Advances in the functionality and accessibility of generative AI chatbots make them an appealing option to assist time-pressed teachers. Our recent activity booth at the 'AI Idea Potluck' event was an exploration of the possibilities of the technology. The booth showcased an interactive digital escape room activity (H5P) designed to encourage the use of AI for materials design assistance and demonstrated the gamification of AI to push educators to collaboratively reflect on and discuss their knowledge and practice. The AI interface used here was the XIPU AI tool developed by our university. With the help of the Chatbot, visitors raced to complete multiple preparation tasks in order to “escape” a digital classroom. Tasks included generating a reading text, comprehension questions, a warmer activity, and AI-constructed images. Users were then required to assess the quality and usability of these materials for their practice, and textual feedback was provided for each type of activity. The activities prompted very positive responses from teachers, whilst also highlighting the limitations of AI and the necessity for careful prompting and oversight. The activity has potential as a teacher training tool to encourage appropriate AI use.

 

Keywords: Generative AI, Materials creation, AI teaching assistant, H5P digital escape room.

 

Teachers need to stay abreast of any current technological trends that could enhance their students’ learning experience, and it would be difficult to ignore the seemingly ubiquitous discussions about substantial recent advances in Generative Artificial intelligence in the popular press, research journals, and education conferences. In the field of language teaching, much of the conversation centres around the impact of AI systems on language acquisition and assessment (Huang et al., 2023; Liang et al., 2023). There is also, however, growing interest in exploring how AI can support and augment the teacher in lesson creation and materials design (Grassini, 2023; Zirar, 2023). In December 2023, The School of Languages Xi'an Jiaotong-Liverpool University (XJTLU) team hosted an 'AI Idea Potluck' event, bringing together teachers to share ideas and trial activities on the above topics. 

 

The event allowed us to experiment with and showcase an interactive digital escape room activity designed to encourage the use of AI for materials design assistance. This also demonstrated the gamification of AI pushed educators to collaboratively reflect on and discuss their own knowledge and practice. The AI interface used here was XIPU AI, an AI chatbot developed by the university specifically for educators and students, accompanied by an H5P digital escape room we created and hosted on XJTLU Learning Mall (XJTLU Learning Management System). The booth drew an excited crowd and led to several important insights. Before describing the activity and making any observations it is appropriate to briefly review the relevant literature.

 

 

 

Image 1: teachers trying out the activity at our booth

 

Liang et al. (2023) conducted a large review of research on AI in education over the past two decades and noted that the use of AI for materials design garnered relatively little attention until recently.  Some of the first studies refer to it more as an advanced search engine to recommend suitable reading materials and investigate the psychological acceptance by educators of AI as a tool (Liang et al., 2023). In the past 5 years interest in the area has increased further, with uses spanning generations of texts, activities, quizzes, and more (Grassini, 2023; Zirar, 2023). From research so far, correct prompting of the AI interface (Zirar, 2023) and sufficiency of human oversight (Wrigely, 2018) have emerged as key aspects to be mastered by practitioners. The past year has seen huge advances in generative AI chatbots such as ChatGPT in terms of functionality and accessibility making it an even more interesting option for time-pressed teachers who can dedicate less time to mundane or lengthy tasks and more to the social and creative aspects of teaching (Grassini, 2023; Wrigley, 2018). The present activity is therefore a timely exploration of the benefits and pitfalls of the technology. 

H5P (HTML5 Package), is a relatively new framework for creating interactive digital materials which is growing in popularity due to its accessibility and proven benefits in terms of increasing engagement with the content. Users can interact with multimedia content (text, video) through a wide range of activities such as open and closed questions, gap fills or drag and drop and receive immediate feedback on their answers which can open alternate learning paths (Homanová, & Havlásková, 2019; Doran & Briggs, 2024). This format was chosen as a way of engaging the participants and guiding them through a set of materials design and evaluation tasks.

 

 
Image 2: The interactive digital escape room

 

Visitors to the booth were greeted with two screens, one showing a digital escape room integrated into an image of a classroom, the other a prompt screen for XIPU AI, the customised Generative AI model used by XJTLU. Participants were presented with the scenario of a teacher who needs to race to complete multiple preparation tasks on the theme of environmental sustainability in order to “escape” the digital classroom. Task completion required the use of the second screen and chatbot, and included generating a text, comprehension questions, a warmer activity, and AI-constructed images. Users were then asked to assess the quality and usability of these AI-generated materials for their own practice, and textual feedback was provided for each type of activity. Each stage completion unlocked different areas of the digital classroom (the whiteboard, a book, etc.) which contained further stages. The three designers were also on hand at the booth to guide and support users throughout the activities. 

Participants were first asked to use the chatbot to assist them in creating a warmer activity for the class, with the following instructions:

“You want a fun and creative warmer activity for the lesson. Ask your AI Teaching Assistant for some suggestions. Remember to be specific about what kind of warmer you want and think about your context. Consider if the warmer will be reading-based, require images or role play, and so on.” 

The activity and feedback highlighted the fact that the more specific the prompt was in terms of teaching context, the type of activity, and the subject matter, the more relevant and useable the output from the AI Assistant. The bot generated a number of interesting and fun warmer activities such as the "Sustainability Strategy Scavenger Hunt" along with detailed teaching instructions.

The second stage involved generating a reading text following these instructions: 

 

“You need a reading text for the lesson. You could spend some time Googling texts; however, the lesson is quite specific. You need:

    - a text that is about 350 words long.
    - which discusses China's approach to sustainable carbon reduction solutions.
    - to include words from the vocabulary list like biodiesel, emissions, carbon offsets, energy efficient and net-zero.

What's more, your students are mixed language level ability, between CEFR A2 - B2, so you want to ensure that the text is achievable for everyone. Approach your AI Teaching Assistant and see what they can do to help you!”

 

This produced some very interesting texts, and revisions prompted by teachers included simplifying the vocabulary and adding more environmental themes. This confirmed the potential of AI text generation and tireless revision, but also the importance of the role of a qualified teacher to prompt and review. 

 

 

Image 3: XIPU the AI chatbot and some example prompts and responses

 

Once teachers had produced a workable reading text, they were tasked with prompting the chatbot to create comprehension questions for a worksheet to accompany it.

“Ask the AI Teaching Assistant to provide you with:

    - 3 multiple-choice questions
    - 3 sentence completion questions
    - 3 overall comprehension check questions
    - 1 open-ended question for group discussion after the activity
    - an answer sheet

Consider all you have learnt about AI prompting so far for this activity. What specifics would need to be included in your prompt to create a good set of questions?”

This was a process of several attempts and re-prompts to produce something satisfactory. Re-prompting centred for example on comprehension of explicit terms, having students think of specific examples, use of synonyms between the text and questions, and any critical thinking points.  

One of the additional tasks was image generation:

“Finding images or photos for classes can often be challenging. You might be teaching vocabulary that you don’t have images to go with or need specific pictures for warmers or other activities. Taking your own photos or finding these online can take hours. What kind of image would you like to go with the lesson you have created today? Prompt XIPU and see the results!”

AI text-to-image generators have been improving over the past few months, but they can be expensive, and the quality of the output varies widely. This was also evidenced here, and it became apparent that even skilfully describing exactly what elements to include did not always produce useable results. 

The above description covers the most important stages of the activity, from which the following observations can be made.

The key takeaway from the activity and discussions was that AI requires creative thinking and a certain amount of skill to be a real asset for teachers, but does have the potential for reducing time spent on the mundane to dedicate more to creative tasks. Teachers were pleased and surprised by the speed and quality of content creation and had a very positive attitude towards interacting with the AI and the digital escape room. Teachers discovered that assigning specific roles to the AI was sometimes necessary to produce more appropriate results (“you are an English teacher”), but developers are aware of this as the latest versions of ChatGPT include individualised chatbots with specific roles. Furthermore, the image generator from XIPU was quite slow and produced only a basic type of image. However, this feature has been greatly improved in recent weeks. 

Overall, this experiment was exceptionally positive as a way to encourage teachers to integrate AI use into their teaching practice, and the activity format could have practical applications ranging from teacher training to in-class use with students.

 

 

Doran, S. & Briggs, C. (2024) Assessing Student Engagement & the Effectiveness of H5P interactive materials, ThaiTESOL Journal (pending publication).
Grassini, S. (2023). Shaping the Future of Education: Exploring the Potential and Consequences of AI and ChatGPT in Educational Settings. Educ. Sci. , 13, 692. https://doi.org/10.3390/educsci13070692
Homanová, Z., & Havlásková, T. (2019). H5P interactive didactic tools in education. In EDULEARN19 Proceedings (pp. 9266-9275). IATED. https://library.iated.org/view/HOMANOVA2019H5P
Huang, X., Zou, D., Cheng, G., Chen, X., & Xie, H. (2023). Trends, research issues and applications of artificial intelligence in language education. Educational Technology & Society, 26(1), 112-131.
Liang, J. C., Hwang, G. J., Chen, M. R. A., & Darmawansah, D. (2023). Roles and research foci of artificial intelligence in language education: an integrated bibliographic analysis and systematic review approach. Interactive Learning Environments, 31(7), 4270-4296.
Wrigley, T. (2018), The power of ‘evidence’: Reliable science or a set of blunt tools?. Br Educ Res J, 44: 359-376. https://doi.org/10.1002/berj.3338
Yan, L., Sha, L., Zhao, L., Li, Y., Martinez-Maldonado, R., Chen, G., Li, X., Jin, Y., & Gašević, D. (2024). Practical and ethical challenges of large language models in education: A systematic scoping review. British Journal of Educational Technology, 55, 90–112. https://doi.org/10.1111/bjet.13370
Zirar, A. (2023). Exploring the impact of language models, such as ChatGPT, on student learning and assessment. Review of Education, 11, e3433. https://doi.org/10.1002/rev3.3433