Artificial Intelligence vs. Natural Stupidity
summary
A set of silly experiments applies practical models to toy problems: a pre-trained TensorFlow.js model predicts FizzBuzz by converting 1–100 into tensors and selecting the highest-probability label, code and model linked; an in-browser trainer at tensorflow-tic-tac-toe.com runs 100-epoch updates from one or two example games and quickly improves play; a face classifier on nickornot.com / declarationofindependencethief.com detects Nicholas Cage–like faces. All projects are open source, runnable in the browser, and demonstrate where AI succeeds and amusingly fails.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:27AI vs Natural Stupidity
- 3:56Speaker introduction
- 9:36AI solving FizzBuzz
- 18:04Training AI for Tic-Tac-Toe
- 23:16Facial recognition demo
- 26:02Closing and resources
related talks
44mArtificial Intelligence and Machine Learning in the browser
56mArtificial Intelligence? - more like Artificial Stupidity!
18mHow to AI in JavaScript
34mThe Future of JavaScript and AI
26mArtificial Intelligence? - More like Artificial Stupidity!
35mAI JavaScript Rocks
34mThe Future of JavaScript is A.I.
41mAI in Web: TensorFlow js