Complexity in global telecommunications markets has been driven by disruptions for a while now. Continuous growth, ongoing and unpredictable changes, fast adaptation needs and convergence makes it even more challengeable.
The world has watched exponential growth of the traffic facing strong requirements for increased investment. While the prices per megabyte is going down, the business model of every telco operator is threatened. Whenever a new investment cycle arises it is heavily controlled, demanding new models of optimisation.
In parallel, AI, one of the most disruptive technologies in 21stcentury is hitting more momentum in various industries. History teaches us that disruption is needed and sometimes more than evolution itself. Henry Ford once said: “If I had asked my customers what they wanted they would have said a faster horse.”
So, where can operators positively impact their business models? There are several basic segments to encounter. The Virtual Assistants aka chatbots here it’s not an option. Sure, they are ok, getting better and used as a typical example of AI logic that should be adopted and customized from other businesses.
The biggest AI related value is reducing OPEX and optimising CAPEX. The most efficient and relatively simple to deploy is energy consumption control and prediction. It can be also (as chatbots) reengineered from other industries. So, opportunities to learn from use cases are greater and the results are real. It’s been 6 years since DeepMind Applied developed AI to optimize cooling of Google’s data centres. What were their expectations? Surely with not such a great success they had, and it was a really high: 40% of energy reduction. Soon after these achievements they started to call themselves as an AI company.
And now, more than ever, we see an importance for energy cost optimization.
Energy consumption control and prediction should be the basis for a huge set of predictive maintenance AI applications which is the only real way to make business models sustainable.
Want a new technology? Investment cycle is a prerequisite! And this is what we are all aware of. 5G deployments are massive across all over the world, probably with no exceptions, networks are geographically distributed around the globe. Smart capex spending exceeding. This is where traffic model prediction is playing the key role. Such models are not only optimising investments but also providing ultimate results improving customers experience.
One of the biggest assets that telco operators have is data sets. There are numerous network sources: fault management systems, performance management systems, passive and active probing systems, CRM systems etc. mostly, operators have huge (a really huge) amount of historic data which helps AI models to be more precise. All those data sets aka sources are playing the key advantage to make AI deployment more efficient.
Considering its criticality, the telco industry has its own ways of running. Availability, even above 5×9 is always imperative. This is one of the main reasons why most of the telco decide to go for an open loop model (which requires human intervention and/or approval of action), and once they get confidence, they transfer it to closed loop (no human intervention required) model.
Biggest initial threat is confidence. How successful and how valuable use cases must be to have only a positive impact? Most of the operators are adopting a one-by-one strategy until a certain level of confidence is reached.
The threats that can be foreseen in the future is lack of orchestration. By deploying more AI logic, especially in a closed loop model, risk of new, complex incidents gets bigger.
When evaluating different deployments, people ask themselves … Is it really AI or is it just an automated process?
My personal expectation is that AI will give the unexpected result (hopefully positive), while automation will do the expected result faster and cheaper. But, if it provides value, is it important in which category it fits?
Last few years brought us an unexpected crisis. First Covid, where the telecom industry proved its importance. And at the moment, the energy crisis brings new challenges to the operating model of Telecom operators. This challenging times should be used to speed up the inevitable adoption of AI. To put it in trekkies words… “Resistance is futile”. Or, as one of my professors from University of Oxford said … “Don’t waste the crisis!” (kudos to Alex, Andrew, Martin Michael and complete DiplAI2022 cohort).