中国工程院院士、浙江大学教授谭建荣在主题发言环节，发表题为"大模型与智能体：关键技术与发展趋势"的演讲，系统拆解了人工智能从底层技术到落地 应用的全链路逻辑。 人工智能技术的迅速发展推动了大模型和智能体的融合发展，成为推动产业革新的核心驱动力。12月9日，由中关村科金主办的"超级链接·智见未来"—— EVOLVE 2025大模型与智能体产业创新峰会顺利举办。当日，中关村科金与华为云、阿里云、百度智能云、火山引擎、亚马逊云科技、超聚变、软通动力 等产业领军企业，共同发布 "超级连接" 全球生态伙伴计划。 谭建荣表示，智能体是人工智能的载体，人工智能的核心由数据、算法、算力三大部分构成，而这三者的融合载体便是智能体。智能体作为人工智能的落地 载体，已广泛应用于智能机器人、无人驾驶汽车、无人机等场景。 责任编辑：山上 关键词阅读：智能体 大模型 中关村科金 为什么近年来人工智能火起来了？谭建荣认为，早期专家系统依赖因果关系编程，而如今大模型则依托大数据挖掘关联关系，实现了技术路径的重大突破。 针对未来发展趋势，谭建荣提出大模型与智能体需要实现云、边、端的协同发展，特别是智能体需要云、边、端的协调部署，实现同步协同， ...

新浪科技讯 12月9日下午消息，今日举办的EVOLVE 2025中关村科金大模型与智能体产业创新峰会上， 中国工程院院士谭建荣分享指出："我们要重视大模型，但也千万不能忽视小模型，没有小模型只有大 模型，人工智能想要落地也很困难。" 谭建荣指出，人工智能模型、算力、算法三大要素之外，知识工程也是实现人工智能的核心关键技术之 一。其中，知识可以分为定性、定量两类，而模型就是定量的知识，大模型需要花费算力对不同数据进 行训练，最终产生知识，因此，大数据、大模型的背后，也需要用到大的算力作为支撑。（文猛） 新浪科技讯 12月9日下午消息，今日举办的EVOLVE 2025中关村科金大模型与智能体产业创新峰会上， 中国工程院院士谭建荣分享指出："我们要重视大模型，但也千万不能忽视小模型，没有小模型只有大 模型，人工智能想要落地也很困难。" 谭建荣指出，人工智能模型、算力、算法三大要素之外，知识工程也是实现人工智能的核心关键技术之 一。其中，知识可以分为定性、定量两类，而模型就是定量的知识，大模型需要花费算力对不同数据进 行训练，最终产生知识，因此，大数据、大模型的背后，也需要用到大的算力作为支撑。（文猛） 责任编辑：杨 ...

Core Insights - The article discusses the implementation of large models in game development, highlighting the challenges and advancements in AI coding tools, particularly in the context of complex game projects [2][3][4]. Group 1: AI Tools in Game Development - Numerous AI coding tools have emerged recently, but their participation in game project coding remains limited due to the complexity and flexibility of game business [2][4]. - A large-scale internal survey revealed that game developers spend more time on code understanding rather than code writing, indicating a need for better tools to facilitate this understanding [4][6]. Group 2: Challenges in Game Development - Three main challenges were identified: lack of clear technical documentation (30%), the complexity of game development pipelines compared to traditional web development, and slow testing and debugging processes [6][8]. - The game development process often leads to accumulated technical debt due to rushed timelines, which complicates the coding and debugging phases [6][8]. Group 3: Knowledge Engineering in Game Development - The company has developed a game development knowledge engineering system to improve code understanding and collaboration among different roles such as planning, art, and development [13][14]. - The knowledge system integrates structured and unstructured data, allowing for efficient retrieval and application of knowledge within the game development context [14][19]. Group 4: AI-Driven Code Generation and Review - A dual-end system was created to enhance code understanding, generation, and quality review, focusing on integrating AI capabilities into the existing development environment [8][11]. - AI-generated code accounted for 30% of the total code produced, with the system contributing approximately 5 million lines of code monthly across various projects [41][44]. Group 5: AI Code Review Process - The company has implemented a combination of traditional static code analysis and AI-driven code review to ensure quality control throughout the development process [44][45]. - The AI review process aims to identify low-level errors that could lead to significant operational issues, enhancing the overall quality of the code produced [45][46]. Group 6: Future Directions and Team Collaboration - The focus is on creating a cohesive team AI agent system that facilitates collaboration across different roles in game development, aiming to enhance efficiency and knowledge sharing [55][56]. - The upcoming AICon event will explore further applications of AI in business growth and development efficiency, featuring insights from industry experts [2][56].

Core Insights - Ant Group and Ningbo Bank's collaboration on the "Agentar Knowledge Engineering KBase" has been recognized as an exemplary case for international financial applications, showcasing its potential to enhance business intelligence in the financial sector [1] - The financial industry faces challenges related to "knowledge silos," where critical information is dispersed across different systems, leading to inefficiencies in service and consultation experiences [1] - The Agentar platform integrates knowledge processing management, logical reasoning engines, and intelligent application scenarios to provide a robust decision-making system for financial institutions [1] Technology and Implementation - The platform manages multi-source heterogeneous data throughout its lifecycle and features capabilities such as intelligent Q&A, knowledge processing, multi-route recall, and knowledge management [2] - A significant technological breakthrough is the knowledge-enhanced generation engine, which utilizes a collaborative mechanism of "planning-retrieval-reasoning" to improve knowledge quality through bidirectional indexing of knowledge graphs and raw text [2] - The system has transitioned from "fuzzy matching" to "precise reasoning," increasing reasoning depth from traditional 1-hop to 3-5 hops, enabling AI to understand financial knowledge and exhibit human-like logical reasoning [2] Performance Metrics - The solution has been implemented across various internal scenarios at Ningbo Bank, including market analysis, product interpretation, dialogue practice, and report writing [2] - Evaluation results indicate that the accuracy of complex Q&A has improved from 68% to 91%, with response times reaching the millisecond level [2] - Content recommendation accuracy has increased by 35%, and recall rates have improved by 40%, leading to a significant enhancement in business efficiency [2] Future Directions - Ant Group and Ningbo Bank plan to deepen their collaboration by expanding the technology to a broader range of financial business scenarios [2] - The partnership aims to actively participate in industry standardization efforts, promoting the regulated and large-scale application of knowledge engineering and large model technologies in the financial sector [2]