Integrative innovation of large language models in industries: technologies, applications, and challenges

Abstract

This paper examines the transformative potential of large language models (LLMs) across diverse industries, emphasizing their ability to enhance natural language processing tasks through pre-training on extensive datasets. Although LLMs offer significant opportunities to automate customer interactions, improve decision-making, and optimize workflows, their rapid adoption also presents challenges such as information security, data quality, model interpretability, ethical implications, and regulatory compliance. To address these issues, this review proposes integrative strategies, including scenario-based applications, methodological innovation, and data-model integration, to boost LLMs’ performance and adaptability. It also explores the evolution of LLMs toward multimodal and multitask general-purpose models, highlighting future trends in sustainable development and human-centric AI. Future research directions include achieving a balance between enhancing model capabilities and managing energy consumption, as well as improving transparency and explainability to strengthen user trust.