The AI/ML landscape for Smart Networks and Services

This document explores the role of Artificial Intelligence (AI) and Machine Learning (ML) in advancing smart networks and services, particularly in the context of the sixth-generation wireless network (6G). Key AI/ML concepts and terminology are clarified, current standardisation approaches are presented, while open implementations and research directions are described.As explained in the document AI is the technology enabling machines to mimic human intelligence, while ML is the toolbox of algorithms and methods that allow machines to learn and improve performance. On this basis the reader is provided with a taxonomy of AI concepts (e.g., AI-native systems, agentic AI, large language models (LLMs), multi-agent systems (MAS), generative AI (GenAI), explainable AI (XAI) etc.) as well as the major ML types (from the fundamental learning methods to Federated Learning (FL).Given the AI/ML taxonomy the document presents standardised frameworks for AI/ML integration in networks, including the AI/ML management framework from the 3rd Generation Partnership’s Project (3GPP), the modular ML pipeline from the International Telecommunication Union - Telecommunication Standardization Sector (ITU-T), the AI/ML architecture from the Open Radio Access Network (O-RAN) Alliance, and AI/ML-related studies and developments in European Telecommunications Standards Institute (ETSI). The converged outcome refers to the importance of modularity, interoperability, and dynamic adaptation in AI/ML operations.One step deeper, intensive AI/ML implementation efforts are currently allocated to the Agentic AI and the Multi-Agent systems, with one major challenge being the devise of efficient multi-agent communication protocols. In addition, since the lifecycle of ML models involves data collection, model training, validation, deployment, and continuous monitoring and retraining, the DevOps principles have been expanded to the Machine Learning Operations (MLOps) concept, while MLflow is introduced as key enabler for implementing MLOps practices in dynamic and distributed environments.Finally, in the research and innovation domain the document underscores the importance of AI/ML in driving innovation, enhancing network performance, and addressing challenges such as security, trust, and sustainability in next-generation networks.