- SK Telecom will debut the telco-specific large language model (LLM) in June.
- Developed an ‘intelligence platform’ that facilitates generative AI applications for the telecom sector.
- Multi-LLM strategy fine-tunes AI for specific telecom functions.
- Telco LLM promises efficiency gains in customer care & network operations.
- Broader applications across marketing, legal, HR, and more anticipated.
- The intelligence platform offers multi-LLM support, multimodal functions, and retrieval augmented generation (RAG).
Main AI News:
In a bold move set to revolutionize the telecommunications landscape, SK Telecom is poised to unveil its pioneering telco-specific large language model (LLM) in June. Spearheading this initiative is the development of an innovative ‘intelligence platform’ tailored to catalyze the advancement of generative AI (GenAI) applications within the telecom industry.
Following a strategic roadmap laid out since last summer, SK Telecom has diligently pursued its “multi-LLM strategy,” aimed at crafting bespoke generative AI systems optimized for specific telecom functions. Suk Geun Chung, the operator’s Chief AI Global Officer, shared insights on this groundbreaking endeavor with TelecomTV during the Mobile World Congress (MWC) 2024 in Barcelona.
Having cemented its identity as an AI-centric entity since 2022, SK Telecom has meticulously trained its LLMs on datasets pertinent to the South Korean telecom sector. This comprehensive approach encompasses diverse facets such as industry terminology, tariff structures, contractual frameworks, subsidies, offers, and ethical considerations in AI deployment.
Central to SK Telecom’s methodology is the aggregation of telecom-centric data, followed by rigorous curation and refinement processes. Leveraging a “human feedback-based reinforcement learning (RLHF)” framework, the collected data undergoes meticulous evaluation culminating in a robust benchmarking phase.
Eric Davis, Head of AI Tech Collaboration at SK Telecom, emphasized the indispensability of tailoring AI models to suit telecom-specific exigencies. He underscored the need for nuanced adjustments informed by domain expertise and telecommunication datasets, encapsulating SKT’s distinctive approach to deploying a diverse array of telco LLMs.
Foreseeing tangible efficiency gains, SK Telecom anticipates streamlined customer care operations through the integration of Telco LLMs. By empowering customer service representatives with real-time AI assistance, the company envisions expedited query resolution and consultation processes, thereby enhancing overall service delivery.
Moreover, SK Telecom envisions leveraging Telco LLMs to bolster network operational efficiencies. Real-time query resolution capabilities promise to expedite troubleshooting and maintenance tasks, ensuring seamless network performance.
Beyond operational enhancements, SK Telecom envisages broader applications spanning various facets of telecommunication operations. Jeong Min-young, overseeing SKT’s AI platform, highlighted the potential for Telco LLMs to augment efficiency across marketing, distribution networks, legal affairs, HR functions, and beyond.
In tandem with its Telco LLM initiative, SK Telecom unveiled an ‘intelligence platform’ poised to redefine AI application development. Designed to cater to enterprises with analogous business profiles, this platform offers a suite of features encompassing multi-LLM support, multimodal capabilities, and retrieval augmented generation (RAG), empowering GenAI models to harness external data sources for enhanced functionality.
Conclusion:
SK Telecom’s groundbreaking initiatives signify a paradigm shift in the telecom industry, leveraging advanced AI technologies to enhance operational efficiency, streamline customer service, and unlock new avenues for innovation. As SK Telecom pioneers the integration of telco-specific LLMs and advances in AI application development, competitors are compelled to adapt or risk falling behind in an increasingly AI-driven market landscape.
