南方财经 21世纪经济报道记者吴斌 上海报道 对于人工智能泡沫的担忧一度引发市场大跌，人们在忐忑中迈向未来。 澳洲会计师公会华东和华中区委员会会长、澳洲资深注册会计师金科近日在接受21世纪经济报道记者专 访时表示，人工智能产业目前确实存在局部过热迹象，但是否会演变成类似2000年互联网泡沫的系统性 风险，还需观察几个关键信号，如CAPEX增速拐点、债务融资占比、盈利兑现率、政策监管动向等。 综合考虑人工智能产业目前的发展状况以及与实体经济的融合度，金科认为，重现互联网泡沫时期的剧 烈破裂风险可能性较小，AI与互联网泡沫存在本质差异。 人工智能已经迈向现实，在千行百业生根发芽。澳洲会计师公会近日发布的《2025年商业科技应用调 查》（以下简称《调查》）显示，过去12个月，中国内地企业的人工智能应用率与应用成熟度实现同步 提升。65%的中国内地受访者预计，其所在企业未来12个月将进一步增加对AI技术的应用，较上一次调 查跃升17个百分点。 展望未来，金科预计，人工智能会更广泛地在不同行业加快垂直化深度发展，更紧密地与行业特性和企 业自身的商业模式结合。随着"十五五"规划的开展，"人工智能+"驱动的新质生产力发展会 ...

Core Viewpoint - The article discusses how artificial intelligence, particularly large models, is reshaping the financial services ecosystem, with a focus on how securities firms are embracing technological transformation [1]. Group 1: AI Implementation in Securities Firms - Securities firms are increasingly exploring the application of large models to enhance service efficiency and innovate business models, with significant progress being made [5]. - Shanxi Securities has integrated large models into its existing digital strategy, focusing on successful scenarios such as text generation and compliance retrieval, achieving a tenfold increase in efficiency for certain trading processes [7]. - Guoyuan Securities has developed a six-layer architecture centered on AI, encompassing everything from computational power to customer-facing applications, aiming to enhance various capabilities [9]. - Huafu Securities has allocated approximately 25% of its annual IT investment to AI-related initiatives and has established performance metrics to assess AI project implementation [11]. Group 2: Challenges and Recommendations for Regulation - As the application of large models deepens, there is a consensus on the need to improve regulatory frameworks within the industry [16]. - Recommendations include establishing a certification system for AI financial service capabilities, clarifying responsibilities and disclosure requirements, and promoting data usage standards to ensure customer privacy and data security [16][17]. - There is a call for industry collaboration to create a shared knowledge center and a data-sharing platform to enhance the capabilities of large models [17]. Group 3: Future Outlook and Industry Evolution - The rapid evolution of technology and its integration with business needs may lead to significant changes in service models and operational logic within the securities industry [18]. - There is an expectation that the next couple of years may see a disillusionment phase for large model applications, but they will continue to serve as powerful productivity tools [18]. - The future value of large models may lie in their ability to facilitate intelligent decision-making by abstracting various elements into logical entities and incorporating external market changes [19]. - The industry anticipates a shift towards AI-native applications and an increase in the use of domestic computational power, which is expected to surpass other heterogeneous computing resources [19].

Core Insights - The forum highlighted the application and challenges of AI large models in the securities industry, with several chief information officers from various brokerages sharing their experiences and strategies [1] Group 1: AI Model Implementation - Shanxi Securities has successfully integrated AI large models into specific business scenarios, achieving a tenfold increase in efficiency for bond trading by reducing response time from 30 seconds to 3 seconds [2] - Guoyuan Securities has established a six-layer AI empowerment system, focusing on the application of AI tools for investment banking projects, including intelligent verification and regulatory Q&A capabilities [2] - Huafu Securities allocates approximately 25% of its annual IT investment to AI, implementing performance assessments based on AI project usage and depth [3] - Southwest Securities has initiated AI large model exploration in 2023, establishing a dedicated digital transformation office and implementing applications such as intelligent knowledge bases and investment advisory assistants [3] - Guotai Junan Securities has adopted an "All in AI" strategy, enhancing employee understanding of AI through competitions and developing AI tools for client services [3] Group 2: Regulatory Framework - There is a consensus in the industry on the need to improve the regulatory framework for AI applications, with suggestions for a tiered certification system for AI financial services and clear responsibility definitions for AI services [4] - Recommendations include establishing data usage norms to ensure transparency and compliance when using customer data, as well as promoting standardization of AI technology to enhance service quality [4][5] Group 3: Future Industry Trends - The rapid evolution of technology is expected to significantly change service models and operational logic in the securities industry, with predictions of a "disillusionment period" for AI large models in the next couple of years [5][6] - The potential for AI large models to evolve into decision-making tools is highlighted, with the ability to abstract various elements and incorporate external market changes into algorithms [6] - The industry anticipates a shift towards AI-native applications and an increase in the use of domestic computing power, which is expected to surpass other heterogeneous computing resources [6][7] - The emergence of a comprehensive intelligent agent matrix is predicted, which could transform business models and ethical considerations within the industry [6]

Core Insights - China's artificial intelligence sector is once again in the global spotlight, highlighted by the WAIC2025 held in Shanghai, featuring over 800 companies and more than 3,000 cutting-edge exhibits [1] - JD.com showcased its upgraded model brand JoyAI, emphasizing its comprehensive intelligent solutions across various scenarios, indicating a shift towards deep application of AI in productivity [1] - The trend of embodied intelligence is expanding from traditional industrial applications to more personal and service-oriented roles, with JD.com making significant investments in this area [3] Industry Developments - The WAIC2025 exhibition area exceeded 70,000 square meters, marking a significant increase in scale and participation [1] - JD.com has made strategic investments in three leading companies in the intelligent robotics sector, indicating a strong commitment to this growing market [3] - The future of embodied intelligence is seen as having nearly limitless potential, with advancements in multimodal capabilities and physical interaction [4] Technological Trends - Embodied intelligence is defined as the integration of intelligence with physical machines, including humanoid robots and other forms [5] - Three core capabilities are essential for embodied intelligence: a controllable "small brain" for movement, a "big brain" for enhanced perception and interaction, and I/O (input/output) interaction capabilities [6] - The application scenarios for embodied intelligence are broadening, moving from industrial uses to more general life scenarios, such as companionship and household services [6]