Communication service providers (CSPs) must deploy artificial intelligence (AI) throughout their organizations to thrive in the coming 5G era, and pulling that off will require a radical culture shift.
That was the key message from Aaron Boasman-Patel, Vice President, AI, Customer Experience & Data, TM Forum, who kicked off the ‘AI and Data’ track at Digital Transformation Asia in Kuala Lumpur this week by explaining why AI is a crucial component of 5G.
In the short term, 5G is essentially a faster pipe, but that will change rapidly as the next phase of deployment enables massive internet of things (IoT) connections, ultra-low latency apps, and targeted vertical apps delivered via network slicing and edge computing.
“The complexity is going to increase in our networks, which means we have to now start deploying and using AI to get us over that complexity,” Boasman-Patel explained. “Everything has to now become more zero touch. We need to have open architectures that are based using microservices, where we can use things such as APIs to transfer that information across.”
To date, CSPs have deployed AI technologies in some form or other in the network management space – for example, cellular site planning, reducing truck rolls, and customer onboarding. However, the ‘AI telco’ of the near future will have to deploy AI technology across the entire business, and manage that AI at scale in real time. That includes having the ability to explain to customers why the AI made a particular decision, especially if that decision leads to a network problem of a bad customer experience.
It also includes proper data governance – ensuring datasets are clean, algorithms are properly trained for those datasets, and managing all of that across the lifecycle. That also means opening up all data in different business silos so that it can be easily accessed in the first place – something that many CSPs currently don’t do.
Naturally, getting from here to there is a major challenge. Boasman-Patel highlighted a number of programs TM Forum is currently offering to help CSPs address these challenges, such as a standardized data model that provides a common platform to ingest data in the same format, as well as a training repository, AI management standards to help manage AI at scale, and a maturity model to determine how ready a CSP is to start adopting AI effectively.
There’s also the AI for IT & Network Operations (AIOps) Catalyst which enables telcos and vendors to explore AI-based use cases for optimizing CSPs’ processes and services to improve customer experience, quality and efficiency.
However, the prerequisite step for adopting AI effectively is for CSPs to change their approach to the business. Acquiring the right skillsets is an obvious necessity, but that doesn’t mean much without establishing the right cultural mindset to make the best use of those skill sets, Boasman-Patel said.
“We’ve got to have a cultural shift and a radical rethink of the way we’re doing this,” he insisted. “We’ve got to change to a culture which is much more open; we’ve got to have a culture of experimentation.”
This article was first published on TM Forums’ Inform