They’re old ideas, but machine learning and AI are now the communications industry’s hot buzzwords, judging from last week’s Mobile World Congress.
“For mobile operators around the world, this is no longer an experiment,” said Patrick Ostiguy, CEO of Accedian, during an MWC panel discussion on machine learning.
Machine learning, which involves training a computer by feeding it examples and counterexamples, has been around for decades. The post office’s optical mail-sorting machines are one example. True AI and deep learning, which strive to teach a brain how to teach itself, are also long-standing disciplines.
The technologies have innumerable uses, but analytics, in particular, is gaining traction with service providers. Carriers such as SK Telecom and Reliance are using machine learning by feeding it with analytics from network monitoring, Ostiguy said.
“The impact of analytics and big data will be as large as the impact of cloud and NFV,” she said. “I would put it at that level. It’s massive.” But there’s one key difference, she noted: The impact of the cloud took carriers by surprise. When it comes to AI, they’re determined to get ahead of the curve.
What makes AI hot now, though? One factor is that while many of these ideas have been in the lab, it’s only recently that processor technology has advanced enough to make them practical. The power of an old supercomputer is available commonly, which gives AI-driven products a fighting chance to get to market.
Another reason is that some Tier 1 carriers are focusing on user experience rather than on network performance.
The difference isn’t just marketing. Carriers previously could only guess at user experience by watching statistics such as the number of dropped packets or dropped calls. AI opens the possibility of measuring quality-of-experience directly, said Brendan O’Reilly, CTO of Telefónica UK, during the MWC AI session.
For example, Telefónica has begun talking internally about a predictive version of the self-organizing network (SON). “If customers behave in a different way, what action should you take to get ahead of the game?” O’Reilly said.
The Telefónica mother ship (the Spanish carrier which owns Telefónica UK) made an AI-related announcement at MWC, in fact — a cognitive system named AURA that’s meant to help customers manage their own accounts with the service provider. It’s not really a digital assistant like Google Now or the Amazon Echo; it’s more of an automated help desk, and it addresses the question of how carriers can build a new model for customer relationships, Rainge said.
Part of what makes AI credible is that known names (such as IBM) have been working on known projects (such as Watson) for some time, and not solely in the context of the communications industry. What’s important there is the brand name, the comfort that goes behind telling a large enterprise “we’re IBM,” Rainge said.
NEC, similarly, lays claim to machine learning products that go back decades, to those mail-sorting machines. At MWC, NEC was promoting the WISE, a brand name assigned last July to a collection of more than 15 software suites. (Watson is similarly an umbrella name for a collection of AI technologies.)
Reflecting the breadth of AI, Watson and the WISE aren’t equivalent. “Watson is more precise in the analysis areas, less so visualization,” Akimoto said. NEC, on the other hand, has a long history with machine learning that uses visual technologies (remember those mail-sorting machines).
NEC is already talking about the WISE going beyond machine learning and even beyond deep learning. Deep learning requires less human intervention than machine learning. The next step would be to reduce human involvement even more.
“Learning processes automatically — that is the point of the third generation of AI,” said Ichirou Akimoto, an expert from the company’s big data strategy division.
Even vendors you wouldn’t associate with this kind of pure research are getting into the act.
Ericsson hadn’t talked much about artificial intelligence before, but CEO Börje Ekholm made it a key point of his Monday press conference. Artificial intelligence is being used in purely reactive ways today, but Ericsson and others are looking toward sharpening the technology’s predictive edge. AI will start to suggest actions to improve or fix the network and eventually will execute those actions automatically, he said.
Photo: IBM’s Gaudi-inspired sculpture at MWC, with moving parts controlled by Watson.