Embedding AI in All Aspects of a Telco’s Operations
AI presents significant opportunities in managing customer interactions and delivering best-in-class customer experiences. Gartner even predicts that by 2025, 80% of customer service and support organizations will apply AI to improve both agent productivity and overall customer experience (source). For instance, AI is already used to augment agents in customer service interactions, leading to faster response times and more personalized service. Also, AI-driven marketing tools generate personalized customer communications instead of mass market campaigns, leading to increased efficiency in the marketing department and higher conversion rates.
In the area of network management, AI can be used to predict and prevent network failures, optimize network performance, and automate routine tasks. This not only improves the reliability and efficiency of the network but also frees up human resources for more strategic tasks. Fully automating Level 1 NOC (Network Operations Center) is an example of key objective, and the rewards for the telco can be very high in terms of increased reliability and quality of service, as well as lower costs of operation.
While the AI opportunity for telcos goes beyond those two business areas, it is more interesting to reflect on what it takes to become an AI-native Telco.
First, a dedicated governance and operating model to be able to deliver in a cost-efficient and standardized way across the organization.
Second, a deep transformation of processes and skills, as AI will be full embedded in ways of working, replacing some roles, creating new ones in addition to augmenting employees.
Third, the ability to choose the fit-for-purpose models depending on the business areas and expected benefits.
Lastly, building the right security & compliance capabilities to limit the risks related to AI solutions as they are becoming core capabilities.
To become an AI-native telco, it is crucial to be efficient at designing, developing, and managing AI applications. An AI-native culture and a clear understanding of the priorities are essential to drive adoption. It is also important to make the best use of the scarce resources and skills available. This is best achieved with a hybrid approach, where local initiatives and ideas are encouraged and promoted, integrated within a central governance, in charge of aligning with the organization’s strategy and driving cultural change.
For delivering AI applications, a central AI Factory model has proven to be efficient, especially at the beginning, to facilitate mutualization and re-use, and to increase the learning curve and sharing in the organization. Distributed AI Factories, at department-level or country-level, can make sense within a decentralized organization, but they must be federated by a central governance, to really mutualize and share resources, solutions and knowledge.
While today the first changes brought by AI may be significant, they are still incremental. In the short run, most existing roles will remain but with the need to learn new skills. Agents in a call center are learning to use AI for inquiry response and information gathering. Software developers are learning to make efficient use of coding copilots. And indeed, 47% of global organizations assert that investments in skills and digital training will be an important investment throughout 2024 to meet the changing operational dynamics with AI and GenAI (source).
In the long run, when a Telco becomes more AI-native, there is less of a need for employees to execute and operate actions but more to oversee workflows with a higher degree of automation – whether that be customer experience, marketing campaigns or network operations. It is estimated that by 2027, 40% of current job roles will be redefined or eliminated across global organizations due to GenAI adoption (source). For example, as AI-driven bots handle more customer interactions, human agents have less chats or phone calls to answer. Instead, the organization should develop the skills and the processes to review, supervise and analyze outputs produced by chatbots. This means a key competence in an AI-native telco is the ability to monitor robotic and AI-driven processes, internal facing and customer facing, understand the dynamics, react to change and tune those solutions for maximum efficiency.
A much-debated topic is the need to develop a telco-specific Large Language Model (LLM). Smaller models tuned to a specific vocabulary, or certain types of conversations, have indeed proven to be able to outperform larger general-purpose models. Bloomberg has chosen to develop its own BloombergGPT, trained on financial documents, and has found it “outperforms similarly-sized open models on financial NLP tasks by significant margins — without sacrificing performance on general LLM benchmarks” (source). This could be particularly relevant for applications like customer service, where the vocabulary is quite specific. On the other hand, the cost and difficulty of such an initiative are daunting and the benefits not proven yet, and thus some operators have chosen so far to stick to general purpose models – such as Vodafone who wants to keep the flexibility to choose from the offerings on the market (source).
While AI-driven solutions have the potential to revolutionize telco operations across various domains, without proper regulations, there is a risk of suboptimal implementation, affecting service quality, fairness, and public perception. For instance, AI algorithms can optimize network resource allocation, predict network congestion, and improve overall performance. However, poorly designed or biased models might lead to network disruptions or inefficient resource utilization. AI-powered chatbots and virtual assistants can streamline customer interactions. Yet, if not regulated, they may provide inaccurate information or fail to address complex queries, impacting customer satisfaction.
Becoming an AI-native telco requires a strong focus on AI security and compliance. Indeed, failure to address these aspects could disrupt operations and lead to significant financial losses. AI systems are vulnerable to attacks, adversarial manipulation, and data breaches. Telcos must secure their AI infrastructure, models, and data pipelines. Encryption, access controls, and continuous monitoring are essential. Additionally, AI-specific security audits should be conducted to identify vulnerabilities. New regulations, such as the European Union’s AI Act (source), are shaping the AI landscape. An AI-native telco must deeply understand these regulations and align its practices accordingly. The AI Act emphasizes risk assessments, transparency, and human oversight. By adhering to these guidelines, telcos can confidently embrace AI while maintaining high standards of safety and compliance – and thus protecting their business as well as their public image.
Becoming an AI-native telco involves deliberately applying AI within the organization. It requires a transformation in business processes & skills, the right governance & delivery model, the ability to choose fit-for-purpose models and a strong commitment to security and compliance. By embracing those changes, telcos can fully leverage the opportunities presented by AI and become truly AI-native.
