Page 77 - AI Standards for Global Impact: From Governance to Action

AI Standards for Global Impact: From Governance to Action

11 AI and machine learning in communications workshop

11�1 Introduction Part 2: Thematic AI

The AI and Machine Learning in Communication Networks Workshop marked the third
edition of the MLComm series. The two-day workshop brought together researchers, industry
professionals, and standardization experts to discuss the evolving role of AI and machine
learning (ML) in modern telecom networks, especially in the context of the transition toward
IMT-2030 (6G) networks.

As communication networks become increasingly complex and data-driven, AI and ML
are playing a transformative role in reshaping network architectures, enabling distributed
intelligence, autonomous operations, and dynamic decision-making capabilities.

This year’s workshop built on the momentum of previous editions. The 2023 edition contributed
foundational work to the ITU Focus Group on Autonomous Networks, while the 2024 edition
advanced collaboration towards the ITU Focus Group on AI-Native Networks.

Several ITU Machine Learning in 5G (ML5G) initiatives, such as the competitions of AI/ML
Challenges, demonstrate the importance of regional inclusivity, diverse datasets, and research
collaborations around AI/ML in networks.

This MLComm workshop aimed to support ITU's work on strengthening the integration of
AI research with real-world telecom needs, while fostering collaboration across regions,
promoting open-source toolsets, and supporting the development of AI-driven solutions
through innovation, inclusivity, and standardization.

Exploring the future of intelligent, adaptive, and sustainable networks, the workshop was
organized into four main sessions, each targeting critical aspects of AI in telecommunications.

• The first session focused on innovations in AI models, including advances in reasoning,
inference, and generative workflows.
• The second session addressed the evolving role of standards bodies and open-source
collaboration in AI-native networking.
• This was followed by an announcement track that introduced the Echo toolkit for hardware
design, a Large Wireless Model challenge, the Open Platform for Enterprise AI (OPEA)
challenge, and TelecomGPT-Arabic alongside Large Perceptive Models.
• The third session focused on architectural impacts, emphasizing AI-driven changes in radio
access networks (RAN), core, and edge networks through technologies like federated
learning and semantic communications.

The final session highlighted practical tools, simulators, and datasets enabling AI-native
networks, concluding with a panel discussion on the long-term vision for AI’s integration across
telecom infrastructure and systems.

• Breakfast Session:

Continuing the tradition of an informal breakfast meeting before MLComm workshops,
a breakfast session was arranged to discuss future AI/ML research directions and related
standardization roadmaps and trends.

65