Research in the Age of AI
summary
Overview of search evolution from classical term-frequency methods (TF/IDF, BM25) toward machine-learning–based approaches using vector and hybrid search. Explains what ML models are, how to use them in Elasticsearch for vector/hybrid search, and how LLMs (like OpenAI’s ChatGPT) integrate with Elastic; includes a demo generating music embeddings to find songs similar to a hummed melody.
AI-written summary · about
chapters (12)
- 0:00Vector search intro
- 2:31Vector basics and semantic search
- 5:40Vector indexing in Elasticsearch
- 7:44Native vector search tools
- 10:56ML model selection and limits
- 13:27Score calculation and optimization
- 15:16Memory and indexing best practices
- 17:47Performance and quantization
- 20:17Demo with audio files
- 29:07Hybrid search methods
- 34:06LLM impact and RAG with Elasticsearch
- 40:32Conclusion
related talks
42mResearch in the Age of AI
39mExperiments with OpenSearch and AI
38mNatural Language Search-Lexical and Semantic
22mMachine Learning: The Engine Behind Modern E-commerce Search
30mBeyond Keywords: Building an AI-Powered Search That Reads Your Mind
40mLevel Up: Everything you need to know about search, ...
47mBoosting AI With OpenSearch: Vector Stores, Similarity Search, and Retrieval-Augme...
41mText to Context: How We Introduced Hybrid Search