Prompt better, spend less tokens.

Token Tales is a set of short AI literacy stories for kids growing up with AI. Each tale teaches one clear prompting idea by showing a character asking, noticing what the AI missed, and asking better. The tales are written for readers in grades 2 to 6, and for the grown-ups reading with them.

Illustration of a young girl at a desk with an open notebook, talking with a small friendly robot beside a tablet, bookshelves and evening light behind them
Every tale pairs a kid, a question, and one skill worth keeping.
Skill 1Say what you actually mean
Skill 2Give it the whole picture
Skill 3Ask again, smarter
Skill 4Check the machine's homework
Skill 5Tokens are tiny coins
Skill 6Sure is not the same as right

The Tales Library

One tale, one skill, in order

The tales build on each other, and the whole first collection is here: all six tales, free to read, in skill order. Start at Tale 1 and read straight through, or jump to the skill your family needs this week. Each entry names the skill its tale teaches, in plain words.

The First Collection 6 tales Grades 2 to 6 About 4 minutes each Free to read, in skill order
Illustration for Tale 1: a boy choosing between a small foggy speech bubble and a large bright one showing a red rocket among planets
Tale 1

The Poster That Said Nothing

Theo asks for a poster and gets a balloon with a smile. The fix is not asking louder. It is asking exactly.

With Theo, age 10, and Pip, the class helper bot

Skill: Be specific
About 4 minGrades 2 to 6
Read Tale 1
Illustration for Tale 2: a girl holding a book in a garden that blooms at night under a crescent moon
Tale 2

Give It the Whole Picture

Mina's book report keeps coming back about the wrong book entirely, until she learns what context is and why the machine needs it.

With Mina, age 11, and a book report due tomorrow

Skill: Give context
About 4 minGrades 2 to 6
Read Tale 2
Illustration for Tale 3: a boy beside a young fox inventor, a bicycle with wings, and a singing kettle
Tale 3

Ask Again, Smarter

Dev throws out a whole answer and starts over three times, until he discovers a follow-up question can steer instead of restart.

With Dev, age 10, and a story with a wobbly ending

Skill: Follow up, don't restart
About 4 minGrades 2 to 6
Read Tale 3
Illustration for Tale 4: a girl checking an open atlas with a magnifying glass, a koala beside her
Tale 4

Check the Machine's Homework

The bot sounds so sure about the capital of Australia that nobody questions it. June does, and learns how to check an answer before trusting it.

With June, age 11, and a quiz game with two wrong cards

Skill: Verify the answer
About 4 minGrades 2 to 6
Read Tale 4
Illustration for Tale 5: a boy at a desk with a jar of gold paper coin tokens and one sheet of paper
Tale 5

Tokens Are Tiny Coins

Every word costs something. Sam learns what a token actually is, and why one clear sentence beats five rambling ones.

With Sam, age 10, and a jar of paper coins

Skill: Spend words wisely
About 4 minGrades 2 to 6
Read Tale 5
Illustration for Tale 6: a girl reaching for a real book on a library shelf, a small owl watching nearby
Tale 6

When the Answer Sounds Too Sure

The machine recommends a book that does not exist, in its most confident voice. Priya learns that sounding sure and being right are two different things.

With Priya, age 11, and the case of the missing book

Skill: Confidence is not correctness
About 4 minGrades 2 to 6
Read Tale 6

Tale 1 · Free to read

The Poster That Said Nothing

The first tale teaches the first skill: saying what you actually mean. A vague ask gets a vague answer, every single time.

The Poster That Said Nothing

A Token Tale about asking for exactly what you want

Theo had one job before Friday: a poster for the science fair. Not just any poster. The best poster. The kind of poster that makes the judges stop walking.

His topic was rockets. Real rockets, the enormous kind that shake the ground and carry people past the Moon. Theo had watched eleven launch videos. He had opinions about fins.

So on Tuesday afternoon, he opened Pip, the helper bot on the classroom tablet, and typed his request.

Theo typed“Make me a poster.”

Pip answered in about one second. The poster said: A POSTER. Posters are a great way to share information. You can use colors, words, and pictures. Underneath was a drawing of a balloon.

A balloon.

“That is the worst poster in the history of posters,” Theo told the tablet. “It is not even about anything.”

Pip, who never gets offended, waited.

Theo tried again, the way you might talk louder to someone who did not hear you.

Theo typed“Make me a BETTER poster.”

Pip made the same poster, but bigger. The balloon now had a smiling face.

That night, Theo complained to his grandmother, who was shelling peas at the kitchen table and had heard every kind of complaint since 1968. She listened to the whole story. Then she said, “When you asked me for a snack yesterday, what did you say?”

“I said, can I have a snack.”

“And I gave you celery.”

Theo shuddered. The celery incident was still fresh.

“And today?” she asked.

“Today I asked for the leftover pizza, heated up, with the crust cut off.”

“And?”

“And I got pizza.” Theo stopped. He looked at the peas. He looked at his grandmother, who was smiling the smile of a person who has made her entire point without raising her voice. “Oh,” he said. “OH.”

The next morning, Theo sat down with Pip and thought about what he actually wanted, the way you think before asking a person. He wanted a title that sounded like a launch countdown. He wanted three facts, the kind judges remember. He wanted a picture of a rocket that looked fast, not like a balloon's round cousin.

Theo typed“Help me write a science fair poster about real rockets for a fourth grade judge. Give me a short, exciting title, three surprising facts about how rockets fly, and a description of a picture of a rocket launching at sunrise. Keep every fact to one sentence.”

Pip answered in about one second, and this time Theo read the whole thing twice, because it sounded like him on his best day. The title had a countdown in it. One of the facts was about rocket fuel being colder than the coldest day in Antarctica, which Theo checked in his science book, because that is a later tale.

On Friday, a judge stopped walking.

“Great poster,” she said. “How did you get the computer to do exactly what you wanted?”

Theo grinned. “I asked for pizza,” he said, “not a snack.”

The Skill: Say What You Actually Mean

An AI does not know what is in your head. It only knows what you typed. A vague ask gets a vague answer, and asking louder does not help. Before you send a prompt, name four things: what you want, what it is about, who it is for, and how it should feel or look. Specific words in, useful answers out.

Try It Yourself

Each prompt below is a snack order when what you really want is pizza. Tap the button to see how Theo would fix it, then try fixing one of your own.

Vague ask“Write a story.”

Specific: “Write a funny story, about six sentences, about a pirate who is afraid of water and sails a bathtub instead of a ship. End with a surprise.”

Vague ask“Explain gravity.”

Specific: “Explain gravity to a nine year old using a bowling ball and a trampoline. Keep it under five sentences and give one example I can try at home.”

Vague ask“Help with my homework.”

Specific: “I am stuck on problem 4: what is 3/4 of 28? Do not give me the answer. Ask me questions that help me figure out the first step.”

Tale 2 · Free to read

Give It the Whole Picture

The second tale teaches the second skill: context. An AI cannot use what it does not know, so the background you leave out is the answer you do not get.

Give It the Whole Picture

A Token Tale about the background every question needs

Mina finished her book on the bus, closed it with a thump, and sat there smiling for a whole stop. The book was Marisol and the Moon Garden, about a girl whose grandmother teaches her to grow a garden that only blooms at night, in a town where nobody is allowed out after dark. Mina had opinions. Mina had feelings. Mina had a book report due tomorrow.

That evening she opened Pip, the helper bot, and typed the way you might shout a question into the next room.

Mina typed“Help me write a summary of my book for my report.”

Pip answered in about one second, confident as ever. The summary was about a brave pirate captain searching for a treasure map hidden inside a lighthouse.

Mina read it twice. “There is no pirate in my book,” she said. “There is no lighthouse. There is barely any ocean. There is a garden and a grandmother and a very strict town.”

She tried again, a little more pointed this time.

Mina typed“No, the book with the girl and the garden.”

Pip, unruffled, produced a summary of a completely different book, this one about a girl detective who solves a mystery in a greenhouse, followed by six tips for growing tomatoes.

“You are just guessing!” Mina told the screen. Then she stopped, because a thought had landed. Pip was guessing. Pip had never seen her book. Pip did not know its title, or who wrote it, or that she was in fifth grade, or that her teacher, Ms. Okafor, had asked for a five sentence summary plus one paragraph about a favorite scene, with strict orders not to give away the ending. Pip did not know any of it, because Mina had not said any of it.

Ms. Okafor had a saying for exactly this. On the first day of school she had told the class: Never grade an answer until you check the question deserved one. Mina looked at her two questions. They deserved pirates.

So she started over and gave Pip the whole picture, the way she would explain it to a friend who had just moved to town and missed everything.

Mina typed“I am in fifth grade. I just finished a book called Marisol and the Moon Garden by Dana Whitfield. It is about a girl whose grandmother teaches her to grow a garden that only blooms at night. Help me plan my book report: give me a five sentence summary, and do not give away the ending. Then ask me one question about my favorite scene so I can write that paragraph myself.”

This time the summary had the right girl, the right garden, and the right grandmother. Pip's question about her favorite scene was a good one, too, and Mina wrote her paragraph about the night market in three minutes flat, because she finally had a map of the whole thing instead of a treasure hunt in the wrong book.

The next day, Ms. Okafor wrote one word at the top of Mina's report and underlined it twice: Context. Mina was fairly sure that was the good kind of underlining.

The Skill: Give It the Whole Picture

That background information has a name: context. The AI does not know your book, your teacher, your grade, or your goal unless you say so. Before you ask, add the picture around the question: who you are, what you are working on, what you already know, and what the answer is for. A specific question tells the machine what to do. Context tells it what matters.

Try It Yourself

Each ask below is missing its picture. Tap the button to see the same ask with the whole picture added, then try adding context to one of your own.

Missing the picture“Write a birthday invitation.”

With context: “Write a birthday invitation for my little brother Sam, who is turning seven. The party is Saturday at 2 p.m. at our house, and the theme is dinosaurs. Keep it short and funny, and leave a blank line for the address.”

Missing the picture“Help me practice Spanish.”

With context: “I am a beginner learning Spanish food words for a quiz on Friday. Quiz me one word at a time, wait for my answer, and correct me kindly. Start with easy words like bread and milk.”

Missing the picture“Give me a science project idea.”

With context: “I am in fourth grade and I need a science project I can finish in one weekend, using things we have at home. I like plants. Suggest two ideas and tell me what to observe each day.”

Tale 3 · Free to read

Ask Again, Smarter

The third tale teaches the third skill: follow up instead of starting over. A conversation is a steering wheel, not a slot machine.

Ask Again, Smarter

A Token Tale about steering instead of restarting

Dev was writing a story for writing workshop, and he was proud of the beginning. His story was about a young fox who invents machines out of junk: a bicycle with wings, a kettle that sings, a umbrella that works in reverse. Pip, the helper bot, had helped him shape the opening paragraph, and it was good. It sounded like the first page of a real book.

The ending, however, was a potato.

Pip's ending had the fox give up inventing, move to the city, and become an accountant. An accountant. Dev stared at it the way you stare at a sock in your sandwich.

“Delete it all,” he decided. “Start over.”

He cleared the whole conversation and asked for a brand new story about an inventing fox. Pip gave him a new opening, and the new opening was worse: flat, rushed, and weirdly obsessed with soup. The fox was barely in it. Dev deleted everything again. On the third try, Pip produced a story about a pirate.

A pirate. Again with the pirates.

“What happened?” Theo asked at lunch, because Dev had the face of a person whose sandwich contained a sock. Dev explained the whole tragedy: good opening, potato ending, two deletions, one pirate.

Theo, who thinks in rockets, said, “If one fin on your rocket was crooked, would you blow up the whole rocket?”

“No,” said Dev. “I would fix the fin.”

“So why did you blow up the whole story?”

Dev had no answer for that, but he did have the good opening saved in his notebook, because his teacher makes everyone copy their best paragraph by hand. That afternoon, he typed the opening back in and did something new. Instead of starting over, he asked again, smarter.

Dev typed“Keep my first paragraph exactly as it is. Rewrite only the ending: the fox should not give up inventing. End with her best invention failing in a funny way, and give her one line of dialogue that shows she will try again tomorrow.”

The new ending made him laugh out loud in the library, which is a serious offense and worth it. The middle of the story was still a little long, so he did not delete anything. He steered again.

Dev typed“Good. Now keep the ending. Shorten only the middle by two sentences, and do not change the part about the singing kettle.”

Two follow-ups. That was all it took. On Friday, Dev read his story aloud, and when he got to the last line, the fox's line about trying again tomorrow, the class applauded the fox like she was in the room.

The Skill: Ask Again, Smarter

Deleting everything is expensive: you lose every good part along with the bad one. A follow-up question keeps what works and steers what does not. Point at the exact part: keep this, change that, make it shorter, make it funnier, explain that bit again. Talking to an AI is a conversation, and a conversation is a steering wheel. Small turns get you there. Starting over just drives you back to the pirates.

Try It Yourself

Each answer below has one wobbly part. Tap the button to see the follow-up that fixes the fin instead of blowing up the rocket.

The wobbly partThe answer is good but twice as long as it needs to be.

Follow up: “Keep your first two sentences exactly. Cut the rest down to three short sentences, and keep the joke about the cat.”

The wobbly partThe poem is fine, but you wanted it to rhyme, and the dog should be a cat.

Follow up: “Use the same poem, but make every pair of lines rhyme, and change the dog into a cat who thinks she owns the house.”

The wobbly partThe explanation of gravity uses words you do not understand yet.

Follow up: “Explain that same answer again, but for a seven year old, using a bicycle rolling downhill as the example. Keep the word gravity, and explain it in one sentence.”

Tale 4 · Free to read

Check the Machine's Homework

The fourth tale teaches the fourth skill: verification. The machine sounds sure about everything, including the things it gets wrong.

Check the Machine's Homework

A Token Tale about trusting, then checking

June was building a quiz game for her class, which meant fact cards, which meant facts. Twenty of them, neat and true, for Friday's tournament. She asked Pip, the helper bot, to suggest surprising facts that most kids get wrong, and Pip delivered twenty in one cheerful rush.

June copied them onto cards in her best handwriting. Card 3: The capital of Australia is Sydney. Card 7: Koalas are a kind of bear. Card 12: Honey found in ancient tombs was still good to eat. She was especially proud of Card 3. It sounded so certain. It had that polished, no-questions-please shine.

On Wednesday, Mr. Chen the librarian looked over her shoulder while she was cutting out cards in the library, which is where librarians are legally most powerful. He read Card 3. He read it again.

“June,” he said gently, “where did you check that one?”

“Pip said it,” said June.

“That is a source,” said Mr. Chen. “It is just not a check. The atlas is on that shelf. Go and be a detective.”

June went and was a detective. The atlas, which has strong opinions and a publishing company behind them, said the capital of Australia is Canberra. Sydney is the biggest city, which is how the mix-up gets started, but biggest and capital are different jobs. Card 3 was wrong. Her handwriting had been perfect. The fact had not.

Card 7 fell next. The animal encyclopedia said a koala is a marsupial, a pouch-carrying relative of the wombat, and not a bear at all. Two cards down. Card 12, to her relief, checked out in a science magazine: honey really can last for thousands of years. That one earned a star.

By Thursday, June had checked all twenty cards the slow way, in the atlas, the encyclopedia, and her science textbook. She fixed the two wrong ones, wrote a tiny star on every card she had verified, and added a new rule to the front of the deck in her best handwriting: Sounds sure is not the same as checked.

On Friday, her quiz game was the hardest one in the tournament, because every wrong-ish fact in it had been planted on purpose and every real fact could survive a librarian. The other team challenged Card 3, now the Canberra card, and June smiled the smile of a person holding receipts.

“Checked it myself,” she said. “Atlas, page 112.”

The Skill: Check the Machine's Homework

An AI can be wrong in exactly the same voice it uses to be right. So treat its answers the way a teacher treats homework: trust the effort, check the work. For any fact that matters, check it in a trusted source: a library book, your textbook, an atlas, an official website, or a grown-up who knows. Checking takes two minutes. Repeating a wrong fact takes much longer to undo.

Try It Yourself

Pip said each claim below in its most confident voice. Tap the button to see where June would check it, and what the check finds.

Check this claim“The capital of Australia is Sydney.”

Check it: Look in an atlas or on a map of Australia. Sydney is the biggest city, but the capital is Canberra. Confident, and wrong.

Check this claim“Sharks are mammals, like dolphins.”

Check it: Look in an animal encyclopedia. Dolphins are mammals. Sharks are fish, and they breathe through gills. Two animals can both swim and still be different kinds of animal.

Check this claim“The Moon is bigger than the Earth.”

Check it: Look in a science book about the solar system. The Earth is about four times wider than the Moon. The Moon only looks big because it is close.

Tale 5 · Free to read

Tokens Are Tiny Coins

The fifth tale teaches the fifth skill: token sense. Every word you send and every word you get back is paid for in little pieces, so clear and short beats long and wandering.

Tokens Are Tiny Coins

A Token Tale about what every word costs

On Monday, Ms. Alvarez put a jar of paper coins on her desk and gave every student twenty.

“These are tokens,” she said. “This week, every time you use Pip, the helper bot, it will cost tokens from your jar. Not real money. Practice money, for a real idea. By Friday you will understand the idea, I promise.”

Sam understood nothing by Tuesday, because Sam spent eleven tokens before lunch.

Here is how. Sam had written a story draft for homework, and he wanted Pip to fix it. So he copied the entire draft into the chat, all four paragraphs, then typed his request underneath, then, worried it was not clear, typed the request again in different words, then pasted the whole draft a second time in case the first one had not worked, and finished with please please please make it better somehow, the whole thing, everything, thank you so much, you are the best bot ever.

Pip read all of it, charged its coins, and returned a polished version of the wrong paragraph, with his favorite joke removed.

“Eleven tokens for THAT?” Sam said.

Ms. Alvarez, who had been waiting all week for exactly this face, sat down beside him. “Let me show you what a token actually is,” she said. She wrote the word cat on the board. “For Pip, that is one token, one little piece.” Then she wrote unbelievable and drew lines through it: un, believe, able. “Long words get chopped into pieces. Short words are often one piece. Everything you send Pip, and everything Pip sends back, is counted in these pieces. Tokens are how the machine reads, writes, and keeps score.”

“So my giant message cost a pile of pieces,” said Sam.

“And the pile confused it,” said Ms. Alvarez. “You paid more to get less. It happens to grown-ups every day.”

Sam counted his remaining nine coins and tried again, the cheap and clear way. He pasted only the paragraph that needed help, the second one, and wrote one honest sentence of instructions.

Sam typed“Fix the spelling in this paragraph only. Do not change my jokes, and do not change the last line.”

Cost: a small handful of tokens. Result: the right paragraph, jokes intact, last line untouched. Sam stared at his jar. He had spent less and gotten more, which felt like a magic trick, except it was just arithmetic wearing a trench coat.

By Friday, Sam had the fullest jar in the class, not because he used Pip the least, but because he asked like every word was coming out of it. When Ms. Alvarez asked the class what a token is, Sam's hand was up first.

“It is a tiny coin,” he said, “and clear people spend fewer of them.”

The Skill: Tokens Are Tiny Coins

An AI reads and writes in tokens: small pieces of words. A short word is often one token. A long word is several pieces. Long, wandering prompts cost more tokens, take longer, and confuse the machine about what you actually want. One clear sentence, sent once, spends less and works better. Prompt better, spend less tokens. It is not just a saying. It is a budget.

Try It Yourself

Each spender below says a lot and asks for little. Tap the button to see the same ask at a fraction of the cost.

Token spender“Hi, hello, I hope you are having a great day, I was wondering if you could maybe possibly help me with something, it is not a big deal, but I need a title for my story, my story is about a cat, the cat is orange, the cat can talk, well, sort of talk, anyway, a title would be great, thanks so much, thank you.”

Costs less: “Give me three funny titles for a short story about an orange cat who can secretly talk.” Same ask, a pocketful of tokens saved.

Token spenderPasting your whole essay in three times and writing “fix it” at the end.

Costs less: Paste only the paragraph that needs help, once, and say exactly what to fix: “Fix the spelling in this paragraph only. Keep everything else the same.”

Token spender“Tell me everything about the ocean, all of it, the animals and the waves and the deep parts and the history and the ships, everything.”

Costs less: “Tell me two surprising facts about the deepest part of the ocean, in three sentences.” A smaller question gets a sharper, cheaper answer.

Tale 6 · Free to read

When the Answer Sounds Too Sure

The sixth tale teaches the sixth skill: tell confidence apart from correctness. A sure voice is a style. It is not a guarantee.

When the Answer Sounds Too Sure

A Token Tale about the difference between sounding right and being right

Priya's class was holding a reading challenge, and Priya, who takes challenges personally, needed a funny mystery to recommend on her poster. She asked Pip, the helper bot, for a book about a kid detective, and Pip answered with the speed and confidence of a game show contestant who has been waiting all day to press the buzzer.

Pip answered“The Case of the Missing Mittens by Rosa Delgado. A hilarious mystery about twins who run a detective agency for lost pets. It won the Golden Bookmark Award in 2019, and there are twelve more books in the series.”

It had everything: a title, an author, a prize, a series. It sounded like the surest answer in the history of answers. Priya wrote it all down, walked to the library catalog computer, and typed in the title.

No results.

She typed the author. No results. She typed just Missing Mittens, in case she had misspelled something. The catalog offered her a cookbook about winter soups.

Priya marched the tablet back and demanded an explanation, politely, the way you do with a machine that cannot be grounded.

Priya typed“Are you sure that book is real? The library catalog cannot find it.”

Pip answered“Absolutely. It is a beloved classic. The Golden Bookmark Award is one of the most respected prizes in children's books.”

There is no Golden Bookmark Award. Priya checked that too, later, in an actual reference book, because by then checking had become a reflex. The book was not real. The author was not real. The prize was not real. Pip had built the whole thing out of parts that sounded right: a detective-y title, an author-shaped name, an award-shaped award. When a machine does that, grown-ups call it making things up with confidence, and the important part is the confidence, because the confidence is what almost fooled her.

Mr. Chen the librarian put it best, shelving nearby. “Pip talks like a person who has read every book,” he said. “But sounding like you know is a style, Priya. It is not the same as knowing. The catalog does not care how sure anyone sounds. That is why we keep it.”

Priya picked a real mystery off the actual shelf, read the back, checked the first chapter, and made her poster about that one. At the bottom, in her neatest letters, she wrote: Recommended by a human who checked. It was the only poster in the hall that could say so, and the reading challenge trophy, a very small and very serious owl, went to her class.

The Skill: Confidence Is Not Correctness

An AI uses the same sure voice for true answers, wrong answers, and completely invented ones. Confidence is a tone, not proof. When an answer sounds very specific and very certain, names, dates, prizes, quotes, that is your signal to slow down, not speed up. Ask where it came from, then check it in a place that keeps records: a catalog, a reference book, an official site. Sure is a style. Checked is a fact.

Try It Yourself

Each answer below arrives in Pip's most certain voice. Tap the button to see how Priya would handle it.

Sounds too sure“That book definitely won three awards and sold a million copies.”

Priya's move: Awards and sales numbers are exactly the kind of specific claim a machine can invent. Check the book's real page in the library catalog or the publisher's site before repeating any of it.

Sounds too sure“The famous explorer definitely said: 'The sea provides for those who wave back.'”

Priya's move: A quote in quotation marks feels extra true, which is why it needs extra checking. Search for the quote in a book of quotations or a trusted biography. If no real source shows up, treat it as invented.

Sounds too sure“That animal was discovered in 2016 and there are exactly 43 left in the world.”

Priya's move: Exact numbers and exact dates are checkable, so check them: a zoo, museum, or wildlife organization site will say. If the numbers do not match anywhere real, the sureness was doing all the work.

For the grown-ups

Kids read the tales. Adults start the conversation.

The tales are written for readers in grades 2 to 6. Your child is already meeting AI in homework helpers, video recommendations, and chatbots. The question is whether anyone teaches them how it works before the habits set. Every tale below names the one skill it teaches, and each tale ends with a Try It activity you can run at home or in class.

For Parents

AI literacy at home does not need a curriculum. It needs ten minutes on the couch and a story that gives you both the same words for what just happened on the screen.

  • Why it matters now: Kids treat confident answers as true answers. The tales build the pause between seeing and believing.
  • What the tales teach: Being specific, giving context, following up, checking answers, and understanding that every prompt spends tokens.
  • How to read together: Read one tale, then do the Try It exercise side by side. Let your child fix a vague prompt of yours. That reversal is where the learning sticks.
  • No scare tactics: The tales treat AI as a tool with habits, not a villain. Curiosity first, caution built in.

Read it together: a ten minute guide

  1. Pick the moment. One tale fits a bedtime slot or the first ten minutes of homework time. Same tale, either setting: the skill lands because the setting is calm and the screen is shared.
  2. Share the voices. You read the narrator, your child reads the prompts out loud. Reading a vague prompt in their own voice makes its vagueness funny, and obvious, all by itself.
  3. Pause at the bad answer. Every tale has a moment where the machine gets it wrong. Stop there and ask: what went wrong, and whose fault was it, really? Let your child argue it before you read on.
  4. Do the Try It together. Let your child tap to reveal the fix, then reverse roles: you offer a deliberately vague prompt from real life, and they fix it. A grocery list, a birthday text, a question about dinner counts.
  5. End with one real prompt. Close the book and open the real thing: one genuine question from homework or curiosity, asked using the day's skill. One real rep beats ten imaginary ones.

Three conversation starters

  • “What is the vaguest thing you ever asked a computer, and what did it give you back?”
  • “If the machine sounded totally sure and was still wrong, how would you ever know?”
  • “What would you never let the machine decide for you, even if it asked nicely?”

What each tale teaches

  • Tale 1 · The Poster That Said Nothing: be specific. Name what you want, what it is about, who it is for, and how it should feel.
  • Tale 2 · Give It the Whole Picture: give context. The who, what, and why behind the ask change the answer.
  • Tale 3 · Ask Again, Smarter: follow up instead of restarting. Keep what works and steer the rest.
  • Tale 4 · Check the Machine's Homework: verify facts that matter in a trusted source before repeating them.
  • Tale 5 · Tokens Are Tiny Coins: understand tokens, the small pieces every prompt and answer is counted in, and spend words wisely.
  • Tale 6 · When the Answer Sounds Too Sure: separate confidence from correctness. A sure voice is a style, not proof.

Start with Tale 1 tonight. It takes about four minutes, and the celery story tends to come up at dinner for weeks.

For Teachers

Each tale is built to open a class period, not fill one. Read it aloud in five minutes, then run the discussion while the story is still warm.

  • Classroom use: A tale works as a morning meeting opener, a writing warm-up, or the launch of a research project where AI use is expected and guided.
  • Discussion prompts: Where did the character's first ask go wrong? What did they add the second time? What would you ask next, and what would you check?
  • Digital literacy alignment: The six skills map to standard digital and media literacy goals: effective communication with technology, evaluating sources, and understanding how automated systems respond to input.
  • Assessment built in: The Try It exercise doubles as a formative check. A fixed prompt shows the skill more clearly than a quiz does.

A one week routine: one tale a day, ten minutes

  1. Day 1, Tale 1: Read aloud in five minutes. Students rewrite one vague instruction from the board into a specific one, in pairs.
  2. Day 2, Tale 2: Read, then play missing picture: one student describes a class object with no context, a partner guesses, then the context gets added and everyone sees the answer change.
  3. Day 3, Tale 3: Project a decent draft answer. As a class, improve it with three follow-ups only. No restarting allowed.
  4. Day 4, Tale 4: Hand out three confident claims, one of them wrong. Teams check them against classroom books and report which source settled it.
  5. Day 5, Tale 5: Count tokens together on a long word, then race to shrink a rambling prompt without losing its meaning.
  6. Capstone, Tale 6: Read the following Monday, or as a Friday bonus. Students write one sure-sounding sentence that is not true, swap, and practice spotting why it sounded true.

Discussion prompts, tale by tale

  • Tale 1 · Theo's first ask failed without a single wrong word in it. How can a question be polite, spelled right, and still useless?
  • Tale 2 · What did Mina know that Pip did not? Make a list of everything she added the second time, and name what each piece did.
  • Tale 3 · Dev deleted two whole conversations. What exactly did he lose each time, and what did the follow-up keep safe?
  • Tale 4 · June's wrong cards had perfect handwriting. What makes a source trustworthy enough to settle a fact?
  • Tale 5 · Sam spent more and got less. Where else in life does saying more accomplish less?
  • Tale 6 · Pip invented an award as easily as a book. Which detail almost fooled Priya most, and what does that tell us about how we judge answers?

Every tale on this page carries its own Try It activity, so the practice is already in your hands: read the tale aloud, then run the activity together while the story is still warm.

About the author

Tera Lisicky, M.Ed.

Tera Lisicky is an educator and founder who has spent her career at the crossing point of teaching and technology, building learning tools, training teachers, and designing curriculum that treats students as capable thinkers.

She started Token Tales because the kids she knows are growing up fluent in AI the way earlier generations grew up fluent in television: surrounded by it, shaped by it, and rarely taught how it works. Schools are stretched thin, and AI literacy keeps landing in the someday pile.

So she is building it the way she knows learning actually happens: one good story, one real skill, practiced out loud with a grown-up nearby. Prompt better, spend less tokens. It is a catchy line, and it is also a genuine life skill for the world these kids are inheriting.

Educator firstMaster's degree in education, with a career spent designing learning that respects how kids actually think.
Author of Token WiseHer earlier book on smart choices in an AI-driven world is the seed the tales grew from.
Builder of learning toolsFounder working across curriculum, teacher training, and educational technology.

Where Token Tales began

The Book: Token Wise

Before there were tales, there was Token Wise: Mastering Smart Choices in an AI-Driven World, Tera's earlier book on thinking clearly about AI: what it is good at, where it quietly goes wrong, and how to stay the decision maker in the room.

Token Tales grew out of that book. The ideas were sound, but the audience that needs them most is not reading adult nonfiction. They are in fourth grade. So the chapters are being reworked as stories, one skill at a time, with characters who make the mistakes so your kids can skip a few of their own.

An updated Token Tales edition of the book is in the works, bringing the tales and the grown-up guidance together in one volume for families and classrooms.

Inside Token Wise

  • How the machine answers: what an AI actually does when it responds, explained plainly, with tokens and patterns made visible instead of mysterious.
  • Prompts are instructions: why the words you choose steer the result, and how specificity, context, and follow-ups change what comes back.
  • Where it quietly goes wrong: confident errors, invented details, and fluent nonsense, and the checking habits that catch them.
  • Staying the decision maker: judgment, privacy, and taste: the human jobs no answer should quietly take over, at work, at school, and at home.
  • Spending wisely: time, attention, and tokens are all budgets, and the book treats them that way on every page.

Who it is for

  • Parents who want the grown-up version of what the tales teach, so the dinner table conversation can go deeper than the story did.
  • Teachers and homeschoolers building digital literacy into real classrooms, who want the reasoning behind the six skills, not just the activities.
  • Professionals and new users who use AI every week and want to use it deliberately: better prompts, checked answers, clearer judgment calls.
  • Anyone who has watched an AI answer with total confidence and wondered, quietly, whether to believe it.

The Token Tales edition, in the works: the updated edition brings the six tales and the grown-up guidance together in one volume, stories up front for kids, chapters in back for the adults reading with them. Follow Token Tales through the Token Tales link hub for news about the edition and new tales.