4 Simple & Easy Ways to Use AI to Differentiate Instruction
Designing for All Learners with AI and Universal Design Learning
One path doesn’t fit all—AI-powered differentiation opens multiple ways forward for every learner.
By Dani Kachorsky, PhD
This year, I’ve been working with some students who—in a public school setting—would have IEPs. At my school, they have learning plans. I’ve worked with students on IEPs and learning plans before, and honestly, I’ve been pretty lucky. The things I’ve needed to do for them as a teacher weren’t particularly difficult or challenging for me to accomplish. That said, I know I can always do better. So this year, I’ve been exploring new ways that AI can help support students with disabilities—students on IEPs, learning plans, or 504s—and, honestly, it’s changing the way I think about differentiation in general.
As a quick note, a lot of what I’m finding applies just as well to English language learners or really to any students. One of the big ideas behind Universal Design for Learning (UDL) is that accommodations and strategies designed for students with disabilities are often just good teaching practices. When we plan instruction that’s accessible to the widest possible range of learners, everyone benefits. For example, UDL encourages explaining things in multiple modes—written, visual, auditory, kinesthetic—because people access information differently. I hear students say they’re “visual learners,” but I think everyone is a visual learner, and an auditory learner, and a kinesthetic learner. The more ways we present information, the more likely it is to stick.
So, with that in mind, here are four ways I’ve been using AI to differentiate instruction for students with disabilities (and, really, everyone else too):
1. Rewriting Materials for Accessibility
Let’s start with the basics. AI can rewrite material for you—anything from a handout you made yourself to a news article or academic passage that’s above a student’s reading level. If you’ve got a 12th grader reading at a 7th-grade level because of a disability, and one of their accommodations is that content should be rewritten or reframed, AI can handle that. Just give it the source material and ask it to make it appropriate for that reading level. (Personally, I’ve found Claude to be the best choice for this task.)
You can do the same thing for language learners—tell the AI to adapt the text for an L1, L2, or L3 English proficiency level. It helps to give the AI examples or definitions to guide it—for instance, show it what “7th-grade reading level” looks like or provide the official language proficiency descriptions. The more context you give it, the better job it does.
2. Providing Alternative Access to Information
Not every student can or wants to read a text. Maybe you’ve got a few articles students need to read and synthesize, but one student gets an audio accommodation. If those articles don’t have audiobook versions, AI can help here, too.
NotebookLM—one of Google’s newer AI tools—is great for this. You can upload the articles or source materials and then create an AI-generated podcast version of them. NotebookLM uses only the sources you specify and produces a natural-sounding conversation summarizing and discussing the key ideas from those texts. This means you can give students options–they can read the original articles OR listen to the podcast version.
And honestly, this helps everyone. Some students will read because it’s faster, others will listen because it helps them process better. Either way, they get access to the same information in a format that works for them.
3. Creating Step-by-Step Instruction Videos
This is my new favorite use case. NotebookLM recently added a video feature, and I’ve been using it to create short, step-by-step instructional videos for assignments.
In class, I always explain directions out loud, have written instructions on Canvas, and include them in slide decks. But I’ve noticed that even with all that, many students struggle to process detailed instructions. And, if I simplify them too much, they miss important nuances.
So now, I upload my assignment overview, step-by-step directions, evaluation criteria—everything—to NotebookLM. It then generates a short video (around five minutes) walking students through the process. The language is a little more general than mine, but the structure is excellent. Students can pause after each step, do the task, then hit play for the next one. I still post the detailed written instructions on Canvas, but I embed the video alongside them.
I’ve found this especially helpful for students who have accommodations for multiple explanations or reminders—but it’s great for everyone. Kids can rewatch the video anytime, whether it’s during class or at 11 p.m. when they suddenly remember an assignment detail.
4. Building Q&A Chatbots for Assignments
This last one is brand new for me, but I’m really excited about it. I’ve started creating Q&A chatbots for assignments using either ChatGPT (if you have a Plus account) or Flint AI, which is built specifically for schools and offers free teacher accounts.
Basically, you upload the assignment description, instructions, rubrics, and any supporting materials. Then, you set guardrails—what the AI can and can’t do. For example, mine can answer questions like “What’s the first step in this process?” or “Can you explain this in a different way?” but it can’t give feedback on student work or write anything for them.
Both ChatGPT and Flint let you test the bot before sharing it with students. Flint even lets you run simulations to see how it might respond to typical student questions, which is super useful.
The idea here is to give students a “just-in-time” helper—something that can clarify instructions when I’m not available. It’s like having a patient, knowledgeable friend who’s available 24/7. As long as I set the right parameters, it can give students the support they need without overstepping into doing the work for them.
Wrapping Up
Each of these strategies connects to the principles of Universal Design for Learning. They don’t just make things easier for students with IEPs or learning plans—they make learning better for everyone.
I’d love to hear how other teachers are using AI to differentiate or support diverse learners. What’s working for you?
AI Transparency Statement
This post was created using the AI-assisted workflow I described in a previous essay. I began by audio recording my thoughts and experiences, then used AI to transcribe and synthesize my reflections while maintaining my voice. I added and revised material through a few additional prompts in the LLM interface before copying the content into a document, where I made further revisions.
ChatGPT prompt for image
Create a square, watercolor-style digital illustration featuring a central, bold blue arrow pointing upward. The arrow should split cleanly into four curved branches or paths near the top, each also pointing upward and slightly outward, symbolizing diverse but progressive directions. Use soft, blended background tones—primarily greens, blues, and subtle yellows—to create a calm, intentional atmosphere. The style should feel modern and professional, not elementary or cartoonish. The overall composition should suggest clarity, inclusivity, and educational growth.
This is great! I’m really interested in the Q&A chatbot for assessments - that could be really useful for my students (and me), definitely something I’ll think about doing. Thanks for sharing your ideas!
This is very interesting and echos something Im working on currently. The future is recognising when and how to use Ai!