Generative AI (GenAI) brings groundbreaking opportunities to telecommunications. It was one of the pivotal topics of discussion at Digital Transformation World (DTW) 2024, as Dr. Sudhir Kumar Mittal, EVP & Chief Architect of Jio Platforms and Anshul Bhatt, Chief Product Officer, OSS Business Unit of Rakuten Symphony, shared their insights on the unique challenges faced by telecom companies in its development and deployment. The conversation, moderated by Abe Nejad, Publisher of the Network Media Group, explored the intricacies of implementing GenAI in a highly regulated and data-intensive industry.
Telcos’ journey to GenAI adoption requires surmounting numerous challenges around AI models. The discussion underlined the need for a closed-loop system where data is not only analyzed for insights but also used to enhance and refine the AI models continuously. Various other factors, such as the financial implications of using GenAI, need to be considered, and cost-effective workarounds, such as balancing between predictive AI and GenAI or large language models (LLMs) should be onboarded to manage massive cost overruns. In the case of a large telecom operator with an extensive customer base generating vast amounts of data daily, the challenge lies in defining and executing the right use cases for GenAI.
"AI-assisted coding and automation are areas that will change the way the telecom industry works.”
- Anshul Bhatt, Chief Product Officer, OSS Business Unit, Rakuten Symphony
Similarly, while GenAI can revolutionize customer network use cases, it must do so within the bounds of regulatory compliance and data governance. Security and scalability are critical themes in a sector where data protection is paramount. Telcos must ensure that AI implementations adhere to stringent security standards.
“It is important to make GenAI output more reliable, traceable and accurate. It should be capable of providing genuine information to the customer and doing exactly what the organization wants it to do.”
- Dr. Sudhir Kumar Mittal, EVP & Chief Architect, Jio Platforms
A focal point of conversation was the differing challenges faced by enterprises and telecom companies in leveraging GenAI. While commonalities exist, enterprises place a higher emphasis on data partitioning and privacy. Ensuring that data remains secure and that AI models do not inadvertently use sensitive user data for training is a significant concern for enterprises. As enterprises increasingly rely on telco platforms beyond their connectivity needs, operators must make sure that their models learn contextually and grow into “large action models” that are capable of decision-making and execution – all within the confines of regulatory data protection norms.
Making GenAI outputs reliable, accurate, and aligned with organizational policies will be one of the priorities for enterprises and telcos in the future. This alignment is crucial to prevent AI from generating responses that could contravene company guidelines or mislead customers. An approach where AI agents and sub-agents are tasked with specific roles and grow more intelligent and effective over time will help.
The discussion also covered Rakuten Symphony's application of GenAI for critical use cases, such as AI-assisted incident management. These applications have led to substantial improvements in operating efficiency, with AI significantly reducing the time needed to resolve incidents.
Watch the full interview here.
