BEIJING, June 3 (Xinhua) — Chinese artificial intelligence (AI) researchers have released an open source framework and real-world scenario competition platform that significantly improves industrial applications of AI.
As AI’s computing power increases and large-scale language models become increasingly sophisticated, Chinese researchers have focused on how best to apply them to real-world scenarios.
To achieve this objective, research initiatives of the Artificial Intelligence Innovation Center of the Yangtze River Delta Regional Research Institute of Tsinghua University have standardized human-machine interaction, task-set mechanisms, and human feedback systems, resulting in improved efficiency of industrial applications and expanded enterprise deployment.
The team leader said that the global AI industry currently faces a structural contradiction in which the capabilities of models and tools are growing exponentially, while the rate of industrial adoption is increasing linearly. The core paradox in AI development has shifted from “enhancing model intelligence” to “closing the adoption gap.”
real world framework
To address the gap between AI capabilities and real-world deployments, the team released the Real World AI (RWAI) open source framework, expanding the scope of the open source effort from code and tools to include role definition, workflow design, human-machine interaction, and human-human collaboration as integrated practices.
This framework reimagines the interaction between AI and humans in real-world tasks through three core elements: restoring real-world task sets, capturing authentic human feedback from real-world interactions, and standardizing human-machine interaction protocols.
According to the team, real-world testing has proven that RWAI outperforms traditional software development models in practical efficiency, real-world effectiveness, and resolution time, reducing pre-project validation timelines from two to three months to less than two weeks.
The team also launched the AI Arena platform. Unlike traditional benchmarks and model leader boards, the platform focuses on evaluating the real-world effectiveness of AI solutions in real-world business operations, including metrics such as organizational costs, time efficiency, computing costs, and compliance requirements.
The platform employs a “challenger-champion” mechanism, where competing entities are not single models, but complete solutions that include team configurations, workflows, agent combinations, and context engineering. Best practice workflows for successful solutions are published and available for replication.
industrial use
China Southern Power Grid’s Internet Services subsidiary leveraged the RWAI platform to address end-to-end safety management challenges for power grid infrastructure projects, from planning to on-site execution. Faced with complex compliance requirements, traditional manual oversight of infrastructure projects was reaching an efficiency bottleneck.
The company used the platform to develop an intelligent risk management solution for site and subcontractor management, increasing hidden risk detection rates by approximately 40 percent and risk alert accuracy to 92 percent. The company and its research team are currently preparing to advance a demonstration project of generative AI across the lifecycle of a construction project.
Hu Rui, a senior engineer at the company, said the RWAI platform has succeeded in bridging the gap between AI technology and implementation while significantly reducing the cost of trial and error. The system transforms engineering management from reactive to proactive intelligent control and uses AI to support high-quality power grid construction.
Jiangsu Eastern Shenghong Co., Ltd. also used the RWAI platform. As a petrochemical manufacturer, the company has long faced challenges such as integrating traditional process industry knowledge, applying general purpose AI to mission-critical operations, and lack of control over decision-making with large-scale models.
Leveraging the RWAI platform, Eastern Shenghong integrated 30 years of production process knowledge and data from across the industry chain to overcome the lack of corpus and high compliance barriers in the chemical industry to build a large-scale industrial model that truly understands the business.
Through multimodal fault monitoring and prediction, the company can significantly reduce unplanned downtime on key production lines and dynamically recommend optimal production schedules to reduce costs, increase efficiency, and optimize processes.
Yang Tianwei, vice chairman and general manager of the company’s AI business unit, said that by using the evaluation capabilities of the RWAI platform, the company has translated high-quality, internally validated model capabilities into a library of reusable, billable, and composable products.
“This not only activates Eastern Shenghong’s unique intelligent development, but also provides field-proven best practice solutions for deploying large-scale models in the process industry,” Yang added.
The RWAI platform currently covers multiple application scenarios, including industrial forecasting systems, document review and risk management, and research report generation. Its implementation has already been implemented in some projects of Fortune Global 500 companies.
The research team said the platform also provides real-world human-computer interaction data to support large-scale model development and academic research. ■
