KT Holds 'AI Joint Research Results Workshop' with Seoul National University and KAIST

Research on Prompt Compression and Optimization Technologies with KAIST

Attendees taking a commemorative photo at the event. In the front row, the fourth from the right is Jaehyung Park, Senior Executive Director at KT AX Future Technology Institute; the seventh from the right is Moonsik Jung, Professor at Seoul National University AI Research Institute; and the eighth from the right is Kieung Kim, Professor at KAIST Graduate School of AI. Photo courtesy of KT.

On August 30, KT announced that it had held the 'Seoul National University-KT-KAIST Industry-Academia Joint Research Final Workshop' at the KT Umyeon Research Center in Seocho-gu, Seoul, along with research teams from Seoul National University and the Korea Advanced Institute of Science and Technology (KAIST).

At the workshop, a total of about 50 researchers from KT, Seoul National University, and KAIST gathered to share the final results of the joint industry-academia research conducted over the past year and discussed how KT could utilize the related technologies.

KT focused its collaboration with Seoul National University on the following areas: understanding and reasoning of contextual situations for autonomous agents; establishing evaluation standards and improving reliability for Responsible AI (RAI); securing Korean-specific datasets and developing behavioral prediction models; and creating a user feedback reinforcement learning framework.

In cooperation with KAIST, the research focused on prompt compression and optimization technologies. These efforts are expected to reduce the input processing load of large language models and lay the foundation for improving inference efficiency in AI models and agent services.

KT will work to verify and internalize the key research achievements obtained through this joint effort, with the aim of strengthening their practical application to models, platforms, and services. In addition, KT plans to identify follow-up projects that reflect business and technology needs and to expand R&D cooperation in fields such as agentic AI, RAI, physical AI, and token optimization.

Jaehyung Park, Executive Director of KT AX Future Technology Institute, stated, "Through collaboration with Seoul National University and KAIST, we have jointly developed core AI technologies essential for KT. Based on these results, we will continue to expand our R&D cooperative model that grows together with universities and ensure that our research outcomes drive KT's competitiveness in AI."

This content was produced with the assistance of AI translation services.