Postgres and the Artificial Intelligence Landscape
summary
Explains how to run machine learning inside Postgres using a PL/Perl demo that builds an 11-neuron tensor and trains it with a 10,000-number training set to predict whether a number contains the digit zero. Implements weight updates, a softmax step, and evaluation inside the database; results run at roughly ~72% accuracy ±15% in the demo. Also surveys PGVector, GPU acceleration and when to keep preprocessing or model calls inside Postgres.
AI-written summary · about
chapters (10)
- 0:00Introduction and conference context
- 1:12What is artificial intelligence
- 5:30History of AI and AI winter
- 9:07Machine learning basics
- 15:44PL Perl AI demo in PostgreSQL
- 29:59Hardware and software for AI
- 33:49AI use cases and fraud detection
- 38:54Why use a database for AI
- 43:39Generative AI and future outlook
- 48:29Q&A and closing
related talks
01h 00mPostgreSQL and Artificial intelligence
50mFrom Transactions to Intelligence: Postgres for Open AI & Analyti
37mRound Table Discussion
35mPostgres in Artificial Intelligence
41mPostgreSQL for AI. Real Vector Search in PostgreSQL
48mHow I found my Pokémon cards thanks to Postgres: an AI journey
33mScalable in-database machine learning with PL/Python
32mSimplify End-To-End MLOps with PostgresML