Data, visualization, and designing AI

on YouTube ↗01h 00m

" summary

Hands-on demos show concrete tools and findings: Facets visualizes CIFAR-10 training images and surfaces mislabels (a frog labeled cat) and softmax confidence patterns; the What-If Tool (code-free, open source, integrated into Cloud AI) lets users probe thresholds, fairness presets and counterfactual nearest neighbors; an embedding projector and TCAV extract human-meaningful directions from high‑dimensional activations exposing gender bias in word vectors; a pathology system called SMILY adds sliders for medical concepts and wins a study with 12 pathologists who prefer it and report higher trust. Tools are JavaScript/D3/WebGL and TensorFlow.js.

AI-written summary · about

" chapters (14)

1. 0:00Start
2. 0:13Martin Watenberg and Fernando Villegas
3. 2:23Why Are We Talking about Human Ai Interaction
4. 5:08Demo
5. 5:24Data Visualization
6. 7:16Confusion Matrix
7. 18:01Demographic Parity
8. 24:11Data in High Dimensions
9. 28:32Word Embedding
10. 37:34Concept Activation Vector
11. 52:16What Tools Did You Use To Make these Visualizations
12. 54:45Ai Literacy
13. 57:25Adversarial Images
14. 59:28Tensorflow Js

" related talks

• 46mData Discussion - AI for GoodPeter Morgan, Tarry Singh, Alisa Piterova · Data Science Conference 5.0
• 11mAI-powered JavaScript applicationsAsim Hussain · The Lead Developer London 2019
• 01h 03mIt's All About AIJonas Mayer, Martin Förtsch, Thomas Endres · YOW! Melbourne 2022
• 17mAI/ML demo derbyZack Akil, Kaz Sato, Markku Lepisto · Google I/O 2021
• 41mModels Schm-odels: Why you Should Care about Data-Centric AIMarysia Winkels · PyData London 2022
• 39mAI + JavaScript RocksAsim Hussain · FrontEnd Con 2018
• 50mExplainable AI Analysis & Visualization: Applications from Brain Computer Interfaces to ChatGPTSean M Phillips · Devoxx Ukraine 2023: AI – Friend or Foe?
• 40mResponsible AI Practices: Fairness in MLAlex Hanna · PyData Miami 2019