大华股份软件研发部副总裁周淼：AI技术正驱动企业数字化全面升级 | 2025 AI Partner大会

Core Insights - The year 2025 is anticipated to be a pivotal moment for AI applications, marking a significant technological transformation in various industries as the global AI race enters a "China moment" [1] - The 2025 AI Partner Conference, hosted by 36Kr, focused on the disruptive changes brought by AI applications across multiple sectors, featuring discussions on the emergence of the next groundbreaking AI super application [1] Industry Trends - The global AI industry is experiencing rapid growth, drawing parallels to the evolution of smartphones, particularly the transformative impact of the iPhone 4 in 2010 [3] - Current AI advancements are hindered by two critical conditions: cognitive ability and intelligent agents, which are essential for the next revolutionary phase in AI [3] Cognitive Ability in AI - Cognitive ability refers to AI's understanding of complex scenarios and abstract concepts, which is crucial for applications in various fields such as energy and security [4] - Enhanced cognitive capabilities allow AI to move from precise recognition to accurate understanding, enabling it to analyze dynamic behaviors and context [4] Development of AI Models - The company has introduced the Xinghan large model series, focusing on three main capabilities: - V Series for visual tasks, excelling in small target detection and complex scene recognition - M Series for multimodal tasks, integrating visual and language processing - L Series for language tasks, facilitating workflow management and task coordination [5] Intelligent Agents - The intelligent agent framework is categorized into four levels, ranging from basic Q&A support to fully autonomous agents capable of independently completing complex tasks [5][6] - These agents are designed to adapt to various industry needs, enhancing operational efficiency and decision-making processes [6] Practical Applications - In practical scenarios, intelligent agents have been integrated into management platforms to automate report generation and provide data-driven insights, significantly improving management efficiency [6] - In the energy sector, specialized intelligent agents enhance safety by monitoring worker proximity to hazardous equipment and generating incident reports [6][7] Challenges and Future Considerations - A major challenge in developing industry-specific intelligent agents is the vast differences in business logic across sectors, necessitating a flexible workflow engine for real-time adjustments [7] - As AI technology progresses towards autonomous intelligent agents, there is a need to rethink IT architecture, potentially positioning AI as a central component in information systems [7]