The effects of AI-based visual instruction on the reading comprehension of students with dyslexia in Saudi Arabia: a single-case experimental study

Students with learning disabilities (LD), particularly dyslexia, often face significant challenges in reading comprehension that traditional instruction may not fully address. Generative Artificial Intelligence (GenAI) offers emerging opportunities to provide personalised and visual instructional support to bridge these gaps. This study investigated the effectiveness of a GenAI-based visual instruction to support the reading comprehension skills of students with dyslexia. A single-case experimental design using a multiple-probe across participants approach was employed with three male students aged 9–11 years diagnosed with dyslexia. All three students were attending a public elementary school in Yanbu city, Saudi Arabia. The intervention utilised ChatGPT-generated visual explanations aligned with the Grade 4 Arabic language coursebook. Five-point quizzes were implemented with the students to assess their reading comprehension of paragraphs of text taken from the Arabic language coursebook. All participants started with an initial low and stable baseline (0%–10%). Mastery in the treatment phase was determined by a score of 80% or higher on the assessments in three consecutive sessions. Using this criterion, two participants showed immediate level changes and the third showed a clear accelerating trend. Maintenance probes were conducted several weeks following the conclusion of the intervention phase, during which all three students showed sustained performance at or above mastery. Visual analysis indicated a strong functional relation, minimal or no phase overlap, and durable effects. These findings provide preliminary evidence for the utility of GenAI-enhanced visual instructional tools to support reading comprehension for students with dyslexia.