of LLMs in biomanufacturing. First, we summarize recent developments and compare the capabilities of US and Chinese language models in addressing fundamental SynBio questions. Second, we discuss the application of LLMs in extracting SynBio information from unstructured data, constructing knowledge graphs, and enabling retrieval-augmented generation. Third, we anticipate that LLMs will not only revolutionize the design-build-test-learn (DBTL) cycle in metabolic modeling and engineering but also enable self-driving laboratories in future biomanufacturing. Finally, we emphasize the need for establishing benchmarks for LLMs, fostering trustworthy knowledge synthesis, developing biosecurity frameworks to prevent misuse, and encouraging collaboration among artificial intelligence (AI) scientists, SynBio researchers, and bioprocess engineers.