AI can help telcos evolve their customer offering, reports Jayne Brooks from the Total Telecom Congress 2018
Telcos should look to use Artificial Intelligence (AI) as a tool kit to build a great user experience, rather than focusing on ‘data science,’ Google’s EMEA Customer Engineering said today.
Speaking during a panel on “Capitalising the AI opportunity” at Total Telecom Congress, Andy Kennedy described how Google has used AI and explored how these methods could help telcos take advantage of the technology.
“Rather than focus on the data science piece, telcos should have an application relevance focus,” he said “If you can consume something and build experiences for your customers, isn’t that more important than being a data science expert? Focusing on how to consume services and using AI as a tool kit to build great user experience and services is something that will give you a higher return on investment over time.”
Kennedy also called on telcos to think more like retailers during the discussion which also featured presentations from Mustafa Ergen, Chief Technical Advisor at Turk Telecom; Dr. Andreea Hossmann, Head of Data Science at Swisscom; and Henri Korpi, Executive Vice President, New Business Development, at Elisa.
“Retailers have done a great job of focusing on the customer buying journey and monetizing information about what customers put in their basket but don’t check out,” added Kennedy. “There is no reason why [telcos] can’t use the same approach.”
Hossmann agreed that telcos need to think beyond just being telcos. To emphasise this point, she outlined Swisscom’s use of AI, highlighting how the technology has been applied to different areas in the business, including Customer Experience, Security, Human Resources, Legal, Customer Service and Marketing. She concluded that impact is the important measurable when applying AI but also warned that it will not necessarily be clear which area AI would have the highest impact on.
“If it doesn’t have a graspable impact, it doesn’t matter,” she said. “Whether the impact is in the customer experience or making things more efficient, there has to be an impact. Ideally, we’d be able to learn from the data where the highest impact is but we don’t always have complete data and we don’t always have labelled data. This means there is still an element of intuition and relying on trial and error to find the highest impact areas.”
As the session drew to a close, data transference and data selection were also debated by the panellists, with Kennedy highlighting the benefits of the former and Hossmann describing the latter as a currently under-tapped method. While Kennedy saw the advantages in this method, he concluded the session with a warning that overfitting the data could lead to more challenges.