LLM-assisted literature analysis for plastic upcycling

Plastic waste poses a global challenge, driving interest in upcycling strategies to convert waste into value-added products. The interdisciplinary nature of plastic upcycling research-spanning fields such as chemistry, material science, and environmental science-has led to a surge in publications, making it challenging to synthesize key insights. Large language models (LLMs) offer transformative potential for literature analysis, enabling rapid, scalable, and consistent analysis across vast datasets. In this perspective, we evaluated the use of LLMs in 883 research articles about plastic upcycling, demonstrating their efficiency and accuracy in classifying plastics, identifying upcycling pathways, and visualizing trends. LLMs achieved performance comparable to human experts in well-defined tasks while completing analysis in a fraction of the time. We highlight the value of LLM-driven insights for guiding future research and propose a collaborative framework among researchers, publishers, and technology developers to optimize LLM applications. By integrating LLMs into workflows, the scientific community can accelerate innovation in tackling environmental challenges.