Lesson 1

Patterns and Predictions

How does AI know what to say next? It guesses — really well.

Zoe was playing a word game with an AI. She typed: "The cat sat on the..." and waited.

The AI finished: "...mat."

"How did you know that?" she asked.

"I've seen that phrase thousands of times," the AI explained. "In books, poems, and sentences online. After 'the cat sat on the,' the next word is almost always 'mat.' I predicted what came next based on a pattern."

Zoe tried a harder one: "The purple elephant carefully..." The AI finished: "...stepped over the sleeping dog."

"That's different every time," she said. "It's not just one answer."

"Right," said the AI. "For unusual phrases, there are many possible next words. I pick based on what makes sense."

AI Learns Patterns

AI language tools work by learning patterns from enormous amounts of text. When you ask a question or give it a prompt, it predicts what words should come next — based on everything it has read.

Common phrases have predictable completions
Unusual phrases have many possibilities
The AI picks the most likely completion based on context

Not Magic — Patterns

AI doesn't "understand" language the way you do. It predicts what words fit together based on patterns in millions of examples. That can look like intelligence — and sometimes it is useful in the same way!

Back Take Quiz

Quiz 1

Patterns and Predictions

3 questions — free, untracked, retake anytime.

did the AI know to complete 'The cat sat on the...' with 'mat'?

✓ Correct — ✅ ✓ AI learns from patterns in millions of examples. It predicts what comes next based on what it has seen before.

❌ ❌ AI doesn't look things up — it learned the pattern from seeing the phrase thousands of times and predicts the most likely next word.

did the AI give a different kind of answer for 'The purple elephant carefully...'?

✓ Correct — ✅ ✓ Common phrases have predictable completions. Unusual ones have many possibilities — so the AI picks something that makes sense in context.

❌ ❌ Unusual phrases don't have one obvious completion, so the AI has more options to choose from based on what makes sense.

is the most accurate description of how AI language tools work?

✓ Correct — ✅ ✓ AI language tools predict the next likely word based on patterns learned from enormous amounts of text. It's pattern prediction — not thinking or searching.

❌ ❌ AI language tools predict next words based on patterns from millions of examples. They don't search the internet or think like a human.

Back to Lesson Start Lab 1

Lab 1

Pattern Detective

Explore how AI uses patterns.

Lab 1 — Pattern Detective

Explore how AI uses patterns with your AI guide.

The AI asks you to think of a phrase that has an obvious completion — and one that doesn't.
Discuss: why does one have an obvious answer and the other doesn't?
Think about: what does this tell you about how AI works?

Try to think of a phrase so unusual that almost any word could come next. What would that tell you about AI's limits?

⭐ AI GuideLab 1

Back to Quiz Next: Lesson 2

Lesson 2

Learning from Examples

AI doesn't start knowing things — it learns from enormous amounts of text.

Ben's teacher asked the class to imagine teaching a robot to recognize cats. "You can't just tell it 'cats have four legs and fur,'" she said. "You'd have to show it thousands of pictures of cats — and thousands of pictures of things that aren't cats. Over time, the robot would figure out the patterns on its own."

"So it learns like we do?" Ben asked.

"In some ways," his teacher said. "But you learn from a few examples and can explain your reasoning. The robot needs many more examples — but it can process them much faster than you can."

Training Data

AI learns from examples called training data. For language AI, this means billions of words from books, websites, and other text. For image AI, it means millions of pictures.

More examples usually means better learning
The quality of examples matters — bad examples teach bad patterns
AI can only learn what's in its training data

How Humans and AI Learn Differently

You can learn from one or two examples and generalize
AI usually needs many examples to learn a pattern
You can explain your reasoning; AI often can't
AI can process data much faster than any human

Key Point

AI doesn't "know" things the way you know things. It has learned statistical patterns from many examples. That's powerful — and different.

Back Take Quiz

Quiz 2

Learning from Examples

3 questions — free, untracked, retake anytime.

is 'training data'?

✓ Correct — ✅ ✓ Training data is the collection of examples AI learns from — billions of words for language AI, millions of images for image AI.

❌ ❌ Training data is the examples AI learns from — not instructions or tests. For language AI that's billions of words; for image AI, millions of pictures.

does the quality of training data matter?

✓ Correct — ✅ ✓ AI learns whatever patterns are in its training data. Bad examples → bad patterns. Biased data → biased AI.

❌ ❌ AI learns whatever is in its training data — including errors and biases. Bad examples teach bad patterns.

is one key difference between how humans and AI learn?

✓ Correct — ✅ ✓ Humans generalize from a few examples. AI usually needs many more examples — but can process them at speeds no human could match.

❌ ❌ Humans generalize quickly from a few examples. AI typically needs many more examples to learn a pattern, though it can process them much faster.

Back to Lesson Start Lab 2

Lab 2

Teaching a Robot

Think through what AI needs to learn.

Lab 2 — Teaching a Robot

Think through what it would take to teach an AI to do something.

The AI asks: if you wanted to teach an AI to recognize a sandwich, what examples would it need?
Think about: what would confuse it? What examples might teach it wrong patterns?
Discuss: what does this tell you about why AI sometimes gets things wrong?

Think about edge cases — what about a hot dog? A wrap? An open-faced sandwich?

⭐ AI GuideLab 2

Back to Quiz Next: Lesson 3

Lesson 3

What AI Knows (and Doesn't)

AI has a knowledge cutoff — and some things it simply can't know.

Nia asked an AI assistant: "Who won the championship last week?"

The AI answered: "I don't have information about recent events. My knowledge has a cutoff date, which means I only know about things that happened before a certain point in time. For recent sports results, you'd want to check a news site or sports app."

Nia tried another question: "What's my favorite color?"

"I don't know that," the AI said. "I can only know things that were in my training data, or things you tell me in our conversation. Personal things about you aren't in my training data."

Nia was impressed. "So you know your own limits?"

"Sometimes," the AI said. "But I don't always know what I don't know."

What AI Knows

Facts and patterns that were in its training data
Information you share with it in your conversation
Things that were common or well-documented before its training cutoff

What AI Doesn't Know

Events that happened after its knowledge cutoff date
Personal information about you (unless you tell it)
What's happening right now in the world
Things that were never written down or documented

The Tricky Part

AI doesn't always know what it doesn't know. Sometimes it will give a confident answer about something it actually got wrong. That's why checking important facts from other sources always matters.

Back Take Quiz

Quiz 3

What AI Knows (and Doesn't)

3 questions — free, untracked, retake anytime.

couldn't the AI tell Nia who won the championship last week?

✓ Correct — ✅ ✓ AI has a knowledge cutoff date. Events after that date weren't in its training data — so it genuinely doesn't know about them.

❌ ❌ AI has a knowledge cutoff — everything that happened after it was trained is unknown to it. Recent events require checking a current news source.

is the 'tricky part' about what AI doesn't know?

✓ Correct — ✅ ✓ AI can confidently state things that are wrong — it doesn't have a reliable signal for what it's uncertain about. That's why checking important facts matters.

❌ ❌ The tricky part: AI sometimes gives confident answers about things it got wrong. It doesn't always know what it doesn't know.

is one thing AI can know during your conversation that it didn't know before?

✓ Correct — ✅ ✓ AI only knows about you what you tell it. Information you share during the conversation becomes part of what it can use — for that conversation.

❌ ❌ AI learns about you from what you share in the conversation. It doesn't have personal information about you unless you provide it.

Back to Lesson Start Lab 3

Lab 3

Know Your Limits

Build a guide for what to trust AI about.

Lab 3 — Know Your Limits

Explore what AI knows and doesn't know with your guide.

The AI asks: what's something you'd trust AI to know — and something you'd always double-check?
Discuss why some types of questions are more reliable than others.
Build a simple guide: what to trust, what to check.

Think about: facts from history vs. recent events, general knowledge vs. personal details, common topics vs. specialized fields.

⭐ AI GuideLab 3

Back to Quiz Next: Lesson 4

Lesson 4

Talking to AI

How you ask a question shapes the answer you get.

Lucas asked an AI to help him write a birthday card for his grandma. He typed: "Write a birthday card."

The AI wrote something generic: "Happy Birthday! Wishing you a wonderful day..."

Lucas tried again: "Write a short, funny birthday card for my grandma who loves gardening and bad puns."

This time the AI wrote: "Happy Birthday! Hope your day is un-be-leaf-able! May your garden grow as beautifully as you have!"

Lucas laughed. "That's much better! Why did adding details help so much?"

"The more specific you are," the AI explained, "the better I can match what you actually want. Vague questions get vague answers. Specific questions get specific answers."

Prompting: Talking to AI

A prompt is what you type to an AI. Better prompts get better answers. Here's what makes a prompt more effective:

Be specific: Include details about what you want
Give context: Tell the AI who it's for and why
Specify the format: Short? Long? Funny? Serious?
Give examples: If you have a style in mind, describe it

You're in a Conversation

If you don't like the first answer, ask again differently. You can also ask the AI to change something specific: "Make it shorter," "Make it funnier," "Now write it for a 5-year-old."

Remember

Vague prompt = vague answer. The more you tell AI about what you want, the better the result. It's a skill you can get better at with practice.

Back Take Quiz

Quiz 4

Talking to AI

3 questions — free, untracked, retake anytime.

did Lucas's second birthday card prompt get a much better result?

✓ Correct — ✅ ✓ Specific details give the AI much more to work with. Who is it for? What do they like? What tone? More details = better results.

❌ ❌ The second prompt worked because Lucas added specific details — who the card was for, what she liked, and the style he wanted. Specifics matter.

is a 'prompt'?

✓ Correct — ✅ ✓ A prompt is what you type to an AI — your question, instruction, or creative direction. Better prompts get better results.

❌ ❌ A prompt is what you type to give the AI — your question, instruction, or request. The quality of your prompt shapes the quality of the answer.

should you do if an AI's first answer isn't what you wanted?

✓ Correct — ✅ ✓ AI is a conversation. If the first answer isn't right, ask again with more details or ask for a specific change. You can keep refining.

❌ ❌ If the first answer isn't right, refine your prompt — add details, specify what to change, or try a different approach. It's a conversation.

Back to Lesson Start Lab 4

Lab 4

Prompt Workshop

Practice writing and improving prompts.

Lab 4 — Prompt Workshop

Practice improving prompts with your AI guide.

The AI gives you a vague prompt and a specific prompt on the same topic.
Discuss what made the specific one better.
Write your own prompt for something you'd actually want AI to help with — then improve it together.

Think: what details would help the AI understand exactly what you want?

⭐ AI GuideLab 4

Back to Quiz Module Test

Module 4 Test

5 questions covering all lessons. Free, untracked, retake anytime.

does an AI language tool decide what to say next?

✓ Correct — ✅ ✓ AI predicts next words based on patterns from training data — not thinking or searching.

❌ ❌ AI language tools predict likely next words based on patterns from millions of examples. No searching, no thinking like a human.

is training data?

✓ Correct — ✅ ✓ Training data is what AI learns from — billions of words, millions of images, etc.

❌ ❌ Training data is the examples AI learns patterns from — for language AI, that's billions of words.

does AI not know about things that happened recently?

✓ Correct — ✅ ✓ AI has a knowledge cutoff date. Events after training aren't in its data.

❌ ❌ AI has a knowledge cutoff date — everything after it was trained is unknown to it.

makes a prompt more effective?

✓ Correct — ✅ ✓ Specific details — who, what, tone, format — help AI understand exactly what you want.

❌ ❌ Specific details make prompts better: who it's for, what you want, what tone, what format.

should you do when AI gives a confident answer about something important?

✓ Correct — ✅ ✓ AI can be confidently wrong. Always verify important facts from a separate source.

❌ ❌ AI can be confidently wrong. Important facts should always be verified from another source.

Back Academy Home