Generative Artificial Intelligence (AI) is perhaps the biggest technology trend of our time, with business leaders and investors rapidly seizing the opportunities it can present. ABI Research's latest report, Generative AI Telco Strategies, discusses the case for generative AI in telecommunications.
It’s still early days for generative AI in telecom operations, but its potential to revolutionize the industry is impossible to ignore. In these early days, it’s clear that telcos are still hesitant toward company-wide implementation. Generative AI is almost exclusively being used by Communication Service Providers (CSPs) for low-hanging business functions, notably customer service, marketing, and sales.
However, as will be pointed out later, generative AI solutions can also support telcos’ ambitions to automate 4G/5G network management. The streamlining benefits of contextualized Large Language Models (LLMs) are especially welcomed by the telecommunications industry, which works in one of the most complex and data-driven industries. Generative AI may serve as a step toward greater automation by carrying out a range of analytical tasks, such as guiding network planning through natural language, summarizing network incidents, and recommending solutions. Generative AI-driven solutions also support network management by generating synthetic data, which can improve data quality for network modeling and expand the range of personnel who can be involved in network operations while maintaining data security. Overall, its potential for the network is increasingly acknowledged as telcos initiate Research and Development (R&D) in these areas.
Improving the Customer Experience
One of the earliest ways the telecom industry uses generative AI is to improve customer service engagements. While AI-driven chatbots are generally well-received, recent findings from AI platform provider Cyara indicate that they still frustrate 50% of consumers. Traditional AI-based chatbots struggle to effectively provide a resolution to complex inquiries. Generative AI can quickly overcome these challenges and take telcos one step closer to a fully automated customer chat service with natural language.
Deutsche Telekom Leading the AI Charge
As a prominent industry example, Deutsche Telekom (DT) launched its Ask Magenta chatbot in 2016 to assist customers. However, the AI-powered application could only accept limited information regarding service malfunctions. DT has since invested in generative models Anthropic Claude and Meta LLaMa to expand the use cases of Ask Magenta. These LLMs allow the chatbot to resolve billing and service issues, such as service requests, while handling customer data.
Of the many public LLMs available to the telecom industry, Anthropic Claude is especially well-suited for customer care. This is because Claude AI models are trained to detect a user’s intent and emotions. In this way, the chatbot can respond more appropriately and resonantly to network service requests. DT recently partnered with Anthropic (Claude 2) and Meta (LLaMa2) to develop a multi-lingual, telecom-specific LLM through the “Global Telco AI Alliance” alongside e&, Singtel, and SK Telecom. This generative AI-based platform will ensure telco chatbots are tailored to the specific needs of a CSP’s customers and potential customers.
The superiority of generative AI-based chatbots plays a key role in telco companies providing timely, context-based, and personalized responses to customers. This is all done with a human touch. Moreover, a good customer experience will keep the consumer on your website longer and increase stickiness. Ultimately, this will lead to more sales, as much as 67% more, according to Intercom's recent survey of 500 business leaders. Lastly, these chatbots reduce the workload for live human support agents.
Taking Marketing and Sales To New Heights
In terms of telecom business operations, some of the top use cases of generative AI are employee guidance, sales support, and content production for marketing. Generative AI-powered chatbots used for customer service can be extended to internal employee support. Service providers like Vodafone and Orange deploy dual customer and Business Support System (BSS) chatbots powered by generative AI LLMs. For example, Vodafone used the same Bard LLM to build both a customer service chatbot (TOBi) and a chatbot for employees (ASKHR). Meanwhile, Orange’s Google Cloud-powered LLM chatbot provides sales agents with a transcription/summarization of a call and offers follow-up recommendations. From customer journey mapping to lead identification, generative AI-based content has many benefits on the sales front.
As we’ve recently pointed out, generative AI excels at creating fresh content, such as text, images, and videos. This allows marketing teams at telco companies to streamline the process of prototyping new branding and advertising ideas. Besides content production, generative AI can also provide recommendations based on past marketing campaigns. To summarize, generative AI is being used by marketing and sales teams to accelerate the time spent creating campaigns and optimizing outreach strategies.
ABI Research forecasts as much as a 40% productivity increase for telecom companies using generative AI in customer care and BSS applications.
Figure 1: Generative AI Use Cases in Telco Network Operations
(Source: ABI Research)
Streamlining Network Operations
So far, this article has discussed generative AI use cases that can technically be applied to any industry. However, the technology can also be used to enhance network operations, which is unique to the telecommunications industry. While the potential for generative AI automating cellular networks is undeniable, its application remains in the Proof of Concept (PoC) stage. Network operations are far more complex and prone to risks than customer service or business operations use cases.
Telco network operators are primarily using generative AI for producing summaries and recommendations regarding network orchestration or coding. AI can identify network anomalies, provide intent-based development or code assistance, automate incident response, and provide step-by-step support for field maintenance workers. LLMs that excel in mathematical and logical reasoning, such as Palm, are the best options for automating network operations.
However, it should be noted that network operations outputs are based on exceptionally complicated datasets and nontransparent procedures. For these reasons, network operators refrain from fully automated network management and instead must oversee the guidance provided by LLMs. In other words, telco network operators cannot take everything the generative AI suggests at face value. Outputs must be scrutinized and verified for assurance.
For now, network operations is a more nascent area for generative AI in telecom. Once LLMs become more trustworthy, the technology has immense potential to automate network optimization and predictive maintenance (e.g., anomaly detection). Generative AI currently sits between manual operations and fully automated networks. According to ABI Research forecasts, telcos that leverage generative AI for network design, optimization, and testing can increase productivity by 15% to 25%.
Will Generative AI Disrupt the Telecom Industry?
From the telecom perspective, generative AI has strong potential to be an industry disruptor. As CSPs deploy AI-based workloads in the cloud, other companies like hyperscalers, network vendors, and open-source contributors will play a more prominent role in network management. These companies will service assets that telcos require for generative AI enablement. More network assets will be controlled by non-CSP entities, such as cloud infrastructure providers. Moreover, the CSP monetization models and business strategies will inevitably veer toward AI-driven, intelligence-based offerings and services (e.g., network slicing).
Applying AI to business is a convoluted process, with most projects failing to deliver a Return on Investment (ROI) across industries. Telcos may face an even lower success rate, considering the complexity of cellular networks. There are many challenges that service providers must be aware of as they aim to adopt generative AI, such as the risk of business strategy misalignment, building “explainable” LLMs, and customer data protection. For further guidance on the latest AI advancements in the telecom industry, download ABI Research’s Generative AI Telco Strategies report.
