ITU-T Study Group 21 (SG21) is driving a coordinated and forward-looking programme on Artificial Intelligence (AI), bringing together work across technical domains under a unified framework. Through Q5/21, AI-related activities are structured into nine thematic categories, complemented by relevant recommendations, best practices, and expert insights to support consistent understanding and implementation. This integrated approach enhances the clarity, visibility, and coherence of SG21’s AI portfolio, fostering cross-sector collaboration and enabling stakeholders to engage with evolving AI standards and emerging directions in an efficient and accessible manner.
In ITU-T Study Group 21, the activities related to AI is mostly being considered under Question 5/21 (Q5/21)– “Artificial intelligence-enabled multimedia applications”, which focuses specifically on AI-enabled systems within multimedia contexts and handles methodologies for evaluating AI models, datasets, and services, including testing frameworks and trustworthiness indicators. In addition, other SG21 Questions also study AI-related topics, for instance, AI-enabled accessibility (Q1/21), AI-enabled digital health (Q2/21), AI-enabled digital culture (Q3/21), AI-enabled visual systems (Q7/21), etc. See all Questions of SG21 and their work items here.
In particular, one of the AI categories under Q5, Embodied AI, has expanded into a Focus Group on Embodied Artificial Intelligence for multimedia technologies, which is studying how EAI can support human‑centric multimedia applications by examining its core technical abilities, such as understanding the physical world, learning from trial and error, and collaborating with humans and their environments. It is investigating foundational technologies including real and virtual datasets, closed‑loop control, world‑model training, and advanced human–machine interaction.
All the standardization related to AI under SG21 could be viewed at ITU-T Landscape for ICT standards, or searched at ITU-T Recommendations Webpage.
