[Interview] Kim Tae-woo, Head of the Digital Center at the Jeju Agricultural Research and Extension Services

[Korea Agricultural Technology = Reporter Baek Cheol-hyun] “Numerous local governments and agricultural technology centers across the country, including the Korea Agency of Education, Promotion and Information Service in Food, Agriculture, Forestry and Fisheries, have come knocking on our door. They all envy us. But on the ground, we constantly struggle with the practical limitations of insufficient budgets and manpower.”

Kim Tae-woo, head of the Digital Center at the Jeju Agricultural Research and Extension Services, who is leading Jeju’s digital agricultural innovation and AX (AI transformation), made this remark while stressing that organizational and budgetary support is urgently needed to revitalize agricultural AX. Recognized for outstanding achievements and having swept up numerous institutional awards, Kim is regarded as setting the standard for digital agriculture in Korea. We met with Director Kim on the 14th.

Jeju differs from the mainland in both topographical characteristics and cultivated crops. What is Jeju’s differentiated strategy in the process of adopting advanced technologies and AI?

“Jeju’s farmland is fragmented into small plots, making precision mechanization difficult. So instead of hardware, we took the approach of densely securing ‘plot-level data.’ In particular, citrus gross farm receipts surpassed 1.3 trillion won last year. To protect this core asset, rather than relying on macro statistics from outside institutions, Jeju independently designated 320 sample plots. Every year in May, August, and November, surveyors have gone into the field and manually counted flowers and fruit to forecast output, but starting this year we are demonstrating a technology in which AI captures 360-degree images of citrus trees, converts them into 3D models, and then automatically counts the fruit. We are currently preparing a patent application.”

Pest and disease problems caused by climate change must also be serious. How is AI supporting the field?

“In the past, when pests were found in the field, it took at least 15 days for investigators to identify them visually and provide technical guidance. As a result, we often missed the golden time for control. To solve this, we installed ‘digital traps’ throughout Jeju. We built a system in which captured pests are analyzed in real time through AI information labeling, and farmers can immediately check the results on their smartphones through ‘JejuDA,’ an app we developed ourselves.” 

There is much criticism that no matter how good an app is, it is useless if farmers cannot use it due to the aging of rural communities. If there is a technology that stands out for its consideration of elderly farmers, what is it?

“The aging of farm households is an unavoidable reality. To conduct management analysis and make sound decisions, farming logs must be kept well, but elderly people find smartphone typing itself difficult. So last year we introduced a ‘voice-recognition digital farming journal’ function.

The problem was the ‘Jeju dialect (Jeju language).’ Existing AI engines from major corporations could not recognize Jeju speech at all. In the end, we endlessly trained the Jeju dialect based on Google’s engine. We have now reached a level where it can accurately convert the dialect spoken by farmers into text and video, and in the future we plan to expand it into a conversational AI system.”
 

What is the current distribution status of the ‘JejuDA’ app, and what data are most popular among farm households?

“Parcel-specific weather information in 500-meter grid units and the AI pest assistant are the most popular. Because the AI assistant responds based on training from specialized books, it has a lower misdiagnosis rate and higher reliability than ChatGPT.

This app initially had only about 300 registered users, but we found a breakthrough through the convergence of administrative data. Once we began accepting applications for farmer allowances and the happiness voucher for women farmers through this app, membership surged to 1,800. Going forward, we plan to integrate all document applications into this platform.”

It seems the work of integrating and refining agricultural data fragmented across the public and private sectors and among companies could not have been easy.

“Building the data hub was truly a painful process. That was because each institution had completely different standards (formats) for defining data, so even after collecting all the documents, they could not be used right away. At first, we prioritized collecting 34 types of data from 16 institutions, and this year we plan to expand that to around 40 types. Since this refinement work cannot be done by the power of the public sector alone, we are currently operating a ‘private-sector collaboration governance’ framework involving 22 private companies and research institutes to untangle the problem. Every day, I realize why convergence and collaborative cooperation are so important.” 

Where is the Digital Center’s focus for the second half of this year?

“It is on advancing the seven major practical services currently incorporated into the platform. In particular, by this coming October we plan to complete an observation system service project to expand the yield forecasting system to all seven of Jeju’s major crops. Ultimately, we need to move toward ‘Stage 3 fully intelligent AI,’ in which data make decisions on their own, but from that point we run into the wall of budget constraints.”

I believe budgetary and policy support must accompany this project if it is to lead to results.

“Our center is currently operated with an annual budget of about 3 billion won. However, to move to the next-generation model we envision—that is, large-scale platform advancement and the stage of receiving the government’s ‘agriculture and forestry satellite’ data to conduct time-series vegetation analysis of Jeju’s farmland—we need budget support on the scale of at least 8 billion to 10 billion won.

More important is ‘people.’ No matter how good the data collected are, they are useless without specialized personnel to analyze them. We are requesting the establishment of an ‘Intelligent Analysis Team’ dedicated to software and data within the institute. Field organizations and manpower must absolutely back this up.”