AI model predicts kimchi flavor, fermentation level

Offices of the World Institute of Kimchi in Gwangju / Courtesy of the World Institute of Kimchi

Korea’s leading kimchi research agency has developed a new artificial intelligence (AI) model that can predict how kimchi will taste based on fermentation conditions, including temperature and bacteria present.

The World Institute of Kimchi under the Ministry of Science and ICT said on Monday the new model can be used to introduce new scientific guidelines that can replace older, more subjective methods that rely on individual human experience and knowledge.

The agency believes the breakthrough will allow for more standardized and predictable taste in mass production kimchi, improving the ability of companies to cater to global consumers.

The AI model, jointly developed by a team of researchers from the institute and Chonnam National University, found nine key elements that help predict the fermentation progress and how the resulting kimchi will taste. The signal markers include different types of organic acids, sugars, amino acids and other compounds. Collectively, they help determine kimchi's unique flavor profile.

Samgwang Temple in Busan hosts a kimchi making event where volunteers prepare kimchi for donation to local anti-poverty groups, Dec. 10, 2025. Yonhap

The model was developed after lead researcher Whon Tae-woong innoculated cabbage with 10 different types of lactobacillus and monitored bacterial development in different temperature conditions to help the AI better predict bacteria growth and fermentation processes.

The institute said the AI model already tallies with existing statistical data about kimchi fermentation that is publicly available. According to the agency, the model can be implemented for mass production or other industrial purposes immediately.

“The new AI model will contribute to minimizing unexpected changes in kimchi quality, which often resulted from unbalanced fermentation. It will help uniform distribution of fermented kimchi with the degree of maturity preferred by a majority of global consumers,” Whon said.

“This breakthrough is a critical turning point for commercial fermentation, advancing from human experiences to prediction based on AI analysis. The standardized procedures for detailed control of environmental conditions will help strengthen K-food's consumer trustworthiness and global competitiveness.”

Whon's latest AI model was introduced in the March issue of the science journal "Food Chemistry."