Ericsson launches Explainable AI in Cognitive Software

• New capabilities enable shorter time to value from AI-based solutions 
• Based on AI models trained on the largest, most diverse global data sets
• Improves efficiency in RAN optimisation

Ericsson (NASDAQ: ERIC) announces new capabilities leveraging Explainable AI (XAI) within its Cognitive Software portfolio for communications service providers (CSPs) to significantly improve time-to-value from AI adoption in network design and optimization.

With XAI, CSPs will have full explainability on actions recommended by the AI-powered solution that identifies the root causes of events affecting network performance and end user experience. This gives optimization teams visibility on the highest contributing factors to the issue, the impact on network performance, and recommended actions.

The solutions rely on AI models trained on the largest, most diverse global data sets in the market, which can also be retrained locally for fast and flexible use case deployment. 

In addition to the new XAI capabilities, an intuitive and adaptable user interface is also being introduced to enable CSPs to boost productivity.

Jean-Paul Arzel, Executive Vice President and CTIO of Bouygues Telecom: “At Bouygues Telecom we are constantly willing to improve the quality of service provided to our consumer and business customers, and actively exploring how best to leverage AI across our operations with speed and scale. We have recently achieved very promising results on our network with Ericsson in applying AI to network optimization, achieving congestion reduction, increasing capacity and comfort of use, as well as improving spectral efficiency. Ericsson is a long-term partner for Bouygues Telecom, we are delighted by their focus on accelerating operators’ AI journey and providing capabilities aimed at improving our network to meet the expectations of our customers.”

Perihane Elhamy Ahmed Metaweh, Chief Technology Officer of Robi Axiata: “With more than 56 million subscribers in Bangladesh, we are continuously looking for ways to improve our customer experience while increasing efficiency with automation. In our recent experiences with Ericsson we have seen AI-driven network optimization bringing improvements of over 11% in both user throughput and data volume for urban areas, and we believe the evolution announced for Cognitive Software will help us take these improvements to the next level”.

Adaora Okeleke, Principal Analyst at Analysys Mason: “CSPs are beginning to realize the value of AI but are hindered by several factors including lack of transparency of their AI solution, limited access to high quality data and the difficulty of scaling AI solutions. Ericsson's focus on addressing CSPs' current concerns, by offering solutions that drive higher trust and flexibility, aligns well with current market needs and will help accelerate AI adoption.”

Jean-Christophe Laneri, Vice President & Head of Cognitive Network Solutions at Ericsson: “The complexity and sophistication of AI-powered systems is evolving at a fast pace, and facilitating the easiest and most agile adoption of these capabilities for CSPs is key to minimize time to value. Explainable AI provides methods and techniques to understand AI models, which brings transparency to otherwise opaque decision-making processes. We are very excited with making these capabilities available in our Cognitive Software applications and consider this a fundamental step in the automation journey.”

Ericsson’s Cognitive Software architecture is cloud-native. It provides for container-based automated deployments, ensuring a secure environment with timely software updates, increasing efficiency. Being modular, the Cognitive Software components can be scaled separately, depending on the current need. This also facilitates integration with CI/CD pipelines for continuous integration and deployment and encourages DevOps ways of working.

To learn more about Ericsson’s Cognitive Software portfolio, click here.