Humans, AI and machine intelligence – a match made in heaven?

Back in 2016, a Google computer program known as AlphaGo beat Lee Sidol, one of the world’s top Go players. Perhaps that doesn’t sound so impressive, after all, it is only a boardgame. I mean, how hard could it be, right?

Connected robots in the home Purpose

Head of our Global AI Accelerator

Head of our Global AI Accelerator

Category & Hashtags

Well, the thing is, Go is a super complex boardgame. Google claims there are more potential positions in a Go game than atoms in the universe. In fact, it’s favored by AI researchers because of its complexity and the potential for a huge number of different outcomes when compared to chess, for example.

Despite the complexity it only took AlphaGo 40 hours to master the game. It’s important to note, it required zero human assistance.

This game, and subsequent ones, captured headlines the world over. By many it was heralded as one of the first major breakthroughs for artificial intelligence and machine learning, since Deep Blue beat Garry Kasparov. The victory over the world’s top players – which many thought would take decades to achieve – underlined the capability of artificial intelligence to compete with humans at mastering complex tasks.

Since then further feats and milestones have been surpassed. But is there more to AI and machine learning than thrashing humans at boardgames?


Some people treat these kinds of developments in artificial intelligence and machine learning as a huge threat. At Ericsson, we prefer to take a more holistic and optimistic view of the future.

Future impact of machine learning and AI on work in the telco industry

Right now, most AI is described as narrow because they can only perform a single task, such as doing face recognition in digital phone albums, for example. But in the future, general-purpose AIs could potentially outperform humans at complicated tasks. Future iterations of AlphaGo could work alongside humans as experts. In theory, they could even be used to teach humans. Ultimately, because AlphaGo learns on its own, its focus can be turned to help solve real-world problems. And that’s something to get truly excited about.

In the future, machines and humans will need to work in tandem and share the same goals, rather than competing independently with one another. After all, we’re much stronger when we work together.

So, how are we at Ericsson and the ICT industry planning to work with AI and machine learning?
Well, in our case the narrow AI problem is in reality very, very broad. Our 5G systems have thousands of parameters. We process 1 terabyte of data for every customer every day. Think about it like this: that is one trillion bytes of data for just ONE network.

When you multiply that with the fact that we manage hundreds of networks with millions of subscribers, it is easy to understand the magnitude of the complexity and responsibility that we have in our hands. Our telecom domain expertise and understanding of how to manage this data, combined with trust, puts us in a unique position to make this marriage work.

Our algorithms develop intelligence through continuous learning. As a result, humans are assisted by machines in their decision-making processes as well as actions.

We use a 3-dimensional capability frustum (LEARNING-REASONING-AUTONOMY) to describe the entire system at work. In this collaboration, humans develop algorithms and software logic to enhance our systems (machines).

This allows our machines to capture (SENSE to achieve LEARNING), organize and process data to develop meaningful insights (THINK to achieve REASONING) that humans or machines can act (ACT to achieve AUTONOMY) on. These correspond to three crucial axes of capabilities that encapsulate varying levels of actionable intelligence and autonomies that our systems are equipped with.


In the end, actionable knowledge and insights are created from structured and unstructured data. Models are created and continuously evolved to make informed predictions and enhance automated decision making. In other words, the realization of zero touch networks is an iterative process in which machines and humans collaborate – like in a symbiotic relationship.

Over the years, Ericsson has built up deep knowledge about managing data. With investments in data science, research, innovation and infrastructure, Ericsson is well on its journey to enhance its offerings and processes, catalyzed with artificial intelligence, automation and analytics.

Informed predictions and enhanced automated decision-making are what we have explored in collaboration with our R&D teams.

We believe that AI based automation, with algorithms that can adapt to the complex environment and the constant changes in user and traffic behavior, is a necessity to manage networks of the future.

In close collaboration between Business Area Managed Services and Research we’ve been building an inhouse Managed Services Networks AI Platform. Now it’s being proofed with our customers and we already see game changing results. We are on the verge of predictive, data-driven operations. And we have built the solution ourselves. We have 13 AI patents already filled. Imagine what we will do next!

The future is bright

So, what can we see on the horizon?

We will see a shift from a system that recommends and applies changes to solve issues or optimize performance to an autonomous system that deals with potential problems, keeping us humans in the loop and seeking input when needed.

In addition, we’re likely to see critical but routine tasks like radio site validation, configuration and maintenance tasks to decrease in time from a few hours to a few minutes. More importantly, human involvement will shift from an operational to a supervisory role. This will allow for smaller teams to have oversight over a larger group of radio sites, effectively reducing OPEX for operations, network rollout, Network Design & Optimization and field services.
But how will we get there?

Automation is not a new practice. When tasks can be shifted from human employees to robots, work can be completed faster and without the risk of human error. This dramatically improves efficiency levels, which means a better bottom line for an organization as a whole. So, there is a net-gain across the whole chain – for all stakeholders involved.

I have had the good fortune of being part of several crucial trend-shifts in various industries over the years. But the magnitude of how machine learning and AI technologies can positively enhance our industry is unmatched – for the sheer pervasiveness of their applicability. And the fact that we are a foundational component (connectivity) of major on-going transformations (digitalization, automation, etc.) in various industries (healthcare, education, transportation, retail, automobile, etc.) leaves me very excited about the opportunity to make a broad and meaningful impact at a societal level – through what we do here at Ericsson.

We intend to create exponential value from the data which we either own or have access to. As leaders in this space, we have access to substantial data across various dimensions (network operational plane, subscriber plane, etc.) – a gold mine for any ambitious data scientist who joins us in any of our three advanced R&D centers around the world.

The future of work will be a hybrid that involves both human and machine intelligence working in conjunction towards meaningful and profitable goals. Much of the new skills of the future will be very dependent on getting access to a lot of data, and the right type of data. With the exponential growth of data, this could be a real opportunity for data scientists.

Are you interested in learning more? Come join our team and make a difference!

The Ericsson Blog

Like what you’re reading? Please sign up for email updates on your favorite topics.

Subscribe now

At the Ericsson Blog, we provide insight to make complex ideas on technology, innovation and business simple.