Generating Differentiated Geometry Problems With AI

Geometry is the branch of mathematics where differentiation matters most — and also where it's hardest to execute manually. A student who can name a triangle's vertices and another who needs help even identifying shapes cannot reasonably share the same worksheet. AI makes it practical to generate three distinct tiers of geometry problems in under ten minutes, personalised to where each cluster of students actually is.

Quick Answer: Use AI to write tiered geometry problem sets by specifying three things in your prompt — the precise skill target (e.g., "classify quadrilaterals by their properties"), a concrete number constraint appropriate to each tier, and a real-world context that gives the problems life. Generate Tier 1 (concrete/visual), Tier 2 (procedural), and Tier 3 (reasoning/proof) separately for maximum quality control.

---

Why Geometry Differentiation Is Uniquely Difficult

Geometry sits at the intersection of spatial reasoning, vocabulary, and procedural skill — and students diverge sharply across all three dimensions at once. A student might have strong vocabulary (can name a rhombus) but weak spatial reasoning (cannot visualise a rotation), while their desk partner is the reverse. That makes geometry one of the most cognitively varied units a K–9 teacher will deliver.

According to NCTM (2025), geometry and measurement consistently show wider within-class variation than number operations, because visual-spatial ability does not track closely with numerical ability. Students who struggle with fractions sometimes excel at transformations; strong arithmeticians sometimes find proofs and spatial arguments difficult.

The practical problem for teachers is time. Creating three genuinely different problem sets — not the same problem with adjusted numbers, but problems that ask different cognitive questions — typically takes forty-five minutes to an hour per lesson. Multiply that by a geometry unit of twelve lessons and you are looking at nearly twelve hours of differentiation work. AI can collapse that to under two hours total, freeing the teacher to focus on what AI cannot do: circulate, question, and respond to students in real time.

This article shows you exactly how to do it — what to put in your prompts, how to structure tiers that represent genuinely different cognitive demands, and how to review AI output quickly before distributing it to students.

---

The Three-Tier Framework for Geometry

Tier 1 — Concrete and Visual

Tier 1 problems are designed for students who need anchor experiences: naming, identifying, and sorting shapes before they can compute or reason abstractly. The key design feature is that the cognitive demand stays at the recognition-and-description level. These problems rely heavily on visual cues, use simple whole-number measurements, and avoid multi-step procedures.

What Tier 1 problems do:

• Ask students to identify, name, or classify shapes given a diagram
• Require measuring or counting (e.g., count the sides, measure one angle with a protractor)
• Match shapes to real-world objects ("which road sign is an octagon?")
• Sort given sets of shapes by a single attribute (acute vs. obtuse triangles)

Sample Tier 1 prompt for AI:

"Write 6 geometry problems for Grade 4 students who are still building shape-recognition skills. All problems should involve classifying or identifying quadrilaterals — square, rectangle, rhombus, parallelogram, trapezoid — using labeled diagrams. Keep measurements as whole numbers under 20 cm. Avoid multi-step calculations. Include one sorting task and one real-world identification task."

Tier 2 — Procedural and Property-Based

Tier 2 is where the majority of a class typically sits during a geometry unit. Students at this tier can name shapes and understand basic properties — they now need to apply those properties to calculate, construct, or compare. The cognitive demand shifts from recognition to procedure.

What Tier 2 problems do:

• Calculate perimeter and area using formulas
• Apply angle properties (angles in a triangle sum to 180°)
• Identify types of angles formed by parallel lines and a transversal
• Use coordinates to find distance or midpoint

Sample Tier 2 prompt for AI:

"Write 6 geometry problems for Grade 7 students working on angle relationships. Include two problems using the angles-in-a-triangle rule, two using co-interior and alternate angles with parallel lines, and two multi-step problems combining both. Use integer angle measures. Include an answer key with working."

Tier 3 — Reasoning, Generalisation, and Proof

Tier 3 problems are for students who have mastered the procedural layer and need to extend into justification and generalisation. These problems require mathematical argument — "always, sometimes, never" reasoning, informal proof, and connections between properties.

What Tier 3 problems do:

• Ask students to explain why a property must be true ("why must the exterior angle of a triangle equal the sum of the two non-adjacent interior angles?")
• Present near-miss scenarios where a common error looks correct but is not
• Ask for generalisation: "Will this method always work? Explain using a different case."
• Require multi-step reasoning combining multiple geometry concepts

Sample Tier 3 prompt for AI:

"Write 5 reasoning problems for Grade 8 students who are confident with angle properties and quadrilateral classification. At least two problems should require students to justify their answer in writing (not just calculate). Include one 'always, sometimes, never' prompt about parallelograms. Do not repeat skills used in lower-tier problems."

---

Building Your Geometry Problem Set: A Step-by-Step Process

Step 1: Pin Down the Skill Target

Before opening any AI tool, write one sentence that names the geometry concept, the expected cognitive level for most students, and the grade. This becomes the anchor of every prompt you write.

Example anchor: "Grade 6 students applying the area formula for triangles and parallelograms, expected at procedural fluency but some still need visual support."

That anchor tells you where to pitch Tier 2, which then lets you construct Tier 1 (one cognitive step back) and Tier 3 (one step forward).

Step 2: Generate Each Tier Separately

Resist the urge to ask the AI to produce all three tiers in one prompt. When you bundle tiers, the AI tends to drift — Tier 1 becomes too procedural, Tier 3 becomes only marginally harder than Tier 2. Generating them separately with explicit cognitive-level descriptors produces much sharper differentiation.

Run three separate requests:

1. Tier 1 prompt with explicit "concrete / visual / recognition" language
2. Tier 2 prompt with explicit "procedural / apply formulas / multi-step" language
3. Tier 3 prompt with explicit "justify / generalise / reason / proof" language

Each prompt should also specify: number of problems, grade level, number constraints, whether diagrams are needed (and how complex), and whether answer keys should be included.

Step 3: Apply the Maths Teacher's Review Checklist

AI geometry problems have two common failure modes: incorrect numerical answers and conceptually muddled questions. Before handing anything to a student, check each problem against this checklist:

• Does the problem have a unique, correct answer (or a clearly justified open-ended response)?
• Are all given measurements sufficient to solve the problem — not over-determined, not under-determined?
• Is the geometry terminology accurate and grade-appropriate?
• Does the diagram, if described, actually match the numerical constraints?

This review typically takes two to three minutes per tier once you know what to look for.

Step 4: Format for the Classroom

Decide whether each tier gets a separate sheet (cleaner for independent work) or whether all three appear on a single sheet with visual tier markers. For most classes, separate sheets work best — students do not need to see what tier they are on, and teachers can hand them out without visual comparison between peers.

---

Prompt Engineering Specifics for Geometry

The difference between a generic geometry prompt and a useful one often comes down to four specifics that teachers instinctively know but often forget to include when typing.

Adding all six elements to your prompt takes thirty seconds and dramatically improves output quality. Education Week Research Center (2024) found that teachers who used specific AI prompt structures for differentiated materials reported a 40% reduction in editing time compared with teachers using open-ended prompts.

Handling Geometry Diagrams

This is the trickiest part of AI-generated geometry problems: most classroom-ready LLMs cannot reliably produce accurate geometric diagrams. The practical workaround is one of three approaches:

1. Text-described diagrams — ask the AI to write a precise description of the diagram ("Draw a triangle with vertices A, B, C where angle B is the right angle") and you sketch or digitally create it. Works well for straightforward shapes.
2. Desmos Geometry — for transformations, coordinate geometry, and constructions, build the diagram in Desmos and pair it with your AI-generated questions. Desmos is free and produces accurate, exportable geometric figures.
3. Geogebra — for secondary-level problems involving circle theorems, proofs, and dynamic geometry, Geogebra's diagram creation is more precise than anything text-based AI will produce.

Matching AI-generated questions with teacher-created or tool-created diagrams is a five-minute task that produces a professional-quality worksheet.

---

Classroom Scenario: A Grade 6 Area Unit

Say you teach a mixed-ability Grade 6 class of 28 students in a UK primary school. Your geometry unit covers area of rectangles, triangles, and compound shapes over three weeks. Based on your pre-assessment, you have grouped students into three tiers:

• Tier 1 (6 students): Still need visual support; comfortable with rectangle area but struggle with the ½ × base × height formula for triangles
• Tier 2 (17 students): Can apply the triangle area formula but make errors in compound shape decomposition
• Tier 3 (5 students): Confident with compound shapes; ready to work backwards from area to find unknown dimensions and to reason about which decomposition strategy is most efficient

For the unit's third lesson, you could use the following workflow:

Monday morning (15 minutes): You write three prompts, one per tier, using the step-by-step process above. Your Tier 2 prompt reads: "Write 8 area problems for Grade 6 students. Problems 1–3: triangles with integer base and height values between 4 and 18 cm. Problems 4–6: compound shapes made of one rectangle and one triangle, with all dimensions labelled. Problems 7–8: two-step problems where students find area and then calculate how many tiles of a given size cover the shape. Include an answer key with working. UK curriculum vocabulary."

The AI produces a set in about forty seconds. You review it: one problem has a dimension that makes the compound shape mathematically awkward to decompose, so you edit it. Total preparation time: roughly eighteen minutes.

During the lesson: Each tier works on their set independently while you circulate and run a small guided group with your Tier 1 students.

What this can change: Compared with running the unit on a single differentiated worksheet, this approach can give your Tier 3 students more intellectually challenging work and reduce the frustration Tier 1 students feel when problems are pitched above their current skill level. Because each set is aimed at the right level, both groups are less likely to need teacher intervention just to clarify the task.

---

Tool Comparison for Differentiated Geometry Content

EduGenius is particularly practical for this use case because its Class Profile system lets you store your Tier 1, Tier 2, and Tier 3 configurations as separate profiles — so next time you generate geometry content, the tool already knows the ability range, grade level, and format preferences for each group. Generating a new three-tier set becomes a two-minute task rather than a fifteen-minute one.

---