运营商最佳实践：人工智能大模型赋能垂直用例（2025）

AI Large Model Empowering Verticals Use Cases Industrial Upgrading &Intelligent ManufacturingPublic Services &Social GovernanceHealthcare &Smart EducationInnovative CustomerServices & Operations 2025 Foreword Sihan Bo CHENHead of Greater ChinaGSMA In 2024, GSMA Greater China released the "Operator Practice:2024 Benchmark Cases of AI Large Model-Empowered Verticals,"presenting typical cases in fields such as government services,customer services, industrial applications, healthcare, education,cultural tourism, and urban governance from operators'perspectives. Focusing on the innovative applications andexploratory practices of large AI models across multiple industries,this report demonstrates how technology is breaking throughindustrial boundaries, reveals the challenges behind success andthe value of innovation, and offers profound insights and Chineseexpertise to global operators and vertical industries. specific capabilities like Open Gateway and cloud-networkintegration have also been incorporated into solution architectures,promoting ecosystem collaboration with governments, hospitals,enterprises, and AI partners, and moving from pilot phases toroutine operations. These practices suggest that Chinese operators shifted theirfocus from "application breadth" to "implementation depthand industrial value reconstruction." They are not only drivingintelligent industrial upgrading and unleashing the value of datafactors but also playing an indispensable role in promoting high-quality social development. Through engineering methods anddeep integration with business scenarios, large model technologyforms a sustainable innovation closed-loop, bringing replicable,measurable, and iterative implementation models to industries, andbringing the development of Digital China to new heights. Entering 2025, the industry landscape is evolving at anaccelerated pace. The deep integration of large AI models with5G-Advanced, cloud computing, and computing power networkshas been a key driving force for digital economic development.This year, the practices of Chinese operators have presentedmarked progress and transformation compared to last year:the case studies no longer merely emphasize the breadth ofapplications and exploratory pathways, but instead focus moreon replicable, measurable, and iterative industrial implementation.The convergence of cloud-network-intelligent computing,industry-specific large models, and agent platforms has becomemainstream. Model capabilities have moved beyond single-pointtools, and a value-closed loop integrating training and inferencehas been formed through engineering methods and iterative datafeedback. Operators and ecosystem partners are co-innovatingto advance AI capabilities from pilot projects to systematicdeployment at scale, truly translating technological bonuses intolong-term value for society and industry. At the same time, responsibility and security are the foundationof technological advancement. As large AI models penetratedeeper into various sectors, challenges such as data security,privacy protection, model generalization, and integration ofdomain knowledge persist. Achieving deep synergy betweentechnology and business models, and building open, win-winecosystems remain areas requiring continuous effort. Besides long-term investment and strategic resolve, operators and ecosystempartners need to jointly advance AI, within a framework of opencooperation and responsible innovation, to support sustainableeconomic and social development. Looking ahead, large AI models will continue to evolve, andapplication scenarios will expand. Operators will unleash innovativevitality across more domains, joining hands with ecosystempartners to co-create a new intelligent era. We believe that largeAI models will play significant roles in more industries, drivingdigital transformation and upgrading, fostering new applicationsand business models, and creating greater value for industries andsociety at large. This year's case studies collection concentrates on four key areas:innovative customer services and operations, healthcare and smarteducation, industrial upgrading and intelligent manufacturing, andpublic services and social governance. They systematically outlinethe pathways and outcomes of large AI models, empoweringdigital transformation across sectors. These cases not only covera wider range of industries but also emphasize quantifiableresults and clearer business models. For instance, China Telecom'seducation platform serves over 2,000 schools in 31 provinces,increasing grading efficiency by 70% with over 95% consistency;China Unicom's healthcare quality control solution has coveredpetabyte-level data volumes of 230 institutions. Millions of dailyquality checks are conducted, shortening the quality control cyclefrom "years" to "months," and saving nearly RMB10 million inmedical insurance costs; China Mobile's civil aviation customerservice solution has reduced order processing time