Solving the Extensive Reading Material Problem with AI: Custom Articles at Your Level
Extensive reading is one of the most reliably effective things you can do for a second language. Read a lot, in your target language, without stopping to look up every word. Pick material you can mostly understand, mostly enjoy. Stay in that zone long enough and vocabulary, reading speed, grammatical intuition, and listening comprehension all improve at the same time.
The catch is that extensive reading has one big practical problem: finding the material. AI changes that almost entirely. Here’s how.
Why Most People Give Up on Extensive Reading
It’s rarely the reading itself. It’s the search for things to read.
Too easy and the content is boring — kids’ stories written for adults learning a language. Too hard and you’re looking up a word every line, which kills the “extensive” part. Even when the level fits, the topic might not. Graded readers exist, but the catalogue is finite and the topics are conservative. Buying books at scale is expensive.
AI text generation solves all of these constraints at once.
Getting AI to Generate Reading Material at Your Level
Step 1: Tell the model your level concretely
Vague level descriptions get vague results. Specific descriptions get usable text. Try something like:
“Write at a level where roughly 5% of the words are unfamiliar to a CEFR B1 learner. When you use slightly harder vocabulary, make the meaning recoverable from context. Use medium-length sentences, with occasional complex clauses.”
That kind of prompt produces text you can actually read at speed.
Step 2: Choose a topic you genuinely care about
This is the most important step. The whole point of extensive reading is volume, and volume requires interest.
“Football” is too vague. “Manchester City’s tactical approach under Guardiola in 2024” is good. “Climate change” is too vague. “How insurance companies are responding to wildfire risk in California” is good. Specificity in the topic gives the model enough to write something engaging rather than generic.
Step 3: Add a follow-up
Reading without a follow-up still works, but a little extra effort multiplies retention. After you read the AI-generated piece, ask the same model: “Pick five expressions from this article that a learner at my level should remember, and give an additional natural example sentence for each.” You now have an annotated reading session, not just a reading session.
Sample Prompts by Level
Beginner (CEFR A2)
Short text, simple structure. Around 400 words. Mostly present tense. Short sentences (5–8 words). Vocabulary limited to the 1,000 most common English words. Repeat key vocabulary naturally through the text. Make the topic concrete and visual.
Intermediate (CEFR B1–B2)
800–1,000 words. Mix of simple and complex sentences. Academic vocabulary fine, but explain through context. Include some passive voice and relative clauses. Use examples, anecdotes, and a clear narrative arc.
Advanced (CEFR C1+)
1,500–2,000 words. Newspaper or magazine register (think Wired or The Atlantic). Include nuanced arguments and counterarguments. Specialized vocabulary appropriate to the topic. Discourse markers and varied sentence rhythm.
Techniques That Make This Work Long-Term
Serialize it
Cliffhangers create return visits. Ask for a five-part story where the protagonist’s situation evolves and the vocabulary level gradually increases. You’ll find yourself opening the next part voluntarily — which is exactly what extensive reading needs.
Have the AI rewrite news articles
Paste a Reuters or AP article and ask the model to rewrite it at your level while preserving the substance. You get level-appropriate exposure to current events. This is one of the best feedback loops for keeping motivation up when the world is changing fast.
Vary the genre
Same level, different genres. News article today, personal essay tomorrow, how-to piece on Wednesday, interview format on Thursday. Each genre has its own vocabulary and rhythm patterns. Variety here is what makes you a flexible reader, not a narrow one.
The Rules That Make Extensive Reading Work
Don’t look up words
This is the hard rule of extensive reading. If a word is unfamiliar, infer from context or skip it. Looking up every word turns extensive reading into intensive reading — useful, but a different activity. When you ask the AI to write at your level, this rule becomes much easier to follow.
Read for speed
Don’t crawl. Push for fluency. Intermediate readers aim for 100+ words per minute. As reading speed goes up, listening comprehension goes up too — your brain’s processing speed for the language is generalizing.
Hit volume targets
Research on extensive reading suggests effects show up reliably after 100,000+ words per year of input. That sounds like a lot. It’s 300 words a day, every day. One AI-generated article gets you there. A year of one article a day puts you well past the threshold.
Track the volume
Keeping a simple log of pages or articles read gives you a feedback loop that’s otherwise invisible. “I read 30,000 words this month” sounds different in your head than “I read a few things,” even if they’re the same thing.
What AI-Generated Material Doesn’t Replace
AI text is sometimes subtly unnatural — slightly off collocations, sentences that no editor would have left in. Graded readers and published articles are professionally edited; AI output isn’t. The best approach is to use AI for volume (where the marginal cost of generation is near zero) and supplement with edited material a few times a month.
Start Today
Open whatever AI chatbot you use. Ask it to write a 500-word article at intermediate English level on something you genuinely want to read about. Read it. Five minutes from now you’ll be done.
Do that every day for a year. The compounding does the rest.
If you’d prefer a single place that generates leveled reading, tracks your WPM, extracts vocabulary into spaced repetition, and gives comprehension quizzes — that’s what SpeakSmart’s Reading and Free Learning modules are for. The free plan gives you a few sessions per day to try.