[Policy Pulse] Four Key Issues on the Path to an AI Government

Calling for an AI-driven Administrative Shift,
But Institutional Frameworks Remain Inadequate
Systems and Philosophy Must Keep Pace with Technological Progress

It is encouraging news that Minister of the Interior and Safety Yoon Ho-jung announced at his first press conference after taking office that he would consider the concept of an artificial intelligence (AI) democratic government. This signals a commitment to actively pursue government innovation using AI, drawing on South Korea’s accumulated experience as a global leader in e-government. However, it must also be clearly acknowledged that, compared to countries like the United States and China-which lead the AI industry ecosystem-South Korea’s vision for an AI government remains in its early stages.

The government has not yet revealed the specific extent to which it will pursue administrative innovation through AI. If AI technology is simply added as an extension of e-government, this would amount to administrative advancement rather than true innovation. Some examples have emerged, such as the National Tax Service’s AI-based system for selecting tax audit targets and the public assistant chatbot 'Gubbi.' However, it is important to avoid excessively highlighting individual projects in a way that creates the impression that a full-fledged AI government is already in place.

A genuine AI government requires a bold transformation that goes beyond automating repetitive and standardized work, demanding a complete reconfiguration of the administrative system around AI. To achieve this, a mid- to long-term strategy, sustained execution efforts, and the role of competent civil servants with creative imagination are essential. Above all, there are several urgent and critical issues the government must address as it transitions to an AI government.

First, there is the problem of data silos and disconnection between ministries. Data is managed by each ministry according to different standards and formats, resulting in an environment where AI cannot effectively perform integrated analysis. This slows down real-time decision-making and policy response.

Second, there is the risk of unexplainable administration. If the criteria and reasons behind AI-driven decisions are not clear, citizens will be skeptical and lack trust in those judgments and outcomes. It is also essential to establish a transparent system that allows citizens to verify results in the event of errors.

Third, there is the issue of citizen exclusion due to the digital divide. Many elderly people and low-income individuals are not familiar with digital devices. If the AI government does not take their needs into account, social inequality could worsen as these groups fail to benefit from technological advancements.

Finally, the lack of AI ethics and legal and institutional frameworks is also a serious problem. When AI intervenes in administration, clear legal standards must be established regarding the legal validity of its decisions and how to determine responsibility in the event of errors. This will be a crucial foundation for building trust between the government and citizens in the future.

The AI government is not simply a technological issue. At its core, it is closely tied to the philosophical question of what kind of relationship the government will form with its citizens through data. South Korea’s AI government is currently undergoing a trial phase. While technology is advancing rapidly, the systems and philosophies to support it are still lagging behind. Rather than being fixated on the legacy of having one of the world’s best e-governments and creating a superficially impressive AI government, it is hoped that the focus will be on building a truly reliable government for all citizens-one that earns their satisfaction and trust.

Oh Cheolho, Professor Emeritus of Public Administration, Soongsil University