How to AI in JavaScript
summary
Five live demos show AI running entirely in the browser: a Teachable Machine banana classifier trained and inferred offline with Tensorflow.js; a natural-language to-do parser using Transformers.js that answers in ~0.8s and a zero-shot labeler at 1.4s; a streamed rhyme generator that takes ~12s; a 20-questions loop implemented with LangChain.ts on Node; and a Chrome-embedded Gemini Nano session that performs offline conversations. Models cache locally, require no servers, and the demos successfully execute end-to-end.
AI-written summary · about
chapters (8)
- 0:00Introduction to AI in JavaScript
- 2:53Why run AI in the browser
- 5:01TensorFlow.js overview
- 6:06Demo: Text to speech and speech to text
- 7:11Demo: Hugging Face Transformers JS
- 13:45Demo: LangChain in Node.js
- 15:10Demo: Chrome’s built-in AI model
- 17:20Future of AI in JavaScript
related talks
44mArtificial Intelligence and Machine Learning in the browser
46mArtificial Intelligence with JavaScript
34mThe Future of JavaScript and AI
34mAI and JavaScript, no You're Not Dreaming
34mThe Future of JavaScript is A.I.
22mBoring but Effective: WebAI That Ships
39mMachine Learning Magic for Your JavaScript Application
35mHow to AI in JS?