Generative AI (Gen AI) is gradually becoming a trend in most industries in Asia and around the world. This innovation is also slowly being implemented in the telecommunications industry, which could potentially result in waves of transformation throughout the region.
In an exclusive interview with Telecom Review Asia, Alan Flower, Global Head for Cloud Native and AI Labs, HCLTech, spoke about the emerging trends of deploying GenAI in various industries, including the telecommunications sector and how this technology will impact the industry in the coming years.
What is the current status of Gen AI applications within the telecommunications industry in Singapore and Asia?
In terms of the impact of GenAI on telcos, one would expect the benefits to be evident in various regions. The observed trend is that companies across different industries are actively exploring Gen AI with intent. At this point, it is clear that clients in the telecommunications sector are already recognizing the potential advantages of deploying GenAI.
Clients engaged in exploration seem to possess a clear understanding of the expected outcome of their journey into GenAI. This ongoing exploration serves as a means for clients to reassure themselves that their anticipated outcomes align with the actual results.
During this early stage of the journey, most telcos are focusing on experimenting with what is known as horizontal use cases, including broad applications of AI across their entire business, encompassing network operations and customer-facing aspects. Several areas of experimentation within operators are notable at this moment.
Primarily, there is a considerable deployment of Generative AI in the realm of software engineering. This is viewed as a low-hanging fruit, with operators experiencing significant productivity gains. The adoption of AI-augmented software engineering is enhancing operators' capacity, quality, and velocity in delivering software-defined services to the market.
The second notable area of impact involves customer-facing use cases, particularly in the context of contact centers. GenAI, despite its intricate technical underpinnings, proves relatively easy to implement. Operators are experimenting with conversational AI, creating chatbots to alleviate the traditional contact center workload. This experimentation aims to enhance the consumer experience, providing better responses and efficiency.
Moving beyond contact centers, operators are exploring the sales and marketing domain. Early evidence indicates improvements in generating marketing materials using AI. Intelligent search and summarization, especially for complex contracts, has emerged as another area of interest, making sense of large and intricate documents.
In addition to these horizontal use cases, operators are considering GenAI-augmented automation, exploring the possibility of using AI to create and fine-tune automation processes. In the network domain, where complexity is high and reconfigurations are frequent, there is anticipation that Gen AI can bring about significant impact by automating tasks such as script creation for network reconfigurations.
While these areas of experimentation show promise, clients are cautious, seeking confidence before widespread deployment. The current focus remains on relatively safe use cases, such as software engineering and IT help desks, as operators navigate the early stages of integrating GenAI into their operations.
Could you provide more details on the current partnerships between the government and industry aimed at promoting the progress of Generative AI in Asia?
The observation regarding GenAI revealed that many use cases are shared among operators. When one applies AI to their contact center or tailors a large language model to their business needs, it often mirrors efforts undertaken by other organizations within the same industry. Gen AI presents an ideal opportunity for collaboration among operators on common use cases.
This collaborative approach does not negate competition; rather, it fosters a spirit of cooperation. The emergence of new AI alliances in the industry is an encouraging trend. Collaboration has historically been challenging for many telcos, as it isn't their natural inclination. Despite the complexities involved in running collaborative initiatives, Gen AI stands out as the perfect opportunity for operators to derive genuine benefits through collaboration. For instance, the prospect of fine-tuning large language models specifically for the industry could be a shared effort, as opposed to individual operators training their own models. The formation of collaborations, like the global AI Alliance, signifies a positive step in this direction.
Shifting our focus to the relationship between telcos and government, particularly in a heavily regulated industry, the impact of regulatory bodies and government directives becomes crucial. The telecommunications sector's history of tight regulation requires careful consideration of governmental influence. The encouraging aspect is that, at a country level, governments are beginning to issue regulatory guidance on responsible AI usage within society. This trend is visible globally, from the U.S. to the recent AI Act in Europe, indicating a commonality in emerging regulatory frameworks.
From the perspective of the telco industry, there appears to be no undue cause for concern. The regulatory landscape doesn't seem more challenging for operators than for any other industry. In fact, the work governments are doing to establish boundaries and frameworks is viewed as immensely helpful. Such guidance allows operators to focus their efforts more effectively during this experimental stage, ensuring they avoid wasting resources on irrelevant areas. There's a belief that these government frameworks, rather than being a source of fear for the industry, might actually accelerate collaboration between operators by providing a clear and shared set of guidelines.
How can Gen AI or LLMs be effectively incorporated in telecoms and what is the impact?
It appears that the incorporation of large language models (LLMs) into telcos is already underway. The journey of utilizing LLMs began with the notable impact of ChatGPT, which has been widely experimented with and perceived as somewhat magical. Based on an exceptionally large LLM, ChatGPT possesses the ability to provide answers to a wide array of questions. Operators emphasize the necessity for Large Language Models (LLMs), customized to comprehend their particular business and industry in an optimized fashion.
Looking at the challenges and opportunities, it becomes evident that each operator will leverage a diverse selection of LLMs, with no single LLM dominating their landscape. Specific LLMs will be employed for distinct use cases and purposes. For instance, in the domain of software engineering, co-pilots, driven by LLMs trained solely on software code, are successfully augmenting the productivity of software engineers.
In network operations and maintenance for telcos, the potential for creating conversational agents equipped with in-depth knowledge of a specific network is highlighted. Operators may seek to fine-tune LLMs based on their network data to optimize performance and configuration rapidly. The trend is shifting towards more specialized, efficient, and rapid LLMs focused on single domains, such as telco-specific LLMs.
The introduction of telco-specific LLMs is expected, with operators likely adopting them as foundational models for further fine-tuning. However, the increasing use of multiple LLMs poses a management challenge for operators. The concept of LLM operations (LLMOps) emerges, emphasizing the importance of ensuring that the intelligent software or contact center is guided by the most appropriate LLM and benefits from continuous improvements. Managing the quality of the AI foundation and deploying the right models, in the right context, within the operator's domain will be crucial in this context.
What challenges do you anticipate in the deployment of Generative AI in the telecommunications sector, and what strategies can be employed to address them?
In terms of the future, there will definitely be remarkable and unpredictable developments in the AI landscape. Reflecting on a substantial number of client conversations within AI labs, there has been a significant shift from experimentation to production, occurring more rapidly than anticipated. The prediction is that this trend will further accelerate, with clients deploying AI solutions into production at an unprecedented pace.
Currently, most organizations, including telco operators, embark on their AI journey in the cloud, leveraging platforms like OpenAI, AzureAI, Google, and AWS. However, there seems to be a growing inclination among operators to own their AI stack, while desiring greater control and innovation. This involves putting a management innovation wrapper around platforms like OpenAI and Google to enhance customization.
Furthermore, Gen AI deployments put emphasis on proximity to quality data, which is deemed crucial for successful AI journeys. And many operators are fortunate to possess abundant, high-quality data in their traditional data center estates. As experimentation begins in the cloud, the journey is expected to transition into an on-premises one. Gen AI is envisioned to become a hybrid multi-cloud reality, requiring operators to build and manage a rich ecosystem for a comprehensive AI environment.
Finally, high-quality data will play an indispensable role in a successful GenAI journey. Operators are anticipated to make substantial investments in effectively connecting their extensive data estates with emerging GenAI platforms. This integration is deemed key to supporting innovative solutions and new use cases, making the connection between an operator's data and the AI stack a pivotal area of investment in the coming year.
