AI in Telecoms – a Valuable Journey

The last time I wrote a paper about Artificial Intelligence was 25 years ago. It was my Telecom Master’s thesis about the use of Neural Networks to improve the quality of digital images by post-processing after their compression, transfer and decompression. At that time, the power of the self-learning nonlinear algorithms was already well established, but their implementation required very large computing power. In order to complete my thesis, I had to live in Switzerland to use the Cray computer at the CERN!

25 years later, I am still in the telecom world and again addressing AI; but this time the combination of the exponential increase of data with almost unlimited cloud-based computing power opens possibilities. At this point, AI has the potential to impact almost every aspect of a communications service provider – and in the process, help traditional telecom players become true Digital Service Providers (DSP). While the potential appears vast, there are four specific areas where I believe AI stands to change the game. They are:

1) Customer Engagement: AI can improve real-time analysis of the customer-related data in order to identify buying patterns and predict future needs. This capability is critical for DSPs that want to propose the right service at the right moment (e.g. the “Digital Moment”) to their customers.

2) Customer Care: AI is used to power different chatbots and other customer interaction mechanisms, to address the requests for support and enhance self-care activities. It can also be used to automate the resolution of network and service issues, based on customer impact. AI can even prevent issues from arising by analysing continuously the evolution of network and services health and triggering preventive actions. This is all thanks to nonlinear algorithms that can represent the subjective perception of the customer experience with the DSP.

3) Operations: In Network and Service Operations Centres, AI is helping automate issues resolution by identifying root causes and triggering corrective actions. This can be applied in service assurance but also in end to end security management. AI can also help predict future outages and capacity constraints. I fully expect this capability will be used to automate the management of the 5G network slices and their hundreds of configuration parameters.

4) Networks: AI is already used in the most advanced Self Organising Networks (SON) applications, allowing mobile networks to continuously adapt – including adding needed capacity and making associated network configuration changes. This function will be even more critical with the coming roll-out of 5G networks. AI can also be applied in the virtual networks, with the auto-scaling of the Network Functions Virtualisation platforms and Software Defined Networks. Finally AI can also help design new networks such as 5G, allowing more complex nonlinear modelling of the network behaviours in case of design changes or subscriber and data growth.

Overall, AI can also be used to better predict customer satisfaction in the form of a projected Net Promoter Score (NPS), or a Customer Experience Index (CEI) which embeds natively NPS. CEI is relevant for all departments to empower service providers users with real-time ‘customer happiness scores’ in marketing, care, operations and networks, and also to automate business processes.

This applicability of AI to so many facets of the telecom space is also reflected in the way Nokia has embraced AI. It is not about the implementation of one big AI application that would be filtering, processing, analysing and exploiting all the network, services and customer data, but rather the introduction of AI algorithms in almost every application in the Nokia software portfolio. This progressive and growing use of AI in digital networks, digital operations, digital experience and digital intelligence applications is already delivering very concrete and measurable benefits to our customers including OPEX reduction through more efficient automation, new revenue streams by data monetisation, enhanced security and improved customer experience.

At Nokia we believe that AI implementation should be approached as a journey: at first it is important for an organisation to embrace the AI culture, being open to consider AI everywhere it could add value. Then by initially introducing AI to simplify the complexity of telecom operations by processing faster the growing amount of data. In a third stage, by automating some business processes. After that one could solve more complex problems through a non-linear combination of events, predictive models and root cause analysis. And the ultimate stage of the journey is to augment human intelligence. AI is not, and should not, be about replacing humans. Instead it is about augmenting the capabilities of the customer care agents, field engineers, sales representatives, network operations engineers to make the most of their time and better serve customers in the digital era.

What an exciting journey in front of us!

About the Author

Olivier Bruyndonckx is Vice President of Business Development and Sales Enablement for the Nokia Software Business Group. Olivier’s organisation has the responsibility to identify and qualify new business opportunities in close collaboration with the local and regional Nokia sales team, as well as to define the go-to-market strategies for the Nokia Software portfolio. Olivier started to work 24 years ago with AT&T as software engineer, and then pursued his career in Lucent, Alcatel-Lucent and finally Nokia, occupying various senior management and leadership functions in the software divisions, including roles in marketing, sales, delivery and operations. Olivier is based in Antwerp, Belgium, has a Master in Engineering from University of Louvain and a Master of Business Administration from the Boston University.