The analysis identifies four stable categories of concept objects: architectural methods, network systems, applications and services, and environments. Across these concept objects, two types of characteristics emerge. Activity-oriented characteristics describe how AI is integrated—whether as a peripheral tool, a core component, or a foundational element embedded across the entire lifecycle of design, deployment, operation, and maintenance. Result-oriented characteristics capture the performance consequences of integration, including autonomy, adaptability, efficiency, and real-time responsiveness.

Building on this dual structure, the study proposes an “activity–result” maturity evolution framework consisting of three progressive levels: AI-Assisted, AI-Enhanced, and Fully AI-Native. At the AI-Assisted level, AI functions as an auxiliary analytical tool, improving efficiency without altering the system’s decision authority. At the AI-Enhanced level, AI becomes a core component in multiple lifecycle stages, enabling partial autonomous interaction among system components. The Fully AI-Native level represents a qualitative transformation: AI is embedded as a foundational design principle from inception, enabling self-configuration, continuous learning, and zero-touch autonomous management. Here, AI evolves from a data processor into a decision-making and evaluative agent within the data intelligence ecosystem.

To validate the framework, the study examines representative AI-Native use cases in telecommunications, including collaborative multi-agent coordination and vertical industry integration scenarios. The mapping of conceptual characteristics alignment with use  cases verifies that deeper lifecycle integration of AI directly drives higher levels of autonomy and adaptive capacity. Importantly, the analysis shows that governance challenges intensify alongside maturity. As AI systems gain greater operational agency, traditional performance-based evaluation becomes insufficient. Instead, governance must increasingly prioritize semantic accuracy, decision transparency, and ethical quality.

The practical implications extend beyond telecommunications. As AI-Native architectures expand into smart grids, intelligent transportation, industrial digital twins, and generative AI services, they reshape the relationship between data, decision-making, and accountability. The structural separation between computational agency and semantic understanding—widely discussed in contemporary philosophy of technology—introduces new risks of automated yet semantically misaligned decisions. Therefore, governance models shall evolve in tandem with technological maturity.

The study recommends a tiered regulatory strategy aligned with system maturity and risk profiles. In high-risk physical operation contexts, such as collaborative intelligent agents interacting with the physical world, human-in-the-loop mechanisms remain essential to ensure meaningful human control. In contrast, in high-frequency and privacy-sensitive infrastructures, AI-Native regulation shall rely on endogenous mechanisms such as federated learning architectures, distributed audit logs, and dynamic knowledge bases to enable continuous digital supervision without compromising real-time performance.

By constructing a standardized conceptual system and a maturity-based evolution framework, this research provides a shared cognitive coordinated system for understanding AI-Native transformation. It shifts the analytical focus from fragmented technological applications to a structued integration logic and governance alignment. In doing so, it contributes not only to theoretical clarification but also to the development of operational evaluation tools for AI-driven digital transformation initiatives. As AI-Native systems increasingly function as autonomous actors within data intelligence ecosystems, establishing such conceptual and regulatory foundations provide harmonized approaches to ensuring trustworthiness, ethically aligned, and sustainable technology evolutions with AI.

About author: Kewei Zhang is a Ph.D. candidate specializing in interaction design. And a member of the International Telecommunication Union (ITU) Focus Group on AI-Native for Telecommunication Networks. His research focuses on human-computer interaction, AI-native systems, and standardization for artificial intelligence technologies.

Cite this article in DAKD as: Zhang Kewei, Yin Jing, Wen Fuquan, et al. AI-Native for Data Intelligence: Constructing a Conceptual System and Evolution Framework from an International Standardization Perspective[J]. Data Analysis and Knowledge Discovery, 2026, 10(1): 48-60 https://doi.org/10.11925/infotech.2096-3467.2025.0810