SKT to Launch Telco LLM in June... "Applied to Call Centers Within the Year" (Comprehensive)

SKT Telecom Industry Specialized Telco LLM Briefing on the 30th
Collaboration with OpenAI 'GPT' and Anthropic 'Claude'
Customer Center Application Within the Year... Generative AI Platform Also Established

SK Telecom will launch a telecommunications service-specialized 'Telco Large Language Model (LLM)' that has learned the telecom company's services, products, memberships, and more in June. The plan is to apply this to telecom customer consultation tasks within the year to significantly improve work efficiency.

On the morning of the 30th, Eric Davis, head of AI Tech Collaboration at SKT, is giving a presentation at SKT's 'Telco LLM Briefing' held at T Tower in Jung-gu, Seoul. Photo by Seoyul Hwang chestnut@

On the 30th, SKT held a 'Telco LLM' briefing session at the Supex Hall in T Tower, Jung-gu, Seoul, announcing that the development of Telco LLM will be completed in June and it will be applied to call center operations within the year. Telco LLM is a telecom industry-specialized LLM that has learned telecom-specific terms such as 5G plans, T Membership, official subsidies, as well as internal guidelines of the telecom company including AI ethical values.

Jung Min-young, head of SKT AI Platform, said, "The completion of Telco LLM development by work unit is expected in early June," adding, "There are preparations such as testing and evaluation before internal application, so it is expected to be applied to the customer center within the year."

For Telco LLM, SKT collaborates with Open AI, Entropic, and others to collect data, select structured and unstructured data, train these on general-purpose models such as 'A.DotX', 'GPT', and 'Claude', and then perform fine-tuning by additionally training on telecom-related data.

Afterwards, Telco LLM undergoes reinforcement learning through human evaluation to verify if it provides useful answers, followed by model evaluation (benchmarking) on language ability, reasoning ability, and telecom-specialized task performance. If Telco LLM receives low scores in consultation type selection, related data will be further built and trained to advance the technology.

Eric Davis, head of SKT AI Tech Collaboration, said, "It will not be easy to solve the services and problems telecom companies want with a single general-purpose LLM," adding, "SKT's unique multi-LLM strategy is to create various Telco LLMs through fine-tuning and benchmarking tailored to telecom data and domain know-how, and to select and use them according to the situation."

Jung Min-young, in charge of SKT AI Platform, is explaining about 'Telco LLM'/Photo by Hwang Seo-yul chestnut@

SKT expects that customer consultation tasks at telecom companies will be greatly improved based on Telco LLM. This is because while the consultant is on a call with the customer, the LLM can assist by summarizing the consultation content and suggesting solutions. Generally, handling one consultation call takes about 3 minutes for customer consultation and more than 30 seconds for post-consultation processing. The Claude version of Telco LLM also has the advantage of accurately understanding contextual nuances such as AI ethical principles, Korean neologisms, profanity, and threatening language.

It is also useful for telecom network infrastructure operation. Since Telco LLM provides situation-appropriate answers based on equipment manuals, response cases, and other information, it can reduce response time compared to operators searching for information manually. In addition, SKT sees that Telco LLM can improve work efficiency in various areas of telecom operations including marketing and distribution networks, legal affairs, HR, and other internal tasks.

On this day, SKT also unveiled an 'Intelligence Platform' that enables telecom companies to efficiently build and develop multi-LLMs specialized for the telecom industry. This is an 'enterprise AI development and operation package' encompassing multi-LLM, multimodal, orchestration, and retrieval-augmented generation (RAG).

SKT is currently applying the Intelligence Platform to services such as A. (A.Dot) and plans to expand its application cases. Services that can be applied include customer center call bots, chatbots, distribution channel assistants, infrastructure assistants, and internal work innovation.

SKT expects that not only global telecom companies but also enterprises with similar work characteristics will be able to reduce costs and time through this platform. Jung said, "While discussing the technical requirements of Global Telco AI Alliance (GTAA) member companies, we are organizing technical details such as what requirements are needed in enterprises and how to prepare, and reflecting them in the platform."