|
Designing AI-Generated and Human-Taught Short-Video Mandarin Lessons: Learner Needs and Instructional Design Quality |
|---|---|
| รหัสดีโอไอ | |
| Creator | Ning Zhou |
| Title | Designing AI-Generated and Human-Taught Short-Video Mandarin Lessons: Learner Needs and Instructional Design Quality |
| Contributor | Kotchaphan Youngmee |
| Publisher | Research and Development Institute, Nakhon Ratchasima Rajabhat University |
| Publication Year | 2569 |
| Journal Title | Research Community and Social Development Journal |
| Journal Vol. | 20 |
| Journal No. | 1 |
| Page no. | 27-45 |
| Keyword | AI-assisted language learning, Human-taught instruction, Short-video learning, Mandarin as a foreign language, Media instructional design quality |
| URL Website | https://so04.tci-thaijo.org/index.php/NRRU/issue/view/18509 |
| Website title | https://so04.tci-thaijo.org/index.php/NRRU/ |
| ISSN | 3027-7515 |
| Abstract | Background and Objective: The growing use of short-video learning and artificial intelligence (AI) ineducation has intensified interest in scalable instructional formats, yet the quality of instructional design in AIgenerated materials remains underexamined. Existing studies often evaluate learning outcomes or technologyacceptance without establishing whether the instructional content itself is pedagogically sound.Methodology: This study adopts a design-oriented approach to examine how AI-taught and humantaught short-video Mandarin lessons can be systematically developed and evaluated. A sequential design wasemployed, including learner needs analysis (N = 180), controlled instructional design, and expert evaluation (N =5). Two sets of videos were developed using identical content to isolate the effect of delivery mode. Instructionalquality was assessed in terms of clarity and suitability.Results: Expert evaluation showed that human-taught videos achieved a higher overall quality rating (M= 4.31, SD = 0.57) than AI-generated videos (M = 4.04, SD = 0.96). Human-taught videos were particularly strongin clarity of explanation and naturalness of language delivery (M = 4.80, SD = 0.45), whereas AI-generated videosexcelled in suitability for the short-video format (M = 5.00, SD = 0.00) and pronunciation consistency (M = 4.80,SD = 0.45). The findings suggest that instructional quality is shaped primarily by systematic design rather thandelivery mode alone, and that the two approaches offer complementary pedagogical strengths.Discussion: This study advances a design validation perspective in AI-supported learning bydemonstrating that |
