Starting Tomorrow: AI Training Data Construction Project Open for Bids... 150 New Types Added This Year

[Asia Economy Reporter Seulgina Jo] The Ministry of Science and ICT announced on the 12th that it will begin a public contest for the ‘AI Training Data Project,’ which involves building and opening a large-scale repository of high-quality AI training data essential for AI development.

This project aims to build 1,300 types of AI training data by 2025, which are difficult for companies, researchers, and individuals to construct individually due to time and cost constraints, and to open them through AI Hub (www.aihub.or.kr). The process of building training data accounts for as much as 80% of AI development projects, making it a crucial task.

Over the past four years, the Ministry of Science and ICT has built a total of 191 types of data in fields such as Korean (speech and natural language) and healthcare, and this year plans to add 150 types of AI training data. These consist of 130 designated public contest projects in six core fields including healthcare, transportation and logistics, disaster, safety, and environment, and 20 free public contest projects (10 regional projects and 10 free projects).

The AI training data built through this project is expected to secure competitiveness in leading AI technologies such as language and visual intelligence, support innovation in industrial fields like autonomous vehicles, and aid in the development of AI services closely related to citizens’ lives, such as improving public health and safety. Additionally, 170 types of data built last year will be sequentially opened in the first half of this year after quality verification.

The Ministry of Science and ICT aims to establish a foundation for AI industry development through this project, which involves a large-scale budget of 292.5 billion KRW, while also mitigating the ongoing job crisis caused by COVID-19. To this end, direct employment by executing organizations is mandated (at least 2.4 persons per 100 million KRW), and the budget allocation evaluation will actively reflect the adoption of crowdsourcing methods that allow any citizen to participate in the data construction process. Furthermore, evaluation will include growth support measures such as training for crowdsourcing participants from executing organizations to encourage career development and stable employment of participants.

The Ministry of Science and ICT plans to provide job information on crowdsourcing platforms for each project on the AI Hub website (www.aihub.or.kr) so that citizens wishing to participate in data construction for this project can check it all in one place. Detailed information related to the project contest will be announced on the Ministry of Science and ICT website (www.msit.go.kr) and the National Information Society Agency website (https://www.nia.or.kr). Institutions and companies wishing to participate can apply by forming consortia for each project.

This content was produced with the assistance of AI translation services.