AI for Telcos: the acceleration imperative
AI soon becomes part of just about every conversation I have with our telco clients. Telco professionals understand that AI and the closely-related areas of machine learning and analytics now become critical to their business. What is not always quite so clear is how and where to apply them – and, most importantly, how to adopt rapidly.
I deliberately steer conversations away from the technology to begin with. It’s easy, for example, to get sidetracked talking about chatbots, natural language processing or any of the other technologies which people associate with AI. All this stuff is fascinating, but it’s not the place to start.
AI development and adoption is all about making business better - and that makes business, not technology, the opening topic.
Every industry has its specialties, and telcos are no different. There are professional divides between network operations, product development and delivery, and customer relations. These divisions are mirrored in the three layers of telco business models with the resource layer, service layer and product layer.
With AI, we are going to use mathematical models and statistics to identify patterns and solve problems – and we can only do this effectively if we are able to make connections between those layers.
In practical terms, what does this mean when planning and implementing AI initiatives?
Opening up the discussion
If AI is going to make a significant contribution, start by actively extending the dialogue. It certainly isn’t limited to the tech decision-makers, and it will always benefit from engaging experts from traditionally distinct telco functions.
We know, for example, that virtual and software-driven networks can only perform as required with a high degree of automation. So clearly, there is a clear need for AI and robotics to help achieve this. But the ability to anticipate load according to predicted uptake of a new product or service crosses the boundaries between resource, service and product layers.
In short, the ability to benefit from joined-up AI will only be as great as the willingness to establish open discussion in the first place.
We love thinking, but …
Quite rightly, telcos are impatient. Nobody can afford to indulge in extended academic discussion about AI strategy and adoption. Telcos want to draw the shortest possible line between making decisions and getting tangible results.
Telcos are at a distinct advantage here – thanks to the fact that, compared to other industries such as pharma and aerospace, they need to analyze and manipulate relatively small data sets. And it’s by establishing the means to analyze and act on network and customer data in realtime, that smart telcos hit the target in terms of product innovation, quality of service, and customer satisfaction.
Atos’ focus on use cases is driven by this desire to accelerate AI implementation and return for our telco clients. By providing a portfolio of AI models which are ready for practical experiment, we can give clients tangible evidence of benefit in weeks not months.
Whether we are importing actual network logs for analysis in NFV proposals or managing customer-consent with AI-driven engagement robots, our focus is on practical approaches to real-world problems.
With digital-by-design as our theme at this year’s Mobile World Congress, AI and analytics are going to be prominent themes. Not surprisingly, we will be maintaining a clear focus on how telcos can establish an effective business focus for their AI initiatives – and then making use cases the catalyst for accelerated adoption.