News
When high-speed networks and AI combine, the results could be beautiful or catastrophic; either way, telcos are in for a wild ride.
If telecoms starred in a Choose Your Own Adventure book, Mobile World Congress presented our protagonist with a dilemma of critical importance: artificial intelligence is coming; do you embrace it knowing that it may end civilisation? Or do you take a more cautious approach, knowing that by doing so, you may go out of business?
I suspect that a clear majority of operators at the show this year would without question take that first option and worry about the civilisation part later.
After all, the application of AI to mobile networks can maximise capacity and solve any problems before end users even notice anything is awry.
Just ask Scott Sumner, director of analytics at Accedian. He and his company are working with India’s Reliance Jio Infocomm to add intelligence to its nationwide 4G network.
Machine learning is becoming critical as operators virtualise their infrastructure, he said, because networks are no longer static and predictable, they are becoming dynamic and inherently unstable, with different elements in constant contact with one another.
"The network becomes exponentially more complicated when it is virtualised because control is delegated to the edge," he explained to Total Telecom this week.
In a virtual environment, if something goes wrong in one part of the network, it is no longer an isolated incident but can have a knock-on effect on a much larger area, affecting a greater number of customers and degrading performance to a much greater degree, he said.
With Accedian’s service assurance solutions, Jio collects 2 billion data points from its network every hour. Information such as packet loss and jitter etc. is presented to engineers who can zoom in from a nationwide overview of network performance to the status of an individual cell site.
The next step is to apply machine learning to identify root causes of problems based on all the information gathered from everywhere in the network. Engineers can solve issues, and feed the solution to the machine learning engine, so that if the same problem crops up again, it will automatically suggest a possible solution. Looking further ahead, the ultimate objective is for Jio’s network to automatically alter its state to head off any potential problems and maximise capacity without any human intervention.
This is what SK Telecom has achieved in its LTE network, Sumner said.
SK Telecom’s network "changes its state every 15 seconds based on the user experience and customer profile, without any human intervention," he said. "It has double the capacity of any other operator with the same amount of spectrum, because of this constant fine-tuning."
This kind of network intelligence will be imperative following the advent of 5G, Sumner said. He went so far as to suggest that operators that don’t have this kind of capability will go out of business.
"Operators need to do this now," he said. "If you’re a dinosaur and you don’t evolve, you’re going to go extinct."
Based on this argument, operators would be mad not to push on with machine learning and artificial intelligence.
Spain’s Telefonica certainly is. It showed off Aura, which it claims will revolutionise the customer experience by imbuing its network, IT, and services with cognitive intelligence.
But as the saying goes, the road to hell is paved with good intentions.
Softbank CEO Masayoshi Son also wants to be at the forefront of artificial intelligence, which is partly why his company forked out €28.8 billion for chip designer ARM and is busy setting up a $100 billion venture capital fund.
He is preparing for the singularity, when computers become more intelligent than humans, which he believes will happen in 30 years.
However, Son made the point that humans must tread very carefully with AI.
"Is singularity good or bad?" he asked. "We need to ask philosophical questions."
Son’s words of caution echo the sentiments of the likes of Stephen Hawking, Elon Musk, and Bill Gates, who have all warned that if AI grows beyond human control it could spell the end of human civilisation.
Before we get to the really apocalyptic stuff though, AI will first pose a serious threat to the employability or otherwise of a lot of folks. The success of Donald Trump and those campaigning for the U.K. to leave the EU is owed in part to the frustration that people feel about there not being enough opportunities to go out and earn a decent living. AI has the potential to exacerbate that frustration. Is it too much of a stretch to consider this might cause a degree of social unrest?
If only telecoms really was a Choose Your Own Adventure book, then we could cheat by flicking forward a few pages and find out the result of each decision before it was taken.
Since that is not an option, the industry must forge a path forward that realises the full potential of AI without accidently eradicating all humanity. Best of luck to everyone!
In the meantime, Total Telecom will return on Monday, unless of course the machines rise up and take over, in which case, all bets are off.
