The increasing capability of computer processing power and recent advancements in cloud storage have made the once sci-fi concept of Artificial Intelligence (AI) an enticing reality. Many industries are currently exploring how to best leverage AI, and the telecommunications industry is no exception. In fact, I wrote about the promise of better Wi-Fi in particular in a previous blog. This article goes beyond Wi-Fi to cover converged operators.
A recent survey from TM Forum (and an excellent paper in my opinion) suggests that in the telecommunications industry, there is a strong desire to begin implementing AI solutions, but no consensus on the best path forward. There are a number of fledgling AI use cases, ranging from network management to predictive maintenance, but an even greater number of challenges to AI adoption. This post will discuss these opportunities and challenges, as well as some strategies for Communications Service Providers (CSPs) to begin adopting AI solutions.
Current uses of AI in telecommunications
Some CSPs have begun implementing AI solutions in several ways, through the help of AI products and services. Some of these services aim to transition Network Operation Centers (NOCs), where human administrators manage the telecommunications network), to Service Operation Centers (SOCs), where analytics and AI deliver closed-loop automation. Telefónica, for instance, has launched pilot SOCs in several markets.
Other CSPs are using AI in different ways. AT&T, for example, is researching how to use AI algorithms to enable drones to inspect and repair base stations. SK Telecom in South Korea is using machine learning (a common type of AI) to analyze network traffic to detect abnormalities and enhance network operations. Hong Kong telecom PCCW is testing AI-powered tools to forecast growth in network capacity and predict network failures.
Opportunities for AI in telecommunications
While CSPs have already begun to explore the potential of AI, there are many more opportunities to exploit. For example, AI solutions can be used to aid in the development of a commercial strategy. This could take the form of generating ideas for products and services, identifying market trends, supporting operation decision making, predicting future trends, and much more.
There are two general categories in which AI can provide unparalleled opportunity: network management and operations, and customer centricity. It is clear that network management would benefit from—in fact, may require—AI solutions to cope with the quickly expanding number of connected devices and users. But even the customer-facing end of CSP businesses stands to benefit enormously from AI techniques, as customers will increasingly demand personalization.
Some of the ways to improve customer centricity with AI involve intelligent customer care. Data and AI analytics could be used to help customers with billing, device onboarding, troubleshooting, and other services to prevent churn. AI could also be used for intelligent marketing, personalizing offers to win back customers, determining platform engagement, OTT upselling, and more.
A final customer centric AI solution is the chatbot. Chatbots are automated customer support, intelligently resolving issues without requiring a human operator. Chatbots benefit both CSPs, who can reduce the number of employees required, as well as customers, who can get personalized help with their issues immediately. Thirty percent of the CSPs surveyed by TM Forum said they’ve deployed chatbots, with one in seven CSPs having redeployed some customer service agents to higher-value tasks.
Some CSPs, including SK Telecom, Telefónica, and Orange, have already begun to move past the chatbot by introducing voice assistants. Consumers will already be familiar with the idea of voice assistants, having quickly grown accustomed to Siri, Alexa, Google Assistant, and others. For CSPs, voice assistants can provide the highest level of personalized, friendly support for customers.
Challenges of implementing AI
Despite the numerous benefits of AI and a strong desire to obtain them, CSPs are having difficulty rolling out effective AI solutions. Due to the relative infancy of AI technology, there are a number of implementation challenges affecting all interested industries, with some challenges specific to the telecommunications industry.
Perhaps the most significant obstacle to AI adoption is a lack of maturity in the technology. With an ever-growing number of commercial AI solutions, it can be hard to keep pace with the best solutions for your business. In fact, 66 percent of CSP respondents indicated a lack of maturity in their top three obstacles to AI.
Another big barrier for AI is a lack of expertise, both in AI software as well as data analytics. While a limited supply of AI experts poses a challenge to all industries, the telecommunications industry may be particularly vulnerable to this problem. Without the proper incentive, computer scientists and data analytics experts may prefer other industries in which to apply their skills. This may mean that CSPs will ultimately have to find partners to help roll out their AI solutions.
Many other concerns are inhibiting AI adoption. 43 percent of CSPs worry about losing control of their network if in the hands of AI algorithms. 37 percent of CSPs are concerned about the lack of standards in AI technology. Finally, 32 percent of CSPs are worried about the possibility of displacing staff—while this is, on the one hand, a clear benefit, it also presents CSPs with the unwelcome prospect of cutting many jobs.
Strategies for AI adoption
With the potential benefits of AI as significant as they are, it is imperative for CSPs to adopt AI solutions as early as possible. By getting an early foot in the AI door, CSPs can position themselves competitively in the changing telecommunications landscape. Fortunately, there are a number of strategies CSPs can use to begin adopting AI technology.
Firstly, CSPs should learn as much as possible about the current possibilities of AI technology. This can entail seeking possible partners, experimenting with AI solutions, and joining standards groups and other relevant bodies.
CSPs should also pave the way for AI solutions by rethinking their data. Data is the seed of artificial intelligence, so CSPs should do all they can to remove siloes and open their data via APIs. By standardizing data collection and management, CSPs will be much better positioned to do business with AI partners and begin using AI effectively.
Finally, CSPs should welcome AI collaborators and open themselves to new and experimental solutions. While it’s not necessary to fully engrain yourself with any particular partners, it is important to keep your options open and explore a variety of AI possibilities. CSPs shouldn’t forget that AI is still in a nascent period, and exploring many options is the best way to discover what works best for your business.
