Abstract:

Artificial Intelligence, especially in its personification of ChatGPT, has been considered a threat to eLearning systems because it lends itself to substitute the student’s effort to learn by simply providing ready-made responses and papers and then submitted as their own. A quasi-experimental study carried out by the author with two Statistics courses aims to prove the hypothesis that, if appropriately used, AI can be an important tool for student success. An elementary and an advanced course of Statistics will be tested with a variant of the Solomon Four-Group Design. In half the groups, a “traditional” approach to eLearning will be used with recorded instructor lectures, modules, and online assignments with feedback. On the other half, the AI systems ChatGPT and Dall-E-2 will be applied in four major functions: instructional design support, surrogate tutoring, questioning, and feedback. The student’s success will be measured for all groups through indicators of achievement, engagement, and satisfaction. Also, a student analytics tool (Starfish) will provide a close follow-up of performance. It is expected that the outcomes would provide guidelines for increasing benefits and minimizing risks of the use of AI in eLearning.

Keywords:

Artificial Intelligence, e-learning, student success, ChatGPT.