Subject playbook

AI for Mathematics Teachers

Go subject-deep: where AI earns its place in a Maths faculty, where it confidently gets the maths wrong, and how to use Lessio so every outcome code and every worked solution holds up.

Modules

Each module: clear learning outcomes → short, accurate, Mathematics-specific input → a hands-on activity using the Lessio generator → interactive knowledge checks. Mapped to the Australian Professional Standards for Teachers, with Working Mathematically (MAO-WM-01) threaded throughout.

Click a module to read it.

1. 1

   Where AI actually helps in Mathematics — and where it fails

   The high-value Maths use cases (varied practice, worked examples, applied contexts, three explanations, misconception probes) — and the one failure that defines this whole course: AI is unreliable at the actual maths.~45 min

   By the end of this module you'll be able to:

   • Identify the high-value uses of AI in a Mathematics classroom and map each to Working Mathematically.
   • Explain, with a maths example, why a general AI model is unreliable at multi-step calculation.
   • Apply the teacher-in-the-loop discipline as 'never trust the maths until I've checked it'.

   Standards2.1 Content and teaching strategies of the teaching area2.5 Literacy and numeracy strategies2.6 Information and Communication Technology (ICT)

   You've done the ethics — now the maths

   You completed the flagship Teaching with AI: Ethical & Effective Practice, so you already hold the policy stack (Australian Framework, NSW DoE + the six ethical checks, NESA integrity), the student-data hard line, and the RICE prompt method. This playbook does not repeat them. It goes where Mathematics is genuinely different — and Mathematics is the subject where AI is simultaneously the most useful and the most likely to be confidently wrong.

   Where AI genuinely earns its place in a Maths faculty

   These are the high-value uses. Notice each one maps onto an aspect of the overarching outcome MAO-WM-01 (Working Mathematically) — communicating, understanding and fluency, reasoning, and problem-solving — which is woven through every focus area of the NSW Mathematics K–10 syllabus.

   Use case	What it gives you	Working Mathematically link
   Endless varied practice	Twelve more questions on the same skill at a chosen difficulty, when the textbook ran dry	Understanding & fluency
   Worked examples with clear steps	A scaffolded model solution you can adapt for the board	Communicating; reasoning
   Applied / real-world contexts	The same skill wrapped in a meaningful situation — bushfire data, casual-job pay, mobile-plan comparison	Problem-solving
   Three explanations of one concept	A symbolic, a visual and a real-world version of the same idea — and the misconceptions each pre-empts	Understanding; communicating
   Misconception-probing questions	Items deliberately designed to expose a wrong mental model (e.g. "is 0.45 bigger than 0.5?")	Reasoning
   Rubric → student-friendly criteria	A marking rubric rewritten as "I can…" success criteria	Communicating
   Draft marking guidelines	A first-pass marking guide with method marks for a problem-solving task	Reasoning; communicating

   For a Maths teacher, the single biggest win is variety on demand: differentiated questions, alternative explanations and fresh contexts arrive in seconds, so you spend your time teaching, not retyping.

   Where it fails — the heart of this course

   A large language model predicts plausible text. Mathematics is not a text-prediction task — it is a computation task — and that mismatch is exactly where a general model breaks:

   • It is unreliable at multi-step calculation. It will produce a confident final answer that is simply wrong, and working that looks like maths but contains a dropped sign, a mis-applied rule, or an arithmetic slip — often in the very steps a student would copy.
   • It mislabels outcomes. It will attach an authoritative-looking NSW outcome code to content that belongs to a different outcome — including labelling Path content as Core (you'll meet the real example in Module 2 and the flagship's case study).
   • It makes notation errors — sloppy index notation, missing brackets, x2 where it means x squared, inconsistent units, or a graph description that contradicts the equation.

   The defining discipline of this playbook: treat every AI worked solution like a student's working at the board — never marked correct until you've checked every line. AI drafts the maths; you verify the maths. This is the teacher-in-the-loop principle from the flagship, made specific to a subject where being wrong is invisible unless you check.

   Why a grounded engine matters here

   A general chatbot writes from general text about maths teaching. Lessio is built to ground every draft in the verbatim NESA Mathematics outcomes and the NSW DoE program template, and is explicitly designed not to relabel Core as Path. That sharply reduces the outcome-mislabelling failure — but it does not remove your job. You still verify the arithmetic, the worked steps and the codes, because in Mathematics a confident wrong answer is the most expensive kind.

   Activity — feel the win and the failure (12 min)

   In the Lessio generator, produce a short set of worked examples for a topic you teach (say, solving two-step linear equations, MA4-EQU-C-01, or Pythagoras' theorem, MA4-PYT-C-01). Then do two things: (1) note how fast the variety arrived — questions and a model you'd actually use; and (2) work every solution yourself and mark it as if it were a student's. Write down any line you'd correct. That second step is your job in this subject, and it never goes away.

   Knowledge check

   1Name three high-value uses of AI in a Mathematics classroom, and the one failure that defines how you must use it.

   Show answer

   2Why is a general LLM specifically unreliable at maths in a way it isn't at, say, rewording a passage?

   Show answer

   3How does the teacher-in-the-loop principle translate for a Maths teacher specifically?

   Show answer

   Mark this module complete
2. 2

   Subject-specific prompt craft for Mathematics

   RICE, grounded in NSW Maths: anchoring prompts to real outcomes, the maths-specific pitfalls (calculation, Core-vs-Path codes, notation), the self-check that surfaces them, and a subject prompt library.~50 min

   By the end of this module you'll be able to:

   • Anchor a Mathematics prompt to the correct NSW outcome, stage and Working Mathematically component.
   • Build a maths self-check into every prompt that surfaces calculation, notation and Core-vs-Path errors.
   • Use the subject prompt library to generate usable maths resources and verify them.

   Standards2.1 Content and teaching strategies of the teaching area2.2 Content selection and organisation3.4 Select and use resources

   RICE, made mathematical

   From the flagship you know RICE — Role, Intent, Constraints, Examples. It doesn't change for Mathematics; it gets sharper, because maths prompts have specific failure points to defend against. The single most important addition for this subject is a self-check that catches the maths.

   • R — Role & context. "You are an experienced NSW Stage 5 Mathematics teacher planning for a mixed-ability Year 10 class." De-identified cohort, never a child (flagship hard line).
   • I — Intent + the syllabus anchor. Name the outcome code, stage and the Working Mathematically focus. "Draft six graduated questions on compound interest addressing MA5-FIN-C-02, building from fluency to problem-solving."
   • C — Constraints & format. Difficulty range, how many, notation conventions, units, whether a calculator is assumed, and the format. "Use correct mathematical notation, Australian currency and metric units; show full working; assume a scientific calculator."
   • E — Examples & evaluation — the maths self-check. "Match the style of this sample. Then re-work every answer step by step, state the final answers separately, and flag any step you are not fully confident is correct so I can verify it."

   The Evaluation step is non-negotiable in Maths. "Show your working and flag any step you're unsure of" turns the model's worst weakness — confident wrong working — into a checklist for you.

   See the difference (Mathematics)

   Weak prompt	Strong prompt
   "Give me some trig questions."	"You are a NSW Stage 5 Mathematics teacher. Write eight graduated questions for a mixed-ability Year 10 class on applying trigonometric ratios to right-angled triangles (outcome MA5-TRG-C-01), from fluency to a worded problem-solving item. Provide an answer key, use correct notation and metric units, assume a scientific calculator, and re-work each answer step by step, flagging any step you're unsure of."
   "Make a worksheet on fractions."	"You are a NSW Stage 3 Mathematics teacher. Create a one-page worksheet introducing equivalent fractions for a mixed-ability Year 6 class (representing quantity fractions). Six graduated questions, one worked example, and an 'explain your thinking' task that targets reasoning. Plain English, Australian context, and list anything you couldn't verify."

   The maths-specific pitfalls a prompt must defend against

   1. Calculation is unreliable. Always demand full working and a separate, restated final answer, then check it yourself. Never paste an AI answer key into a class set unchecked.
   2. Outcome codes get mislabelled — Core vs Path is the classic trap. The real example: a generic AI-generated Year 10 program labelled the sine rule, cosine rule and area of a non-right-angled triangle as MA5-TRG-C-02, a Core outcome. It is wrong. That content is a Path outcome — MA5-TRG-P-01 ("applies the sine, cosine and area rules to solve 2-dimensional problems"). MA5-TRG-C-02 is actually bearings and angles of elevation and depression. The code looked real and was completely mislabelled — so verify every code against the official syllabus, and especially check that Path content hasn't been dressed up as Core.
   3. Notation errors. Watch for x² rendered as x2, missing brackets that change order of operations, dropped negative signs, sloppy index laws, and inconsistent units. Constrain notation explicitly and proofread it.
   4. Difficulty drift. "Hard" to a general model may be off-stage. Anchor to the stage and a worked exemplar of the level you mean.

   Grounding beats prompting — use Lessio for the heavy lifting

   You can defend against code-mislabelling with a careful prompt — or you can let Lessio ground the draft in the verbatim NESA outcomes from the start, so the Core-vs-Path distinction is built in rather than re-typed each time. Use NSWEduChat (or your general tool) for quick one-off items with a tight RICE prompt; use Lessio when you want a syllabus-anchored resource, program or assessment for a focus area. Either way, you still check the arithmetic.

   Activity — build, then break, a maths prompt (15 min)

   Take a maths task you'll set this week. Write your instinctive one-liner, then rebuild it with RICE: add the role, the outcome code + Working Mathematically focus, the notation/units/calculator constraints, and the 're-work and flag' self-check. Run it, then deliberately hunt for the error — check every answer and verify the outcome code against the NESA site. Note which RICE element made the biggest difference, then try the same task in Lessio and see how much syllabus-anchoring is already done.

   Knowledge check

   1Which RICE element matters most in a Mathematics prompt, and what exactly do you ask for?

   Show answer

   2An AI program labels the sine and cosine rules as MA5-TRG-C-02 (Core). Why is that wrong, and what's the correct code?

   Show answer

   3Name two maths-specific things you should constrain in a prompt that you wouldn't bother with for, say, an English task.

   Show answer

   Mark this module complete
3. 3

   Planning & resources for Mathematics with Lessio

   Generating a connected Maths set — scope & sequence → program → resources → assessment — grounded in real outcomes, with a worked Stage 5 Trigonometry example and the maths review-before-use checklist.~50 min

   By the end of this module you'll be able to:

   • Generate a connected Mathematics scope & sequence, program and resources grounded in the real NSW outcomes.
   • Run a maths-specific review-before-use check covering codes, sequence, working and notation.
   • Sequence a Stage 5 focus area (e.g. Trigonometry) so Core and Path are correctly placed.

   Standards2.2 Content selection and organisation2.3 Curriculum, assessment and reporting3.4 Select and use resources

   Maths planning is a chain — generate it connected

   Good Mathematics planning is a connected sequence, not a pile of worksheets:

   Scope & sequence (the year/stage) → Program / unit (the weeks) → Resources (worked examples, practice, applied tasks) → Assessment (the measure) — all pointing at the same outcomes, with MAO-WM-01 (Working Mathematically) running through every one. Lessio is built to produce exactly these four artefacts for a Maths focus area, grounded in the verbatim NESA outcomes and the NSW DoE program template — which is where a Maths faculty saves the most time.

   The DoE template Lessio follows is the five-column teaching-and-learning sequence — Outcomes & content · Activities · Evidence of learning · Differentiation & adjustments · Registration & evaluation — around an outcomes list, a needs analysis and a reflection. For Mathematics, the Evidence of learning column is where you make Working Mathematically visible: not just "can compute", but "can reason, communicate and solve problems".

   Worked example — a Stage 5 Trigonometry program

   Suppose you're programming Stage 5 Trigonometry. The Core focus area carries two outcomes:

   • MA5-TRG-C-01 — applies trigonometric ratios to solve right-angled triangle problems (sin, cos, tan; finding sides and angles in right-angled triangles).
   • MA5-TRG-C-02 — applies trigonometry to solve problems, including bearings and angles of elevation and depression.

   The sine rule, cosine rule and area of a non-right-angled triangle are not in this Core sequence — they sit in the Path outcome MA5-TRG-P-01 (for students continuing toward Advanced/Standard pathways). A defensible program places these correctly: right-angled trig and applications as Core for all; the sine/cosine/area rules flagged as Path for the appropriate cohort. Ask Lessio to generate the Stage 5 Trigonometry program, then confirm the Core/Path split is exactly as above before you adopt it — this is the precise spot where an ungrounded tool slips.

   Worked example — a Financial Mathematics resource set

   For Stage 5 Financial mathematics, the outcomes are MA5-FIN-C-01 (simple interest, earning and spending money) and MA5-FIN-C-02 (compound interest and depreciation). A strong Lessio resource set here gives you: graduated practice from a single interest calculation up to a multi-year compound-interest problem; an applied task comparing two real savings or loan scenarios (problem-solving); and matching marking guidelines. Generate it, then work the compound-interest answers yourself — these are exactly the multi-step calculations a general model gets confidently wrong.

   Review-before-use — the Maths edition (90 seconds)

   Before any AI-drafted maths resource or program is real, check:

   1. Codes — every outcome code is current and correctly placed; Core vs Path is right (the MA5-TRG trap); the focus area matches the content.
   2. Working & answers — you have worked every solution and answer key yourself, and they're correct.
   3. Notation & units — notation is clean (indices, brackets, signs), units and currency are consistent, calculator assumptions are stated.
   4. Sequence — concepts build sensibly (e.g. fractions before algebraic fractions; right-angled trig before the sine/cosine rules) for your students.
   5. Working Mathematically — the set isn't all rote fluency; it includes reasoning, communicating and problem-solving (MAO-WM-01).

   If you couldn't defend it in a faculty meeting and mark every answer correct, it isn't ready.

   Activity — generate a connected maths set, then verify (15 min)

   In Lessio, generate a program or resource set for a Maths focus area you teach next term. Then: (1) verify two outcome codes against the official NESA Mathematics syllabus — and if it's a focus area with a Path, confirm the Core/Path split; (2) work two of the harder solutions yourself; and (3) edit one week of the sequence for your class. Your edits are the visible proof of your professional judgement.

   Knowledge check

   1What are the four connected artefacts of a Mathematics plan, and what runs through all of them?

   Show answer

   2When programming Stage 5 Trigonometry, where exactly should you check the Core/Path split?

   Show answer

   3Name three checks in the Maths review-before-use list.

   Show answer

   Mark this module complete
4. 4

   Assessment, feedback & integrity in Mathematics

   Building valid maths assessments with graduated questions and marking guidelines, AI-framed feedback you make specific, and assuring authorship the NESA way in a calculation subject.~45 min

   By the end of this module you'll be able to:

   • Draft a valid Mathematics assessment with graduated questions and marking guidelines, then verify it.
   • Use AI to frame mathematical feedback you make specific, accurate and student-ready.
   • Assure authorship and integrity in Mathematics the way NESA expects — by design, not detection.

   Standards5.1 Assess student learning5.2 Provide feedback to students on their learning7.1 Meet professional ethics and responsibilities

   Drafting a valid maths assessment — fast first pass, verified result

   AI can draft a maths assessment quickly; validity and correctness are yours. A strong workflow:

   1. Anchor to outcomes. Name the focus area and codes (e.g. a Stage 5 Financial mathematics task on MA5-FIN-C-01 and MA5-FIN-C-02).
   2. Ask for graduated questions that span Working Mathematically — fluency items, a reasoning item ("justify…"), and a problem-solving item set in a real context — with a clear mark allocation.
   3. Ask for marking guidelines with method marks, so partial credit rewards correct working, not just the final answer.
   4. Then verify everything. Work every solution. Check the marking guideline's answers and the mark totals. Confirm the task actually measures the named outcome (validity) and is accessible (fair).

   A maths assessment an AI drafted is a proposal. It becomes an assessment when you've worked every solution, fixed the marking guide, and confirmed it measures the outcome.

   Feedback — frame from AI, mathematics from you

   AI is good at structuring feedback; it does not know your student or always know the maths. Use it to draft a frame against your criteria — a strength, a priority, a next step — then make it mathematically precise: name the specific error (a sign error in line 3, not "check your working"), the misconception behind it, and the next step. Never paste an identifiable student's response into a general tool (flagship hard line) — de-identify, or use your school's approved secured environment. And check any worked solution the AI offers the student: a feedback comment that "corrects" with wrong maths is worse than none.

   Convert a rubric to student-friendly success criteria

   Turn your marking rubric into "I can…" statements students understand — "I can choose the right trig ratio for a right-angled triangle and show my working" — keeping them faithful to the standard. This is a genuinely good, low-risk AI use (it's language, not calculation) — but still confirm it matches your rubric.

   Integrity in a calculation subject — NESA's way

   Schools decide whether AI is permitted for a given task, and you uphold HSC/RoSA integrity. In Mathematics the integrity questions are specific:

   • AI can solve many routine problems, so a take-home set of standard questions is weakly authored. Design for visible process: in-class checkpoints, "show and justify your working", multi-step problems in novel contexts, and short oral/viva questions ("talk me through line 4").
   • Assure authorship by design, not by 'AI detectors' (which NESA does not endorse and which are unreliable). Watch the working: a correct answer with absent or incoherent working is the maths-classroom signal to have a conversation — model that for students rather than accusing.
   • Teach students to use AI with integrity — as a tutor that explains a method they then practise and check, not a solver that hands over answers. Be open that you use AI to draft resources you verify; that honesty sets the classroom norm and builds the AI literacy the Australian Framework asks for.

   Activity — build and stress-test a maths assessment (12 min)

   In Lessio, generate a short assessment with marking guidelines for a focus area you teach. Then stress-test it: (1) work every question and check the marking guide and totals; (2) judge whether it validly measures the named outcome and is accessible; and (3) ask "could a student get full marks using AI without understanding?" — and redesign one item to make the process visible. Note one place your judgement overrode the draft.

   Knowledge check

   1Before an AI-drafted maths assessment goes to students, what three things must you do?

   Show answer

   2How does NESA expect you to assure authorship of maths work — and what's the maths-specific tell?

   Show answer

   3Why is rewriting a rubric into 'I can…' statements a lower-risk AI use than generating an answer key?

   Show answer

   Mark this module complete
5. 5

   Capstone — build a real Mathematics resource and critique it

   Build a connected Maths program + resource + assessment with Lessio, verify every answer and every code, self-assess against the Ethical-Use Checklist, reflect, and log it as PD.~60 min

   By the end of this module you'll be able to:

   • Build a connected Mathematics program, resource and assessment with Lessio, end to end.
   • Verify every worked solution and every outcome code (Core vs Path), and fix what's wrong.
   • Self-assess against the Lessio Ethical-Use Checklist, reflect, and record the hours as PD.

   Standards2.3 Curriculum, assessment and reporting6.2 Engage in professional learning7.1 Meet professional ethics and responsibilities

   The task — a real, connected, defensible maths set

   Choose a Mathematics focus area you'll teach next term — Stage 4, Stage 5 (Core and/or Path), or Stage 6 (Standard, Advanced, Extension 1 & 2). Using Lessio, build and then critique:

   1. A program / unit of work for the focus area, with outcomes and a teaching-and-learning sequence.
   2. A resource — a set of graduated practice and worked examples (and, ideally, one applied/real-world task).
   3. An assessment with marking guidelines for the unit.

   Then apply your professional judgement as a mathematician:

   • Verify every outcome code against the official NESA Mathematics syllabus — and where the focus area has a Path, confirm the Core/Path split is correct (remember the MA5-TRG trap).
   • Work every solution and answer key yourself — every worked example, every practice answer, every marking-guide solution. Fix any error.
   • Check notation, units and sequence, and make the assessment valid (it measures the outcome) and fair (accessible, with reasonable adjustments under the Disability Standards for Education 2005).
   • Confirm Working Mathematically is present — not all rote fluency, but reasoning, communicating and problem-solving too (MAO-WM-01).

   What good looks like

   A connected, syllabus-accurate, mathematically correct set you'd actually use — drafted by AI, unmistakably shaped, checked and owned by you. The errors you caught and the edits you made are the evidence of your professional judgement, and exactly what teacher-in-the-loop means in a subject where a wrong answer is invisible until someone checks.

   Self-assessment — the Lessio Ethical-Use Checklist

   Run your capstone against all five items on this page. The two that define this playbook — every worked solution and answer checked, and every outcome code verified (Core vs Path) — must be honestly tickable. If one isn't, fix the artefact; that is the learning.

   Reflection — write a short response

   • What did AI genuinely save you time on (variety? a first draft? marking guidelines?) — and what maths did you have to fix?
   • Which outcome code(s) did you verify, and did any need correcting? Was Core/Path ever at risk?
   • Where did working a solution by hand catch an error you'd otherwise have shipped?
   • One rule you'll keep for using AI in Mathematics responsibly from now on.

   Log it as professional learning

   This capstone is your assessment: a complete, critiqued maths set plus your ethical-use reflection — keep it as evidence of practice. Since NESA removed the Accredited/Elective PD distinction in 2024, Standards-relevant learning like this counts toward your 100 maintenance hours — record it in your eTAMS PD log against the Standards it addresses (especially 2, 3, 5, 6 and 7). Your faculty can also run this playbook as part of its professional-learning plan or a staff development day.

   Activity — build, verify, reflect, log (20 min)

   Produce your connected Mathematics set in Lessio. Verify the codes (Core/Path included), work every solution, edit for your cohort, then self-assess against the Ethical-Use Checklist and write your reflection. The reflection plus the critiqued artefact is your eTAMS evidence.

   Knowledge check

   1What turns an AI-generated Mathematics unit into defensible professional work?

   Show answer

   2Which two checklist items are the non-negotiable heart of this maths playbook?

   Show answer

   3How can this playbook count toward your NESA professional-development hours?

   Show answer

   Mark this module complete

Take-away prompt library

Ready, RICE-shaped prompts for common NSW teaching jobs (Module 3). De-identified — copy one, swap in your details, and use it today.

You are an experienced NSW [stage] Mathematics teacher. Write one fully worked example for [topic / outcome code], showing each step of the working with brief reasoning, correct mathematical notation and consistent units. State the final answer separately. Then re-work the solution a second way (or check it) and flag any step you are not fully confident is correct. Verify the maths yourself before using it.

You are a NSW [stage] Mathematics teacher. Write eight practice questions on [topic / outcome code] for a mixed-ability [year] class, graduated from fluency to a worded problem-solving item, with a separate answer key. Use correct notation, Australian context and metric units, and state whether a calculator is assumed. Re-work each answer step by step and flag any step you're unsure of so I can verify it. Verify the maths yourself before using it.

You are a NSW [stage] Mathematics teacher. Write one real-world, multi-step applied problem on [topic / outcome code] for a [de-identified class], set in an authentic Australian context (e.g. casual-job pay, a mobile-plan or loan comparison, sport or bushfire data). Include the data students need, a clear question, and full worked solution with the final answer stated separately. Flag any step you're unsure of. Verify the maths yourself before using it.

You are a NSW [stage] Mathematics teacher. Explain [concept] in three different ways for a [de-identified class]: a plain-English/symbolic explanation, a visual or concrete representation, and a real-world analogy (Australian context). Then name one common misconception students hold about [concept] and how to pre-empt it. Keep notation correct and flag anything I should double-check. Verify the maths yourself before using it.

You are a NSW [stage] Mathematics teacher. Write six short questions on [topic / outcome code] for a [de-identified class] that are deliberately designed to surface common misconceptions (e.g. for decimals, that 0.45 > 0.5 because '45 > 5'). For each, state the misconception it targets and the correct answer with brief reasoning. Flag any item where the answer could be ambiguous. Verify the maths yourself before using it.

You are a NSW [stage] Mathematics teacher. For the problem-solving task below [paste], write marking guidelines aligned to [outcome code]: a sample full solution with the final answer stated separately, and a mark allocation that awards method marks for correct working as well as the final answer (partial credit). Keep notation and units correct. Then re-check every answer and mark total, and flag anything you're unsure of. Verify the maths yourself before using it.

Standards-relevant professional learning, mapped to the APST · content verified against national and NSW frameworks, June 2026 · self-log the hours in eTAMS.