The government's science and AI policy is moving from "setting strategy" to "delivering results". After laying the groundwork with a national AI strategy and expanded infrastructure, it is now focusing on delivering results by embedding future technologies such as physical AI and agentic AI in industrial sites.

The Ministry of Science and ICT and other relevant ministries said on Wednesday that 12 meetings have been held through Aug. 24 since the first ministerial meeting on science and technology on Nov. 24 last year. The meeting is a pan-government body that coordinates science, technology and AI policy, with the deputy prime minister for science and technology as chair and participation from relevant ministries.

Early on, it was about "setting the stage" - from AI and AX strategies to GPUs.

AI was at the centre of policy from the first meeting. The first meeting discussed the "10 AI projects for people's livelihoods" as well as defence AI transformation (AX), manufacturing AX and a science and technology-by-AI national strategy. The focus was on positioning AI as a growth strategy for the entire country rather than the responsibility of a specific industry or ministry.

A month later, the second meeting detailed the infrastructure and industrial strategy to support that push. The government set directions for distributing about 10,000 high-end graphics processing units (GPUs) it had secured. It also put on the agenda an AI semiconductor industry leap strategy, a national AI bio strategy, a network strategy for the AI era and measures to train AI talent for the labour market.

It was a phase of laying the foundation across the ecosystem, from computing resources needed for AI development to semiconductors, networks and talent. Later meetings continued infrastructure support to back earlier strategies, including allocating GPUs to national AI projects.

This year, work began in earnest to specify a pan-government AX 추진체계 and national-level technology goals. The fourth meeting in January dealt with a "Korean Genesis Mission K-Moonshot promotion strategy", a one-stop, full-cycle support plan for government AX projects, measures to expand the use of public works for AI training and a comprehensive information security plan.

After the K-Moonshot and data strategy, the focus shifted to "creating results".

From March, the direction took shape to select technologies needed at the national level and broaden the foundation for AI use into the public sector.

The fifth meeting confirmed 12 missions for the "K-Moonshot" project. It sets ambitious technology goals to solve national challenges and concentrates cross-ministry capabilities. It also discussed measures to strengthen AI utilisation capabilities for all citizens and make AI use routine, presenting a policy direction to pursue technology development and expanded adoption at the same time.

In April, the meeting's key theme moved further from "strategy" to "results". The seventh meeting presented an "advanced R&D commercialisation system strategy" to connect government research and development results to the market. It decided to pursue startups based on research results and to introduce investment-type R&D. It also discussed applying technology to actual demand, including AI care and public-sector AX.

The eighth meeting brought data and manufacturing sites to the fore. Along with the direction of data policy in an era of a major AI transformation, it put on the agenda support measures for manufacturing AX that incorporate "tacit knowledge" - such as skilled workers' experience and judgement - into AI. It also discussed building an AI materials R&D platform and support measures for national strategic technologies.

It was a shift from merely setting strategy to preparing the data, on-site know-how and tools for commercialising research results needed for AI to work in practice. At the time, Deputy Prime Minister and Science and ICT Minister Bae Kyung-hoon (배경훈) said, "It is time to show the public the results of government AI investment and support," and signalled he would closely review each ministry's AX progress and future plans.

In May, the ninth meeting reviewed pan-government AX progress while discussing a strategy to secure core competitiveness in physical AI and the "Manufacturing AI 2030 Strategy". It also checked AX progress and on-site difficulties in 10 areas including manufacturing, healthcare and agri-food.

R&D policy also shifted toward spreading results. The 10th meeting discussed expanding investment-type R&D so that outcomes generated with government support can be reinvested into R&D. It also covered follow-up steps on research security and on abolishing the project-based system (PBS).

Physical, agentic and autonomous-driving AI put pilots and commercialisation front and centre.

Recently, the execution stage has been moving down to more concrete industries and technology areas.

The 11th meeting last month covered a physical AI port strategy and an agentic AI initiative as major agenda items. It decided to push the automation of port cargo-handling equipment and apply physical AI technologies to Jinhae New Port and other sites. It also decided to promote industrial pilots of agentic AI and link them with a regulatory sandbox. It also prepared measures to supply additional GPUs to major AI projects including "AI for Everyone".

At the 12th meeting on Aug. 24, a "roadmap for piloting and commercialising autonomous-driving AI" for early commercialisation of domestically developed autonomous-driving AI was put on the table. The focus was on creating a pilot base that can verify technology on actual roads and linking it to commercial services.

It also discussed a strategy to revitalise deep-tech startups. It aims to overhaul support systems so technologies developed in laboratories can lead to startups, investment and market entry. It also prepared South Korea's AI ethical principles and moved to build a trust base needed for the spread of AI.

The industry agrees that it is time for the government to connect its science and AI strategies to concrete results. With policy moving beyond strategy-setting and foundational work into pilot testing and commercialisation, the success or failure of future government AI policy is likely to be decided in the field.

Bae also stressed at the 12th meeting, "The true result of AI competition lies in implementing AI that is used in industrial sites and in people's daily lives, beyond developing excellent technology."