[Young journalist Cheon Ye On] Korean researchers developed AI that finds out the slum areas only by satellite photos. It is a technology that automatically adapts in a new city, increasing its accuracy without the need of a person marking the location in advance, and is expected to basically change urban policy planning and public resource allocation in developing countries lacking sufficient data.

KAIST announced on the 6th that they made satellite photo based slum-detecting AI technology through combined research with Mee-Young Cha, School of computing professor and professor Ji-Hee Kim in Department of Technology Management and the research team, also with professor Jae-Seok Yang, from Chonnam National University.

This research was awarded with the Best Paper Award in ‘AI for Social Impact’ area in ‘The Association for the Advancement of Artificial Intelligence (AAAI) 2026’. It is the greatest fame that only 2 of the 693 research submitted in this area were selected, and it is an achievement that made sure that the Korean research team’s AI technology is on the top of the world not only in technology but in the social value area as well.

Until now, there were several studies using satellites to detect slums, the accuracy dropped extremely in new areas due to massive difference of building shapes and density. It was especially difficult for AI training because of the lack of non-automatically marked data of slum locations in lots of developing countries. 

The research team, in order to address this problem, trained an AI model to learn different area features, and implemented ‘Mixture-of-Experts’ (MoE) structure that automatically selects the best model whenever a new city is inserted.

The core of this research is ‘Test-Time Adaptation’ (TTA) technology. Without the need for humans to mark the new slum city in advance, AI reduces the error by itself by comparing and validating various model’s prediction results and trusting the areas that only match together. Like this, it maintained its quality in areas that didn’t have lots of data.

After the research team adapted this technology in major cities such as Africa’s Kampala and Maputo, it was proved that the new technology can intricately distinguish slum areas from the old one.

This technology is expected to be used in various policy areas such as △ urban infrastructure expansion planning △ early detection of areas weak on disaster · infectious disease △ selection of company to enhance dwelling environment △ UN Sustainable Development Goals (SDG) implementation check, and more.

Professor Mee-Young Cha said “This research showed that AI is not only a basic detection tool, but can contribute to solving real social problems even in data-lacking areas.” Professor Ji-Hee Kim revealed that “It will help in effectively distributing limited resources in the most needed area by reducing the cost of the field investigation.”

In this research, KAIST School of Computing PhD researcher  Lee Soo-Min and KAIST PhD researcher Seong-Won Back participated as Co-first authors, and the research result was presented in AAAI 2026 that opened in Singapore, January 25th.


▶원문 기사: KAIST, 전 세계 슬럼가 찾아내는 AI 개발..AAAI 2026 최우수논문상

◆이코노미스트가 글로벌 차세대 경제 리더, 청소년 기자단 영 저널리스트와 함께합니다. 영 저널리스트 기자단은 프리미엄 경제지 이코노미스트, 논술 전문 기관 Ni 에듀케이션과 함께 주요 시사 이슈를 팔로우업하고 직접 기획, 취재, 기사 작성 활동을 하며 사회적 문제를 고심하고 해결 방안을 제시하는 과정을 경험하게 됩니다. 이번 기사는 영 저널리스트 기자단이 이코노미스트 기사를 영문으로 번역하며 이슈를 팔로우업한 기사입니다. 차세대 글로벌 경제 리더, <영 저널리스트 with  이코노미스트> 영문 기사는 매주 연재됩니다.

ⓒ이코노미스트(https://economist.co.kr) '내일을 위한 경제뉴스 이코노미스트' 무단 전재 및 재배포 금지