Artificial intelligence (AI) may be the next big thing for communications service providers (CSPs), but it’s not clear yet exactly how they will use it or where it will have the biggest impact on their business.
“Let’s break it down a bit – it can be misleading,” Telefónica Global Group’s CIO, Phil Jordan, told attendees at the Executive Summit during TM Forum Live!. “We see clear use cases and value in cognitive and machine learning. In other areas of IT, we’re looking at all decisioning and that’s interesting. Any decision we take in a systemic way, I have asked for a plan for when and where does that become a machine-learned activity?
“It’s the next transformation wave that is going to hit all of us – converting decision-making into something that isn’t static rule-based,” he adds. “I don’t think that’s a technology problem – it’s here or it’s coming.”
But is Telefónica making extensive use of AI today?
“We made no use of it in the transformation,” Jordan emphasized in discussing the company’s massive digital transformation. “AI isn’t a magic trick,” he says, and it won’t be useful unless operators transform their existing IT systems first. (Check out this article to read more of Jordan’s comments at the Executive Summit.)
Indeed, that’s the message we’ve been hearing from many of our members: AI is promising but it isn’t reality – yet. What do you think?
In November we will publish an extensive Trend Analysis Report on AI and machine learning, analyzing the results of our surveys of CSPs and suppliers (choose the right one for you). Take the survey and you’ll be entered into a draw for a $250 Amazon gift voucher.
Certainly, new virtualized network functionality and new operational and business support systems are needed to take advantage of AI and machine learning (a form of AI), in customer facing applications such as virtual agents and chatbots and for end-to-end network and service management.
“You have to teach it; you have to give the machine context all the time,” Jordan explains. “You have to have a business that is ready and able to understand outcomes and go back and feed it into machine learning. The business change associated with AI and cognitive learning is much bigger than the technology and it’s a much bigger transformation for the business.”
AT&T aims for automation
Other telcos also are beginning to embrace AI and machine learning as they work towards automation. As we noted in our recent Quick Insights report Data Analytics & AI: Key to end-to-end management, AT&T is an undisputed leader in the push to automate. The company has an ambitious plan to virtualize 75 percent of its network by 2020.
The Optimization, Reliability and Customization Analytics practice within AT&T Labs is designing and implementing algorithms to optimize the company’s software-defined network (SDN) network and is also responsible for the deep learning that helps the company identify and predict faults. This work also includes security analytics and machine learning for optimizing performance and increasing revenue.
Like most operators, AT&T sees three important financial benefits to automation: Saving on capital and operational expenditures (CapEx and OpEx) and increasing revenue opportunities.
“OpEx opportunities result from the ability to self-heal without manual intervention,” says Kathy Meier-Hellstern, Assistant Vice President of Inventive Science, AT&T Labs. “The other big area for OpEx savings is the ability to automatically pinpoint the source of problems.”
CapEx savings result from virtualizing network functions and running them on commodity hardware instead of buying new specialized devices, and from using network capacity more efficiently.
“I think of my analogy as a Tinker Toy sculpture with rods representing the network and the little nodes, locations in the network,” Meier-Hellstern explains. “With SDN we can dynamically reconfigure the Tinker Toy sculpture. Whereas in the pre-SDN days you’d have to add extra capacity so that you could ride through a variety of failures, in the post-SDN world the network can just reshape itself so that everyone gets taken care of.”
An important goal is to be able to predict problems before they happen and this requires development of applications that ‘learn’ based on flexible, evolving analytical models. Every operator we spoke with for our report said machine learning in network and service management is going to be critical, but, again, it’s still very early days.
At BT, data scientists are ‘training’ systems to look for patterns that indicate a router may be about to go down or congestion is about to occur.
“You can learn from previous patterns where you saw failures or congestion and then match those patterns in real time to the analytics you’re gathering in real time to say, ‘This looks like a repeat of something that has happened previously and could be indicative of a router going down, a circuit going down or congestion’,” explains George Glass, Chief Architect, BT. “By building that in with machine learning you can start to take proactive interventions in the network to fix the service before the customer is impacted. It’s very close to a self-healing network.”
AT&T is using machine learning in its core IP network to define optimal network configurations and automatically reconfigure them, and like BT, the company is working on using analytics to proactively identify network events that could impact performance.
“In the VNF area we have been looking at algorithms to dynamically adjust capacity and network response in response to DNS attacks,” Meier-Hellstern says. “And we have some of our first instances of VNFs that can do self-healing in response to a failure. That’s very important in order to scale our business – that we can identify failures through closed loops and policies and restart, move or rebuild a VNF as needed. We will have our first trials of that coming out soon.”
The end goal is to be able to change policy automatically by using machine learning and AI. Today, humans fine-tune policies and scripts based on data analytics, but in the future the policies will be able to update themselves.
Using AI to interface with customers
Operators also are excited about the potential for AI in customer-facing applications. An ongoing TM Forum Catalyst has been exploring the use of avatars and chatbots for customer service. The most recent phase of the project demonstrated at TM Forum Live! in May looked at using AI tools from the customer service agent’s point of view, examining how they can be used to reduce stress, map all interactions within a customer service center and reduce fraud.
Ultimately, the real benefits of AI in customer experience and network management will come from machine intelligence, not just machine learning.
“Today, machine learning is only as good as the data you use to train it,” Guavas’ Chief Scientist told us in advance of his presentation at TM Forum Live! “Machine intelligence is the next step and it represents the ability of the machine to extrapolate from raw, disparate data to create new valid information that cannot be discovered by applying a machine-learned model. For CSPs, the full value of machine intelligence solutions goes beyond detecting and prescribing a solution to a problem – it’s about systems becoming intelligent.”