In the latest DGT podcast episode, the burgeoning intersection of languages and technology within the EU is explored, focusing on significant strides in AI language evaluation. The episode, released on October 23, 2026, provides insights into the European Union's (EU) innovative approach to multilingual technology through the EU MMLU project. This initiative sees European Master’s in Translation students actively translating an AI evaluation dataset into the 24 official languages of the EU. Such an endeavor not only allows students to gain invaluable practical experience but also contributes to creating a robust multilingual benchmark necessary for the advancement of large language models (LLMs). As EU Translation Centre notes, this effort underscores the critical role human expertise plays in enhancing the efficacy of AI systems.

During the episode, hosted by Anna Holmén, guests Pilar Sánchez Gijón from the Autonomous University of Barcelona and Susana Valdez of Leiden University engage in a dynamic discussion about the challenges of adapting AI evaluations to diverse languages and cultures. They emphasize not only the linguistic intricacies involved but also the broader cultural context that must be considered in AI development. Anna Kokkinidou, representing DG Translation, contributes her perspective on the importance of integrating human expertise into AI processes. She emphasizes that understanding cultural nuances is essential for the accurate deployment of AI technologies in multilingual settings, which presents both challenges and opportunities for ongoing development within the EU.

The guests also highlight plans to expand this multilingual benchmark to encompass all 24 EU official languages, a task that involves significant commitment to linguistic and cultural diversity. The anticipation around this benchmark is indicative of the growing recognition of its value in supporting inclusive communication across the EU. Expanding to all official languages is not only technically demanding but also calls for collaboration across various fields of expertise, stressing the importance of human-AI synergy in overcoming language barriers.

Ultimately, this episode of the DGT podcast sheds light on how collaborative efforts in education and technology can contribute to a more interconnected and understanding multilingual society. The EU's forward-thinking approach, as illustrated by the EU MMLU project, sets a remarkable precedent for integrating language services with advanced technological frameworks. Through fostering such initiatives, the EU reiterates its commitment to linguistic diversity and cultural preservation while advancing technological innovation.