As artificial intelligence (AI) continues its move to become a mainstream discussion point, relevant to all industries, George Malim interviews Ravi Palepu, the vice president and global head of telco solutions at Virtusa, to learn how the telecoms industry is harnessing AI for its own benefit.
George Malim: What are communications service providers (CSPs) hoping to achieve with artificial intelligence?
Ravi Palepu: This very much depends on which area a CSP wants to apply AI to. Six months ago CSPs were struggling to identify use cases that would have a return on investment (ROI) for AI. Therefore, so far, adoption has been very low. However, that is now changing and ideas around the customer facing areas of a CSP are being targeted in which AI can determine buying patterns and enable recommendations.
In general, though, a lot of effort has been put in but, in terms of AI, there are still not many use cases. What has been achieved is more machine learning-based and in this sense AI is not fully mature – even on the customer facing side of the telecoms industry.
GM: Which areas do you see CSPs applying AI to?
RP: Away from the customer facing side, people have looked at telco operations in the network and billing. I haven’t seen big AI activity in billing operations but 2-5% of top line revenue is being lost on revenue leakage each year by CSPs so eventually I could see AI being used to address this, although I don’t see much AI-related effort going into this today. The potential for this is not yet mature.
On the network operations side, I am seeing more and more companies moving from 4G to 5G and rolling out more fibre networks. Their areas of focus are on quality of service and improving efficiency both in terms of operations and network utilization.
These are areas where CSPs are investing a lot but they are also not mature yet. However, the number of proofs of concept projects are growing although this area is not yet at the point of defining what AI can deliver for CSPs.
There is a third area to consider and that’s how a CSP can embed AI into a product or customer lifecycle. From when a customer is planning an order to analyzing when a customer is placing an order presents and opportunity to embed AI as part of the business process. This takes AI to the next level but there are two approaches being taken. One involves keeping AI within a siloed environment and the other involves embedding AI as part of the entire business process so you can make decisions faster.
GM: Do CSPs actually have too much information to be able to apply AI effectively?
RP: People have never taken data that seriously in the past so the concept of sifting data to gain meaningful insights from it is pretty new for CSPs. They are hampered by a lot of legacy systems and data and that mean AI-enabling it will be difficult.
In addition, CSPs need to change their skillset to deploy AI effectively. In the past, the response might have been to hire a data scientist but, in general, they don’t have a great understanding of how to deploy and AI or knowledge of the outcomes they can expect.
The knowledge of the industry, the business and the data alongside AI expertise is what is missing. You therefore see a lot of initial trials being done but few get to the light.
GM: Do you see AI as a tactical or strategic commitment for CSPs?
RP: Without doubt it has to be strategic with the full commitment of the business.
This was originally published on Vanilla Plus