QuestAI is an AI-driven educational platform designed to enhance learning efficiency and engagement. By leveraging state-of-the-art language models and responsible AI practices, QuestAI provides tailored educational support, including material summarization, quiz generation, content creation, and chatbot-based clarification.
- Learning Material Summarization: Generate concise summaries of lengthy academic materials, emphasizing key ideas and conclusions.
- Interactive Chatbot: Answer user queries related to summaries and provide further clarification.
- Learning Content Generation: Create structured and engaging educational materials based on specific topics and difficulty levels (High School, Undergraduate, Graduate).
- Quiz Generation: Generate multiple-choice questions with explanations to reinforce understanding of key concepts.
- Responsible AI Practices: Filters to prevent harmful or inappropriate content and ensure privacy by not retaining input documents.
- React.js: Component-based architecture for dynamic rendering.
- Chakra UI: Comprehensive and customizable user interface library.
- Flask: Lightweight and agile backend framework hosted on Render.
- AI Integration:
- Gemini v1.0: For multimodal learning content and quiz generation.
- PaLM-2: For summarization and chatbot capabilities.
- Google Cloud Vertex AI: For scalable and reliable AI performance.
Visit the live application at QuestAI.
- Clone the repositories:
- Follow the setup instructions in each repository's
README.md.
- Material Summarization:
- Upload a PDF document to generate a summary.
- Chatbot Interaction:
- Query the chatbot for further clarification on summaries.
- Content Creation:
- Specify a topic and difficulty level to generate learning materials.
- Quiz Generation:
- Create interactive quizzes to test your understanding of a subject.
- AI Models:
- Gemini v1.0: Fine-tuned for academic content generation and quiz creation.
- PaLM-2 Text-Bison: Specializes in summarization and natural language understanding.
- Hyperparameter Tuning:
- Temperature, Top-P, and Top-K optimized for task-specific outputs.
- Prompt Engineering:
- Custom prompts ensure high-quality and structured responses.
- User Surveys:
- 96.7% of users found content generation beneficial.
- 83.3% agreed quizzes reinforced their understanding.
- Average user satisfaction score: 4.6/5.
- Safety and Trust:
- 0% reported harmful or inappropriate content.
- Responsible AI practices ensure safe and accurate outputs.
- Multimodal Content:
- Extend support for visuals and videos.
- Personalization:
- Advanced algorithms for tailored learning experiences.
- Continuous Model Improvement:
- Feedback loops for dynamic updates and refinements.
- Kumar Prabhat: Backend architecture, model tuning, Google Cloud deployment.
- Lin Weilin: Frontend design, prompt engineering, hyperparameter tuning.
- Wang Haitao: Interactive quiz design, experimentation, prompt engineering.
This project is licensed under the MIT License.