Core Insights - The 2025 World Artificial Intelligence Conference in Shanghai showcased over 800 companies and more than 3000 innovative technology exhibits, emphasizing the rapid evolution and application of domestic AI models across various industries [2][3] - AI is transitioning from technical breakthroughs to deep integration with industries, significantly enhancing operational efficiency in sectors like supply chain, logistics, and human resources [2][4] Group 1: AI Development and Application - Domestic AI models have seen rapid iteration and deployment in various sectors, with AI employees automating tasks such as order processing and resume screening, leading to notable efficiency improvements [2][3] - The data production in China is projected to reach 41.06 zettabytes in 2024, a 25% year-on-year increase, providing a rich foundation for deep AI applications [2][3] Group 2: JD's AI Innovations - JD's upgraded JoyAI model has been integrated into hundreds of business scenarios, demonstrating replicable industrial practices and significantly enhancing operational processes [4][5] - The intelligent replenishment system at JD exemplifies how continuous data feedback optimizes supply chain models, reducing inventory turnover days to nearly 30 [3][4] Group 3: Consumer and Business Impact - JD's digital human anchors have shown remarkable commercial viability, serving over 20,000 brands and increasing user engagement significantly in live streaming [6][7] - The collaboration with leading brands and the integration of AI in consumer products have transformed traditional business models, enhancing productivity and consumer interaction [5][6] Group 4: Future Directions - JD aims to leverage its advantages in scene scale and data resources to drive technological innovation and set industry standards, contributing to high-quality development in the real economy [7][8]

近日，京东全新推出750B深度思考大模型、行业首批高商业可用数字人，并发布了角色智能体与具身 智能融合的最新进展，展示出AI认知能力与商业应用的最近进展。 "大模型的能力光谱在变、参数尺寸在变，不变的是让产业用好大模型。只有深耕产业，让大模型在产 业里跑起来，才是最有价值的事情。"京东集团探索研究院副院长、京东科技人工智能业务部总裁何晓 冬表示。 技术拓宽大模型能力边界，支持企业构建专有模型 在AI进化的"马拉松"中，京东大模型展现出了独特的产业基因。 目前，京东已经具备全尺寸大模型，满足多样化的产业需求：3B和10B模型可提供极致响应效率；81B 主力模型可兼顾效果和性能。最新推出的750B超大规模模型，则兼具"深度思考"和"非深度思考"双通 道能力，能满足各行业对"即时响应"和"深度推理"的双重需求。 具体来说，750B超大规模模型在训练过程中，使用了动态分层蒸馏、跨领域数据治理等京东创新技 术，降低大模型的训练和部署成本他，同时兼顾大模型效果，保证大模型能力"大而精"。 数字人实现高商业可用，618面向商家全免费 在大模型"硬实力"的支撑下，数字人等"软实力"应用，也迅速实现了商业化可用。 基于通用数字 ...