A colleague of mine spends Sunday evenings writing math word problems. Not because she enjoys it — because the textbook problems are either too easy, too hard, or so disconnected from her students’ lives that they check out before finishing the first sentence. “When will I ever need to know how many watermelons Train A is carrying?” as one of her fourth graders put it.

AI won’t replace your understanding of what your students need. But it can generate 20 differentiated word problems in the time it takes to write 3 by hand.

The Base Prompt

“Create 10 math word problems for [grade level] students practicing [skill — e.g., multi-digit multiplication, fractions, percentages]. Each problem should: use real-world scenarios relevant to [age group], include the answer and a brief solution explanation, and vary in difficulty (3 easy, 4 medium, 3 challenging). Avoid unrealistic numbers. Use names and situations that reflect diverse backgrounds.”

Differentiation by Level

Below Grade Level

“Simplify these word problems for struggling learners. Use: smaller numbers, fewer steps, simpler vocabulary, and add a visual hint or diagram suggestion for each problem. Keep the same real-world scenarios so students don’t feel singled out.”

Above Grade Level

“Create extension versions of these problems. Add: an extra step, larger or decimal numbers, or ask students to explain their reasoning. Include one open-ended problem where multiple approaches work.”

Subject-Specific Variations

Money and Financial Literacy

“Create 8 word problems about money for [grade] students. Scenarios: shopping with a budget, making change, comparing prices, saving for a goal, splitting costs with friends. Use realistic prices for items students actually buy. Include problems that require: addition, subtraction, multiplication, and one multi-step problem.”

Measurement and Geometry

“Create 8 word problems involving measurement for [grade] students. Scenarios: cooking (fractions, conversions), building something (area, perimeter), sports (distance, time), travel (maps, scale). Include diagrams where helpful. Mix metric and imperial units based on [your preference].”

Data and Statistics

“Create 6 word problems using data interpretation for [grade] students. Include a small data set (table or described chart) with each problem. Ask students to: find the mean/median/mode, make predictions, compare groups, or identify trends. Use topics students care about: sports stats, social media usage, school survey results.”

Making Problems Culturally Relevant

This is where AI actually shines. Instead of generic “John has 5 apples” problems:

“Rewrite these math problems using scenarios from [specific context — e.g., a school fundraiser, a family cooking for a holiday celebration, planning a class field trip to a local museum, a student starting a small business]. Use names that reflect the diversity of a typical [region] classroom.”

One teacher I talked to generates problems set in her actual school — the cafeteria, the playground, the library. Her students solve problems about real distances in their building, real prices from their book fair, real data from their class surveys. Engagement went through the roof.

The Answer Key Prompt

“For each problem above, provide: the correct answer, a step-by-step solution showing all work, the most common mistake students make on this type of problem, and a hint you could give a stuck student without giving away the answer.”

That last part — common mistakes and hints — saves you time during class when students are working independently.

Batch Generation for the Week

“Create a week’s worth of math warm-up problems for [grade level]. Monday through Friday, 3 problems per day, focusing on [skill]. Monday: review/easy. Tuesday-Wednesday: grade-level practice. Thursday: challenging/extension. Friday: mixed review of the week’s concepts. Format as a simple worksheet with space for student work.”

The Honest Limitation

AI-generated word problems need a quick review before you hand them out. I’ve seen AI produce problems with impossible answers (negative quantities of physical objects), unrealistic scenarios (a 7-year-old buying a $500 item), and occasionally incorrect answer keys. Spend 2 minutes checking the math. It’s still faster than writing everything from scratch.

What the Output Actually Looks Like

Here’s a real example. I asked ChatGPT for 4th-grade multiplication word problems using real-world scenarios:

Easy: “Maya’s class is making friendship bracelets for a school fundraiser. Each bracelet needs 6 beads. If Maya makes 8 bracelets, how many beads does she need?” (Answer: 48 beads)

Medium: “The school cafeteria orders 24 cartons of milk every day. How many cartons do they order in a 5-day school week? If each carton costs $2, what’s the total cost for the week?” (Answer: 120 cartons, $240)

Challenging: “Three classes are collecting canned food for a food drive. Ms. Rivera’s class collected 47 cans, Mr. Chen’s class collected 3 times as many as Ms. Rivera’s, and Ms. Johnson’s class collected 28 fewer than Mr. Chen’s. How many cans did they collect in total?” (Answer: 47 + 141 + 113 = 301 cans)

Notice how the problems use real names, real school scenarios, and build in complexity. That took about 15 seconds to generate. Writing those by hand would take 10-15 minutes.

Which AI Is Best for Math Problems?

ChatGPT is the best for creative, varied word problems. It comes up with scenarios students actually find interesting.

Claude is better when you need precise difficulty calibration and detailed answer explanations. It’s also better at following complex formatting instructions.

Gemini works fine for basic problems but tends to produce more generic, textbook-style scenarios.

For most teachers, ChatGPT’s free tier is enough. If you need to generate problems in bulk (a full week’s worth), Claude handles longer outputs more consistently.

Grade-Level Prompt Adjustments

Grade	Key adjustments to your prompt
K-2	Single-step only, numbers under 20, include picture cues
3-4	Two-step problems, multiplication/division, numbers under 1,000
5-6	Fractions, decimals, multi-step, real-world budgeting
7-8	Ratios, percentages, negative numbers, algebraic thinking
9-12	Functions, statistics, financial literacy, SAT-style

Add the grade-specific details to your prompt and the AI will calibrate accordingly. The more specific you are about what “4th grade” means (what they’ve learned, what they haven’t), the better the output.

Related reading: 10 AI Prompts for Math Teachers · AI for Science Lessons — Experiments, Explanations, and Lab Reports · AI for Differentiated Instruction — Reach Every Learner

🛠️ Need a quick lesson outline? Try our Lesson Plan Generator — free, no signup.