LLM—Search Engine Killer?
summary
Compares LLMs to keyword/semantic search using a live RAG pipeline: OpenSearch + KNN for vector retrieval, MiniLM embeddings for indexing document chunks, and Llama 3 via AMA for generation. Builds a book/PDF Q&A demo, splits texts, indexes embeddings, retrieves top k and conditions the LLM; the demo mostly works despite Docker hiccups. Claims LLMs don’t replace search—retrieval quality still drives correctness—and notes a local 8B model is cheaper but less reliable than cloud giants.
AI-written summary · about
chapters (11)
- 0:00Introduction and background
- 2:13Will chat GPT kill search?
- 5:06History of information retrieval
- 9:23BERT revolution in search
- 18:42From search results to answers
- 19:24Generative models era
- 23:47Limitations of LLMs
- 29:03Retrieval augmented generation
- 32:38Live demo walkthrough
- 40:18Does search engineering change?
- 41:40Q&A session
related talks
46mLLM—Search Engine Killer?
01h 30mUsing LLMs to improve your Search Engine
24mConversational Search with LLMs
15mSearch Engineering Reimagined: Leveraging LLM in ML Workflows
01h 19mBuilding an LLM on your Laptop to...
51mQuery Understanding With Large Language Models: Techniques and...
36mOptimizing LLM Performance With Caching Strategies in OpenSearch
22mAgentic Hybrid Retrieval: Orchestrating LLM Agents, Hybrid Search & Explainability in OpenSearch