LLMs Applied to Education & EU's Artificial Intelligence Act

Large Language Models in Education: Evaluating Model Performance for Exam Question Difficulty Assessment by Thiago Quaresma Chaves Brant

This presentation explores the application of large language models for educational assessments, focusing on classifying the difficulty of exam questions. By comparing both open-source and commercial models alongside varied prompt structures, the analysis uncovers how context sensitivity impacts classification accuracy, highlighting both the capabilities and current limitations of these models in education. The study emphasizes the importance of exploring foundational applications, such as prompt optimization and fine-tuning, to enhance model performance in real-world assessments. Practical insights on refining context strategies will be shared, offering guidance for educators and data scientists interested in advancing AI-driven evaluations in education.

Introduction to the AI Act by Fräntz Miccoli

The Artificial Intelligence Act (AI Act), like the GDPR, is a major move toward the regulation of information technologies in the European Economic Area (European Union and some other european countries). While it will gradually come into force by August 2026, the magnitude of the changes it requires and the consequences of non-compliance make it relevant for many actors to start think about it soon.

Where the GDPR required mostly contractual work, user information and consent collection, the AI Act impacts heavily on the engineering process of uses classified as "high risk" ; the definition of artificial intelligence in the regulation is broad enough to include everything starting from an Excel model or a linear regression.