Spring Assessment: AI Support or Malpractice?

Clear boundaries for revision, coursework and take-home tasks

A teacher explaining responsible AI use during spring assessment season

Spring assessment season always sharpens existing questions about fairness, independence and support. AI has not created those questions, but it has made them more urgent. A pupil using a chatbot to build a revision timetable is doing something very different from a pupil submitting AI-written coursework, yet both may describe it as “just getting help”. That is why schools need clearer language this term, not louder panic. If your team is refining policy, it may help to pair this guide with assessment design thinking and a broader review of current assessment integrity trends.

Why boundaries matter

In spring, stakes rise. SATs preparation intensifies, nationella prov revision becomes more focused, coursework deadlines close in, and take-home assignments begin to feel less like learning and more like performance. In that atmosphere, families often want to help, pupils want reassurance, and teachers want consistency. Without clear boundaries, AI use drifts from support into substitution before anyone notices.

The central problem is simple. AI can either strengthen a pupil’s own thinking or replace it. If it helps a learner organise revision, clarify misconceptions or practise recall, it may be entirely reasonable. If it generates answers, drafts paragraphs, solves unseen problems or invents evidence, it undermines the validity of the work. The line is not whether AI was used. The line is whether the submitted performance still belongs to the pupil.

The core test

A useful question for any teacher, pupil or parent is this: is the AI supporting thinking, coaching thinking or substituting for thinking? Support usually means helping a learner prepare. Coaching means prompting the learner to improve their own work while remaining in control. Substitution happens when the system produces material that the learner presents as their own.

That distinction matters across contexts. A Year 6 pupil using AI to generate ten arithmetic practice questions is in a very different position from a pupil copying an AI explanation into a take-home task. A Swedish lower-secondary pupil asking for a checklist of likely errors in lab write-ups is doing something different from uploading a draft and accepting a rewritten version. In both cases, the product may look polished. The ethical difference lies in authorship and independence.

A red-amber-green model

Schools often need something more practical than abstract principles, so a red-amber-green model works well.

Green use supports revision and understanding. It includes making flashcards from class notes, creating a study timetable, generating low-stakes quiz questions, or asking for an explanation of a concept in simpler language. The pupil still does the learning and still produces the final work independently. Articles on revision workflows and holiday study planning can help staff explain these safe uses.

Amber use needs caution, teacher judgement and, often, disclosure. This includes asking AI to suggest an essay structure, identify weak reasoning in a paragraph, or provide model questions in the style of an exam board or national test. These uses can be legitimate, but only if the pupil remains the author and the task rules permit that support. Amber cases are where “ask first” expectations matter most.

Red use crosses the line. This includes hidden drafting, answer generation, solving assessed problems, translating whole passages for submission, writing code for marked tasks, fabricating quotes, references or data, and using AI outputs that are then lightly edited and handed in as original work. Red use is malpractice because the system has replaced the learner’s assessed performance.

SATs preparation

For SATs preparation, the safest rule is that AI may help pupils practise, but not perform. A pupil can use AI to create spelling quizzes, arithmetic drills or reading retrieval questions based on a teacher-approved text. A family might ask for a simple revision schedule or a set of mental maths prompts for the week. Those are green uses because they build readiness.

What pupils cannot do is use AI to answer comprehension tasks that they are then expected to complete themselves, generate written responses for grammar practice and copy them, or rely on chatbot explanations in place of actually attempting the work. In SATs season, younger pupils can struggle to see this difference, so adults need to make it concrete: “AI can make practice for you; it cannot do the practice for you.”

Nationella prov realities

In Swedish classrooms, the same principle applies, though local practice and subject expectations may differ. During nationella prov revision, AI can help pupils review vocabulary, create self-quizzing prompts, or summarise teacher notes into manageable revision points. It can also support multilingual understanding when a pupil needs a concept explained more clearly before independent practice.

The problem begins when AI becomes a hidden co-author. If a pupil uses it to draft a response in Swedish, reshape a literary analysis, or generate lab conclusions that are then submitted, the work no longer reflects the pupil’s own attainment. For schools working through these issues in Sweden, practical guidance on Lgr22 and AI tools can help frame local conversations.

Coursework and take-home tasks

Coursework and take-home assignments are where malpractice most often begins because the boundary feels less visible. Pupils tell themselves they are only “getting started” or “improving the wording”. Families may think they are helping by encouraging a chatbot to tidy expression. Yet once AI starts generating the structure, examples, argument or phrasing of the assessed work, authorship becomes blurred.

A practical rule is this: if the task is meant to show what the pupil can do, AI may support planning and reflection but must not produce the assessed content. Brainstorming possible angles for a history essay may be amber. Asking AI to draft the introduction is red. Receiving feedback on whether a paragraph has a clear topic sentence may be amber. Asking for a rewritten paragraph to “make it better” is red.

Ready to Revolutionise Your Teaching Experience?

Discover the power of Automated Education by joining out community of educators who are reclaiming their time whilst enriching their classrooms. With our intuitive platform, you can automate administrative tasks, personalise student learning, and engage with your class like never before.

Don’t let administrative tasks overshadow your passion for teaching. Sign up today and transform your educational environment with Automated Education.

🎓 Register for FREE!

Subject examples

English

Green use includes vocabulary revision, retrieval quizzes on set texts, and asking for practice questions about themes. Amber use includes requesting a checklist for analysing language or comparing two characters. Red use includes generating essay paragraphs, interpretations, quotation analysis or creative writing for submission.

Maths

Green use includes extra practice questions, worked examples on similar problems, and explanations of a method already taught. Amber use includes hints on where an error may lie in a completed solution. Red use includes solving assigned homework questions, generating final answers, or walking through a take-home assessment step by step so that the pupil can copy it.

Science

Green use includes revision cards, definitions, and self-test questions on processes such as photosynthesis or forces. Amber use includes feedback on whether a method description includes variables and controls. Red use includes writing conclusions, inventing results, fabricating sources, or producing evaluation sections for assessed practical work. Staff teaching older pupils may also find it useful to reflect on research and evidence evaluation in science.

Humanities

Green use includes timelines, key-term recall and practice source questions. Amber use includes checking whether an argument is balanced or whether a paragraph answers the title. Red use includes generating essays, source analyses, case-study write-ups or references that the pupil has not genuinely used.

Languages

Green use includes vocabulary practice, verb drills and pronunciation support. Amber use includes checking whether a self-written sentence is grammatically secure. Red use includes translating whole assessed responses, generating extended writing, or rewriting a pupil’s work into more advanced language for submission.

Computing

Computing deserves special care because AI can produce plausible code quickly. Green use includes quizzing on syntax, debugging concepts and explaining what a line of code does. Amber use includes asking for hints about why a program fails. Red use includes generating code for assessed tasks, completing coursework logic, or pasting in a solution the pupil cannot explain. Departments may want to connect this with current thinking on computing and code verification.

Acceptable help

Teachers can reduce confusion by explicitly naming acceptable help. In most schools, pupils can safely use AI for revision planning, low-stakes self-testing, simplifying explanations, identifying topics to revisit, and receiving general feedback that does not rewrite the work. Families can also be told that asking AI to produce practice materials is usually fine, while asking it to complete school tasks is not.

This is where disclosure helps. If a pupil used AI to create revision questions, there is no need for secrecy. If they would feel the need to hide the interaction, that is often a sign that the use is drifting into red territory. Clear expectations after holidays or before exam blocks are especially valuable, and some schools may want to revisit reset routines for AI boundaries.

What crosses the line

The clearest red flags are hidden drafting, answer generation and fabricated evidence. If AI writes the first draft, the “improved” draft, or the final version, that is substitution. If it supplies the answer to a take-home task, that is substitution. If it invents a quotation, source, data point, experiment result or bibliography entry, that is fabrication. None of these can be defended as harmless support.

Teachers should also watch for softer forms of substitution. A pupil may ask AI for “just a model”, then copy its structure, examples and phrasing with minor edits. A parent may paste in the assignment and ask for “guidance”, receiving a near-complete response. Those are still red cases because the assessed thinking has been outsourced.

Explain without panic

The best messages to pupils and families are calm, short and repetitive. Say that AI is like a revision coach, not a stand-in. Say that it can help pupils prepare, but it cannot do their assessed thinking. Say that, when in doubt, they should ask first and disclose use. This avoids the unhelpful impression that every AI interaction is cheating, while still protecting integrity.

Departments can use quick scripts such as: “Permitted means revision support and self-testing.” “Prohibited means AI-generated answers, drafts or evidence.” “Ask first means any use that shapes assessed work.” If schools need stronger wording, policy clause packs can speed up the process.

A one-page routine

Before deadlines hit, schools can issue a one-page disclosure routine. Keep it simple. Ask pupils to state whether AI was used, what it was used for, and whether any output appears in the submitted work. Add three examples of permitted use, three prohibited uses, and one line that says, “If you are unsure, ask before submitting.” That single page can prevent many avoidable problems.

The goal is not to catch pupils out. It is to protect fairness, reduce anxiety and preserve trust in what assessment evidence actually shows. When boundaries are clear, most pupils will stay on the right side of them.

May your assessment season be fair, calm and clearly understood.
The Automated Education Team

Table of Contents

Categories

Assessment

Tags

Ethics Safety Strategies

Latest

Alternative Languages