Open Category Submission

941 – An AI Chatbot Prototype to Supplement Traditional Learning Management Systems

There is an increase in student diversification in terms of mother tongues, study modes (full-time/part-time/blended) and demographics. Many of these students require continuous, personalized attention, which teachers cannot provide beyond class hours. Thus, many students’ academic needs often go unmet due to limited interaction. The customizable chatbot is a potential solution by providing round-the-clock personalized support and fostering engagement with students.

Key Innovations:
While AI is powerful, it may go off-topic or generate false information known as hallucinations. The project’s key innovation is the integration of a chatbot interface with Natural Language Processing (NLP) and Retrieval-Augmented Generation (RAG) to ensure the delivery of accurate, contextually relevant information tailored to each student’s needs.

Design Concepts:
Some students are overwhelmed by the information and choices on a course website. The chatbot can supplement this by providing bite-size information in a dialogue format. It is cloud-based to ensure scalability and security. The contents for structured communications, like quizzes and motivational messages, are table-driven, allowing non-IT teachers to manage them easily. For answering free-text, open questions from students, teachers just need to upload notes and slides to the knowledge base. Responses to students are enriched with emojis to enhance engagement and reliability.

Technical Principles:
The chatbot operates seamlessly in the cloud and is accessible any device with an internet browser (PCs, tablets, smartphones). It allows students to ask free-text questions flexibly, even when a menu is displayed. The NLP will answer questions in the language in which questions are asked. Thus, students can interact with the chatbot in their preferred language, without rigid menu structures.

Potential Impact:
The chatbot has the potential to strengthen teacher-student relationships by addressing both academic and emotional needs. It supports students from diverse backgrounds, fosters reflection, and cultivates skills that are invaluable for their future studies and professional lives.

Problem Identification and Relevance in Education
The project was motivated by the increasing challenge of providing personalized, immediate, and adaptive learning support to students in diverse educational contexts. Hong Kong is seeing an increasing proportion of non-local/foreign student in higher education, and students studying in part-time or full-time modes, online and face-to-face modes. Traditional classroom settings often struggle to cater to individual mother-tongues, learning styles and paces, leaving some students feeling unsupported and disengaged. This gap inspired us to explore the potential of a chatbot prototype to address these challenges and improve educational outcomes.
Our hypothesis is that an AI-powered chatbot can enhance learning experiences by delivering personalized, adaptive, and immediate support to students. The chatbot, called SPEED AI Buddy (Buddy), launched in January 2025 this year, is designed to cater to individual learning needs by sending them personalized messages based on their progress in a course. Buddy reminds students the quizzes they haven’t completed and provides encouragements during the quizzes. Buddy also lets students ask questions using free-text input in their preferred language.

We hypothesize that our chatbot will enhance student outcomes due to its intuitive, human-like dialogue interface, fostering natural interactions. Students increasingly accept AI chatbots, such as ChatGPT and DeepSeek, as a convenient way to access vast varieties of information. However, these generic GenAI-powered chatbots may provide incorrect information in a phenomenon known as hallucination. Furthermore, these generic chatbots lacks the ability to guide students towards a study goal. Buddy’s custom-built design mitigates off-topic discussions and hallucinations, ensuring accurate, relevant, and context-specific educational support tailored to learners' needs.

Through this project, we aim to contribute to the competition’s goals of promoting innovation, fostering responsible AI practices, and supporting skill development, while addressing a critical need in education.

We will leverage existing large language models (LLMs) to build and enhance the prototype. The AI Buddy runs on a cloud computing platform called Botpress. The resource required will be the money to keep the performance of AI Buddy to use AI to understand student enquiries in free-text questions. The prototype has been developed to an extend that further development is only required to input the teaching materials for each specific course. The requirement for computer programming is minimal. A teacher who is competent with dragging and dropping icons should be competent enough to tailor AI Buddy to suite his/her course. Additional effort, however, is need to add multimedia contents such as more interesting images as avatars.

The core functionalities of the solution is to provide round-the-clock support to students who may want to raise their questions in free-text format and interact with the learning and teaching materials by issuing queries. Positive user experience is ensured by monitor the unanswered questions that are keep in a specific table. Then the teacher and fine tune the documents in the knowledge base and the AI engine by providing the desirable answers to those unanswered questions. The primary performance metrics to evaluate its effectiveness will be the student performance in the course, the low ratio of unanswered questions and student satisfaction in a survey.

not applicable

The SPEED AI Buddy represents an innovative and creative solution by addressing the challenges of providing personalized, round-the-clock student support. Unlike traditional learning management systems (LMSs) such as Blackboard or Moodle, the chatbot utilizes AI technologies, including Natural Language Processing (NLP) and Retrieval-Augmented Generation (RAG), to deliver contextually relevant, human-like interactions tailored to each student’s needs.

The NLP is driven by a large-language model, such as ChatGPT-4o, to understand the free-text question entered by students. For example, to know the date for the group project presentation, the student can ask in various ways such as “when is the group project presentation?”, or “when is the group project presentation?”. The NLP will convert the question into tokens so that the RAG can retrieve the information from the instructions, notes and slides uploaded by the teacher. Then AI Buddy will answer the question in the language what was used by the student. If the student asked the question in simplified Chinese, the answer will be in simplified Chinese.

Teachers can easily upload their notes and slides to the knowledge base without requiring technical expertise, ensuring that the chatbot remains up to date with the latest course materials. Moreover, the chatbot’s responses are enriched with emojis and conversational styles that can be customized to suit different student demographics, making interactions more relatable and engaging.

Creativity is demonstrated in the chatbot’s design, which goes beyond providing information and quizzes. When students are doing a quiz, AI Buddy will show encouraging messages, which are not available in LMSs like Blackboard. At the end of each quiz, the AI Buddy shows an animated congratulations message, and a summary of the students’ progress in terms of a progress bar. Furthermore, AI Buddy offers motivational messages through the Yoda character from the Star Wars movie.

To ensure scalability, the AI Buddy uses cloud-based architecture that can dynamically allocate resources to meet fluctuating user demands. As the number of students grows, the system scales up without compromising performance. Potential bottlenecks, such as increased server loads or delayed response times, are mitigated through load balancing and distributed data storage. These are managed by the cloud service provider through a “pay-as-you go” subscription. Conversely, if the number of students goes down, the expense can be reduced. The school does not need to commit to fixed costs of hardware and manpower.
The AI Buddy was designed with sustainability in mind. The retrieval augmented generation (RAG) feature allows teachers to easily upload their notes, slides, and project instructions to the knowledge base without requiring technical expertise. This ensures that the chatbot remains up to date with the latest course materials without the need for additional technical support. The RAG will use a large-language model (LLM) to understand free text questions from students and retrieve the related information accordingly. The platform allows the teacher to switch to the most suitabl LLMs after using the AI Buddy for a certain period.
In case LLM and RAG cannot understand students’ question, the unanswered questions are stored in the HistoryTable. The Excel-like interface allows the teacher to easily filter out the unanswered questions and update the knowledge base. This is known as training the chatbot in an on-going basis. The teacher can also export the tables as Excel file for further analysis.
The quizzes and motivational messages are stored in the QuizTable and MotivationTable respectively. The teacher can easily create and update the quizzes and motivational messages using an Excel-like interface. The teacher can also create the quizzes and motivational messages using in Excel format and then import them into the respective tables.

Our chatbot solution addresses critical social issues in education by providing access to personalized learning support to that student from diverse socioeconomic and cultural backgrounds. The teacher can import the performance of the students from the learning management system (LMS) into the User Table. Then AI Buddy will show personalized messages according to the student’s performance in the course. It will also ask the student to reflect and enter his/her improvement plans.
The solution aligns with broader social goals of equity and inclusion by empowering teachers with tools to deliver customized support without requiring technical expertise. This enables underserved communities, where resources and teacher availability may be limited, to benefit from high-quality, scalable educational assistance. Additionally, by fostering self-reflection, emotional awareness, and critical thinking, the chatbot prepares students for future challenges in both professional and personal contexts.
To measure social impact, we will track key metrics such as student usage of the AI Buddy. When students interact with AI Buddy, their actions are recorded in the HistoryTable. We will also collect student satisfaction ratings through surveys. Additional insights will be collected using focus group discussions to gain further understanding on the numbers returned by the survey statistics.
ML-GP-Buddy, the predecessor of AI Buddy was implemented for students’ use in September 2025 to December 2025. An online survey was conducted. There were 47 valid responses, and the average satisfaction with the pre-prototype was 3.7 out of 5. In the survey, 5 means “Strongly Agree” and 1 means “Strongly Disagree” to the satisfaction statements.
Responsiveness to community needs is ensured through a continuous feedback loop. Teachers, students, and the applicant will have opportunities to provide input, enabling iterative improvements to the chatbot’s functionality and content.

Personal Information Collection Statement (PICS):
1. The personal data collected in this form will be used for activity-organizing, record keeping and reporting only. The collected personal data will be purged within 6 years after the event.
2. Please note that it is obligatory to provide the personal data required.
3. Your personal data collected will be kept by the LTTC and will not be transferred to outside parties.
4. You have the right to request access to and correction of information held by us about you. If you wish to access or correct your personal data, please contact our staff at lttc@eduhk.hk.
5. The University’s Privacy Policy Statement can be access at https://www.eduhk.hk/en/privacy-policy.

• I have read and agree to the competition rules and privacy policy.