Learning about AI from kids
summary
Hands-on classroom work uses Scratch blocks and a child-friendly trainer to teach the full ML workflow: predict labels, collect examples, train models, test on unseen data, iterate. Projects include chatbots that map questions to answers, a Pac-Man model that improves from random to evading ghosts as examples increase, image classifiers that reveal spurious cues (left foot, background) and a postal OCR app; students build, break, visualize confidence scores and fix bias.
AI-written summary · about
chapters (15)
- 0:00Introduction to AI with kids
- 1:10Scratch and AI for young learners
- 2:57Training chatbots with students
- 4:31Pac-Man and training data quantity
- 7:27Crowdsourcing training data
- 8:53Separating training and test data
- 11:05Sentiment analysis with compliments
- 12:10Virtual assistants and edge cases
- 15:05Confidence scores in AI models
- 17:51Interpreting black box models
- 20:28When AI gets it wrong
- 23:39Fixing unexpected AI behavior
- 26:51Recognizing handwritten postcodes
- 28:27Judging books by their covers
- 29:30Key lessons from kids about AI
related talks
36mLearning about machine learning from kids
06mCreating and Teaching Games in Python to Teach Artificial Intelligence Concepts
44mArtificial Intelligence and Machine Learning in the browser
26mHow Machine Learning Can Help
23mKeynote: What is Learning in the AI Era?
42mMachine Learning Crash Course
24mDon't want to miss out on the AI wave?
54mMachine Learning explained with Examples