Best AI for Teaching Mathematics Problem Solving in 2026

Quick Answer: AI for teaching mathematics problem solving generates non-routine, rich mathematical tasks at calibrated complexity levels; Polya-framework problem-solving scaffolds that develop metacognitive strategy use; three-act mathematical task setups (Dan Meyer style) that develop mathematical modeling and motivation; number talk prompts for mental calculation discussion and number sense development; error analysis activities that turn mistakes into learning opportunities; productive struggle facilitation guides with teacher moves that extend rather than rescue struggling students; worked example sequences that support schema development; mathematical discourse discussion protocols; and formative assessment tasks that reveal mathematical reasoning, not just procedural accuracy. Platforms like EduGenius help mathematics teachers (Grades 1-9) design lessons that develop genuine mathematical problem-solving capacity alongside procedural fluency.

The most persistent gap in mathematics education is the gap between procedural fluency (the ability to execute algorithms accurately and efficiently) and mathematical problem solving (the ability to reason about unfamiliar situations, choose appropriate strategies, persevere through difficulty, and construct and evaluate mathematical arguments). Students can be taught to execute standard algorithms—long division, the quadratic formula, integration by parts—without developing any ability to decide which algorithm is relevant to a novel problem, whether the answer makes sense, or how to approach a situation that doesn't match a familiar procedure.

Research on what differentiates high-performing from low-performing mathematics students (Schoenfeld, 1992; Boaler, 2016; Hattie, 2009) consistently identifies three factors—not procedural accuracy—as the primary predictors of long-term mathematics achievement:

• Problem-solving orientation: approaching mathematics as a sense-making activity
• Metacognitive strategy use: choosing and applying strategies deliberately
• Mathematical disposition: viewing struggle as part of the process rather than evidence of inadequacy, and being able to judge whether an answer is reasonable

Students with these traits consistently outperform procedurally stronger peers on non-routine problems.

The implications for teaching are profound and largely ignored: the dominant mode of mathematics instruction in most classrooms—teacher demonstrates a procedure; students practice similar problems; the cycle repeats—develops procedural fluency but systematically fails to develop the mathematical reasoning that research identifies as most important for long-term success.

AI supports mathematics teaching by making reasoning-focused, problem-solving-centered instruction more accessible: generating the rich tasks, structured discussions, and metacognitive scaffolds that develop genuine mathematical thinking.

Research Foundations of Mathematical Problem Solving

Polya: Problem-Solving Heuristics

George Polya's How to Solve It (1945)—which has sold over a million copies and has been continuously in print for over 75 years—introduced the most influential framework for mathematical problem solving in mathematics education:

Polya's Four-Phase Model:

1. Understand the problem: What is the unknown? What data is given? What is the condition? Can you draw a figure? Can you introduce suitable notation? Polya emphasized that students frequently attempt to solve problems they haven't clearly understood—working from a misunderstanding of the problem statement. Teachers ask: "Can you restate the problem in your own words?" "What are you trying to find?" "What information do you have?"

2. Devise a plan: Have you seen this problem before, or the same problem in a slightly different form? Do you know a related problem? Could you use its result? If you can't solve the problem directly, could you solve a related simpler problem? Could you use analogy? Polya catalogs 19 problem-solving heuristics: working backwards; finding a pattern; setting up equations; drawing a diagram; solving a simpler problem; making a list or table; guessing and checking. These heuristics are the strategies that expert problem-solvers use and novices rarely do—not because novices are less intelligent, but because they haven't been explicitly taught the heuristics.

3. Carry out the plan: Execute the plan you've devised; check each step; can you see clearly that the step is correct?

4. Look back: Can you verify the result? Can you derive the result differently? Can you see it at a glance? Can you use the result or method for some other problem? Polya's "looking back" phase is the most consistently omitted phase in student problem solving and the most important for learning transfer: examining not just whether the answer is correct but why the strategy worked and how it generalizes.

Heuristic Instruction Research: Alan Schoenfeld's research (discussed below) found that simply teaching Polya's heuristics without also teaching how to choose between heuristics and monitor their effectiveness during problem solving produces limited improvement. The heuristics are necessary but not sufficient; metacognitive strategy regulation is also required.

Schoenfeld: Mathematical Metacognition

Alan Schoenfeld's research—most extensively reported in Mathematical Problem Solving (1985) and "Learning to Think Mathematically" (Handbook of Research on Mathematics Teaching and Learning, 1992)—established the critical role of metacognition in expert mathematical problem solving:

Resource, Heuristic, Control, Belief: Schoenfeld identified four categories of factors that distinguish expert from novice mathematical problem solving:

1. Resources: Mathematical knowledge—facts, algorithms, procedures, concepts—that can be brought to bear on a problem. Experts don't know more facts than competent novices; they're better at organizing and accessing what they know.
2. Heuristics: Problem-solving strategies (Polya's heuristics and others). Experts use heuristics systematically; novices either don't know them or don't use them consistently.
3. Control: Metacognitive processes—monitoring and regulating one's own problem-solving process. This is where expert-novice differences are most dramatic: experts frequently pause to evaluate their approach, abandon unproductive paths, and redirect; novices tend to "run off the end" of an initial approach—pursuing a strategy even when it's clearly not working—rather than stepping back to reassess.
4. Belief system: Students' beliefs about mathematics and their own mathematical ability profoundly affect their engagement: students who believe mathematics is a set of rules to be memorized approach problems by searching for the right rule to apply; students who believe mathematics is sense-making approach problems by trying to understand the situation. Students who believe mathematical ability is fixed ("I'm just not a math person") give up on difficult problems quickly; students who believe mathematical ability develops with effort persist.

Control as the Key Variable: Schoenfeld's most significant finding is that control—metacognitive monitoring and regulation—accounts for more variance in problem-solving success than knowledge or heuristic use. Students who know the right strategy but don't monitor whether it's working fail more often than students who know fewer strategies but monitor more effectively. This finding directly argues for explicit instruction in metacognitive monitoring during problem solving, not just strategy instruction.

The Expert's Internal Monologue: Schoenfeld documented that when expert mathematicians problem-solve, they maintain an ongoing metacognitive commentary: "What am I doing? Why am I doing it? How does this connect to what I'm trying to find? Is this working? Should I try something different?" Teaching students this internal monitoring dialogue—and providing opportunities to practice it explicitly—develops the control metacognition that is the critical missing element in novice problem solving.

Jo Boaler: Mathematical Mindsets and Productive Struggle

Jo Boaler's research—synthesized in Mathematical Mindsets (2016) and Limitless Mind (2019)—extends Carol Dweck's growth mindset work (Dweck, Mindset, 2006) into the specific domain of mathematics education:

• Math trauma and fixed beliefs: Boaler's research documents that a significant proportion of adults carry "math trauma"—persistent beliefs from early mathematics education that they "are not math people," that mathematics is about speed and right answers, and that mathematical ability is fixed. These beliefs, developed in early childhood (often from timed tests, public humiliation from wrong answers, tracking into "low math" groups), persist into adulthood and reproduce in parents' communications to children. The consequence: even students with strong mathematical potential disengage from challenging mathematics before they discover what they can do.
• The myth of the "math brain": Boaler synthesizes neuroscience research (Maguire et al.; Park et al.) to challenge the common belief that mathematical ability is innate and fixed. Neuroimaging studies show that mathematical engagement changes brain structure—mathematics learning is not a revelation of pre-existing capacity but construction of new neural pathways through practice. The practical implication: "I'm not a math person" is a conclusion about past experience, not a fact about fixed capacity.
• Low-floor, high-ceiling tasks: Boaler advocates for mathematical tasks with low floors (accessible entry points that allow all students to engage productively from the beginning) and high ceilings (significant extension possibilities that challenge the most mathematically confident students without establishing a "finished" point). This is fundamentally different from tiered tasks, where different students do different work: low-floor, high-ceiling tasks let the same task productively engage all learners, with natural differentiation emerging from how deeply different students explore it.
• Productive struggle: Drawing on research showing that struggle (cognitive effort at the edge of competence) is necessary for learning, Boaler argues that the well-intentioned teacher response of immediately rescuing students who show difficulty ("let me show you how to do this") prevents the productive struggle that develops genuine understanding. Students who are given procedures when they struggle don't develop the problem-solving tolerance or the independent strategy repertoire that produces durable mathematical competence.
• Speed and mathematics: Timed mathematics tests—intended to measure fluency—actually measure anxiety in students who don't have strong automatic recall, produce trauma for those students, and signal to all students that mathematics is about speed rather than reasoning. Boaler's review of the evidence (including OECD PISA data showing negative correlations between timed practice and mathematical achievement in several countries) argues for eliminating timed tests and replacing them with fluency activities that don't activate anxiety.

NCTM Process Standards and Mathematical Practices

The National Council of Teachers of Mathematics (NCTM) Principles and Standards for School Mathematics (2000) identified five process standards that complement content standards and describe mathematical ways of knowing:

1. Problem solving: Solving novel problems is how mathematical knowledge is built and deepened; problem solving is not a separate topic but the mode through which mathematical learning happens
2. Reasoning and proof: Mathematical ideas are justified through logical reasoning; students should make and investigate mathematical conjectures; develop arguments; evaluate arguments made by others
3. Communication: Mathematical thinking must be communicated—in spoken and written language; students develop understanding by explaining their thinking to others and by interpreting others' explanations
4. Connections: Mathematical ideas are interconnected; connections within mathematics (how algebra and geometry relate; how fractions connect to division) and connections to the world outside mathematics deepen understanding
5. Representation: Mathematical ideas can be represented in multiple ways (numerical; algebraic; graphical; geometric; verbal); facility with multiple representations deepens understanding and enables problem solving through translation between representations

Common Core Standards for Mathematical Practice (2010): The eight Standards for Mathematical Practice (SMP) in the Common Core State Standards align closely with NCTM process standards while also drawing on Polya's heuristics and Schoenfeld's metacognition research:

1. Make sense of problems and persevere in solving them
2. Reason abstractly and quantitatively
3. Construct viable arguments and critique the reasoning of others
4. Model with mathematics
5. Use appropriate tools strategically
6. Attend to precision
7. Look for and make use of structure
8. Look for and express regularity in repeated reasoning

Dan Meyer: Three-Act Mathematical Tasks

Dan Meyer's three-act task framework—developed through his teaching practice and blog (dy/dan) and widely adopted in middle and high school mathematics—reframes mathematical problem solving around student engagement and real-world mathematical modeling:

1. Act One: A video or image presents an intriguing, open-ended situation. Its purpose is to generate student curiosity and mathematical questions before providing any mathematical information. Students observe the situation and ask: "What do you notice? What do you wonder? What question is most interesting to investigate mathematically?" Classic examples: a video of someone piling pennies on a paper (students wonder when the paper will collapse); a video of someone filling a water tank (students wonder how long it will take); a photograph of a staircase (students wonder the slope).
2. Act Two: Students request the information they need to answer the mathematical question they identified. This reversal of the standard information-then-problem structure is pedagogically significant—students who have generated a question and then identify the information they need to answer it have much more ownership over the mathematical structure than students who are given information and told what to do with it.
3. Act Three: The answer is revealed (usually through a video or image showing the real-world resolution), letting students compare their mathematical predictions to reality and reflect on the accuracy and sources of any discrepancy. This check-against-reality step develops mathematical modeling habits central to STEM practice.

The Value of Incomplete Information: Meyer's argument is that standard textbook problems are "pseudo-context"—they use real-world situations as thin wrappers for mathematical procedures without genuine engagement with the real world's mathematical messiness. Three-act tasks engage students with genuine mathematical modeling: real-world situations where some information is missing, some is irrelevant, and the answer matters.

AI Applications in Mathematics Problem Solving Teaching

Rich Task Generation

"Create a set of 10 non-routine mathematics problems for 7th-grade students on the topic of proportional reasoning. The problems should be:

• Non-routine: not solvable by directly applying a single memorized procedure
• Contextually rich: set in situations students find interesting or relevant (sports statistics, social media, food, gaming)
• Low-floor, high-ceiling: accessible to students still developing proportional reasoning, with extension opportunities for those who grasp the basic relationship quickly
• Open-ended where possible: admitting multiple solution strategies
• Estimation-first: inviting an estimate before calculation

For each problem, provide: the problem statement; the proportional reasoning concept targeted (unit rate, scale factor, percent, part-to-whole ratio); 2-3 solution strategies students might use; common errors or misconceptions the problem surfaces; and an extension question.

Sample topics: comparing water usage across countries, scaling recipes for different group sizes, analyzing track team improvement, and comparing price-per-ounce across container sizes."

"Design a number talk sequence for a 5th-grade class working on multiplication with multi-digit numbers. A number talk is a 5-10 minute whole-class mental calculation activity where students mentally solve a problem and explain their strategies without paper. Design a 5-day sequence:

1. Day 1 — 15 × 4: strategies students might share include decomposing 15 (10×4 + 5×4) or doubling (15×2=30; 30×2=60)
2. Day 2 — 24 × 5: strategies include halving and doubling (12×10) or decomposing (20×5 + 4×5)
3. Day 3 — 36 × 25: strategies include using the relationship to 100 (36×25 = 36×100/4 = 900) or decomposing (30×25 + 6×25)
4. Day 4 — 18 × 99: strategies include the near-number approach (18×100 − 18)
5. Day 5 — student-generated problems

For each day, include: the problem; anticipated student strategies; facilitation questions to draw out connections between strategies ('Mariana used decomposition; Kaito used doubling. How are their strategies related?'); key mathematical ideas to highlight; and the connection to the formal algorithm.

Also provide a teacher facilitation script covering how to record student strategies on the board, how to respond when students give wrong answers without correcting immediately, and how to honor diverse strategies."

Productive Struggle Facilitation

"Design a productive struggle facilitation guide for a 9th-grade algebra class working on a challenging systems of equations problem. The guide provides teacher moves that support students in persevering without rescuing them. Context: students are working on a problem with no obvious starting point, so they need to make strategic choices about approach.

The productive struggle support system:

1. Pre-task cognitive preparation: ask students to first write down "What do I know? What am I trying to find? What's confusing me?" before attempting the problem—this Polya-aligned initiation activates metacognitive monitoring
2. Monitoring prompts (teacher circulates and uses these when students are stuck): "What have you tried so far? What happened?"; "What do you know for sure?"; "Is there a simpler version of this problem you could try first?"; "What information haven't you used yet?"; "Can you draw a diagram or make a table?"
3. Rescue prevention — what NOT to say: "Let me show you how to start"; "You should try substitution"; "Did you set up the equations yet?" These rescue moves prevent productive struggle
4. Struggle-to-breakthrough facilitation: when students say "I give up" or "I can't do this," normalize struggle ("This problem is supposed to be hard. What have you tried?"), focus on partial progress ("You figured out how to set up the variables—that's a significant step"), and redirect to available resources ("What strategies do you have in your toolkit for this type of problem?")
5. Post-struggle debrief: "What was hard about this problem? What strategy worked for you? What would you do differently next time?" This metacognitive debrief builds strategic self-awareness

Include: the challenging problem with solution notes; facilitation moves organized by phase of struggle; common student responses and teacher responses; and assessment notes for identifying students who need more support than productive struggle allows."

Mathematical Error Analysis

"Create an error analysis activity for 6th-grade students on fraction division. Common student errors in fraction division include:

• Multiplying instead of dividing (not inverting the second fraction)
• Inverting the first fraction instead of the second
• Dividing numerators by numerators and denominators by denominators (treating division like multiplication)
• Converting to decimals incorrectly and then dividing
• Getting the operation right but making arithmetic errors

Activity structure:

1. Students examine five "student work" samples showing different errors (fictional student work that you generate showing each error type authentically)
2. For each sample: identify whether the student's work is correct; if incorrect, find the error; explain why this error makes sense (what might the student have been thinking?); fix the error
3. Discussion: "Which error is most understandable? Which one might you make yourself? What would you tell each fictional student?"
4. Metacognitive reflection: "What are the most important things to remember when dividing fractions? What check could you use to catch errors in your own work?"
5. Connection to concept: "Why does 'invert and multiply' work? Can you explain it using diagrams or by connecting to measurement division?"

For each error type, provide: a fictional student work sample (mathematically accurate error); teacher analysis (what the error reveals about student understanding); and instructional response (what concept needs reinforcement)."

"Design a classroom discussion protocol for making mathematical discourse genuinely productive in a middle school class. Protocol: 'Strategy Sharing and Critique.' After students have independently attempted a rich problem, the class comes together to analyze multiple student approaches. Structure:

1. Gallery Walk: 4-5 student solution approaches posted around the room (either projected or on chart paper); students walk the room with sticky notes—green for "I understand and agree," yellow for "I have a question," pink for "I see an error or have a different approach"
2. Author explanation: the student who created each solution approach briefly explains their thinking (1-2 minutes)
3. Structured critique: class offers observations using sentence frames such as "What I notice about this approach is…," "A question I have is…," "A similar approach would be…," and "What I would change is… because…"
4. Teacher synthesis: highlight connections between approaches; ask "Which approach is most efficient? Most generalizable? Most interesting?"
5. Metacognitive closure: "What did you learn from seeing different approaches? Which approach would you use on a similar problem and why?"

Discussion norms: all approaches are shared respectfully; identifying errors is productive; explaining your own reasoning is required (not just a correct-wrong binary); and connecting approaches is valued.

Include: student sentence frames for discourse; teacher facilitation moves; how to select student work to share (not always the "best"—include strategic errors and partial solutions); and the connection to SMP 3 (construct viable arguments and critique the reasoning of others)."

EduGenius (edugenius.app) helps mathematics teachers (Grades 1-9) design rich problem-solving tasks, number talks, error analysis activities, productive struggle protocols, and mathematical discourse discussions—with credit-based access from $7.99/month and 25 free welcome credits.

Classroom Scenario: Mathematics Problem Solving in Buenos Aires, Argentina

Imagine you teach 8th-grade mathematics at a public secondary school in Buenos Aires's San Telmo neighborhood—one of the oldest and most historically layered neighborhoods in Argentina's capital, known for its tango milongas, antique markets, and colonial-era architecture. Buenos Aires is a city of approximately 3 million people (metropolitan area 15 million)—one of South America's major cultural and economic capitals and home to one of the highest concentrations of university-educated professionals in Latin America.

Argentina's educational context is distinctive in several ways relevant to mathematics teaching:

• A relatively well-developed public university system: the Universidad de Buenos Aires (UBA) is one of the largest and most respected universities in Latin America, and tuition-free for students who complete the preparatory course
• Secondary school completion rates that are high by regional standards
• A strong tradition of mathematical culture—Argentina has produced Nobel-level economists (Leloir, Milstein), prominent mathematicians, and a strong engineering tradition

However, Argentina also has significant educational inequality between private and public school sectors, between Buenos Aires and more rural provinces, and between schools in different neighborhoods of the capital itself. San Telmo is a gentrifying neighborhood with economic diversity—your classes might include students from professional families who have lived in San Telmo for generations alongside students from families who migrated from Buenos Aires's more peripheral neighborhoods or from other provinces or countries.

Argentine Mathematics Culture and Polya's Framework

Argentine secondary mathematics education has historically been characterized by a rigorous, proof-oriented mathematical culture—particularly in elite schools and university preparatory programs—that aligns with the formal mathematical reasoning Polya and Schoenfeld describe. Students in strong programs are expected to prove mathematical results, not merely apply algorithms; mathematical understanding is valued over mere procedural fluency in the academic culture many Argentine mathematics teachers are trained in.

This cultural orientation toward mathematical proof and reasoning gives a problem-solving approach strong cultural support in the school's academic expectations. But it also creates a challenge — a gap between:

• Students who come from mathematically enriched home environments (books, board games, families that discuss mathematical ideas, access to supplementary programs)
• Students who come from environments where mathematics is encountered only in school

This gap can be significant, and the school's assumption of mathematical cultural capital can disadvantage students who haven't had those enriching experiences.

You can address this explicitly by providing the cultural tools—problem-solving heuristics; mathematical vocabulary; proof-writing structures—that some students bring from home as explicit curriculum for all students.

Number Talks in Spanish-Language Mathematics

You could adapt the number talk tradition (primarily developed in English-language educational contexts) for Spanish-language mathematical discourse. The linguistic adaptation matters: Spanish mathematical vocabulary (multiplicar, dividir, descomponer, el producto, el cociente) is precise and rich, and mathematically precise language use in Spanish is itself a content objective in this approach—not just a medium for discussing mathematics but a cognitive tool for mathematical thinking.

The oral mathematical culture of Argentine classrooms—where verbal explanation and debate are expected as normal parts of classroom life—supports the number talk structure more naturally than in contexts where mathematics class is primarily silent individual work. Students in an Argentine class are often culturally comfortable debating, explaining, and evaluating each other's mathematical reasoning in oral discussion.

Error Analysis in Argentine Assessment Culture

Argentina's national assessment system (the Aprender national assessments) and university admission pathways create pressure toward procedural accuracy in standardized contexts. You can navigate this by explicitly connecting your error analysis activities to assessment performance: "Understanding why this error makes mathematical sense—and why it's wrong—is how you prevent yourself from making it on the exam." The error analysis approach is pitched not as alternative to high-stakes performance but as the deepest preparation for it.

Productive Struggle in an Argentine Classroom

Argentines have a cultural norm of directness and intellectual confidence that affects how students engage with mathematical struggle. When you first introduce sustained productive struggle—asking students to work on problems without immediate teacher rescue—several students with strong academic self-concepts may push back: "This is the teacher's job. Why aren't you teaching us?"

A culturally calibrated response frames the productive struggle as what university mathematics (and engineering, and economics—aspirational pathways for these students) actually requires. "In the Universidad de Buenos Aires ingeniería program, no one will rescue you from a difficult problem. The question is whether you develop the tools here to handle that, or whether you discover that need in the first UBA exam." This appeal to the genuine academic challenge ahead—rather than abstract pedagogical arguments—tends to land with students who are highly motivated by university aspirations.

The Tango Mathematical Metaphor

In a classroom in San Telmo—the spiritual home of tango—you could use the tango as a metaphor for productive mathematical struggle that students immediately grasp: "Tango is not learned by watching; it is learned by dancing—making mistakes, feeling where the balance is wrong, adjusting. Mathematics is the same."

"I can demonstrate every procedure; you can watch every solution. But you learn mathematics by doing it badly first, figuring out what went wrong, and adjusting. The struggle is the learning, not the obstacle to it." This locally resonant metaphor—students have seen tango danced in the streets of San Telmo since childhood, and many have tangoed themselves—makes the productive struggle principle viscerally accessible in a way that abstract pedagogical language would not.

Mathematical Discourse and Argentine Intellectual Culture

Buenos Aires has a strong café culture of intellectual debate—cafés like La Biela and El Federal have historically been gathering places for intellectuals, writers, and political thinkers. You can explicitly frame your classroom's mathematical discourse culture in this tradition: mathematical discussion becomes genuine intellectual engagement, not just procedure display. This reframing—from "showing you followed the steps" to "defending a mathematical argument"—shifts the social meaning of classroom discourse from compliance to genuine intellectual activity.

Students in such a class learn to use phrases like "Estoy de acuerdo con tu razonamiento, pero no con tu conclusión" ("I agree with your reasoning, but not with your conclusion") and "¿Por qué usaste ese enfoque en lugar de...?" ("Why did you use that approach instead of...?")—the language of genuine mathematical critique and debate.

Internal Resources

• Best AI for Teaching Algebra and Functions
• Best AI for Teaching Geometry and Spatial Reasoning
• Best AI for Differentiated Instruction
• Best AI for Formative Assessment and Feedback
• Best AI for STEM Education