Why ITU?｜为何该问题应由 ITU 讨论

中文版（可作为正式正文）

为何这一问题应由 International Telecommunication Union 予以讨论，根本原因不在于该问题涉及人工智能，而在于其所处的技术层级与治理属性，与 ITU 的历史职责高度一致。

从国际标准体系的角度看，ITU 的核心使命并非推动具体技术实现，也并非评判单一模型或算法的优劣，而是在多种实现路径并存的前提下，确立能够支撑全球互操作、可验证与可扩展部署的协议性框架。这一使命在通信网络的演进过程中已反复得到验证，从早期电信系统到 IMT-2000、IMT-Advanced，再到 IMT-2020，ITU 始终承担着定义代际边界与协调全球一致性的角色。

当前人工智能系统的发展，正逐步呈现出与早期移动通信类似的结构性特征：多种技术路线并行演进，不同组织和平台在能力与实现方式上高度异构，而系统之间的互操作性、可审计性与跨域治理能力却尚未形成统一基础。在此背景下，单纯依赖市场竞争或个别组织的技术选择，难以形成具备长期稳定性的全球性基础设施。

因此，本研究所关注的问题并非“如何改进人工智能模型”，而是是否有必要在模型之上、应用之下，引入一个可被标准化的协议层，用以定义最小一致的语义状态、交互边界与治理约束。这一问题天然具有跨组织、跨实现、跨司法辖区的属性，超出了任何单一企业或技术联盟的治理能力范围。

正是在这一意义上，该问题与 ITU 的制度定位高度契合。通过提供一个中立、包容且以共识为导向的讨论平台，ITU 能够促成不同技术路径之间的结构性对齐，为未来智能系统的互操作部署与长期治理奠定基础，而不预设具体实现或商业模式。

English Version (for Paper Body)

The relevance of this topic to the International Telecommunication Union does not stem from the fact that it involves artificial intelligence per se, but from the technical layer and governance characteristics at which the problem is situated.

Historically, ITU’s mandate has not been to advance specific implementations or to endorse particular models or algorithms. Rather, its central role has been to establish protocol-level frameworks that enable global interoperability, verifiability, and scalable deployment across diverse implementations. This role has been consistently demonstrated throughout the evolution of telecommunications systems, from early telephony to IMT-2000, IMT-Advanced, and IMT-2020.

Today, AI systems are increasingly exhibiting structural similarities to the early phases of mobile communication development: multiple technical approaches coexist, implementations are highly heterogeneous across organizations and platforms, and yet shared foundations for interoperability, auditability, and cross-domain governance remain underdeveloped. Under such conditions, reliance on market-driven convergence or isolated technical leadership is insufficient to support long-term global infrastructure.

Accordingly, the question addressed in this work is not how to improve AI models, but whether a standardizable protocol layer is required above model implementations and below application-specific logic—one capable of defining minimal shared semantic states, interaction boundaries, and governance constraints. By its nature, this question transcends individual vendors, technologies, and jurisdictions.

In this respect, the issue aligns directly with ITU’s institutional role. By providing a neutral and consensus-driven forum, ITU is uniquely positioned to facilitate structural alignment across heterogeneous implementations, thereby enabling future intelligent systems to evolve as interoperable and governable global infrastructure, without prescribing specific technologies or commercial outcomes.