KAIST team develops AI communication app for minimally verbal autistic children and their parents

A Korean research team has developed an AI communication tool app that facilitates meaningful conversations between minimally verbal autistic children and their parents.

A diagram on how the AACessTalk tool works [KAIST]

A Korean research team has developed an AI communication tool app that facilitates meaningful conversations between minimally verbal autistic children and their parents.

Professor Hong Hwa-jung’s team from the Department of Industrial Design at KAIST created “AACessTalk,” according to a press release on Sunday. The tool was made in collaboration with Naver AI Lab and the Dodakim Child Development Center.

AACessTalk is a tablet-based AI communication system designed to foster meaningful dialogue that overcomes the limitations of conventional augmentative and alternative communication (AAC) tools, which have been limited to simple card-based communication that fails to fully reflect children's interests or subtle emotions.

To address this, the team developed a system that uses AI to recommend personalized word cards in real time based on the child’s interests and situational context, while offering parents specific conversation guides tailored to each situation.

The system also includes features that give children more control over the conversation via a button that allows children to choose when to start or end a dialogue. They can also press a “What about you, Mom?” button to ask for their parents’ thoughts.

Most children used these features to ask their parents questions for the first time in their lives, according to families who participated in the research.

From left: Professor Hong Hwa-jung of the Department of Industrial Design, doctoral student Choi Da-som of the Department of Industrial Design, researcher Park So-hyun at Naver AI Lab and lead research scientist Kim Young-ho at Naver AI Lab. [KAIST]

In a two-week field study involving 11 families, parents broke away from repetitive, routine communication patterns by using the AI-guided conversation suggestions and engaged in richer conversations. Children also expressed themselves more actively and took more initiative in conversations.

“I was surprised when my child used an unexpected word, and it helped me understand their language abilities more deeply,” said one parent.

“Creating an environment where children can express their own voices is more important than anything else,” said Prof. Hong. “This study confirmed that AI can not only improve the efficiency of communication but also serve as a tool to foster genuine connection and understanding within families.

“Our team will continue developing human-centered technologies that respect and embrace neurodiversity,” she said. “In particular, we aim to expand research focused on user experience and the practical application of technology for marginalized communities.”

This research was the result of doctoral student Choi Da-som’s internship at Naver AI Lab while studying at the KAIST Department of Industrial Design.

The research received the Best Paper Award at the ACM Conference on Human Factors in Computing Systems 2025 held in Yokohama, Japan from April 26 to May 1.

BY KIM MIN-YOUNG [kim.minyoung5@joongang.co.kr]