Faced with the need to improve operational efficiencies quickly process an ever-growing volume and variety of data, and prepare their networks for the imminent rollout of 5G, mobile operators everywhere are embracing Artificial Intelligence (AI) technologies. Some operators do not believe in taking baby steps.
According to a survey conducted at this year’s Mobile World Congress in Barcelona, 93 percent of telcos believe AI will be a ‘game-changer’ for their business, and 76 percent will have incorporated the technology into their business to some degree. Such is the level of interest that global telecoms industry investment in AI-related software, hardware, and services is expected to reach US$36.7 billion by 2025.
Many of the key players in the market are already exploring innovative ways of using AI. Verizon’s Exponent initiative, for example, uses AI and machine learning (ML) techniques to help operators analyze and monetize huge tracts of data, while AI plays a significant role in the virtualization of AT&T’s Domain 2.0 software-defined network.
A fundamental concept is sometimes lost in all the excitement, however. With telcos implementing AI in a bid to innovate, lower costs, and boost efficiencies through automation, there is no argument that AI has considerable potential for transforming the network. But AI needs to be trained and nurtured until it reaches full maturity; otherwise, it will only ever remain as a “potential”.
Operators therefore need to have realistic expectations of just what adding AI to their network will really mean for them over the coming years. It is important to learn to walk – before you can run.
The pace at which the telecoms sector has advanced in recent years means that humans are no longer able to manage the processes required to meet the demands of customers and the industry at large. As part of the resulting digital transformation of telcos’ IT systems and business processes, AI will play an increasingly important role in addressing a number of issues across various aspects of the ecosystem.
It can be applied, for example, as a means of improving the deployment and operation of ever more complex mobile networks. To this end, ETSI, the European Telecommunications Standards Institute, is working on the definition of a context-aware AI system that is able to automatically adjust the services offered by an operator based on changes in user needs, environmental conditions and the operator’s business goals.
Operators can also use AI to improve their margins. By combining current and historic customer data such as purchase patterns and usage history with information from ERP, BSS and OSS systems, operators can apply AI algorithms to the highly detailed insights to create individual plans for each of their subscribers. These plans can then be integrated with CRM systems to allow sales staff to offer them to customers, thereby improving the customer experience as well as optimizing and increasing margins.
AI can optimize the subscriber experience too. 24/7 customer service chatbots, such as Vodafone’s AI-based assistant TOBi, can handle a range of user queries and provide instant responses and resolution to many different issues. By using historic information such as service ticket and network log data to recognize fault models, these chatbots can identify common solutions for connectivity problems, allowing for quick and easy resolution. What’s more, by identifying which failure patterns require a technician to be dispatched and which don’t, they can help to avoid the need for on-site maintenance, resulting in significant cost savings.
While solutions such as these will improve an operator’s efficiency and cost-effectiveness, however, it’s worth remembering that none of them are plug-and-play fixers to all of a network’s problems. Instead, they will take time and require human input before they’re fully functional.
Training, feeding and nurturing
For its implementation to truly succeed, any AI employed by an operator should be considered as an additional part of the organization, and its capabilities must match those of the wider team within which it sits.
The organization’s data-gathering and management processes must be aligned to meet the needs of the AI, for example. Indeed, given the importance of data to the technology’s functionality, it’s perhaps unsurprising that the number of data scientists has risen by 650 percent since 2012, with the biggest growth seen in the particular role of machine learning engineer, preparing the data ready to feed the AI and ML’s algorithms.
As with any new member of a team, AI must be trained. It must learn how the wider business operates and, in turn, the business itself must adapt its processes to align with those of its latest additions. A mutual process of continual improvement will eventually see ever closer alignment over time and, once the AI is able to handle most of the organization’s heavy lifting, the wider team will have more time available to focus on key business drivers.
AI is not child’s play
The level of investment in AI in the telecoms industry shows confidence in the promise of AI to address a number of the challenges faced by the sector today. However, while it’s true that AI has the potential to be more efficient, more comprehensive, more accurate and faster than human operators – it’s still a relatively nascent technology and will not be smarter for some time to come.
Until that time, operators must work patiently to train and nurture the technology, adapting their own business processes and feeding it the data it needs. Stay positive in the belief that, in time, AI will mature and transform their business for the better.