AI animation course series
AI without the mystery
Brains, models, tools, and why AI can guess wrong.
10 friendly short lessons
- Start with familiar questions
- Learn through recurring comic characters
- Leave each lesson with something to try
Pick what you're curious about
AI without the mystery
10 friendly short lessons
-
Is AI magic or a word game?
Story setup: BIT predicts the next word and jumps from watermelon to the moon.
What you will learn: Learn that generative AI makes patterns-based outputs; it cannot read minds.
Open this lesson's animated storyboard → -
Who does what: model, app, or tool?
Story setup: Gulu puts a brain on a drill but forgets to plug it in.
What you will learn: Tell apart the AI model, the app interface, and connected tools.
Open this lesson's animated storyboard → -
Why can AI sound sure and still be wrong?
Story setup: Goose mistakes “sounds true” for “has evidence.” The meeting vanishes into fog.
What you will learn: Spot invented claims and practice checking sources and cross-checking facts.
Open this lesson's animated storyboard → -
How are AI, machine learning, and generative AI related?
Story setup: BIT packs a picture-making task into one tiny box. Director Goose draws an umbrella and overlapping circles, then BIT gets flattened under four hats.
What you will learn: AI is the broad field; machine learning and deep learning describe methods; generative AI describes creating new content. These categories can overlap.
Open this lesson's animated storyboard → -
Training vs. answering: what is the difference?
Story setup: BIT drops Gulu’s “no peanuts” breakfast order into a brain blender, and Director Goose rushes over to hit pause.
What you will learn: Training uses many examples to adjust a model; inference uses a trained model to answer. Check a service’s settings and policy to learn how it uses chats.
Open this lesson's animated storyboard → -
What are tokens and a context window?
Story setup: BIT tries to stuff an entire encyclopedia into a tiny suitcase. The zipper pops, and books fly everywhere.
What you will learn: A model can use only a limited amount of context at once; that is not permanent memory. Share relevant sections and check what the AI actually read.
Open this lesson's animated storyboard → -
Training data: How does AI learn patterns?
Story setup: After ten photos, BIT mistakes yellow sticky notes for signs of a ripe watermelon.
What you will learn: Incomplete, biased, or mislabeled examples can teach the wrong clues; test with varied examples.
Open this lesson's animated storyboard → -
How does a neural network spot clues?
Story setup: BIT stacks cat-clue cards into layers, then gives every card a graduation cap.
What you will learn: A neural network is connected, layered calculation inspired by brains—not a tiny human brain.
Open this lesson's animated storyboard → -
Does AI ace practice but fail a new test?
Story setup: BIT has only seen cats in red hats, so it says a cat without one is not a cat.
What you will learn: Hold out new examples to check whether a model learned a pattern that generalizes.
Open this lesson's animated storyboard → -
Three common ways for a model to learn
Story setup: BIT spins between labeled fruit, mystery-sock groups, and a maze that awards stars.
What you will learn: Supervised learning uses labeled examples; unsupervised learning finds structure; reinforcement learning improves through action feedback.
Open this lesson's animated storyboard →