philip lelyveld The world of entertainment technology

How a Decades-Old Technology and a Paper From Meta Created an AI Industry Standard

... Vector databases have been around for decades, but are now emerging as something of an industry standard for AI businesses to use alongside a technique called retrieval-augmented generation, or RAG.

When combined, businesses can link their private data with large-language models like OpenAI’s GPT-4, allowing the AI to perform data analysis, summarization and other tasks on their data. Without them, AI models are limited to what they have learned from their initial training on public data online, up to a certain point in time, and are more prone to factual errors called “hallucinations.”

New York-based startup Pinecone was an early entrant in the vector database AI space. ...

Vector databases are finding a new use by enterprise and AI developers because they power the RAG technique. Originating from a 2020 paper by an AI research group at Meta Platforms, RAG is commonly used by enterprises to build chatbots for employees to reference company policies, or for customer service and salespeople to pull information from knowledge bases.

Vector databases are different from traditional databases with columns and rows because they are designed to store a massive amount of data as “vectors,” or numerical representations of the raw data. That makes them ideal for RAG, the process where generative AI models pull from large amounts of vector data to improve their responses with the additional information.

RAG vs. fine-tuning

Compared with fine-tuning—a technique used to create a custom AI model based on an existing large language model—RAG is cheaper because it doesn’t require massive amounts of computing power and advanced AI expertise, and is easier and faster for developers to implement. Roughly 80% of enterprises are using RAG, compared with 20% using fine-tuning, Chandrasekaran said. ...

See the full story here: https://www.wsj.com/articles/how-a-decades-old-technology-and-a-paper-from-meta-created-an-ai-industry-standard-354a810e