Evaluating an AI Educational Advisor for Supporting Personalized Physics and Electrical Engineering Learning Among High School Students on the Autism Spectrum

This study evaluates changes in educational outcomes during a year-long programme integrating generative artificial intelligence (AI) as an educational advisor in secondary-school physics and introductory electrical engineering. The study involved 35 students diagnosed with autism spectrum disorder (ASD), who used ChatGPT, Gemini, Grok, and DeepSeek during regular classroom instruction under teacher supervision. The findings indicate substantial pre–post improvements across several educational outcomes, including subject knowledge, conceptual understanding, problem-solving performance, interpretation of physical simulations, learner autonomy, digital competence, and confidence in using AI technologies. The study employed a single-group pretest–posttest design without a control condition. Therefore, the observed changes should not be interpreted as causal effects of AI alone. The findings provide evidence of educational change during an AI-supported programme and demonstrate the practical feasibility of integrating multiple generative AI systems into inclusive secondary STEM education within a teacher-supervised instructional framework.