Effectiveness of Using AI-Generated Texts as Reading Skill Instruction Materials: The Case of DeepSeek and Copilot

Artificial Intelligence (AI) has been revolutionizing educational settings, including language teaching classrooms, where it transforms teaching and learning skills. This study aimed to investigate the effectiveness of having AI chatbots generate texts to be used as reading comprehension practice materials. To this aim, a total of ۶۰ students were randomly selected from those enrolling in a BA level general reading course at Ilam University, and three homogeneous classes of ۲۰ students (۱۰ females, ۱۰ males) were formed following an entrance level exam and proficiency test. Two classes received reading skill instruction with AI-generated passages by DeepSeek and Microsoft Copilot used as the only source and teaching material, while in the third class a common reading textbook (Select Reading, Elementary Level, Oxford University Press) was used. All other factors, including session length and frequency, course period, instructor, time of the day classes were held, etc. were held identical. Upon completion of the course (۱۳ sessions), a reading comprehension test was given to the students. The test results showed that the learners who received training with texts generated by DeepSeek achieved the highest scores, followed by those instructed via passages generated by Copilot, and learners from both AI-enabled classes outperformed peers from the class where the textbook was used. Results confirm the effectiveness of applying AI-generated texts in reading skill instruction. Further studies with different groups of learners of different age groups and levels, as well as diverse contexts and settings, and AI tools are suggested to improve insights into AI effectiveness.

کلیدواژه ها:

AI-generated texts ، DeepSeek ، educational technology ، Microsoft Copilot ، reading comprehension ، reading instruction