Operationalising AI in Telcos
By Pratik Bose, Head of Mobile Big Data and Digital Solutions, BT
Operationalizing AI in Telcos depends on three things- understanding the importance of business context in decision making, having a clear adoption and implementation strategy.
Context awareness is key for Telcos
In our environment of constant change, Telcos are seeing something very interesting. Increasing network scale & complexity combined with further service abstraction is making it increasingly difficult to understand real customer experience. But at the same time, some of these very changes are also making the network more intelligent, autonomous and smart. Combination of this new network paradigm and advanced analytics presents better opportunities to understand and manage our network. AI and ML techniques allow us to understand and predict few things better-
-Where are our customers and what services they are using?
-How can we assure the right coverage and quality?
-How can we make our network more reliable from reactive to proactive management?
Embedding this context in decision making allows Telcos implement closed loop planning, design and management. And helps deliver the right customer experience and perception, while aligning its investment decisions and commercial strategies.
Journey of Adopting AI
AI and ML is a journey from the world of descriptive analysis (what happened in the past and why) to a world of understanding trends and predicting a potential future behaviour. AI supports three major functions:
1) “Identify” the current patterns and behaviour from a large amount of data (big data).
AI is vital for delivering our next generation automated, self-optimising, self-healing networks
2) Analyse the trends and time characteristics to “Predict” future tendencies.
3) “Execute” an optimum plan which is based on the identified/ predicted trends.
The above techniques can be applied in multiple planning, design and operations process. And it is important to chart a clear adoption journey based on key priorities and challenges.
Trend analysis and anomaly detection in usage, performance and capacity utilisation across the network can deliver proactive identification of issues and bottlenecks. Network virtualisation and software defined orchestration has created a new opportunity for leveraging AI for automation and self-optimisation of networks. Use of real-time data and advanced analytics can help identify anomalies, predict issues and deliver automated corrective actions. Pro-active optimization such as traffic shaping, dynamic capacity management can help deliver the right experience and service perception.
Advanced marketing and customer lifecycle management is another domain where AI techniques can help deliver context based personalised interactions.
Implementation of AI: From F1 to A Family Sedan
Sorry, not going to dwell much on Formula1 but think of how technology developed by experts in F1 eventually find their way into our mass market family sedan. Similarly, to operationalize AI, it needs to be made mainstream; available to large user base and making it easy to use. AI should empower our operational and business teams with proactive and dynamic insights, enabling faster and more accurate actions. There are some key challenges though, when it comes to leveraging the power of AI.
• Telcos often have multiple islands and pockets of data-preventing it from creating a holistic view of customer and network context. Creating the data pipelines and joining data sets can be the boring bit but involves significant effort
• Need for both data science and data awareness-every organisation needs the brilliant minds of data scientists to create the amazing algorithms, and insights. But to become truly data driven, Telcos will need to make AI easily accessible and consumable to its operational, engineering and support teams. Organisations will need more people who are data aware and have the skills to apply insights in their day to day jobs
• And finally, we need the right tools for the right jobs- focus sometimes is on the cutting-edge data science toolsets and the latest AI/ ML frameworks. But AI operationalisation fundamentally depends on delivering the insights to our users in a manner that is easy to understand and use in the decision-making processes. Algorithms help create the gold nuggets of insights, but the tools required to embed these in our operations, engineering teams and customer service agents have different dynamics to the AI/ ML data science toolsets.
AI is vital for delivering our next generation automated, self-optimising, self-healing networks. This journey can only be made possible if Telcos can enable its people with the right skills and tools. Manage the people and skills part well, and we take a giant step toward operationalizing AI!
Leading and Getting Started with AI
Fredrik Backner, Telia Company
Data Analysis has Come of Age But There are Still Challenges Ahead
Dr Rex Davis, Chief Data Specialist, dunnhumby
Creating Superior Commercial Performance in Retail Banking through Advanced Data Analytics
Bjørn Büchmann-Slorup, Head of Portfolio Management & Pricing, Danske Bank
Capitalizing on Data Analytics Using Automation
Ashish Bansal, Senior Director, Data Science, Capital One [NYSE:COF]