The Real Question Every Math Teacher Should Be Asking About AI
AI tools are arriving in classrooms faster than most professional development programs can keep up with. Math teachers, in particular, are being pulled in two directions at once: pressure to modernize instruction with new technology on one side, and the professional instinct to protect rigorous, student-centered thinking on the other. The conversation often gets stuck on whether AI belongs in math class at all. But that is already the wrong question.
The right question, as math educator Karen Levin — founder of Math for Humans and a nationally recognized voice in mathematics instruction — frames it, is this: can AI support your students without lowering the bar for their thinking? And just as importantly, do you know when to put it down?
In a recent EdTechTeacher webinar, Levin walked educators through a grounded, practical approach to that exact challenge. What follows are the key ideas from that session and how you can bring them into your own classroom or coaching practice.
One Classroom, Wildly Different Learners
Any experienced math teacher can describe the scene without hesitation. You have one class period, one lesson plan, and students who are in completely different places. Some came in already confident with the concept. Others are multilingual learners navigating both language and mathematical vocabulary simultaneously. Several are carrying math anxiety that quietly shuts down their ability to access even familiar content. And somewhere in the room, a handful of students need an extension challenge before you have finished introducing the idea to everyone else.
Differentiating meaningfully for all of those learners, on the same day, every day, is one of the most demanding and exhausting parts of the job. It requires not just instructional creativity but significant time — time to build materials, adjust tasks, and personalize feedback. This is the real differentiation problem, and it is the space where AI tools have genuine, practical potential.
The danger is not that teachers will start using AI. Most already are, or soon will be. The danger is using it in ways that quietly remove the cognitive work from students while creating the appearance of engagement and productivity.
Start With an Assignment Audit
One of the most valuable moves Levin introduced in her session is what she calls an assignment audit. Before layering in any AI tool, teachers benefit from pausing and honestly evaluating what a given task is actually asking students to do cognitively. Is it asking students to recall a procedure? To reason through a novel situation? To communicate their thinking? To connect representations?
This audit matters because AI interacts very differently with low-demand tasks versus high-demand tasks. A procedure-heavy worksheet is something AI can essentially complete on a student's behalf with minimal friction. A task that requires a student to explain their reasoning, justify a choice, or interpret a real-world context is much harder to outsource, and much more worth protecting.
When you audit your assignments before introducing AI support, you are making a deliberate decision about where the intellectual weight should stay with the student and where you are comfortable with AI lending scaffolding. That distinction changes everything about how you set up the work.
Practical Ways AI Can Genuinely Support Math Learning
Once you have clarity about which parts of a task require student thinking, AI can be a powerful tool for the surrounding support work. Here are several specific applications Levin highlighted as effective and appropriate.
- Generating differentiated versions of the same task. Teachers can use AI to quickly produce variations of a problem at different entry points — simpler numbers, visual supports, translated language, or additional context — without redesigning the task from scratch each time.
- Creating worked examples and anchor charts. AI can produce clear, step-by-step models that students reference while working, allowing teachers to spend less time building support materials and more time circulating and responding to student thinking in real time.
- Drafting feedback prompts. Rather than writing individualized written feedback for every student response, teachers can use AI to generate a menu of feedback prompts tied to common misconceptions, then select and personalize the ones that fit.
- Building discussion questions. AI can generate higher-order discussion questions aligned to a specific math concept, giving teachers a strong starting point to refine based on what they know about their students.
- Supporting multilingual learners. AI can help translate problems, simplify academic language, or produce glossaries of key math vocabulary in a student's home language, lowering the language barrier without reducing mathematical rigor.
Where AI Has No Business in a Math Classroom
Equally important to knowing how to use AI well is knowing where it does not belong. Levin was direct about this boundary. When the goal of a task is for a student to develop mathematical reasoning, to struggle productively with a problem, or to build the kind of conceptual understanding that only comes from working something out independently, AI should not be a participant in that process.
Productive struggle is not inefficiency. It is the mechanism through which durable learning happens. When a student uses AI to skip that struggle — to get an answer or a worked solution before wrestling with the problem themselves — the learning opportunity is gone. You might see completed work, but the thinking that makes math stick simply did not occur.
This is why classroom norms and explicit conversations with students matter so much. Students need to understand not just how to use AI tools, but why certain moments call for putting them away entirely.
Keeping the Teacher at the Center
Perhaps the most important takeaway from Levin's session is a reframe of the teacher's role when AI enters the room. AI does not replace teacher judgment. It creates more room for it. When AI handles some of the time-intensive preparation work — building differentiated materials, generating scaffolds, producing draft feedback — teachers get back something valuable: attention. More time to observe student thinking, ask better questions, notice who is stuck and why, and make the instructional decisions that only a human who knows those students can make.
That is the version of AI integration worth pursuing. Not AI as a shortcut, but AI as a tool that frees up the professional capacity of the teacher to do the most irreplaceable parts of the job even better.
Bringing This Into Your Practice
If you are a math teacher or an instructional coach working with math teachers, the best first step is not to find the most impressive AI tool on the market. It is to audit one upcoming lesson or unit with Levin's core question in mind: where does the student thinking need to stay protected, and where can AI provide support without displacing that thinking?
Start there. Build the habit of asking that question every time before introducing AI into a task. The technology will keep evolving, but this foundational question will remain the right one regardless of which tools exist at any given moment.
To explore Karen Levin's full session and hear her walk through specific examples and classroom applications, you can view the complete EdTechTeacher webinar on the Math for Humans approach to AI integration.