KT will share the results of joint research on core AI technologies with Seoul National University and the Korea Advanced Institute of Science and Technology (KAIST) and move to apply them to actual services.

KT said on Aug. 30 it held a "Seoul National University-KT-KAIST final results workshop for industry-academia joint research" at its KT Umyeon R&D Center in Seocho district, Seoul.

About 50 researchers from KT, Seoul National University and KAIST attended the workshop to share results from the past year of joint research and discuss ways for KT to use the related technologies.

With Seoul National University, KT studied contextual understanding and reasoning for autonomous agents. It also studied evaluation criteria for responsible AI (RAI) and reliability improvements, securing Korean datasets and developing behavior prediction models, and user-feedback reinforcement learning frameworks.

With KAIST, it conducted joint research on prompt compression and optimization technology. This laid a basis to reduce the input processing burden for large language models (LLMs) and improve inference efficiency for AI models and agent services.

KT will push verification and internalization to apply technologies obtained through the joint research to actual models, platforms and services. It will also identify follow-up tasks reflecting business and technical demand. It will expand R&D cooperation to agentic AI, RAI, physical AI and token efficiency.

Jae-wook Lee (이재욱), head of the Seoul National University AI Institute, said, "We will develop an industry-academia cooperation model in which academic achievements lead to AI technology innovation and strengthened industrial competitiveness."

Ki-eung Kim (김기응), a professor at the KAIST Kim Jaechul Graduate School of AI, said, "I hope the technologies secured this time will improve LLM processing efficiency and become a basis for expanding AI use in industrial sites."

Jae-hyung Park (박재형), managing director and head of KT AX Future Technology Lab, said, "In cooperation with Seoul National University and KAIST, we completed core AI technologies that KT needs together." He said, "We will expand the industry-academia R&D cooperation model so that the research results lead to KT's AI competitiveness."