[Trend Report] Why Public-Sector AX Is Slower Than Expected: Structure, Not Technology

Many public institutions are pursuing AI projects, but a significant number remain stuck at the PoC (Proof of Concept) stage. This stems not so much from problems with the technology itself, but from structural issues in data, organization, and infrastructure.

The main obstacles blocking the spread of public-sector AX are as follows:

• Limits in data utilization foundations: Lack of data consistency due to data being scattered across institutions/departments and a high proportion of unstructured data
• Infrastructure and platform constraints: Insufficient cloud infrastructure and platform environments to stably operate AI services
• Complexity of organizational structure and stakeholder interests: Difficulty in resource allocation due to complex decision-making processes and conflicting interests between departments

By contrast, private companies are building data governance, leveraging flexible cloud-based infrastructure, and rapidly advancing AX through practical application based on RAG.

For public-sector AX to actually spread into real services, a shift in the administrative paradigm, establishment of AI governance and ethics frameworks, securing of data infrastructure, and building a public-private cooperation system are essential. Ultimately, the next step for public-sector AX leads to establishing an infrastructure strategy that can stably operate AI services.