The appeal of a detective novel lies less in the brilliance of the detective's mind than in how accurately scattered clues are connected into a single truth. No matter how sharp the reasoning, a missing or wrongly linked clue can lead to an absurd conclusion. Artificial intelligence is no different. However smart AI becomes, it can produce plausible but wrong answers if the latest information is missing or if it bases its judgment on faulty data. In the end, what determines AI's performance and reliability is not model size alone but the data on which it makes judgments and answers.
That is why the axis of global AI competition is said to be shifting from "bigger models" to "better data." The same holds for national administration. However advanced the AI technology, it is hard to put to real use if policy, statistical and administrative data are scattered across ministries. Conversely, if data are systematically connected, AI can become a useful administrative tool that supports policy analysis and decision-making.
Yet gathering all public data in one place is not the answer. There are legal and security constraints, and each ministry bears responsibility for managing and verifying the data under its jurisdiction. Data copied and piled up in a single location also become harder to keep current over time. National data strategy now needs to be redesigned from the perspective of "connection" rather than "concentration."
MODI, the Model Data Intelligence platform for "AI for All" being pursued by the National Data Office, starts from this problem. MODI is not a project to draw data into the center but one to link it through "standards." It aims at a distributed foundation in which data stay where they are at each institution but can be safely connected and used at the moment they are needed, through standard methodologies and interfaces. In short, it connects rather than collects.
The key here is "standards." If each ministry classifies and manages data differently and uses its own connection specifications, data linkage easily remains a slogan. Just as a common 220-volt standard allows a wide range of home appliances to be used anywhere, the AI era needs common specifications to connect different sets of data. That is why MODI emphasizes standard methodologies and interfaces.
Once such a foundation is in place, a homegrown national foundation model can continuously reflect the latest policy materials, and "AI for All," a service for the public, can answer based on trustworthy policy materials. The path to supporting policy analysis by linking data from multiple ministries also widens. A system for analyzing statistical citations and news trends, which verifies the authenticity and accuracy of statistics cited in news articles, can also operate on this basis.
If roads and railways connected people and goods and served as the foundation of national growth in the industrial age, the foundation that connects data will be the new digital social infrastructure in the AI era. A country's AI competitiveness does not depend solely on who has the larger model. More important is whether needed data can be connected and used reliably, in time and securely.
Just as the truth of a case emerges when scattered clues are properly connected, data too create new value when they are linked to one another. It is time to move from "data that are collected" to "data that are connected." This quiet but solid power of connection will be a new starting point for South Korea's AI competitiveness.
