Leveraging Generative AI Tools to Produce Supplemental Videos: Course Instructors’ Reflections

R. M. SINGH, U.T.F. LAM, and F. M. YEONG* 

Department of Biochemistry, Yong Loo Lin School of Medicine, National University of Singapore

bchyfm@nus.edu.sg

Singh, R. M., Lam, U. T. F., & Yeong, F. M. (2024). Leveraging generative AI tools to produce supplemental videos: Course instructors’ reflections [Poster presentation]. In Higher Education Conference in Singapore (HECS) 2024, 3 December, National University of Singapore. https://blog.nus.edu.sg/hecs/hecs2024-rmsingh-et-al/

SUB-THEME

Opportunities from Generative AI 

KEYWORDS

Supplemental Videos, Generative AI, Replacement, Instructor Reflection

CATEGORY

Poster Presentation

INTRODUCTION

In LSM 2233 Cell Biology, we sought to provide supplemental materials to bridge students’ background gaps. With affordances provided by generative artificial intelligence (AI) to enhance teaching, learning, assessments and administration in higher education (Chiu, 2023), we explored the use of AI tools to produce supplemental videos. We targeted the replacement level in the replacement, augmentation and transformation (RAT) model (Hughes, 2000) on the use of educational technology. Here, we reflect on our initial attempts at leveraging AI tools to support teaching without unduly increasing the workload of instructors 

METHODS

Workflow

We prompted ChatGPT-4o to generate basic slides, which we re-organized and inserted diagrams to fit the course (Fig.1). Revised slide decks were used to prompt ChatGPT-4o for narration scripts. We used Descript’s AI voice-generator to create a speaker voice modelled after the YFM’s voice. Descript was used to produce videos using the slide decks, narration scripts and speaker voice.

Figure 1. Flowchart for use of Generative AI in this project

Reflections and Coding

We reflected individually upon our experience using the AI tools, guided by a series of questions (Table 1). The reflections were coded and summarised by R.M.S. and U.T.F.L. (Cohen et al., 2011) into positive or negative comments, and further by categoried based on frequencies of occurrence (Table 2).

RESULTS

Products from Use of AI Tools

With our prompts (Fig 2, left), we obtained slides (Fig 3, left) from ChatGPT-4o. We included different aspects in the prompts as shown (Nazari et al., 2024) to obtain useful slides.  

Figure 2. Example of a prompt to generate a presentation slide deck (left) and narration script (right)

We improved the slides organization and inserted additional information and relevant diagrams (Figure 3, right). We added slides (Fig 4, left) to connect supplemental topics to the main lectures and inserted a slide to declare to students our approach (Figure 4, right).  

Figure 3. Example of a slide before (left) and after (right) revision by instructors

 

Figure 4. Example of an additional slide inserted by instructors to highlight relevance of information to course (left) and a disclaimer slide for use of AI (right)

 

Using the revised slides, we prompted (Fig 2, right) ChatGPT-4o for slide narrations, which we improved by inserting additional points or linking words to make them more natural-sounding (Fig 5). The improved script and slides were uploaded to Descript to generate supplemental videos (Link 1). On comparison, the AI-generated voice (Link 2) sounded close to YFM’s voice (Link 3).    

Figure 5. Example of a part of a transcript before (left) and after (right) revision by instructors

Reflections on Our Approach Using Generative AI Tools

In terms of functionality, we reflected positively (40.9%) on aspects like the speed of ChatGPT responses, and negatively (21.4%) for areas such as writing good prompts. We had concerns on accessibility (21.4% of responses), mainly on the high costs of the AI tools. 

 

We further compared the time-taken for us to create supplemental videos using the AI tools with estimated times needed to manually create them. We estimated it was 6-fold faster when using AI tools (Table 3). 

 

An analysis of mid-semester viewership data for the first four supplementary videos revealed that an average of 30.65% of the cohort accessed the videos uploaded to the Canvas Learning Management System. Of those who accessed the content, students completed at least 70% of the video material (Table 4).  

DISCUSSION

From our experience, we efficiently (Table 3) produced usable supplemental videos (Link 1) using AI tools. Our positive reflections (61.1%, Table 2) were consistent with our high level of perceived usefulness, positive attitude towards using and intention to use the AI technologies (Davis et al., 1989). Our negative reflections (38.9%, Table 2) centered around prompt quality in ChatGPT, unfamiliarity with the platforms and subscription costs. These can be overcome by improving prompts, consulting manuals and using trial versions for the platforms. We suggest that leveraging technology for replacing (Hughes, 2000) fully-instructor-created supplemental videos can possibly be achieved by using AI tools. The preliminary viewership analytics indicated that a third of the cohort viewed the videos, suggesting that the videos could have served students’ needs. (Table 4). Future studies will involve surveying students for acceptance of the AI-generated videos. 

REFERENCES

Chiu, T. K. (2023). The impact of Generative AI (genai) on practices, policies and research direction in education: A case of chatgpt and Midjourney. Interactive Learning Environments, 1–17. https://doi.org/10.1080/10494820.2023.2253861 

Hughes, J. E. (2000). Teaching English with technology: Exploring teacher learning and practice (dissertation). Teaching English with technology: exploring teacher learning and practice 

Hughes, J., Thomas, R., & Scharber, C. (2006). Assessing technology integration: The RAT–replacement, amplification, and transformation-framework. In Society for Information Technology & Teacher Education International Conference (pp. 1616-1620). Association for the Advancement of Computing in Education (AACE) 

Mackie, K. and Aspenlieder, E. (n.d.) “Evaluating AI Tools”. The Curious Educator’s Guide to AI. Open Library 

Nazari, M. and Saadi, G. (2024). Developing effective prompts to improve communication with ChatGPT: a formula for higher education stakeholders. Discover Education, 3(45). https://doi.org/10.1007/s44217-024-00122-w 

MULTIMEDIA LINKS

a37-QR-link1

(Link 1) Demo video: https://shorturl.at/TMWjM

a37-QR-link2

(Link 2) AI voice demo: https://shorturl.at/Bgl05

a37-QR-link3

(Link 3) Authentic voice demo: https://shorturl.at/vYRgy 

Viewing Message: 1 of 1.
Warning

Blog.nus accounts will move to SSO login, tentatively before the start of AY24/25 Sem 2. Once implemented, only current NUS staff and students will be able to log in to Blog.nus. Public blogs remain readable to non-logged in users. (More information.)