Reach Every Student, Every Learning Style

Differentiating Instruction with AI

It's Wednesday afternoon. You have 28 students in your Year 8 English class. Six of them read two years below grade level. Four are English language learners at different stages of fluency. Three finished last week's novel study two days early and are visibly bored. The rest fall somewhere in between. You have one prep period, roughly 45 minutes, to plan tomorrow's lesson. Differentiated instruction is the right approach, every teacher knows this, but creating four or five versions of the same lesson material, each calibrated to a different reading level or learning need, can take hours you simply don't have. This is exactly the kind of problem AI tools were made for. Not to replace your professional judgment, but to do the time-consuming production work so you can focus on teaching.

What Differentiation Actually Means (and Why It's Hard)

Differentiated instruction means adjusting what you teach, how you teach it, or how students show what they know, based on their readiness, interests, or learning profile. The theory is straightforward. The execution is exhausting. A single lesson on, say, the causes of World War One might need to be delivered as a simplified graphic organizer for struggling readers, a standard text-based activity for on-level students, a primary-source analyzis for advanced learners, and a vocabulary-supported version for ELL students. That's four separate materials, all teaching the same content, all needing to be accurate, age-appropriate, and engaging. Most teachers manage this by choosing one version and hoping it works for most of the room. AI changes that calculation significantly.

The core shift is this: AI tools like ChatGPT Plus, Claude Pro, and Microsoft Copilot can take a single piece of content, a passage, an explanation, a set of questions, and reformat it in minutes for different audiences. Think of it like having a very fast, very patient editorial assistant who never gets tired of rewriting the same paragraph at a Grade 4 reading level, then a Grade 7 level, then with sentence starters added, then with a glossary attached. You still decide what the learning objective is. You still review and adjust the output. But the drafting work, the part that used to eat your Sunday evenings, gets compressed from hours to minutes.

This isn't about producing generic worksheets. When you give an AI tool specific context, the grade level, the subject, the particular learning need you're addressing, the standard you're targeting, the output becomes genuinely useful. The difference between a vague prompt and a specific one is the difference between a bland generic passage and something you'd actually hand to a student. The skill you're building in this lesson is how to give AI tools enough context to produce differentiated materials that save you real time without sacrificing quality.

One more thing before we get practical: differentiation with AI works best when you already know your students. AI can rewrite a text at a lower Lexile level, but it can't tell you that Marcus shuts down when he feels singled out, or that Priya responds well to real-world connections. Your professional knowledge of your students is the irreplaceable layer. AI handles the production. You handle the pedagogy.

The Four Things AI Can Differentiate For You

There are four main ways teachers use AI tools to differentiate instruction, and each one maps to a real classroom problem. First, reading level adjustment: taking a text and rewriting it at a lower or higher Lexile level while keeping the core content intact. Second, language support: adding vocabulary glossaries, sentence frames, or bilingual explanations for ELL students. Third, complexity scaffolding: breaking a complex task into smaller steps with worked examples for students who need more structure, or removing scaffolds and adding extension challenges for students who need more stretch. Fourth, modality variation: converting text-based content into question-and-answer formats, visual description scripts, or audio-friendly summaries. All four of these can be done in ChatGPT, Claude, or Copilot without any technical knowledge, just well-constructed prompts.

• Reading level adjustment, rewrite any text at a specified grade or Lexile level
• Language support, add glossaries, sentence starters, or simplified syntax for ELL students
• Complexity scaffolding, add or remove structure, hints, and worked examples based on readiness
• Modality variation, convert written content into dialog format, visual descriptions, or step-by-step guides
• Assessment differentiation, generate tiered versions of the same quiz or discussion prompt
• Extension tasks, create open-ended challenges for early finishers that deepen rather than just add more

Click to enlarge

Prompt Example: Rewriting a Text for Two Reading Levels

Choosing the Right AI Tool for Differentiation Tasks

Not every AI tool works equally well for every differentiation task. Some are better at long-form rewriting. Some integrate directly into the tools you already use. Some offer better control over tone and reading level. The table below compares the four most commonly used AI tools among educators, based on the specific tasks involved in differentiating instruction. Note that features and pricing change, always check the current offering directly on each platform.

Putting This Into Practice: Your Monday Morning Workflow

Here's what a realiztic AI-assisted differentiation workflow looks like for a classroom teacher. You've identified tomorrow's lesson objective: students will analyze how an author uses figurative language to create mood in a short story excerpt. You have one excerpt. You open ChatGPT Plus or Claude Pro and paste in the excerpt. You write a prompt that specifies three things: the grade level of your class, the specific student groups you need materials for, and the exact format you want. Within two to three minutes, you have three versions of the same passage, one with annotations and a vocabulary sidebar for below-level readers, one standard version, and one with extended analyzis questions for advanced students. You spend ten minutes reviewing and tweaking. Total time: fifteen minutes instead of ninety.

The key to making this workflow fast and reliable is specificity in your prompts. Vague prompts produce vague results. If you write 'make this easier,' you'll get something generic. If you write 'rewrite this for a Grade 4 reading level, use sentences under 12 words where possible, replace any words above a Grade 5 vocabulary level, and add a four-word glossary at the bottom,' you'll get something you can actually use. This is what educators mean when they talk about prompt engineering, it's not a technical skill, it's a communication skill. You're briefing a very capable assistant. The more clearly you describe what you need, the better the output.

Teachers who've built this into their regular practice report saving between three and six hours per week on material preparation. That time doesn't disappear, it shifts into higher-value work: giving more personalized feedback, running small-group sessions, actually talking to students about their learning. The goal isn't to automate teaching. It's to automate the parts of lesson preparation that don't require your expertise, so the parts that do get more of your attention. Differentiation becomes something you actually do consistently, rather than something you intend to do when you have more time.

Click to enlarge

How This Looks Different Across Teaching Roles

Differentiation needs vary significantly depending on what and who you teach. The same AI capabilities apply across roles, but the specific use cases shift. Here's how different educators are applying these tools to their specific contexts.

• Primary / Elementary Teachers: Use AI to create levelled reading groups from a single text, generate phonics-friendly versions of stories, and produce visual vocabulary cards in Canva AI for early readers
• Secondary / High School Teachers: Use AI to produce tiered essay prompts, create scaffolded writing frames for students who struggle with structure, and generate extension research questions for high-achieving students
• ELL / ESL Teachers: Use Claude or ChatGPT to add bilingual glossaries, simplify syntax without losing content, generate sentence frame banks, and create vocabulary pre-teaching materials for upcoming units
• Special Education Teachers: Use AI to break multi-step tasks into single-step instructions, rewrite content in plain language, generate social story templates, and create modified assessment versions aligned to IEP goals
• Corporate Trainers and L&D Professionals: Use AI to create beginner, intermediate, and advanced versions of the same training module, generate role-specific examples within the same content, and produce simplified quick-reference cards alongside full training guides
• University and College Lecturers: Use AI to create accessible summaries of complex academic readings for students with learning differences, generate discussion questions at varying levels of abstraction, and produce annotated versions of primary sources for students new to academic analyzis

Prompt Example: Creating a Scaffolded Writing Task for ELL Students

Key Takeaways from Part 1

1. Differentiated instruction is the right practice. AI makes it feasible to do consistently, not just occasionally
2. AI tools handle the production work (rewriting, reformatting, scaffolding) so teachers can focus on the pedagogical decisions
3. The four main differentiation tasks AI can support: reading level adjustment, language support, complexity scaffolding, and modality variation
4. Specificity in your prompts is the single biggest factor in getting usable output, treat it like briefing a capable assistant
5. ChatGPT Plus and Claude Pro are the strongest general-purpose tools for text-based differentiation; Canva AI works well for visual formats; Copilot and Gemini integrate into existing workflows
6. Always review AI output before using it with students, accuracy, tone, and fit to your specific students require your professional judgment
7. Start with one student group and one lesson before trying to differentiate everything at once
8. Teachers using AI for differentiation consistently report saving three to six hours per week on material preparation

Differentiating by Readiness, Interest, and Learning Profile

It's Tuesday morning. You have a Year 9 English class at 9am. Your unit on persuasive writing has three students who finished the baseline task in 20 minutes and are now bored, six students who are solidly on track, and four who are still unclear on what a thesis statement even is. You have 48 hours to prep three different pathways before Thursday's lesson. In the past, that meant three separate evenings of work. With AI, it means 25 minutes during your lunch break, if you know how to ask.

This is the practical reality of differentiation. The theory, that students learn better when content matches their readiness, interests, and learning preferences, has been established for decades. Carol Ann Tomlinson's framework identifies these three dimensions as the core levers teachers can pull. The problem has never been the theory. It's been the time. Creating genuinely different materials for different learners, at scale, within a real teaching schedule, is an enormous ask. AI doesn't replace your professional judgment about which students need what. But it does collapse the production time dramatically, turning a three-hour task into a 30-minute one.

Readiness-based differentiation means adjusting the complexity of a task or text based on where a student currently is, not where you wish they were. Interest-based differentiation means letting students explore the same concept through a topic they actually care about, persuasive writing about sports rather than historical speeches, for example. Learning profile differentiation means varying how students access or demonstrate knowledge: through visual summaries, structured debates, written arguments, or verbal explanations. AI tools can generate materials across all three dimensions once you know how to frame your request clearly.

The key shift is thinking of AI as a drafting partner, not a search engine. You're not looking up a pre-made worksheet. You're commissioning a specific document for a specific group of students, with your own context baked in. The more precisely you describe your students' needs, the better the output. A vague prompt, 'make a differentiated worksheet on fractions', gives you something generic. A specific prompt, 'create three versions of a fractions task for Grade 5: one for students still building place value understanding, one for grade-level learners, one for students ready for improper fractions', gives you something you can actually use.

Building Tiered Tasks Without Starting from Scratch

Tiered tasks are the workhorse of readiness-based differentiation. The concept is straightforward: all students work toward the same learning goal, but the complexity, support, and scaffolding vary by tier. A Tier 1 task might provide a sentence frame, a vocabulary list, and a worked example. A Tier 2 task provides the vocabulary list but expects students to construct their own structure. A Tier 3 task presents the same challenge with minimal support and an extension element. Writing three genuinely different versions of the same task used to take significant time. Now you can generate all three in a single prompt, then spend your professional time reviewing and refining, which is where your expertise actually matters.

Notice what happened in that output. The AI maintained the same learning objective, understanding energy transfer, across all three tiers. It varied the cognitive demand, the language complexity, and the level of support. You didn't have to invent three different framings from scratch. Your job now is to review each tier against what you actually know about your students: does Tier 1 match what your struggling learners can access? Is the extension genuinely challenging for your high-flyers, or does it need pushing further? That professional judgment layer is irreplaceable. The AI just did the heavy lifting of the first draft.

Click to enlarge

Comparing AI Tools for Differentiation Tasks

Not every AI tool performs equally well for every differentiation task. Some are better at generating structured documents like tiered worksheets. Others are stronger at explaining concepts in multiple ways or adjusting reading levels. Knowing which tool to reach for saves time and produces better results. The table below reflects real-world performance across the tools most accessible to teachers, no institutional IT budget required, though school licensing may give you access to Microsoft Copilot or Google Gemini for Education at no extra cost.

The most effective workflow many teachers land on is a two-tool approach: use ChatGPT or Claude to generate the content and structure, then move it into Canva or Google Docs for formatting and visual polish. This separates the thinking work (which AI is genuinely good at) from the presentation work (where visual tools add real value). If your school uses Microsoft 365, Copilot inside Word is worth exploring specifically because it removes the copy-paste step, you can generate and edit in the same document, which matters when you're working fast.

Adapting Texts for Different Reading Levels

One of the most immediately useful differentiation tasks AI handles well is adapting a source text to different reading levels. This comes up constantly: a Year 10 history class where the textbook excerpt is inaccessible to three students with reading difficulties, an EAL learner who understands the concept but not the academic language, a gifted student who needs the original source text rather than a simplified version. Traditionally, a teacher might spend an hour rewriting a passage at a lower reading age. AI does this in under a minute, and does it well, preserving the key information while changing the sentence structure and vocabulary.

The same technique works in reverse. If you have a straightforward textbook summary and want a version that challenges your advanced readers with more complex syntax, nuanced argument, and higher-level vocabulary, you can ask for that too. The critical step is always to read the output carefully. AI can occasionally introduce inaccuracies when simplifying complex content, it might drop a key detail or slightly misrepresent a concept in the process of simplifying the language. A 60-second review catches these before they reach students.

Differentiation by Interest. A Faster Path to Engagement

Interest-based differentiation is often underused because it feels like a lot of extra work, if you're letting students explore persuasive writing through their own chosen topic, that seems like 30 different lesson plans. AI makes this manageable. The trick is creating a flexible template prompt that you can run multiple times quickly, swapping out the interest area each time. A student who's passionate about football gets a persuasive writing task framed around sports governance. A student who loves gaming gets the same rhetorical skills task framed around gaming regulation. Same learning objective, different content wrapper. Each version takes about 90 seconds to generate.

This approach works particularly well for English, PSHE, Humanities, and any subject where the skill being taught is somewhat content-agnostic. math is slightly different, interest-based differentiation there often means using contextualised word problems that reflect student interests rather than changing the mathematical concept itself. A student who loves baking gets a fractions problem about recipes. A student interested in sport gets the same fractions problem framed around match statistics. AI generates these contextualised versions quickly when you provide the core mathematical task and a list of interest areas.

How Different Educators Use These Techniques

• Primary teachers: Use AI to generate three reading comprehension versions of the same text at different Lexile levels, then create visual vocabulary cards in Canva for EAL learners in the same session.
• Secondary subject teachers: Produce tiered exam practice questions that mirror the actual assessment format, same question structure, adjusted complexity, so all students practice the real skill.
• SEN coordinators: Use Claude or ChatGPT to rewrite standard lesson materials into plain English, add structured prompts, and generate visual step-by-step instructions for students with processing differences.
• EAL/ESL teachers: Generate bilingual glossaries, sentence frame scaffolds, and simplified source texts for specific units, tasks that used to require specializt resource creation time.
• Sixth form and college teachers: Create extension reading lists with AI-generated summaries at different complexity levels, so advanced students engage with primary sources while others access accessible summaries.
• School leaders and curriculum leads: Use AI to review existing schemes of work and identify where differentiation is missing, then generate suggested adaptations for each unit.
• Cover teachers and supply staff: Use a brief AI prompt to generate a fully tiered cover lesson from the topic name alone, reducing the burden on class teachers to over-prepare cover work.

Click to enlarge

Generating Scaffolded Support Materials at Speed

Beyond tiered tasks, scaffolded support materials are one of the highest-value things AI produces for differentiating teachers. Sentence starters, writing frames, vocabulary banks, worked examples, step-by-step process guides, these take time to build well, and many teachers either skip them or reuse outdated versions that no longer fit the task. With AI, you can generate a bespoke scaffold for a specific task in about two minutes. The scaffold is designed for that exact activity, not a generic template dragged in from a previous year.

That scaffold took under two minutes to generate. A teacher building it manually, paragraph frame, sentence starters, vocabulary bank, checklist, would typically spend 20 to 40 minutes. More importantly, the scaffold is specific to this exact task and question, not a generic writing frame that students half-ignore because it doesn't quite fit. Specificity is what makes scaffolding actually useful, and AI makes specificity cheap. You can generate a new, tailored scaffold for every major writing task in your scheme of work without it consuming your planning time.

Click to enlarge

From Theory to Monday Morning. Your First Differentiation Workflow

The most common barrier teachers report isn't capability, it's not knowing where to start. So here's a specific, repeatable workflow you can use this week. Before you plan your next lesson where differentiation is needed, open ChatGPT or Claude and type a three-sentence context block: your year group, your subject, and what the lesson is trying to achieve. Then make one specific request, a tiered task, a simplified text, a scaffold, or an interest-based version. Review the output, make two or three edits based on what you know about your students, and save it. That's the whole workflow. It takes under 10 minutes once you've done it twice.

The second step is building a small personal prompt library. Once you find a prompt that produces a good tiered task for your subject, save it in a Google Doc or Notion page. Swap out the topic each time you need it. You're not reinventing the prompt, you're running a proven template. Teachers who work this way report that within two or three weeks, they have five to eight reliable prompts that cover the most common differentiation tasks in their subject. That's a permanent reduction in planning time, not a one-off shortcut.

Key Takeaways from Part 2

1. Tiered tasks are the most immediately practical differentiation output AI produces, same learning goal, three levels of complexity, generated in a single prompt.
2. Context is the difference between generic output and genuinely usable materials. Three sentences about your class transforms AI responses.
3. ChatGPT and Claude are strongest for generating content and structure; Canva and Google Docs handle visual formatting and distribution.
4. AI adapts existing texts to different reading levels quickly, but always read adapted content before distributing to catch any accuracy issues.
5. Interest-based differentiation becomes manageable with AI because you can run the same template prompt multiple times with different topic contexts in minutes.
6. Scaffolded support materials, sentence frames, vocabulary banks, worked examples, checklists, are high-value, time-consuming to build manually, and fast to generate with AI.
7. A personal prompt library of five to eight reliable differentiation prompts, built over two to three weeks, creates a permanent reduction in planning time.
8. Your professional judgment about which students need what remains irreplaceable. AI handles production, you handle the pedagogical decisions.

It's Thursday afternoon. You have a unit test on Friday, and you already know the results will scatter across the room, some students will ace it, several will scrape by, and a handful will struggle badly. You need three different versions of next week's review activity, a support sheet for your English language learners, and an extension challenge for your advanced group. Normally, building all of that takes your Sunday. This week, you have a different plan. You open ChatGPT, paste in your learning objective, and start generating. By the time you finish your coffee, you have four differentiated materials ready to review. That's not a fantasy. That's what teachers are doing right now.

Turning One Lesson Into Many. Without Burning Out

Differentiated instruction has always been the right idea and the exhausting reality. The theory is simple: meet students where they are. The practice means creating multiple versions of every resource, anticipating different learning needs, and somehow doing it all before the bell rings. AI doesn't replace your professional judgment about what students need, you still make those calls. What it removes is the production burden. You decide a struggling reader needs a vocabulary-supported version of a text. AI builds it in 30 seconds. You decide an advanced learner needs an open-ended extension task. AI drafts three options. Your expertise points the direction; AI does the heavy lifting of writing and formatting.

The most practical shift in thinking is this: stop treating your lesson materials as finished products and start treating them as starting prompts. Your existing worksheet, rubric, or reading passage becomes raw input for AI. You feed it in, specify the learner group, and ask for an adapted version. A grade 5 science text can become a simplified version with sentence starters for struggling readers, a standard version, and a version with extension questions for high achievers, all from one original document. Claude and ChatGPT both handle this well. The key is being specific about who you're adapting for and what support or challenge looks like in your classroom context.

Visual learners, students with attention difficulties, English language learners, gifted students who finish in ten minutes, each group has different needs, and AI can address several of them simultaneously. Canva AI helps you reformat text-heavy content into visual layouts. Grammarly AI simplifies language complexity. ChatGPT generates tiered question sets. Microsoft Copilot, if your school uses Microsoft 365, can draft differentiated versions of Word documents directly inside the tools you already use. The workflow is genuinely teachable, once you've done it twice, it takes minutes, not hours.

Scaffolding, Extensions, and Everything in Between

Scaffolding means giving students structured support to access content they couldn't reach independently. AI is particularly strong at generating scaffolds: sentence frames, worked examples, vocabulary lists, step-by-step instructions, and guiding questions. Extensions push students who've mastered the core content toward deeper thinking, analyzis, synthesis, real-world application. Both require you to know your students. Neither requires you to write everything from scratch. The prompt examples below show how a single learning objective can produce two very different materials for two very different learner groups.

Comparing AI Tools for Differentiation Tasks

Putting It Into Practice This Week

Start with one material you already have, a worksheet, a reading passage, a set of discussion questions. Paste it into ChatGPT or Claude and ask for two adapted versions: one with scaffolding for students who need support, one with extension challenges for students who need stretch. Review both outputs critically. You'll likely adjust the vocabulary level or tweak a question. That editing process takes two minutes, not twenty. The point isn't that AI gets it perfect, it's that AI gets you 80% of the way there instantly, so you spend your time on the 20% that requires your professional knowledge of your students.

The second habit worth building is keeping a short brief about your key learner groups saved in a document. Three or four sentences describing your support group, your grade-level group, and your extension group, their reading levels, what they struggle with, what motivates them. Paste that brief into every AI prompt. This context transforms generic outputs into genuinely useful ones. It also saves you re-explaining your class every time. Think of it as onboarding a new teaching assistant before each planning session.

Over time, you'll build a personal library of prompts that work for your subject and year level. A prompt that generates a tiered question set for year 8 science. A prompt that creates sentence frames for ELL students in English class. A prompt that produces an open-ended extension task from any learning objective. These reusable prompts become your differentiation toolkit, and the investment of writing them once pays off every single week for the rest of the term.

How This Applies Across Teaching Roles

• Primary / Elementary Teachers: Use AI to generate phonics-based scaffolds, picture-supported instructions, and number sentence frames for math, ideal for early readers and ELL students.
• Secondary / High School Teachers: Create tiered essay prompts, differentiated source analyzis tasks, and extension research questions that push advanced students toward independent inquiry.
• Special Education Teachers: Prompt AI to strip complex syntax from any text, add visual cues instructions, and generate IEP-aligned task modifications based on specific learning goals.
• ESL / EFL Teachers: Use Claude or ChatGPT to rewrite texts at CEFR A2, B1, or B2 levels on demand, specify the level and ask for a glossary of new vocabulary with example sentences.
• Instructional Coaches and Curriculum Designers: Build differentiation frameworks and tiered rubric banks that classroom teachers can customize. AI drafts the structure, teachers adapt it for their context.
• School Leaders: Use AI to model what differentiated materials look like across year levels, helpful for professional development sessions and coaching conversations with teachers.

Key Takeaways

• AI removes the production burden of differentiation, you still make the professional decisions about what each student needs.
• Treat your existing materials as prompts, not finished products. Feed them into AI and ask for scaffolded, standard, and extended versions.
• Specific prompts produce useful results. Include the year level, reading level, and what support or challenge means for your actual students.
• Keep a short learner brief saved and paste it into every differentiation prompt, it transforms generic outputs into classroom-ready materials.
• ChatGPT and Claude are both free to start and require no technical skills. You can build a differentiated activity set in under 30 minutes.
• Build a personal prompt library over the term. Reusable prompts that work for your subject and year level become a permanent time-saving asset.
• AI-generated materials need your editorial eye, a two-minute review and edit ensures the output fits your specific students, not a generic classroom.
• The goal is not to automate teaching. It's to spend your limited preparation time on the high-judgment work only you can do.