Exploring the Collaborative Potential of Generative AI Chatbots in Language Education: Enhancing Learning Outcomes through Human-AI Interaction
This article explores the collaborative potential between humans and AI, specifically generative AI chatbots powered by large language models (LLMs), in educational settings. The authors highlight the importance of teachers' active roles in leveraging AI for enhanced learning outcomes rather than viewing it as a substitute for educators. They identify four key roles that AI chatbots can play in language education: interlocutor, content provider, teaching assistant, and evaluator. The case study presented focuses on how chatbots like XIPU AI and ChatGPT enhance the English for Academic Purposes (EAP) learning experience for advanced students.
With the increasing integration of AI in educational settings, scholars have directed significant attention toward the collaboration between humans and AI, emphasizing that collaborative efforts tend to yield more effective learning outcomes than isolated approaches (Kim et al., 2022; Xu & Ouyang, 2022). Such research underscores the importance of teachers' active roles, emphasizing how they can facilitate the positive impact of AI on education rather than viewing AI as a potential substitute for educators. To better harness the pedagogical potential of AI, specifically generative AI chatbots powered by a large language model (LLM), Jeon and Lee (2023) have identified four key roles that such chatbots play in language education: interlocutor, content provider, teaching assistant, and evaluator. In our case study, we illustrate these roles by showcasing how chatbots such as XIPU AI and ChatGPT enhance the EAP learning experience for advanced students with language proficiency levels at B2+ according to the CEFR (Common European Framework of Reference).
We utilized the chatbot's role-play function to enhance students' language skills and cultivate critical thinking by encouraging nuanced perspectives and constructing compelling arguments. Using prompts like "You act as A, while I will act as B," students immersed themselves in discussions spanning topics from business to earth science. By navigating diverse perspectives in AI-facilitated role-plays, students learn to consider various angles, improving their ability to analyze and articulate complex ideas. Additionally, prompts such as “Provide more details, such as real-life examples, expert opinions, and statistics,” helped students develop the ability to articulate robust arguments.
EAP teaching and learning, which requires students to engage in group discussions and make fully developed arguments in both writing and speaking contexts, was effectively enhanced through this AI-based approach. In the class, after demonstrating a sample conversation with the chatbot, we organized students into groups, assigned roles, and had them converse with the chatbot. As note-takers recorded key arguments, students were then regrouped for non-AI-assisted discussions, integrating insights gained from the AI conversations.
The chatbot provides an immersive and supportive environment for the application of language skills through interactive discussions and contextual activities. This enables learners to practice text-based conversations, similar to interacting with a native speaker. However, it has been observed that the data sources behind AI programs like ChatGPT may exhibit biases (Hartmann et al., 2023). Consequently, teachers and students bear the responsibility for the final assessment of AI-generated knowledge in education.
Sample prompts:
  • Put yourself in the shoes of …, what do you think about…?
  • If you were an employee, what do you think about..?
  • Imagine you were an employee, what do you think about?
  • Can you provide more details, such as real-life examples, expert opinions, and statistics?
Content provider:
The chatbot's versatility extends to crafting tailored curriculum and assessments for students. Leveraging AI XIPU, we designed final exam materials for advanced students enrolled in the EAP035 course. The task involves crafting a 400-word integrated summary featuring opposing ideas from a given text and lecture. In assigning this task to AI XIPU, we specified paragraph structures, word counts, and requested the integration of supporting details while adjusting the language difficulty to cater to the B2-C1 proficiency level.
Apart from generating listening and reading materials, the chatbot was instructed to generate model summaries in both point-by-point and block styles, accompanied by a cheat sheet outlining main ideas and supporting details using bullet points. This streamlined process, completed in a single day compared to the previously estimated 7 days without AI assistance, enabled us to generate eight sets of exercises. These exercises, accompanied by model summaries, served as self-study materials for students to practice summarizing skills before the final exam.
Strategically using chatbots enhances efficiency in curriculum and assessment development, aligning with recent research on its significant capabilities in literature generation (Lund & Ting, 2023). However, caution is needed for information accuracy. Crafting accurate prompts can be time-consuming, and the potential for false information, as noted by Rudolph et al. (2023), requires proactive measures. To ensure accuracy, educators should adopt a vigilant approach by regularly verifying AI generated output, cross-referencing information, and providing clear instructions. This collaborative approach allows educators to maintain control over the accuracy and reliability of educational materials, creating a balanced and effective learning environment.
Sample prompts:
  • List at least 5 arguments for xxx(topic) and their corresponding arguments
  • Can you adapt the language level to B2-C1?
  • Can you use bullet points to list the three main ideas and their corresponding main supporting ideas?
Using XIPU AI to generate testing materials was a valuable function performed by the chatbot in its evaluator role. Employing XIPU AI, we crafted quizzes to assess students' comprehension of our self-study materials, primarily comprised of video content covering diverse topics. Since the self-study activities aimed to encourage students to independently engage with the provided materials, we implemented strategies to gauge their learning progress. Instructing the chatbot to create quizzes with diverse question types helped focus students' attention on key information within the videos. Additionally, generating an answer key for each quiz was an integral part of our instructions.
One of the main advantages of using XIPU AI for quiz creation is its efficiency. By automating the process, instructors can save valuable time and effort that would otherwise be spent manually crafting quizzes. The chatbot's ability to generate quizzes quickly and accurately allows educators to focus on other important aspects of their teaching responsibilities (Bekeˇs & Galzina, 2023). Nonetheless, one downside of XIPU AI in quiz creation is its inability to directly process video materials. Since XIPU AI is a text-based language model, it cannot interpret audio or visual cues from videos. This poses a challenge when creating quizzes based on video content, as the chatbot relies solely on transcriptions or written summaries of the videos. This limitation may result in the omission of crucial visual or auditory elements that could be relevant for generating effective quiz questions. Instructors would need to ensure accurate transcriptions or summaries are available for the chatbot to work with, potentially requiring additional time and effort for the preparation of materials essential for quiz creation.
Sample prompts:
  • Please write ten quizzes with different questions types based on the content below:
  • Could you please create ten diverse types of quizzes based on the following content? Please provide the answer key as well.
Teaching assistant
The chatbot assumed the role of a teaching assistant in different distinct ways. One notable application was its function as a grammar checker. As part of the assessed tasks, students were required to write a report of approximately 1300 words. To facilitate targeted feedback on language use, instructors selected a paragraph from each report for closer examination. The chatbot was then employed to further assist in identifying and correcting grammar errors within the selected paragraphs, providing additional support to students in improving their writing skills. This approach allowed for targeted feedback and enhanced the effectiveness of the grammar checking process.
Using ChatGPT as a grammar checker offers both advantages and disadvantages. One of the main benefits is its ability to assist in identifying and correcting grammar errors within written texts. By employing the chatbot to analyze selected paragraphs from students' reports, an additional layer of assistance is provided, contributing to the refinement of their writing skills. However, it is essential to acknowledge certain limitations. ChatGPT may not consistently capture all grammar errors accurately, as it relies on pre-trained patterns and may not fully comprehend the context or nuances of the writing. For instance, Huang et al. (2023) noted that computational semantic analysis primarily emphasizes the denotative meanings of words, potentially overlooking the complete representation of connotative meanings. An illustrative example is evident in the given sample below, where the apostrophe indicating possession is already present in the phrase "students' weaknesses." in the original paragraph. Therefore, while ChatGPT can be a valuable tool for grammar checking, it should be used in conjunction with human feedback and guidance to ensure comprehensive and accurate language improvement.
Sample prompts:
  • Please highlight the grammar issues in the paragraph below, but please do not correct them in the paragraph:
  • Please list grammar errors in the paragraph without corrections:
  • Please point out grammar problems in terms of tense usage:


Chat GPT has significant educational potential, yet its purpose is not to replace human teachers. Instead, it serves as a support tool to enhance their professional expertise. Teachers recognize the power of Chat GPT in automating instructional tasks, but they view it as a tool that improves pedagogical effectiveness, not a substitute for teacher-student interaction. We believe that the teaching profession may need to be redefined with the help of this versatile technology, identifying the four roles that emerged from teaching.
Allouch, M.; Azaria, A. and Azoulay, R. (2021) ‘Conversational Agents: Goals,
Technologies, Vision and Challenges’, Sensors, 21(24), 8448. DOI: 10.3390/s21248448
Bekeˇs, E. R., & Galzina, V. (2023). Exploring the pedagogical use of AI-powered chatbots educational perceptions and practices. In 2023 46th MIPRO ICT and electronics convention (MIPRO) (pp. 636–641). https://doi.org/10.23919/ MIPRO57284.2023.10159734
Hartmann, J., Schwenzow, J., & Witte, M. (2023). The political ideology of conversational AI: Converging evidence on ChatGPT’s pro-environmental, left-libertarian orientation. arXiv. https://doi.org/10.48550/arXiv.2301.01768
Huang, X., Zou D., Cheng, G., Chen, X., & Xie, H. (2023). Trends, research issues and applica- tions of artificial intelligence in language education. Educational Society & Technology 26(1), 112–131. https://www.j-ets.net/collection/forthcoming-articles/26_1
Kim, J., Lee, H., & Cho, Y. H. (2022). Learning design to support student-AI collaboration: Per- spectives of leading teachers for AI in education. Education and Information Technologies, 27, 6069–6104. https://doi.org/10.1007/s10639-021-10831-6
Lund, B., & Ting, W. (2023). Chatting about ChatGPT: How May AI and GPT Impact Academia and Libraries? Library Hi Tech News, 40(3), 26-29. https://doi.org/10.1108/LHTN-01- 2023-0009
Pérez-Marín, D. (2021) ‘A Review of the Practical Applications of Pedagogic Conversational Agents to Be Used in School and University Classrooms’. Digital, 1(1)
  1. 18-33. DOI:10.3390/digital1010002
Rudolph, J., Tan, S., & Tan, S. (2023). ChatGPT: Bullshit spewer or the end of traditional assessments in higher education? Journal of Applied Learning and Teaching, 6(1), 342- 363. https://doi.org/10.37074/jalt.2023.6.1.9
Xu, W., & Ouyang, F. A. (2022). A systematic review of AI role in the educational system based on a proposed conceptual framework. Education and Information Technologies, 27, 4195–4223. https:// doi.org/10.1007/s10639-021-10774-y

Jiashi WANG (Language Lecturer)
Anying WANG (Senior Language Lecturer)
English Language Center
School of Languages

11 March 2024

Related Articles