[Exclusive] National LLM to Be Developed by 2025... "Innovation in Public Services"

Promotion of Private and Public AI Model Adoption
Enhancing Public Service Satisfaction
Improving Civil Servants' Work Efficiency

Will the day come when an artificial intelligence (AI) chatbot instantly answers administrative information that citizens are curious about, and the handover of public officials is automatically carried out, reducing the difficulties faced by businesses suffering from regulations?

The government has confirmed that it is discussing the implementation of a national-level large language model (LLM) to innovate administrative work.

[Image source=Yonhap News]

According to the Digital Platform Government Committee (DPLG) on the 5th, the government is in consultation with private companies developing their own AI technologies, such as Naver and SK Telecom, aiming to complete the introduction of LLM by next year. The plan is to implement a Korean government LLM similar to GPT by utilizing domestic companies' technology and infrastructure. The AI model trained on data scattered across various ministries can be used in diverse ways, including public services and innovation of public officials' work, according to the government's vision.

A DPLG official said, "We are conducting a final review to confirm the introduction of LLM as a government policy," adding, "Since this requires investment from the private sector and is the government's first attempt, everyone is taking a cautious approach."

The government's introduction of LLM is notable as an attempt to integrate AI into public administration. It must learn vast content, including not only press releases distributed by the central government but also laws, ordinances, and regulations. The more high-quality data it learns, the smarter the AI becomes. There is an endless amount of data available for learning, such as presidential speeches, ministry briefing materials, and various statistical information.

However, if sensitive government documents must be handed over to private AI companies, there is concern about security issues arising during this process. Since this project requires close public-private cooperation, it is impossible for companies to simply participate in government bidding and win contracts.

The government is also considering utilizing the PPP (Public Private Partnership) strategy, which Daegu City has jointly promoted with private companies and is regarded as a successful model.

Unification of Public Portals and Customized Services

The government aims to achieve two main effects through the implementation of LLM. First, by unifying public portals, AI chatbots can provide customized information. Until now, citizens had to visit each responsible ministry’s website and search individually to find the administrative information they wanted. This will change to a format where an AI chatbot for public services answers questions immediately. The government expects to significantly increase satisfaction with public services.

When users log in to the public portal, the first page can be changed to customized menus or guidance tailored to the user's needs.

Innovation in Public Officials' Administrative Work

The administrative work of public officials will also become more efficient. Until now, electronic records produced by ministries had to go through a complicated process, passing from document management systems to records management systems and then to the central permanent records management system. This has been pointed out as a cause that makes inter-ministerial collaboration and communication difficult. Therefore, the government’s plan is that by implementing a government-specialized LLM, the entire process?from document processing and records management to AI learning of administrative data?can be solved in one cycle.

If administrative processing is automated and intelligent, companies appealing for regulatory relaxation are expected to welcome it. One of the difficulties for businesspeople is personnel changes every 2 to 3 years, where the responsible public official changes. Each time the head of the department or division changes, they had to visit government offices and explain the background for regulatory reform requests from the beginning. In the case of ‘complex regulations’ involving multiple laws and systems, they had to go through the trouble of dealing with several ministries.

A DPLG official explained, "Once the government LLM is implemented, the handover of public officials can be automatically carried out," adding, "If the industry has made regulatory improvement requests and there are records of reviews, it can immediately move on to the next step."

Advanced Countries Are Already Introducing AI into Administration

Some countries are already using AI to streamline government administrative work and save budgets. They have also established guidelines for the safe use of public AI.

According to the Korea Data Industry Promotion Agency, UK Deputy Prime Minister Oliver Dowden formed an AI task force (TF) at the end of last year to innovate public services. He recruited 30 AI experts and announced plans to prioritize applications such as improving the efficiency of medical services, speeding up refugee and immigration application processing, and eradicating welfare system abuse. Some analyses suggest that this attempt aims to reduce the number of UK government officials, which has expanded due to Brexit and COVID-19.

In January, the Irish government approved guidelines for using AI in public services. According to the EU AI Act, all AI tools used in public services in Ireland must comply with seven ethical principles, including personal data protection, non-discrimination, and accountability. Simon Coveney, Minister for Enterprise, Trade and Employment, said, "AI technology will greatly enhance the potential productivity of public service agencies," adding, "The approved guidelines will protect individual rights and provide safeguards for responsible innovation, supporting public service agencies."