AI-Powered YouTube Video Assistant

This research presents an AI-Powered YouTube Video Assistant designed to improve the way users interact with YouTube video content through intelligent question-answering capabilities. The system successfully integrates transcript extraction, semantic retrieval, and Retrieval-Augmented Generation (RAG) techniques to provide accurate and context-aware responses from lengthy videos. By utilizing technologies such as FastAPI, FAISS, FastEmbedEmbeddings, and the LLaMA 3.1 model via Cerebras API, the proposed system enables efficient transcript analysis and interactive learning support.   The implementation of the Chrome Extension interface and backend APIs demonstrates how Artificial Intelligence and Natural Language Processing can be effectively applied to video-based learning systems. The system reduces the time required to search for important information manually and provides structured responses within a few seconds. Experimental results indicate that semantic retrieval combined with Large Language Models significantly improves response relevance and user interaction compared to traditional keyword-based approaches.   The major contribution of this research lies in developing a lightweight and practical AI-based assistant that enhances accessibility and learning efficiency from YouTube videos. Although the system currently depends on transcript availability and internet connectivity, future improvements such as multimodal analysis, voice interaction, persistent vector databases, and real-time video processing can further enhance its capabilities. The proposed system demonstrates the growing potential of AI-driven educational assistants for interactive and intelligent digital learning environments.

1.            Google AI. (2024). Cerebras API / LLaMA Model Documentation. Available  at: https://docs.cerebras.ai 2.            Python Software Foundation. (2024). Python Language Reference and Documentation. Available at: https://www.python.org/doc 3.            FastAPI. (2024). FastAPI Documentation for Building APIs.  Available   at: https://fastapi.tiangolo.com 4.            LangChain. (2024). LangChain Documentation for RAG Applications.   Available   at: ttps://www.langchain.com 5.            FAISS. (2024). FAISS: Efficient Similarity Search and  Clustering.         Available          at: https://faiss.ai 6.            YouTube Transcript API. (2024). YouTube Transcript API Documentation. Available at: https://pypi.org/project/youtube-transcript-api 7.            Chrome Developers. (2024). Chrome Extension Documentation (Manifest V3). Available at: https://developer.chrome.com/docs/extensions 8.            Brown, T. et al. (2020). Language Models are Few-Shot Learners. Advances in Neural Information Processing Systems (NeurIPS). 9.            Vaswani, A. et al. (2017). Attention Is All You Need. Proceedings of the 31st Conference on Neural Information Processing Systems. 10.       Lewis, P. et al. (2020). Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. arXiv preprint arXiv:2005.11401. 11.       Reimers, N., & Gurevych, I. (2019). Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks. 12.       Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep Learning. MIT Press.