Designing and Developing a Generalized AI Bot for Adaptive Learning Assistance

Abstract

This project developed an AI-powered tutor to support K–12 students by offering personalized, adaptive learning and instant feedback. Using technologies like Retrieval-Augmented Generation (RAG), Large Language Models (LLMs), and tools like Ollama and LangFlow, the tutor delivers accurate, context-aware responses in natural language.Designed to complement—not replace—traditional teaching, the tutor targets areas where students typically struggle. It operates in two phases: ingestion (processing and storing educational content) and runtime (retrieving relevant information to answer student queries).While acknowledging challenges like the digital divide, data privacy, and AI bias, the project demonstrates the tutor\'s potential to enhance learning accessibility, engagement, and effectiveness.

Introduction

1. Background

Education has evolved from traditional systems like India’s Gurukul to industrial-age models focused on standardization. While effective in delivering basic skills, these systems often fail to address individual learning needs. The digital era introduced tools like smartboards and online platforms, and now, AI is revolutionizing education by offering personalized, adaptive learning through technologies such as machine learning and natural language processing.

2. Role of AI in Education

AI-powered tools like DreamBox and Squirrel AI offer real-time feedback, emotional understanding, and tailored learning experiences. These tools are especially promising in K–12 education, where traditional classrooms often lack individualization. However, challenges like data privacy, digital divide, and teacher displacement concerns hinder widespread adoption.

3. Project Objective & Approach

The project aimed to develop an AI tutor focused on helping K–12 students, especially in Physics, starting with the topic “Motion.” The tutor:

• Acts as a supplemental aid, not a replacement for teachers.

• Provides instant feedback, personalized instruction, and adaptive learning paths.

• Integrates Retrieval-Augmented Generation (RAG) with Large Language Models (LLMs) for accurate, conversational learning.

4. Development Methodology

• Phases:

  1. Educational system analysis.

  2. RAG implementation for retrieving trusted content.

  3. LLM integration for natural interaction.

  4. Combined RAG + LLM for contextual, accurate answers.

• Technologies Used:

  • Ollama + LLaMA3 for local LLM capabilities.

  • ChromaDB for storing/retrieving educational content.

  • LangFlow to structure conversations.

5. Student Data Analysis

• 113 students participated in testing.

• Word clouds and correlation analysis showed students prefer:

  • Fun, hands-on, creative, interactive learning.

  • Autonomy and real-world examples.

  • Self-paced and multisensory learning.

• Insights:

  • Some learners enjoy solo, hands-on learning more than group work.

  • Real-life examples and digital tools should be thoughtfully combined.

  • Students support gamified, emotionally intelligent learning platforms.

6. New AI Tutor Framework

• Built on a student-centered, emotion-aware RAG architecture.

• Provides:

  • Natural interaction with emotional understanding.

  • Contextual, tailored responses based on user’s grade, learning style, and prior knowledge.

  • Scaffolded learning, with clear explanations and active recall.

  • Adaptive and emotionally intelligent support—not just Q&A.

7. Implementation

Two key phases:

• Phase 1: Knowledge Setup

  • Loads and embeds textbook content for semantic search.

• Phase 2: Answer Generation

  • Uses ChromaDB and LLMs to create context-rich, accurate, and student-specific responses.

8. AI Bot Interaction Design

• Friendly, emotionally intelligent tone.

• Adapts based on user emotion (e.g., boredom or confusion).

• Offers structured learning plans and tailored explanations.

• Supports interactive, engaging learning through jokes, experiments, and relatable content.

• Politely limits responses to its domain (e.g., avoids unrelated topics like sports).

9. Student Feedback

• Most students described the tutor as friendly and approachable, making them feel comfortable asking questions.

• The “Motion” chapter content was especially well-received for clarity and structure.

• Over 80% would reuse the tutor, and many requested expansion to other subjects.

• Limitations include a narrow academic focus and limited real-time information on non-educational topics.

Conclusion

This project aimed to improve K–12 learning, particularly in Physics, by integrating AI as a supportive tool rather than a replacement for teachers. The AI tutor offers personalized, accessible, and engaging assistance with complex topics like \"Motion.\" Built on advanced AI technologies—including Retrieval-Augmented Generation (RAG), GPT-based LLMs, LLaMA3 via Ollama, ChromaDB, and LangFlow—the tutor delivers clear, context-aware, and tailored explanations aligned with curricula. Its key strength is adaptability, adjusting to each student’s pace, comprehension, and emotions to provide a human-like, encouraging learning experience. Feedback from 113 students helped enhance usability and relevance. Despite challenges like digital accessibility and AI limitations, the project demonstrates AI’s potential as an educational ally for students needing extra support. With ongoing development, the tutor could become a trusted, engaging companion that helps students learn effectively and enjoyably.

References

[1] Akgun, S., & Greenhow, C. (2021). Artificial intelligence in education: Addressing ethical challenges in K-12 settings. AI and Ethics. https://doi.org/10.1007/s43681-021-00096-7 [2] Ali, D., Hassan, K., Shazia, S., & Dragan, G. (2024). Impact of AI assistance on student agency. Computers & Education, 210, 104967. [3] Ayeni, O. O., Hamad, N. M., Chisom, O. N., Osawaru, B., & Adewusi, O. E. (2024). AI in education: A review of personalized learning and educational technology. GSC Adv Res Rev, 18(2), 261–271. [4] Baig, A., Cressler, J. D., & Minsky, M. (2024). The future of AI in education: Personalized learning and intelligent tutoring systems. AlgoVista J AI Comput Sci, 1(2). [5] Baillifard, A., Maxime, G., Pamela, B. L., & Corinna, S. M. (2023). Available at: arXiv:2309.13060v1 [cs.CY]. [6] Celik, I., Dindar, M., Muukkonen, H., & Jarvela, S. (2022). The promises and challenges of artificial intelligence for teachers: A systematic review of research. TechTrends, 66, 616–630. https://doi.org/10.1007/s11528-022-00715-y [7] Chassignol, M., Khoroshavin, A., Klimova, A., &Bilyatdinova, A. (2018). Artificial intelligence trends in education: A narrative overview. Procedia Computer Science, 136, 16–24. https://doi.org/10.1016/j.procs.2018.08.233 [8] Chen, X., Xie, H., & Hwang, G. J. (2020). A multi-perspective study on Artificial Intelligence in Education. Computers and Education: Artificial Intelligence, 1, Article 100005. https://doi.org/10.1016/j.caeai.2020.100005 [9] Chiu, T. K. F., Xia, Q., Zhou, X., Chai, C. S., & Cheng, M. (2023). Opportunities, challenges, and future research recommendations of AI in education. Computers and Education: Artificial Intelligence, 4. https://doi.org/10.1016/j.caeai.2022.100118 [10] Cope, B., Kalantzis, M., &Searsmith, D. (2020). Artificial intelligence for education: Knowledge and its assessment in AI-enabled learning ecologies. Educational Philosophy and Theory. https://doi.org/10.1080/00131857.2020.1728732 [11] Goksel, N., & Bozkurt, A. (2019). Artificial intelligence in education: Current insights and future perspectives. In Handbook of Research on Learning in the Age of Transhumanism (pp. 224-236). IGI Global. https://doi.org/10.4018/978-1-5225-8431-5.ch014 [12] Kochmar, E., Dung, D. V., Robert, B., Gupta, V., Iulian, V. S., & Joelle, P. (2020). Automated personalized feedback improves learning gains in an intelligent tutoring system. In Artificial Intelligence in Education (pp. 140–146). Springer. [13] Rizvi, M. (2023). Investigating AI-powered tutoring systems that adapt to individual student needs. ICRESS, 31, 67–73. [14] Salem, B. I. (2024). Integrating artificial intelligence in personalized learning. Int J Post Axial: Futuristic Teach Learn, 111–119. [15] Singh, A. P., Saxena, R., & Saxena, S. (2024). The future of learning: AI-driven personalized education. Asian Journal of Current Research, 9(4), 207–226. https://doi.org/10.56557/ajocr/2024/v9i49018 [16] Tuomi, I. (2018). The impact of artificial intelligence on learning, teaching, and education. Publications Office of the European Union. https://doi.org/10.2760/12297

Copyright

Copyright © 2025 Siddharth P. Sharma. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.