AI Business officially launched our new series ‘AI Innovators’ in January. This is a series dedicated to featuring interviews with relevant spokespersons across a range of businesses that have entered the world of artificial intelligence. AI Innovators aims to provide our readers with a broad insight into how artificial intelligence is implemented within different industries, anything from finance to medical research.

We recently caught up with Rickard Cöster and Elena Fersman of Ericsson Research to gain an insight into Ericsson’s current AI involvement.

Rickard Coster ericsson

Cöster is a Data Science Expert at Ericsson Research. He is a technical leader in machine intelligence research, using data-driven innovation to create new value in Ericsson’s existing portfolio and explore new areas. Having obtained a PhD in Computer Science from Stockholm University in 2005, Cöster joined Ericsson Research in 2010 and is the co-inventor in more than 30 patent applications. We caught up with him to gain an insight into Ericsson’s current AI involvement.

elena-face

Fersman is a research leader in the area of Machine Intelligence at Ericsson Research and an adjunct professor in Cyber-Physical Systems at the Royal Institute of Technology in Stockholm. While working at Ericsson, Elena has held various positions, including product management, portfolio management and research leadership. She is responsible for setting the research direction for multiple research groups based in several countries. Her current research interests are in the areas of modeling, analysis and management of software-intensive intelligent systems applied to 5G and Industry & Society.

Ericsson are currently involved in a broad range of AI initiatives, the interviewees reveal, and they have developed a dedicated organisation set out to work with machine intelligence. “We use machine intelligence as a collective term for Machine Learning and Artificial Intelligence. This research organization is responsible for developing new technologies and pushing the envelope in the area of machine intelligence”, they said.

This work is done in close collaboration with top universities in the field, as well as Ericsson’s industrial partners. Currently, Ericsson is a cosponsor of a new UC Berkeley lab for Machine Learning, which focuses on developing state-of-the-art technologies to help machines make intelligent decisions based on real-time input.

They tell us that Ericsson’s product portfolio benefits greatly from the latest technological developments in the machine intelligence field, as the new functionality is now included in their products and services, as well as within the internal operations.

“We are a data-heavy company – data is produced, transmitted and consumed at many places in our portfolio”, they said, explaining their progress of work: “we aim to draw insights from this data, organise these insights in a knowledge base, enhance this knowledge base through the application of semantics, machine learning and input from experts, and use it to create actions that contribute to automation and value creation”.

Looking at the technologies you work with today, which are you specifically focused on developing, and how does these AI technologies fit into your existing solutions?

“With 5G we are creating one network for a million needs, and 5G also expands the telecom network to industrial and IoT applications”, the pair said. They explained how they experience an increased use of machine intelligence in 5G to make real-time decisions to optimize network connectivity while efficiently utilising resources.

“Network management and digital decision support systems are also increasingly being automated by machine intelligence”.

“Intelligent algorithms in the 5G network are also applicable to management of our customers’ operational technology processes”, the interviewees explain, elaborating on how this is valid for both telecom operators as well as “Industry and Society”, which is Ericsson’s umbrella term for utilities, transport and public safety players.

“Another important area is human intelligence augmentation. In this area we will experience increased productivity and revenue growth through various support systems that leverage decision assistance and advanced analytics operations”, they said.

One of the examples is Ericsson Expert Analytics, which enables advanced customer experience management for telecom operators. This drives new revenue growth and improves customer experiences.

Which verticals have you experienced that you have gained most traction from in relation to artificial intelligence?

The interviewees reveal that intelligent algorithms operating on data and knowledge are fully applicable to all data-and knowledge-intensive domains. For Ericsson, telecoms is a clear choice due to the domain expertise that they have internally in their company.

“We are using machine intelligence technologies to optimize our services and internal operations, as well as the products we offer our customers. And as Ericsson is moving into IoT and the Industry and Society space, we see great benefits in applying machine intelligence to business processes and operations technology in utilities, transport ,and public safety”, they said.

“Digitalisation of industries is a pre-requisite for successful application of machine intelligence, and as domains mature, machine intelligence can be applied on a broader scale”.

Looking at your enterprise’s way of working, how are your solutions impacting this?

“The main purpose of machine intelligence is to increase automation and create new value”, the interviewees said. “For example, logistics companies that connect their assets and digitize their business processes can not only track their goods in real-time but also gain access to a network slice with the quality of service that they need. In addition, they can optimise the routing of goods given a set of parameters such as cost, deadline and customer satisfaction”.

Another example Ericsson mention is the maintenance of smart meters and electricity substations. With the help of machine intelligence technologies, Ericsson can provide field support engineers with a proposal of the tools required to solve a problem, as well as the most probable solution.

How would you say that Ericsson differs from the other companies out there on the market today?

They emphasize Ericsson’s big research organisation with a diverse set of competences and how they keep a lot of focus on innovation through Ericsson Garage, in addition to intellectual property via patents and scientific publication.

“When it comes to development of machine intelligence methods, Ericsson has the advantage of a solid knowledge base in the telecom domain. This gives Ericsson Research the ability to experiment and test new methods”.

Where do you see Ericsson in five years’ time?

“Ericsson’s Networked Society vision states that everything that benefits from being connected will be connected, and connectivity is the first and necessary step towards automation of industries”, they said.

They both emphasize that if the connectivity is in place, systems will eventually become smarter and more capable of interacting with each other by understanding each other’s semantics.

“Semantic interoperability will allow for inter-domain interactions and the tapping of new values. Safety and security will be key aspects in these interactions”.

In the future, as well as today, Ericsson’s 5G system will continue to depend on machine intelligence technologies to ensure that they fulfill the changing needs of our customers in the best way. There will also be possibilities of optimizing enterprise business processes towards given business objectives while keeping systems safe and secure.

How do you see the rate of AI adoption evolving in 2017?

Cöster and Fersman predict that due to the large development of machine intelligence on such a broad scale, in 2017 we will see a large-scale adoption of these technologies. This is a result to the maturity of the cloud and increased computational resources, in addition to the rapid digitalization of industries.

“The focus has shifted from Big Data – how to manage all the data – to machine intelligence – how we can make software and machines more intelligent through the use of data, algorithms and compute power”.

What would you consider the biggest challenges faced by enterprises who are looking to implement artificial intelligence today?

“The competence gap is still large, and skilled individuals need to work with the technology to really take advantage of the recent advances”, they said. However, they mention that another factor is digitalisation.

“For new start-ups, this process is natural, since the business processes will be machine intelligence-ready by construction. Companies with well-established business processes that are not yet digitised must manage this step before they will be able to use machine intelligence at full scale”, they added.

Cöster and Fersman believe that the three areas where machine intelligence will have a large impact over the next five years are automation, human intelligence augmentation, and in new innovations generally.