- Five telcos have formalized their joint venture to jointly develop telco-focused LLM
- The initiative aims to develop multilingual LLM helping telcos to improve customer engagement
- The GTAA is likely to face stiff competition
One of the highlights of the recently concluded TM Forum’s DTW: Ignite 2024 at Copenhagen was the formation of the Global Telco AI Alliance (GTAA). Several telcos, including e&, Singtel, Deutsche Telekom, SK Telecom and SoftBank, signed a joint venture (JV) to form GTAA to develop multilingual large language lodels (LLMs) and use artificial intelligence (AI) solutions to help service providers enhance customer experience in their respective markets.
This is important because the existing LLMs are unable to address the unique requirements of the telecom service providers. “An industry-specific LLM equipped with comprehensive understanding of our diverse product range will help us present a personalized offer that aligns with the preferences of our customers,” said William Woo, Group Chief Information Officer at Singtel, reported TM Forum’s blog.
“The existing generic LLMs fall short in addressing telco-related issues due to the complexity of local languages and industry terminologies. The telcos in the alliance will specifically collaborate and develop AI platforms tailored to address telecom-specific business issues, by leveraging extensive telecom data,” said Arun Menon, Principal Analyst at MTN Consulting.
This follows the announcement by the GTAA at Mobile World Congress in Barcelona earlier this year. All five partners will invest equal capital in developing telco-specific LLM to help telcos improve customer experience.
“The JV will look at deploying innovative AI applications tailored to the needs of the founding partners in their respective markets, enabling them to reach a global customer base of approximately 1.3 billion across 50 countries,” says the press release.
The LLM will be available in several languages, including Korean, English, German, Bahasa (official language of Indonesia) and Arabic, which will help the telcos improve customer engagement in their markets.
“The alliance aims to develop multilingual telco LLMs with a focus on 'localization'. Through localization of processes using Gen AI, telcos vow to eliminate language barriers and improve customer engagement in their respective operating markets. This is significant as current Gen AI tools have inadequate linguistic capabilities, mostly limited to the world’s dominant languages such as English, French, German, Spanish, Chinese, etc.,” said Menon. Incorporating local languages can go a long way in enhancing customer engagement as well as developing a culturally appropriate user experience.
Essentially, telecom service providers stand to benefit from AI and generative AI (GenAI) in two domains of network management and customer service. “In the network domain, these include topology optimization, network capacity planning, and predictive maintenance, for example. In the customer support domain, they include localized virtual assistants, personalized support and contact center documentation,” noted Menon.
Growing competition
The GTAA is likely to face stiff competition, not just from other telcos and vendors but also from its members who are developing their own LLMs as well. For instance, SK Telecom has developed A.X, trained on its own data to capture the dynamics of the local market. There is also nothing preventing the vendors from developing LLMs targeting the telecom industry.
“While telcos in GTAA are largely using in-house expertise and collaborations to develop their own Gen AI tools, many operators are likely to leverage pre-trained models from vendors that can be customized by feeding internal company-proprietary data to meet telecom-specific needs. That’s especially true for small- and medium-sized operators who are constrained by the lack of technical resources and workforce expertise,” explained Menon. The cost and speed-to-market concerns may push the service providers to opt for pre-trained vendor LLMs in their systems.
“One key consideration for telcos in opting for vendor-based models is to ease concerns surrounding regulatory compliance. Vendor-based models will often offer support for security and privacy requirements that are compliant with major regulations such as GDPR. For instance, Amazon Bedrock, which is a machine learning platform from Amazon Web Services (AWS) to build and scale GenAI applications with foundational models, is compatible with common compliance standards including GDPR,” added Menon.