On 14 November, the third transnational meeting of the European AI4Prosa project took place at the Centre d’Études Internationales de la Propriété Intellectuelle (CEIPI), University of Strasbourg, with CEIPI acting as host.
During the session, the project’s progress was presented, and a specialised workshop was held on trust and the level of hallucination in artificial intelligence assistants, delivered by CEIPI. In this context, CEIPI emphasised a key concept:
“Trust in AI is like building a bridge: it must be safe and ethical, but also passable. To achieve this, we need measurable criteria such as reproducibility and accuracy, while avoiding hallucinations.”
AIJU, in turn, presented the results of the first Beta validation of the AI4Prosa assistant, carried out in Spain. The data were very positive, although minor adjustments were identified and will be incorporated into future tests planned for 2026 in countries such as Germany and the Czech Republic. These results will enable the design of new learning sequences and optimise the assistant, with the aim of feeding into the upcoming version 0.9.
The European AI4Prosa project, led by AIJU and co-funded by the Erasmus+ programme, is moving towards its goal: to provide a reliable and ethical AI-based educational tool that facilitates understanding of the new European regulation on consumer product safety.
Comments are closed.