The first decade of the 21st century was about making the internet mobile. Now the focus is on developing and using artificial intelligence (AI) and machine learning to create exciting new products and services for consumers and businesses. Many of those services will be provided by hyperscale companies like Amazon and Google, so what’s the role for communications service providers (CSPs)? Can they compete?
We explore those questions and more in our first-ever Trend Analysis Report AI: The time is now. During a webinar last week TM Forum’s Chief Analyst Mark Newman summarized the results and offered some suggestions for CSPs looking to adopt AI technology.
Place your bets
“The race is on to build up AI skills and competencies,” Newman said. “The ability of [hyperscale]players to maximize use of AI holds the key to their future success.”
This means it’s going to be difficult for CSPs to attract top AI talent. Recruitment and consulting firm Paysa tracks the number of vacancies for AI and machine learning experts across many companies. The chart below compares the annual number of AI recruits by company, showing Amazon as the clear leader.
“The only company on the list that you would say is telecoms-centric is Huawei, and I think that tells us a lot about where telecoms is today versus other sectors in terms of its creation of competence and skills around AI,” Newman said. “From our calculations, Microsoft and IBM…have 15,000 people working on AI and machine learning. You can see the challenges that CSPs are going to have creating skills and competence in this global race for talent.”
Are CSPs deploying AI?
Nearly 70% of the CSPs we surveyed are at least experimenting with AI.
“AI is helping operators focus their data analytics strategy,” Newman said. “What’s different about AI is that it’s very use case-centric, so it’s easy to start measuring the benefits. It’s helping operators to justify their investment in analytics.”
CSPs are beginning to use AI first in customer centricity, and they also are using it for network management and perhaps most importantly for automation.
“There is a growing view among many large operator groups that brutal network automation is the way to go,” Newman said. “AI is absolutely central to that automation. What we’re really talking about is moving to an intent-based network: I as an operator say what the connectivity is and automation is used to create the path across the network.”
ODA can help
TM Forum has been looking at AI as part of a new architecture called the Open Digital Network (ODA – formerly known as ODES), which is designed to replace operational and business support systems (ODA).
Navigating the roadblocks
In addition to difficulty recruiting talent, CSPs face many other obstacles in adopting AI.
“The first challenge is data,” Newman said. “If you don’t have clean, open data, it will be difficult to start using AI.”
CSPs must ask and answer many pressing questions in order to have any chance at all of competing with hyperscale players.
“Are we as an industry going to be able to create large enough training sets for AI to work its magic? What can we do about operator fragmentation to make AI more effective? What’s best practice for introducing AI progressively? Who is financially, ethically, legally and morally responsible when a machine makes decisions?”
Listen to the webinar on demand to hear Mark’s suggestions for the steps CSPs should take to get started with AI or to improve existing programs, and to hear his answers to questions from the webinar audience.