AI can write texts, generate images, and answer questions. There’s one thing it can’t do: read for your child.
91 percent of young people in Germany already use AI tools. 74 percent use them for homework and learning. That’s according to the JIM Study 2025 by the Media Education Research Network Southwest. When you read these numbers, you might think: reading is becoming obsolete. Chatbots answer every question. Language models summarize, translate, explain.
The opposite is true.
The Interface is Text
When you use AI, you type a question. AI responds with text. You read, evaluate, ask follow-ups. Every interaction with generative AI is an act of reading and writing. No buttons, no swiping, no likes. Language.
A child who doesn’t understand a complex sentence can’t formulate a precise prompt. It can’t evaluate the response. It takes what it gets. That’s just operating a tool. It’s not meaningful use.
Prompt literacy, AI literacy, AI competency: these terms all sound new. But their core is old. If you want to ask precisely, you need to understand what words mean. If you want to check answers, you need to be able to read texts. Not skim. Read.
AI Sounds Convincing. Even When It’s Wrong.
Language models generate texts that sound fluid, structured, and confident. Sometimes they’re wrong. A made-up study, a fabricated quote, a plausible but false date. The problem: the text doesn’t reveal the error. It sounds just as convincing as a correct answer.
To spot that, your child needs more than decoding letters. It needs to question meaning. Is that really true? Where does this claim come from? Is something missing? Reading research calls this reflective and evaluative reading. In the PISA framework model, it’s the highest competency level: not just finding and understanding information, but integrating, reflecting on, and evaluating it.
That’s exactly the skill that separates AI use from AI consumption.
The Numbers Speak Clearly
According to PISA 2022, 26 percent of fifteen-year-olds in Germany don’t even reach the minimum reading competency level. More than one in four young people can’t understand texts well enough to draw their own conclusions. The IGLU Study 2021 shows the same picture one grade level down: one in four fourth graders reads so weakly that basic requirements aren’t met. The Read Aloud Monitor 2024 from Stiftung Lesen adds to the picture: 32 percent of children between one and eight years old are rarely or never read to.
These children are growing up in a world where 91 percent of young people use AI tools. What happens when they operate these tools without being able to evaluate the results?
Strong Readers Use AI as a Tool. Weak Readers Are Used by It.
Reading research pioneer Keith Stanovich described the Matthew Effect in Reading in 1986: those who read well read more, learn faster, and get even better. Those who read poorly avoid texts and fall further behind. The gap widens with each school year. AI accelerates this dynamic.
A strong reader checks the output, combines it with their own knowledge, asks follow-up questions, refines their prompt. They use AI as a lever.
A weak reader accepts the output unchecked. They delegate thinking to the machine. Not out of laziness, but because they lack the tool to do it differently: reading literacy.
AI doesn’t level educational differences. It deepens them. Those who don’t read today won’t be shapers of the AI future tomorrow. They’ll be passive consumers.
What Your Child Really Needs
Not a prompt engineering course in elementary school. Not another school subject called “AI Competency.” The foundation runs deeper.
Neuroscientist Maryanne Wolf describes in Reader, Come Home what happens during deep reading: the brain builds connections that enable empathy, critical thinking, and analytical judgment. Those who only skim, scroll, and swipe don’t train these abilities. Wolf warns explicitly that digital short formats atrophy the exact cognitive processes we need most to work with AI.
Your child needs the ability to understand a text. To distinguish claims from evidence. To spot contradictions. To make sense of what they’ve read. And the confidence to say: I understand this, but I can’t verify whether it’s true.
Of course, it needs more. Source criticism, subject knowledge, logical thinking, responsible handling of their own data. But all of that builds on one foundation. Those who can’t read a fact-check can’t verify facts. Those who don’t understand a privacy notice can’t protect their data.
Reading literacy isn’t the only competency that matters. But it’s the one all the others stand on.
PISA 2029 Asks the Right Question
The OECD gets it. PISA 2029 makes reading literacy the main domain, the focus of the world’s largest education study. At the same time, “Media and AI Literacy” is being assessed as an innovative domain for the first time. The message: reading and AI competency belong together. Not as separate disciplines, but as two sides of the same coin.
We don’t have to wait until 2029 to know what the results will show. Today’s data is clear: those who can read will use AI as a tool. Those who can’t will be used by it.
The Paradox
The technology that seems to make reading unnecessary makes it more important than ever. AI drives the cost of text generation toward zero. The volume of synthetic content explodes. What becomes scarce isn’t information. It’s the ability to evaluate it.
Those who can read can master AI competency on their own. Understand instructions, follow explanations, spot errors, learn new things. Those who can’t read depend on what others summarize. The difference between shaper and consumer doesn’t start with the first prompt. It starts with the first book.
Support Reading, Don’t Replace It
That’s exactly why we built MichMatz ReadMe. Not as a replacement for reading, but as a tool that brings children to reading. Stories that match their interests. Difficulty levels that fit their ability. An experience that shows: reading is worth it.
The children who learn to read today will decide tomorrow what they use AI for. The children who don’t will leave that decision to others.
Reading literacy is not a skill of yesterday. It’s the foundation for tomorrow.
Those who can read master AI. Those who can’t are mastered by it.
Michael Schulz, Founder of MichMatz GmbH