Today’s internet user interfaces are primarily designed for human browsing. The first step, therefore, should be to build an “AI-friendly” data system that is better suited to being read, understood and accessed by AI. By structuring content resources and turning them into organized knowledge, while strengthening ontology systems and knowledge graphs, such a system can improve AI’s ability to understand and apply information. On this basis, integrated hardware and software infrastructure combining computing architecture with data systems can be developed for AI applications.

　　AI-native teams: Reshaping organizations and workflows

　　AI is profoundly reshaping organizational structures and ways of working. AI-native teams should redefine job responsibilities, organizational structures and business processes around the A-To-A model and AI-friendly services, enabling “everyone to have multiple AI agents.” Professional capabilities and business experience can be encapsulated into “digital employees,” promoting autonomous learning by agents, multi-agent collaboration and the evolution of collective intelligence, thereby multiplying per-capita productivity.

　　AI for good: Putting values first, serving social development

　　AI also requires values-based guidance. Guided by the principle of “being people-centered and using AI for good,” efforts should follow the approach of “governing AI with AI and guiding intelligence with wisdom,” ensuring that AI always serves social development, the public interest and people’s needs. The “Mainstream Values Corpus,” whose development was launched by the People’s Daily in May 2023, has already begun to take shape and has been validated through numerous large-model applications.

　　Inclusive AI: Protecting cultural diversity, advancing global sustainable development