- There’s a lot of data telcos have to sift through to make sure their networks can support AI
- Google Fiber is zeroing in on diagnostic, predictive and generative AI to “do more with less”
- C Spire is interested in using AI to make its billing systems more efficient
SCTE TECHEXPO, ATLANTA – Telcos are still very much in the early innings of artificial intelligence (AI) adoption, so how can they ensure their network infrastructure supports it? The first step starts with data, and the telecom industry is “swimming in an ocean” of it, said Hani Elmalky, head of product architecture and access at Google Fiber.
“We have much more data than we can use,” he said at SCTE TechExpo this week. Most of Google Fiber’s systems are “built on the cloud,” Elmalky noted, which helps the company put all the data flowing across its network “where it belongs.”
AI, as we all know, is pretty data-hungry. “You need to have it accurate, you need to have it granular and it needs to be available,” said Netcracker CTO Bob Titus.
From Google Fiber’s perspective, “the most important thing about the data is to figure out a way to attribute it to the right level, to store it in the right way,” Elmalky explained.
It’s also key for everybody to access that data, as opposed to just having a centralized group with “unilateral accessibility.”
Google Fiber’s AI strategy
With AI, the “overarching premise” for Google Fiber is the ability to “do more with less.”
In other words, Elmalky believes with “the right AI tooling and the right connection of AI models with the data,” Google Fiber’s technicians, call center operators and software engineers can accomplish more in their day-to-day work compared to just working on their own.
On the network side, Google Fiber is zeroing in on three buckets of AI: diagnosis, predictive and generative.
Diagnosis, or using AI to determine whether a system is malfunctioning or not, “requires a lot of ability” to get data in real time and “synthesize them in a way that will allow us to understand what is the situation we are dealing with,” Elmalky said, whether it’s a planned or unplanned change in the network.
With predictive AI, Google Fiber is trying to figure out why a problem has happened on the network and what is “the best course of action at that moment," he said.
“We’ve seen a significant amount of improvement when it comes to certain workflows by actually understanding these two together as a predictive model,” Elmalky explained.
For instance, the company could determine a method it’s used in the past to solve an issue in the network “is wrong and now we actually have the chance to fix it," he added.
As for generative AI (GenAI), Google Fiber’s goal is to essentially “automate the creation of certain artifacts, like designs, configurations or even MOPs [methods of procedures].”
C Spire leans on data consolidation
Rabun Jones, chief information officer at C Spire, said his company is trying to vet AI use cases “to make sure that they’re technically going to work” and that they can be operationalized “very quickly.”
One use case C Spire is interested in involves making billing systems more efficient. The more lines of businesses a telco has, the trickier it can be to simplify and consolidate the data at its disposal.
“Just over the past two years we probably did six or seven billing migrations,” Jones said. C Spire offers fiber-to-the-home and wireless service and a few years ago acquired Alabama broadband provider Troy Cablevision to expand its footprint.
“We’ve had a focus on consolidation systems, so picking best-in-class systems and putting those in place…and making migrations in our ability to migrate data,” Jones said.
C Spire is also aiming to create a “persona-focused UX.” Meaning fiber installers or customer support staff can have one place of record that they can access to “pretty much do their entire job.”
“It doesn’t have to actually be one single system, but to them it should feel that way,” said Jones. “And then the same thing with data as a way to basically tie it all back together…and visualize it.”