The Role of AI/ML in Key Value Indicator Analysis

Anastasius Gavras
Eurescom GmbH

Traditionally, technology design in ICT focused on functional performance and market potential. However, with increasing attention on societal challenges and sustainability goals, there’s a call for a shift towards aligning technology development with key values for society. In the last while the concept of Key Value Indicators (KVIs) became a prominent point of attention in research and innovation for next generation ICT solutions. At the same time artificial intelligence and machine learning (AI/ML) have gained traction as a means to cope with the complexity of the future mobile telecommunications network and to optimise resource use for the delivery of advanced services. But, can AI/ML contribute to the design of values-driven technology development?

Key Value Indicators

To date we lack a clear definition of the concept, as well as a framework that serves as a tool for addressing societal challenges and identify value outputs. The Smart Network and Services (SNS) joint undertaking of the EU Horizon Europe programme for research and innovation started to request from research projects to address Key Value Indicators (KVIs). The work programme does not explicitly specify which values should be addressed, but rather refers to the Sustainable Development Goals (SDGs) of the UN and the work of the research projects HEXA-X and HEXA-X II. A recent publication in Telecommunications Policy magazine tries to shed more light into the problem space.

In the case of UN SDGs, the SNS work programme provides examples such as: SDG 8: Promote sustained, inclusive, and economic growth: achieve higher levels of economic productivity through diversification, technological upgrading, and innovation. SDG 9: Build resilient infrastructure, promote inclusive and sustainable industrialization and foster innovation, upgrade infrastructure and retrofit industries to make them sustainable with increased resource-use efficiency and greater adoption of clean and environmentally sound technologies and industrial processes. SDG 11: Make cities and human settlements inclusive, safe, resilient, and sustainable. SDG13: Climate Action: Support smart low carbon lifestyles, monitoring emissions, and shaping demand in transport and energy, enabling resilient mission critical communications in extreme weather (vertical markets: transport, health, and public safety). Furthermore, it provides for complementary societal values in ethics issues related to privacy and EMF (electric and magnetic fields) awareness and reduction.

A first indicative list of KVIs published by HEXA-X names three groups of KVIs, namely Democracy (privacy, fairness, digital inclusion, trust), Ecosystem (sustainability, business value, economic growth, open collaboration, new value chain) and Innovation (safety, security, regulation, responsibility, energy consumption). However, from a more practical AI/ML applicability point of view different KVI areas could be formulated, namely resilience, sustainability, and inclusiveness. In these areas the role of AI/ML can be better formulated and possibly measurable indicators can be derived.


Resilience, defined as the ability to rapid recovery if failure occurs and ability to scale to meet unforeseen demand, is a fundamental property of 6G networks. AI/ML has demonstrated the ability to enhance network resilience by enabling dynamic adjustments to unforeseen disruptions, as well as forecasting unusually high demand and mitigating security attacks. AI algorithms can contribute to developing robust communication protocols and fault-tolerant architectures to handle unexpected disruptions, ensuring continuity of network operations.


Sustainability, defined as meeting present needs without compromising future generations’ ability to meet their own needs, has become a top priority for the design and development of 6G networks. AI/ML may play an important role in enabling environmentally responsible and energy-efficient network capabilities. First and foremost, AI/ML can help optimising energy consumption and overall resource use during service delivery and network operations. The technology may also help prioritising technologies and practices that minimise the network’s environmental footprint. However, there is a notable tussle to address, namely the high energy demand of AI/ML algorithms during the training phase. Hardware accelerators are increasingly being used, like GPUs (Graphics Processing Units) and TPUs (Tensor Processing Units). Although they achieve efficient processing of data during the training phase, they increase the environmental footprint, since increased demand for hardware has a negative impact on the overall life cycle assessment as well.


Inclusiveness, defined as the recognition that every member of society should have equal access to opportunities, resources, and services, is a top priority of policy and regulation in the context of 6G. In this context AI/ML may play a role in steering the development and deployment of 6G towards inclusiveness. Through transparent system capabilities and user-friendly interfaces, AI/ML-powered systems can ensure accessibility for individuals with diverse technical knowledge or disabilities. Moreover, support for under-represented communities can be achieved by providing real-time translation services and access to remote civic services, thereby improving accessibility and enhancing overall well-being. However, there is a tussle to overcome in inclusiveness as well; namely that the measures to address inclusiveness are known since decades, yet they are expensive, because they often imply the deployment of more resources. AI/ML could be used in such a setting, as the tool to help resolving the tussle, i.e. to propose economically viable, yet inclusiveness supportive 6G deployment models.


The telecommunications ecosystem; vendors, operators, and service providers, have started to acknowledge the need to address the pain points and the needs of the customers and the society at large. Such pain points and needs are usually not quantifiable through performance metrics, such as data rate and latency. Additional indicators and metrics are necessary to measure the value output of ICT. Although technology evolution has started to understand how to use AI/ML as a tool to improve “classical” key performance indicators, it must now start to investigate how to use the same tool to address key value indicators and metrics.

Further information