[Policy Pulse] For Administrative AI, Define Where Not to Apply It
"Can AI Handle Welfare, Recruitment, and Fines?"
Success Depends on Accountability, Not on Numbers
Some Areas Must Remain 'Human Reserved Zones'
Artificial intelligence (AI) is rapidly establishing itself as the new standard in government administration. The government is accelerating the transition toward AI across all administrative sectors, going beyond its use in individual ministries. The Ministry of the Interior and Safety has announced a strategy to realize an AI-driven democratic government, and the Ministry of Science and ICT is also supporting the government’s AI transition by positioning AI as a core pillar of national competitiveness.
The direction itself is valid. AI is a powerful tool to reduce repetitive work, analyze data efficiently, and proactively deliver administrative services needed by citizens. Services like benefit notifications and citizen assistant platforms demonstrate the possibilities of AI-powered governance.
The problem arises when the extent of AI adoption becomes the sole yardstick for innovation. Just as the private sector uses the term "AI-washing," it is problematic to exaggerate the mere adoption of AI as if it were synonymous with innovation. If the number of systems increases, but employees’ workload does not decrease, and if the documents and procedures required from citizens remain unchanged, then AI is not an instrument of innovation but merely a new layer added to the existing bureaucracy.
Moreover, there are areas in administration that should never be fully entrusted to AI. In the Netherlands, an algorithm designed for welfare risk detection stirred controversy due to discrimination and privacy violations. In the United Kingdom, an algorithm used to determine college admission scores during the COVID-19 pandemic met with fierce backlash and was ultimately scrapped. Amazon, likewise, discontinued an AI recruitment tool that discriminated against female applicants.
The lesson is clear. If past biases are embedded in data, AI can perpetuate those biases more quickly and precisely, rather than eliminating them. In particular, decisions that have a direct impact on people’s rights and lives—such as welfare eligibility, permits, fines, recruitment, or promotions—cannot be made based solely on numbers. This is because exceptional circumstances, context, and value judgments must also be considered.
Ultimately, the core issue is accountability. The law for promoting AI-based administration, which went into effect in August this year, also stipulates that final responsibility lies with public institutions. When AI makes a wrong decision, who is responsible? Is it the developer, the adopting ministry, or the public official with final authority? If this question is not answered while simply expanding AI adoption, the technology could serve as an excuse for avoiding responsibility. The higher the risk of a decision, the more imperative it is to institutionalize human oversight, record rationales for decisions, and establish procedures for explanation and appeal. We should recall Bill Gates’ idea of a "Human Reserved" zone—leaving certain domains to humans.
What the government needs now is not just speed in adopting AI, but also clear boundaries on what will be entrusted to AI and what remains the responsibility of humans. Repetitive and rule-based work can be entrusted to AI, but decisions involving rights and responsibilities must remain within the human domain. The true measure of success for an AI-powered government should not be the number of systems built, but whether administration actually improves after adoption. The government should develop and release metrics that citizens can tangibly perceive, such as the rate of document resubmission, and discontinue projects that prove ineffective.
The more the government places AI at the forefront, the simpler the core question becomes—it's not about where to apply AI, but rather where not to apply it. Administrative innovation may start with adding technology, but it is completed by streamlining unnecessary processes and customs and by safeguarding areas where humans must remain accountable.
Hot Picks Today
"Two Spend 500,000 Won in a Flash, Yet Wait Two Hours?" ... 'Dreamland' Shunned by Kids and Foreign Tourists
- The Era of Getting 200 Million Won After 30 Years Is Over... A Japanese Bank's "Perpetual Department Head" Who Just Clocked In Cries Out
- Spain Hit by Its Worst Tragedy... Lee Kangin, Moreno and Real Madrid Join Soccer World in Mourning
- They Said, “Make It Your Home for Life” - Then Suddenly Kicked 200 Passengers Off... What Happened to the Around-the-World Cruise?
- "130 Rice Bowls Sold Out in a Day"... 'Unfiltered Face' Adds 8 Million Won to Night Market Sales: China's Bungeunwoo
Cheolho Oh, Professor Emeritus, School of Public Administration, Soongsil University
© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.