Digital transformation by Artificial Intelligence (AI) is not a need; it is a must have. AI can help telecom operators to increase their market value, make better decisions in complex business processes and improve the digital experience of end users. By 2021, expected spend on AI technologies by companies is around 57.6 Billion dollars*. 40 Trillion Gigabytes (zettabytes) of data will be created. It is 300 times more than we had in 2005.
The term Artificial Intelligence encompasses several areas in different contexts, ranging from sensors and robotics to qualitative argumentation. Indeed, AI is a vast field both in terms of cutting-edge scientific research and in terms of its applications. AI methods are applied in several industries to save costs, reduce risk, increase efficiency, make better decisions, deliver optimal solutions. Relatively simple applications like customer service chatbots, personalised recommendation systems, big data management are indispensable for many companies. More complex applications like supply chain network optimisation, investment planning, forecasting, advanced analytics, real-time automated decision-making systems are providing a competitive advantage to companies which can embrace these new technologies. In a narrow sense, digital transformation can be seen as the replacement of traditional processes with digitised counterparts. A more comprehensive definition of digital transformation has to include intelligent decision making at the heart of the newly developed systems. Effective digital transformation must make use of analysis and optimisation to achieve a step change.
The telecommunication sector is a prime candidate for being the pioneers of incorporating advanced Artificial Intelligence methods into their digital transformation journey. Therefore, telecom operators accelerate their digital transformation by integrating new innovative digital services for end-user operations and internal operations. For instance, advanced analytics can help companies to make better investments, develop a sales strategy, improve network design by analysing customer behaviours and also can provide more customers loyalty to their services. As another example, operators can reshape their services based on actual scenarios, simulate their business development models and can have dynamic models compatible with real market needs. By Location Based Targeted Advertisement, mobile operators can understand their customer behaviours even in the indoor locations without any hardware installation, and engage with them with a relevant message at the right time and location. With the help of machine learning technology personalising the experience with high accuracy is possible and crucial to power real-time customer engagement. In addition to chatbot and live supports, analysing customer requests and frequencies together with the network data, and their location can help to find the right solution to solve customer’s problem quickly and more efficiently or fixing potential problems before they occur can also decrease the operational costs. Secondly, customer services can be improved by identifying daily customer requests. For example, identifying potential hot requests from thousands of requests, and send them to the related departments automatically by also providing brief information.
Even though organisations rapidly change to keep up with digitalisation, the main metrics seem to stay the same. Sustainability, integrity and expandability of telecom operators are vital elements to remain in the race. Ability to move fast, to observe and interpret, to survive and operate is crucial. Therefore, telecom operators must not only adapt to today’s AI solutions but also be able to integrate their AI strategy to the overall organisation. However, these integrations might create new challenges. With the 5G plan, networks’ scalability, capacity and flexibility might need to be improved. At the same time, the current systems we have are not as agile as expected. Advanced AI techniques play a pivotal role in this process since transforming the infrastructures are quite costly comparing to the business process transformations. That is why a holistic AI approach should be designed and implemented.
Currently, telecom operators assess many technology solutions. When telecom operators and regulators perform several solutions to make their processes more agile, the next challenge might be making the integrated platforms more agile and elastic. On the other hand, it is not easy to take the right action in a hyper-connected world. When the telecom operators and regulators can take the right action, it might also be costly. If this expense is not an issue for the organisations, the next challenge might be providing the optimal accessible and governable processes. With the capacity of 5G, the systems can be more connected, more accessible and more manageable. To make this journey real, telecom operators need scalable systems. In addition to that, having an automated decision-making AI agent help to fine tune the systems. AI and 5G integration can make nimble manageable network design and implementation.
There is tremendous potential in applying AI to challenging problems in businesses. However, there are also very significant concerns one needs to keep in mind when applying AI methods. I would like to highlight two concerns that can potentially be the most devastating unless they are handled with care.
First significant concern about applying AI methods in a business context is the data-driven nature of these methods. Businesses need to be very wary of using user data in ways that may potentially conflict with the user agreements between the data collectors and users. This issue is particularly challenging when datasets from multiple sources need to be linked together. Companies can control and reduce this risk by working with digital transformation and AI consultants that have significant expertise and legal support.
A second major concern is that of explainability, especially in applications where an AI system replaces a task that was previously performed by a professional. These are tasks like sales/ account management (traditionally performed by a sales/ account representatives), suspect identification (usually performed by security personnel). Some cutting-edge AI methods lend themselves better to explainability whereas some others that produce impressive results are terrible for explainability (such as deep learning). When explainability is a concern in a particular application, this must be identified as a requirement early in the planning phase. Developing a self-explaining AI system requires peculiar expertise that is not easy to come across in the consultancy market. That is why organisations should collaborate with AI experts who fill the gap between the advanced AI theories and the real business cases.