The National Information Society Agency (NIA), an affiliate of the Ministry of Science and ICT and the Ministry of the Interior and Safety, announced on the 26th that it has published the 'Artificial Intelligence (AI) Data Quality Management Guidelines v3.5.' The guidelines, released in collaboration with the Ministry of Science and ICT, include information on the changes in artificial intelligence (AI) technology and industry observed during the 'super-large AI expansion ecosystem creation project' conducted last year.
Since publishing the AI training data quality management guidelines v1.0 in 2021, NIA has been enhancing the relevant content each year. The guidelines v3.1 released last year recorded over 60,000 downloads and are actively utilized to the extent of being referred to as a 'standard' in both public and private sectors.
This guideline presents new indicators and construction methodologies tailored to the characteristics of generative AI, including language models (LLM) as well as multimodal data and synthetic data. NIA noted, "We aimed to secure scalability so that we can flexibly respond to the rapidly changing trends in AI technology." This guideline from NIA is available for anyone to download and utilize from its website and the AI Hub.
NIA Director Hwang Jong-seong said, "The core of artificial intelligence technology is data, and the life of artificial intelligence depends on data quality management," adding, "NIA will continue to enhance and present AI data quality guidelines and standards to ensure the foundation for a high-quality data ecosystem is established to respond to rapidly advancing AI technology."