Core Insights - The focus of the industry is on how companies can implement cutting-edge AI technologies into practical business scenarios for sustainable growth as the AI model technology hype returns to rationality [2] - Tencent's Senior Executive Vice President emphasized that "driving industrial efficiency through intelligence and revenue scale through globalization" are the two core drivers of corporate growth [2] Group 1: AI Infrastructure and Investment - Tencent is significantly investing in AI infrastructure, with a strong emphasis on providing comprehensive support from infrastructure to model training and inference acceleration tools [4] - The shift in the big model industry focus from training to inference has become an industry consensus, leading to a surge in inference demand [4] - Tencent has established 11 regional offices globally and deployed 9 global technical support centers, enhancing its international infrastructure investment [5] Group 2: AI Strategy and Development - Tencent aims to create "human-centered AI," with a clear positioning that embraces AI across all business sectors [5] - The company has released over 30 models in the past year, focusing on achieving stronger model performance at lower deployment and inference costs [6] - Tencent's intelligent agent strategy was officially launched, providing a comprehensive open development platform and support for various application scenarios [6] Group 3: Market Dynamics and User Engagement - The demand for intelligent agents is diverse, with small and medium enterprises seeking more commercial support products from Tencent's intelligent agent development platform [6] - Tencent's AI applications are still in the investment phase, with a focus on enhancing product and service experiences rather than immediate commercialization [7] - User inquiries to Tencent's AI applications have reached the total monthly inquiries from earlier this year, indicating growing engagement [7]

过去一年，腾讯积极拥抱AI，率先接入DeepSeek，此后几乎一周一次的节奏飞快迭代。腾讯几乎所有 业务均涉足AI，如今，腾讯已经从"好用的AI"到"全面开放好用的AI"。 9月16日，2025腾讯全球数字生态大会上，腾讯集团高级执行副总裁、云与智慧产业事业群CEO汤道生 表示，"向智能化要产业效率，向全球化要收入规模"，已经成为企业增长的两大核心动力。腾讯将打 造"智能化"与"全球化"两大效率引擎，助力企业稳健和可持续增长。 智能化方面，腾讯云正式发布腾讯云智能体战略全景图，全面开放AI能力、C端和B端优势场景。通过 智能体解决方案、SaaS+AI、大模型技术三大升级，激发企业的创新潜能。 大会上，汤道生透露，腾讯元宝上线一年多，已经成为国内DAU排名前三的AI原生应用，用户现在每 天向腾讯元宝的提问量，已经达到年初一个月的总量；IMA知识库文件数量已经突破1亿；QQ浏览器的 AI月活数比4月增长了17.8倍。同时，AI也助力腾讯广告、游戏等业务实现双位数增长。 本次大会上，腾讯提到最多的是"以人为本的AI"，汤道生认为，AI的应用场景很多，腾讯还是聚焦在怎 么能让AI服务好人，以人的需求为中心，去推动技术 ...

Core Insights - The core argument presented by Zhou Hongyi is that the new industrial revolution is fundamentally driven by artificial intelligence, transitioning from digitalization to an intelligence-centric empowerment model [1][2]. Group 1: Development of Artificial Intelligence - The current focus of artificial intelligence has shifted from large language models to intelligent agents, marking the transition from the "first half" to the "second half" of AI development [1]. - Zhou Hongyi emphasizes that relying solely on large models is insufficient for driving the new industrial revolution, as they lack the ability to perform tasks directly [1]. Group 2: Intelligent Agents - Intelligent agents are described as "robots in virtual space" with four core capabilities: task decomposition and planning, memory, tool usage, and collaborative division of labor [2]. - The market potential for specialized intelligent agents in the industrial sector is projected to be ten times larger than that of traditional software [2]. Group 3: Impact on Workforce and Organizations - The concepts of "super employees" and "super organizations" are introduced, suggesting that individuals managing multiple intelligent agents can significantly enhance personal productivity, while organizations can achieve greater efficiency and flexibility [2]. Group 4: Platform Development and Collaboration - 360 Group has launched an intelligent agent factory platform that allows users to create intelligent agents without programming skills, facilitating rapid development through natural language descriptions [2]. - The company is open to collaborating with BRICS nations to support the digital transformation of traditional industries by providing comprehensive empowerment from intelligent agent construction to practical application [3]. Group 5: Future Outlook - Zhou Hongyi concludes that the intelligent transformation of traditional industries represents a profound change, with intelligent agents being a key force that will create limitless development opportunities [3].

9月16日—17日，2025金砖国家新工业革命伙伴关系论坛在厦门举行。人工智能是引领新一轮科技革命 和产业变革的战略性技术，在此次盛会上，围绕AI的讨论热度高企。论坛开幕式上，工业和信息化部 党组书记、部长李乐成便提到，要以人工智能、5G、工业互联网、云计算等新型技术赋能工业化，充 分释放数字技术潜力。 中国移动通信集团董事长杨杰关注到，AI赋能新型工业化正由单点突破进入协同推进、群体演进的关 键阶段。为充分释放AI应用潜能，有力促进数字经济和实体经济深度融合，急需构建更大范围、更广 领域、更深层次的产学研用创新联合体。 而在这一进程中，新兴基础治理体系的完善，尤其人工智能领域需要明确的规则正受到关注，在全球产 业数字化转型的背景下，各个国家、各个区域的政策法规要求不一致，也被部分企业视作推动开放合作 与创新驱动的一大挑战。此外，伴随着人工智能的高速发展，智能体也正逐步走入赋能新型工业化的视 线中心。 9月16日—17日，2025金砖国家新工业革命伙伴关系论坛在厦门举行。（冉黎黎/图） AI赋能进入协同 推进阶段 在2025年"人工智能+"被再次写入政府工作报告后，8月底，《关于深入实施"人工智能+"行动的意 ...

21世纪经济报道见习记者冉黎黎 厦门报道 9月16日—17日，2025金砖国家新工业革命伙伴关系论坛在厦门举行。人工智能是引领新一轮科技革命和产业变革的战略性技 术，在此次盛会上，围绕AI的讨论热度高企。论坛开幕式上，工业和信息化部党组书记、部长李乐成便提到，要以人工智能、 5G、工业互联网、云计算等新型技术赋能工业化，充分释放数字技术潜力。 中国移动通信集团董事长杨杰关注到，AI赋能新型工业化正由单点突破进入协同推进、群体演进的关键阶段。为充分释放AI应 用潜能，有力促进数字经济和实体经济深度融合，急需构建更大范围、更广领域、更深层次的产学研用创新联合体。 迪尔玛·罗塞芙提到，任何国家都无法独自弥补这些差距，知识共享、技术联合开发和基础设施共建是加快全球南方创新步伐的 关键。金砖国家具备引领这一进程的独特优势，它们具备规模庞大、多元的资源禀赋，通过合力协作可以加速成员国的发展， 降低对外部的依赖。"我们还应完善新兴基础治理体系，尤其人工智能领域需要明确的规则，确保透明度、公平性和安全性，这 些规则必须通过多边机制制定"。 而在这一进程中，新兴基础治理体系的完善，尤其人工智能领域需要明确的规则正受到关注，在全 ...

Core Viewpoint - Both Li Auto and ByteDance independently discovered a fundamental issue in the exploration of agents, leading to similar solutions and effects based on their respective business characteristics [2][4]. Group 1: Solutions and Algorithms - Li Auto's approach is more focused on efficient and practical engineering solutions, while ByteDance's method is supported by more formal and comprehensive mathematical theorems, considering all possible scenarios [3][27]. - Li Auto proposed the AWE algorithm, while ByteDance introduced the Entropy-Modulated Policy Gradients (EMPG) framework, which consists of two components: Self-Calibrating Gradient Scaling and Future Clarity Bonus [4][10]. - AWE focuses on supervised fine-tuning (SFT) within token-level adjustments, whereas EMPG emphasizes reinforcement learning (RL) at the step level, both addressing gradient issues caused by uncertainty [4][27]. Group 2: Key Components of Algorithms - AWE is designed to dynamically adjust the influence of each token on model parameter updates, allowing the model to learn easier tokens first before tackling more difficult ones [9]. - Self-Calibrating Gradient Scaling in the EMPG framework directly intervenes and calibrates the strength of learning signals based on the model's confidence in its actions [10]. - Future Clarity Bonus serves as an internal reward mechanism, guiding agents to choose paths that lead to clearer future states, thus enhancing learning efficiency [11]. Group 3: Insights on Learning Dynamics - The core insight from both companies is that there exists an undesirable coupling between the strength of learning signals (gradients) and the model's uncertainty state (entropy) [24][25]. - The EMPG framework focuses on the uncertainty at the step level, while AWE emphasizes the token level, with both approaches utilizing the model's internal feedback signals to guide training [27][28]. - Li Auto's AWE primarily addresses gradient size, while EMPG tackles both gradient size and credit assignment issues [6][27].

Core Insights - The article discusses the emergence of general-purpose AI agents in various consumer platforms, highlighting their integration into core products to enhance user experience and streamline decision-making processes [1][2] Group 1: AI Integration in Consumer Platforms - Major companies are embedding AI assistants into their applications, such as Taobao's AI assistant for multi-turn dialogue shopping and Meituan's "Xiao Mei" for food delivery and local services [1][2] - Alipay has launched the first "AI payment" service in China, allowing users to place orders and complete payments with a single command through the AI assistant [1] Group 2: Data Utilization and Personalization - AI assistants utilize user data, including order history and behavior information, to provide personalized recommendations, thereby enhancing conversion rates [2][3] - Taobao's "AI Universal Search" and Meituan's "Xiao Mei" both rely on user data to tailor search results and recommendations based on past interactions [2][3] Group 3: User Consent and Compliance - Platforms must obtain explicit user consent to utilize personal data for AI-driven recommendations, as outlined in their user agreements [3][5] - Compliance with regulations, such as the Personal Information Protection Law, requires platforms to inform users about data usage and provide options to opt-out of personalized recommendations [5][6] Group 4: Regulatory Considerations - The design of AI features must include easy opt-out options for users, as mandated by regulations governing algorithmic recommendations [6][8] - The article emphasizes the importance of user choice and transparency in data usage to mitigate regulatory risks and enhance user trust [8][9]

Core Insights - The article emphasizes that robots should be viewed as "AI colleagues" rather than mere tools, highlighting the importance of creating an efficient human-robot collaboration system to enhance operational efficiency and customer experience in the smart era [1][5][14] Company Overview - Cloudwise Technology, founded in 2014, has been deeply involved in the robotics field for 11 years, witnessing the industry's transformation from "machine tools" to "AI colleagues" [3][6] - As of 2024, Cloudwise robots have been deployed in over 30,000 hotels, completing 500 million service tasks, which is equivalent to traveling around the Earth's equator more than 500 times [3][6] Robot Capabilities - The company defines robots as intelligent entities with mobile and operational capabilities, driven by six dimensions of abilities: learning, adaptability, interaction, autonomy, emotional intelligence, and social integration [3][8][10] - Robots are designed to continuously optimize their strategies based on environmental feedback, maintain stability in various scenarios, and understand emotional cues from guests [3][9][10] Collaboration Strategies - Effective collaboration with AI colleagues requires focusing on four key aspects: clear objectives, controllable processes, traceable results, and responsive feedback [4][10] - Hotel managers set clear KPIs for robots, allowing them to identify subtle customer needs through operational data and drive system optimization through continuous feedback [4][10] Industry Applications - The human-robot collaboration model is applicable across various sectors, including hotels, hospitals, and factories, enabling intelligent upgrades and enhanced experiences [4][10] - The integration of embodied intelligence and disembodied intelligence is expected to create highly adaptable "industry intelligent agents" that can operate in specific scenarios [5][12] Future Outlook - The arrival of more "AI colleagues" should not induce fear regarding job value; instead, they should be seen as partners that extend human capabilities [5][14] - The article suggests that the era of robots will differ from the human population dividend era, as robots rely on research and development to optimize algorithms and enhance their capabilities [11][15]

Core Insights - The core argument presented by Zhou Hongyi emphasizes the transition from large models to intelligent agents in the context of artificial intelligence, suggesting that while large models serve as a foundational element, they lack the practical application needed for real-world tasks [3][4]. Group 1: AI Development - Zhou Hongyi asserts that the focus of AI has shifted from large models to intelligent agents, which are seen as essential for driving the new industrial revolution [3][4]. - He critiques the current use of large models, stating that they often function merely as chatbots or customer service tools, failing to address production and business challenges effectively [3][4]. - The evolution of AI is described as moving from knowledge-based large models to reasoning models, with a notable decrease in operational costs [3]. Group 2: Intelligent Agents - Intelligent agents are defined as specialized digital employees that transform general large models into tailored solutions, possessing capabilities such as memory, tool usage, collaboration, and task planning [4]. - Zhou outlines five levels of intelligent agents, from chat assistants to collaborative multi-agent systems, with the highest level (L5) being self-learning and self-evolving agents [4]. - The emergence of intelligent agents is expected to redefine roles within organizations, positioning humans as planners, leaders, and supervisors of these agents, leading to the concept of "mixed teams" of silicon and carbon-based employees [4]. Group 3: Empowering the New Industrial Revolution - Zhou advocates for a specialized intelligent agent approach to empower the new industrial revolution, claiming that the potential for intelligent agents in industrial applications is ten times greater than traditional software [5]. - The 360 Intelligent Agent Factory aims to develop L3-L4 intelligent agents, providing customizable solutions for small and medium enterprises, with 50,000 agents already available [5]. - The digital transformation of traditional industries is framed as a profound change, with intelligent agents being the key driver for future opportunities, and 360 is open to sharing its intelligent agent factory technology with BRICS nations [5][6].

Group 1 - AI is reshaping the world and human self-perception, but humans remain irreplaceable due to unique capabilities [1][2] - Humans possess three core abilities that AI cannot replicate: accountability, creativity and adventurous spirit, and self-drive [3][4] - Continuous self-drive, learning, and dialectical decision-making are fundamental drivers of human societal progress [4] Group 2 - The rise of AI has caused anxiety among people, with some rejecting it and others feeling overwhelmed [5] - Even if AI technology stagnates, existing models can still trigger profound societal changes, particularly with the emergence of AGI and ASI [5][6] - Humans must remain the decision-makers, especially in specific contexts where AI lacks understanding [6][9] Group 3 - Humans' accountability stems from their unique existence and ability to provide credible experiences, which AI cannot do [9][10] - Decision-making by humans is a concentrated act of judgment and responsibility, which AI cannot replicate [10][11] - The essence of human decision-making is rooted in personal experience and the ability to be held accountable [10][12] Group 4 - Creativity and the ability to take risks are uniquely human traits that AI cannot fully emulate [11][12] - AI can generate solutions based on existing information but lacks the ability to create entirely new concepts [12][13] - The ideal collaboration between humans and AI involves humans making final decisions based on AI-generated options [12][14] Group 5 - Continuous learning and self-driven exploration are essential for personal growth and adaptation in the AI era [28][30] - Reading is crucial for personal development, as it shapes neural connections and influences one's perspective [28][29] - The process of learning and searching for knowledge is scalable, leading to greater personal and professional growth [30][31] Group 6 - In the AI era, humans should embrace their role as decision-makers, particularly in micro-decisions that require personal accountability [23][24] - AI should be viewed as an enhancement to human capabilities rather than a competitor [25][26] - The future of work will involve leveraging AI to improve decision-making and creativity, allowing humans to focus on unique contributions [24][25]