The issue of ai-based support for struggling readers through the lens of UDL principles

One of Universal Design for Learning principles suggests to “design multiple means of representation” (CAST, 2024) and it can be applied to text accessibility by granting 1) readability and legibility, 2) disambiguation of terms and symbols, and 3) support to comprehension through various means. Text simplification is often mentioned as a possibility offered by AI for struggling readers. In this contribution we argue that it can efficiently be combined with intelligent tutoring systems in order to achieve better results that are in line with UDL guidelines. There is in fact evidence in literature of a growing interest in text simplification operated using LLMs, with promising results (Hedlin et al., 2025), also specific to Italian language (Padovani et al., 2024). Some other specific techniques have been developed to support comprehension using AI, like questions generation (Liu et al., 2024) or bidirectional dialog with virtual tutors for reading comprehension (McCarthy & Yan, 2024). We argue that a virtual tutor that is capable of clarifying text using multiple techniques can offer the best interface to text simplification and hence to comprehension.