An Opportunity for Telcos
Artificial Intelligence (AI) has seen a recent hype in terms of technology funding as well as public ethical debates. As AI requires various technologies to enable it, its applications are also diverse. For telcos, the question becomes whether they want to become actively involved in its development and application or if they would rather take a passenger seat – providing simple connectivity as opposed to value added services – an experience that is similar to M2M.
Artificial Intelligence (AI) has been a hype topic for some time now, and also dominant at the Mobile World Congress (MWC) in Barcelona. During the recent MWC Softbank’s CEO Masayoshi described his long-term vision: He is predicting the so called Singularity, i.e. when AI will surpass human intelligence, within the next 30 years. His keynote is well worth watching, but besides his future predictions especially his core message is of interest: the connection between artificial intelligence and connectivity. Connectivity plays a key role in his predicted “birth of super intelligence”, which is why telcos around the world are starting to invest into and apply AI in a growing number of ways.
Let's start at the Beginning - What is AI?
But AI itself is a broad concept, based on a multitude of individual technologies such as image recognition, natural language processing, deep learning, only to mention a few. And opposed to the common perception the ultimate goal is not necessarily to create a human-like robot that will replace human workforce: AI is the imitation of human intelligence processes by machines, especially computer systems – and these processes include learning, reasoning or self-correction (SearchCIO). A focus here is on defining AI in performing tasks or solving problems as humans do, rather than defining how humans think – this puts the emphasis on practical applications rather than a scientific neurological replication of human intelligence. To put it with Masayoshi’s words: AI wants to create super human intelligence to be able to solve problems and issues much faster, more reliably and less error-prone than humans. So it aims to improve our world beyond human intelligence rather than replacing it. Also there is a direct link between intelligence, available data and connectivity: the more data that will be available for decision making, the better the expected outcome can be. And in a world of the Internet of Things, according to Gartner in 2016 already 6.4bn things were connected reaching over 20bn in 2020, connectivity of course constitutes the main ingredient to access this information.
Take network management as an example: where current software solution and the output and interpretation of it are mostly tied to what the programmers pre-designed (including human readable formats such as graphs, tables, etc.), AI will enable the network to be self-optimizing and maintaining by using all available data (also coming from environmental sensors) and make things happen without human interaction (zero-touch). This is already possible today and the quality of the output will enhance every day with additional data being made available, further inputs (such as more sensors), as well as more sophisticated algorithms that will be able to learn and improve without programmers required to hard-code everything (i.e. using machine learning).
Coming back to Masayoshi’s vision at the MWC to emphasize the forecasted enhancement in AI capabilities: He elaborates that the human brain consist of around 30 billion connected neurons – with transistors in chips as the technological equivalent surpassing 30 billion in 2018. In the next 30 years we will have 3 quadrillion transistors in a chip, which is 1million times more than a human brain has neurons. Based on an average human IQ of 100, this would then enable machines to potentially have IQs of 10,000 – the “birth of super intelligence”.
Where do we encounter AI today?
A popular image of AI is a human-like robot that combines the field of AI with that of robotics. And as mentioned before this image causes a vivid ethic discussion about humans being replaced by robots, being plotted in Hollywood movies and daily newspapers alike. Scholars of the University of Oxford have even defined 12 global super crisis that pose a threat to human civilization, being one of them. But while this discussion is important and being addressed for example in interesting approaches such as Stanford’s AI100 (a 100-year study into Artificial Intelligence), there is already an interesting and broad public history of applications of AI: From Deep Blue’s victory over chess grandmaster Garry Kasparov in 1997, Google’s AI beating the Go world champion Lee Sedol in 2016 to virtual digital assistants such as Apple’s Siri.
Today’s applications are diverse and among others include robotics, entertainment, customer service, business intelligence & analytics or security and productivity applications. From a verticals perspective AI is being used in industries such as healthcare, advertising, financial services, retail, manufacturing or utilities. Some of the more prominent examples include IBM Watson, which is for example used healthcare applications for decision support in diagnosis and treatment options for cancer patients. AI is also used to help mobilize autonomous cars such as Google’s self-driving car or Tesla’s Autopilot. An area where a growing number of consumers is getting exposed to AI are consumer electronics, such as vacuum cleaners (e.g. Roomba’s iRobot) or lawn mowers (e.g. Husqvarna’s Auto Mower).
As an example for the telco industry and as we could witness also during this year’s MWC, virtual digital assistants are omnipresent: From Deutsche Telekom’s Tinka or Telefonica’s Aura to SK Telecom’s Nugu. These AI systems are the next step to existing chatbots by adding voice activation as well as plans to further integrate these services into products such as smart homes. More widely known digital assistants include Amazon’s Alexa or Google Home, which continue to implement additional skills to be applied in more and more use cases. Amazon alone has increased the number of Alexa’s skills from 130 in January 2016 to over 7,000 a year later. Today voice assistants are seen as a key enabler for the success of smart home applications.
What is in it for Telcos?
Given that telcos are directly or indirectly involved in a number of AI application fields and verticals, also their potential benefits from AI range across various parts of their value chains: From network planning, optimization and maintenance to customer facing sales and service capabilities (e.g. sales-agent support systems, human-like service interactions). This also indicates the scale of applications from a corporation’s perspective: AI can be used in revenue generation or cost optimization, internally as well as externally towards the customers. As mentioned before connectivity is at the very core of the value proposition of AI applications, enabling it to use the broadest data base available at the (physical) point where it is necessary. And the timing for telcos to get their feet wet is great, as AI is still a young discipline, with practical applications still to be seen on a large scale. But even though the before mentioned AI virtual digital assistants are a first step into that direction, telcos need to come up with a strong strategy and knowledge base to understand and utilize AI. Especially on the revenue side telcos need to ensure that they are driving this development and are not left behind with only the commodity (low margin) connectivity piece as often in M2M.
Figure: AI Annual Global Financing History (CBInsights, 2016)
All this to say that we are still at the beginning of a long journey. As Rodney Brooks, the former director of MIT’s Computer Science and Artificial Intelligence Laboratory, put it though: Extreme AI predictions are “comparable to seeing more efficient internal combustion engines…and jumping to the conclusion that the warp drives are just around the corner”. But the connection between AI, data and connectivity will further drive this development and will keep telcos at the core of its progression – if they choose to do so and are able to come up with a solid strategy how to integrate and monetize AI.
Author Oliver Platzen