[Barcelona, Spain = DigitalToday reporter Jin-ho Lee] KT on Sunday unveiled results from its artificial intelligence business in the public and enterprise (B2B, B2G) sectors at MWC26 in Barcelona. It promoted its Korean-style AI transformation capabilities to the global market.

KT won the Supreme Court's project last year to build an AI platform to support trial work. Earlier, since 2023, it has developed and operated the AI-based teaching and learning platform Hi-Learning with the Gyeonggi Provincial Office of Education.

At its MWC26 exhibition, KT is focusing on proof-of-concept cases and the technology architecture for the two projects. The trial-support AI platform applies AI to core trial tasks such as case-law searches, issue analysis and document drafting to boost productivity. The project is being carried out in four phases over a total of 45 months. It is currently at a pilot open stage for the first phase, in the area of intelligent legal information search.

The trial-support AI platform applies a range of services including intelligent semantic-based search and research, summary and issue analysis, review and analysis of newly filed cases, support for drafting judgments and review reports, and extraction and de-identification of personal information. Of these, KT is showcasing the intelligent legal information search service in particular at MWC26 this year.

The service combines a vector database with retrieval-augmented generation technology to understand the context of legal inquiries and implement similar case-law searches and precise document-based question-and-answer functions. KT used two large language models, Mi-deum: K 2.0 and Llama K. KT pre-trained 20 billion pieces of legal-related data to enhance model quality. It is also building a database based on about 25 TB and more than 8.38 million legal documents, including judgments, complaints and preparatory briefs.

In education, it introduced Hi-Learning, a platform that supports teachers' digital lesson design and operation and students' self-directed learning in public education settings. Its flagship AI Diagnostic Recommendation function helps students diagnose their learning level and set goals on their own.

Based on diagnosis results, AI suggests personalised learning for each user. All learning data are stored and analysed in the Learning Record Store on the international standard xAPI, leading to personalised feedback. Through AI reports, teachers can assess students' levels and design customised questions. It also applies single sign-on, OAuth standard integration and the Education Ministry's Digital One Pass, enabling use of services with one integrated ID in a secure environment.

Youl-gyu Yoo (유용규), head of the Public Business Division at KT's enterprise unit, said, "We will expand services tailored to demand from public institutions by using various AI including Mi-deum 2.0." He added, "We will become a trusted AX partner in the public sector."