Interactive learning · M3-T3-MLC3

AI-Supported Inclusion for Learners with Special Educational Needs

The teacher adapts, AI assists

Author: Zsolt LengyeliTStudy Hungary Training and Research Centre
Reviewer: Mária HartyányiiTStudy Hungary Training and Research Centre

One task branches into three versions offering different levels of support for different learners. — One learning goal, several routes: task variants with different levels of support.

Learning objective

Adapting materials and activities for learners with different support needs is one of the most time-consuming tasks in teaching. AI can help by simplifying texts, creating differentiated activities, providing step-by-step guidance, supporting reading and writing, and improving accessibility.

This course shows how to use AI as an assistant for adaptation. Pedagogical decisions and checking remain with the teacher and the support team.

By the end of this course, you will be able to

describe support needs functionally, without names or diagnoses;
prepare simplified texts, differentiated activities and step-by-step guidance for the same learning goal;
check adapted materials for accuracy, appropriateness and usefulness;
respect privacy and professional boundaries: do not use AI to diagnose, label or automatically assess learners.

The teacher adapts, AI assists

Meaningful participation for every learner

Inclusive education means that every learner can participate meaningfully and make progress. A class may include different reading, attention, sensory, physical and comprehension needs.

Learners with dyslexia, attention difficulties, autism, visual, hearing or physical impairments, or lower reading literacy may need different routes to the same professional competence.

AI drafts; the teacher decides

Preparing several versions of a text, a task or an instruction used to take hours. Generative AI can quickly produce drafts of these adaptations.

The teacher and the support team decide what the learner needs, which goal must stay the same, and whether the adapted material is correct, respectful and useful.

Keep the learning goal the same; vary the route and the support.

Simplifying texts

A dense technical text can become easier to understand while keeping essential content and technical accuracy.

Plain language, complete content

A language model can rewrite a text using short sentences, one idea per sentence, active voice and clear structure.

Ask it to explain technical terms and preserve safety rules, numbers and conditions. Check these carefully after simplification.

With AI support, a dense text becomes a structured version with short sections. — A dense text becomes a structured, simplified version – with the key content kept.

01

Short sentences

One idea per sentence, expressed clearly.

02

Clear structure

Headings, short sections and numbered steps where appropriate.

03

Explained terminology

Keep technical terms and add plain-language explanations.

Example prompt for simplification

The prompt describes what the text should be like. It can be written without a learner’s name, diagnosis or other personal data.

The prompt in the source

Rewrite the attached text about safe use of an angle grinder for 17-year-old vocational learners who need simpler language.

Use short sentences, one idea per sentence and numbered steps. Keep all safety rules and technical terms, and explain each term in brackets the first time it appears.

Do not use a childish tone. Add a 3-point summary at the end.

What does the prompt make clear?

The age group and learning context.
The language and structure of the steps required.
The technical and safety content to preserve.
A respectful, age-appropriate tone.

There is no name, diagnosis or personal data; the prompt describes the required features of the text functionally.

Differentiated activities

Differentiation offers different routes to the same learning goal. The amount of support changes; the professional goal stays the same.

First, define the learning goal

Write the professional competence you want to develop in one sentence. Then ask AI for task variants with different levels of support.

AI can create a version with worked examples and hints, a standard version, and an extension task. The teacher checks that each leads to the same goal.

A shared task branches into three versions with different levels of support. — One learning goal, several routes: task variants with different levels of support.

01

More support

A worked example, helpful hints and intermediate checkpoints.

02

Standard version

A task leading to the shared professional goal with the usual support.

03

Extension

A task linked to the same goal that offers more independent work and further thinking.

Step-by-step guidance

Make a complex task manageable

Many learners, especially those with attention difficulties or autism, benefit from breaking complex tasks into small, clear steps.

AI can turn a workshop task into a checklist, a visual schedule or a sequence of short instructions. Goblin.tools offers a tool for splitting tasks into smaller steps.

A staircase of small steps and ticks shows progress through a task. — Small, clear steps make progress visible and reduce overload.

01

Checklist

Short, followable steps that can be marked as complete.

02

Visual schedule

Cues and pictures make the sequence of activities visible.

03

Short instructions

One manageable action at a time, with a clear checkpoint.

Reading, writing and accessibility

AI can support both access to text and ways of responding. Choose the right option based on the learner’s needs and the learning goal.

01

Reading

Audio versions, reading modes, glossaries with pictures, and questions that guide attention while reading.

02

Writing

Sentence starters, word banks, structure templates for reports, and dictation instead of handwriting.

03

Accessibility

Clear headings, sufficient contrast, alternative text for images and consistent layout.

AI can suggest improvements

Ask for a document to be reviewed for structure, image alternative text or clarity. The teacher checks the suggestions.

Test the version with the learner too: support is useful when it actually makes participation easier in that situation.

Which support serves the goal?

If the goal is to express a technical idea, sentence starters or dictation may support the response. Decide on the adaptation based on the goal and the learner’s feedback.

Coordinate reading and writing support with the support team.

Dignity: support without labelling

Describe the need

The learner’s dignity and privacy come first. Describe the support needed, for example “needs shorter instructions and visual cues”.

Offer adapted materials to everyone who may benefit. Do not separate or single out learners who need support.

A shield protecting a person and a crossed-out label represent dignity and the rejection of labelling. — Support without labelling: learners’ dignity and privacy come first.

Functional descriptions

“Needs short instructions and visual cues.”

“Reads slowly but understands well when listening.”

These descriptions help you plan the task and the support.

Respectful tone

Simpler language should suit the learners’ age. Avoid a childish tone and wording that suggests lower expectations.

Ask the learner which version helps them.

Boundaries: diagnosis and assessment

01

No diagnosis

Do not use AI to decide or suggest whether a learner has a condition.

02

No labelling

Describe the support needed. Do not assign learners to categories invented by AI.

03

No automatic assessment

Decisions about grades, progression and support remain human.

Professional roles remain in place

Identifying special educational needs is the task of qualified professionals. The teacher and the support team plan adaptations.

AI helps prepare versions of learning materials; pedagogical decisions about learners require human expertise and cooperation.

Drafting a task or deciding about a learner?

Drafting task variants for the same goal is different from automatically assigning a grade, level or learning path based on a learner’s performance.

In this course, AI is used for drafting tasks; people decide and check the results.

The EU AI Act’s rules on prohibited practices apply from 2 February 2025. AI emotion recognition based on biometric data is prohibited in educational institutions, except for medical or safety purposes.

Systems intended to evaluate learning outcomes and use them to steer learning are listed among the educational uses in Annex III. Classification depends on the intended purpose and the conditions in the Regulation.

According to European Commission guidance checked on 14 October 2026, the rules for high-risk systems under Annex III apply from 2 December 2027. Commission guidance · Annex III.

Privacy and professional cooperation

Work without personal data

Information about special educational needs may reveal health data. Health data is a special category of personal data under the GDPR.

Never enter names, diagnoses, expert opinions or individual development plans into general AI tools. Describe support needs functionally and anonymously.

Work with the support team

Align adaptations with special education teachers, the learner’s individual plan and parents.

Above all, listen to the learner. Find suitable support together and adjust the material based on their feedback.

A situation to think through

What would you provide to prepare a task variant?

The goal is a task with shorter instructions and visual cues. Choose how you would phrase the prompt.

Choose a prompt description to see the explanation.

Checking the result

Check three things after every adaptation. This is the teacher’s responsibility once the AI response is ready.

01

Accuracy

Did simplification change or remove essential content, especially a safety rule?

02

Appropriateness

Is the tone respectful and age-appropriate?

03

Usefulness

Does it work for the learner? Ask for feedback and adjust if needed.

Tick what you have checked in your own material!

AccuracyThe essential technical content and all safety rules are complete and correct.AppropriatenessThe tone is respectful and suitable for the learners’ age.UsefulnessI have asked the learner for feedback and adjusted the material if needed.

Takeaways for teachers

Five ideas for your next adaptation

AI can quickly draft simplified texts, differentiated tasks and step-by-step guidance; the teacher decides and checks.
Keep the learning goal the same; vary the route and the support.
Describe needs functionally and anonymously; never enter personal or health data.
Do not use AI to diagnose, label or automatically assess learners.
Work together with the support team and listen to the learner.

The teacher decides and checks; the learner gives feedback.

Teacher exercise: three routes to the same goal

35 minutes · Individual work

Practise AI-supported adaptation of a real task using a functional description of needs, an unchanged learning goal and no personal data.

Exercise steps

Choose a task from your subject and write down its learning goal in one sentence.
Describe three fictional support profiles without names or diagnoses, for example “needs short instructions and visual cues” or “reads slowly but understands well when listening”.
Ask AI for a task version for each profile with the same goal, a simplified version of the instructions, and a step-by-step checklist.
Check each version: is the goal unchanged, is the content correct and complete, are all safety elements preserved, and is the tone respectful and age-appropriate? Correct what is needed.

Tools and lessons learnt

ChatGPT, Claude or Gemini, preferably with a school account.

AI saves time in drafting adaptations; pedagogical checking remains essential.
Functional descriptions of needs work well; personal data is not needed.
Differentiation changes the route, not the goal.

Download detailed instructions for both exercisesM3-T3-MLC3-EXC-EN.pdf

Classroom exercise: make it clearer!

40 minutes · Group work, then peer testing

Groups turn a complex workshop instruction into a clear step-by-step checklist with AI support. The aim is to practise accessible communication and experience that clarity helps everyone.

Steps for group work

Each group receives a long, complex instruction for a workshop task.
With AI support, create a checklist of short steps. Add visual cues, symbols or drawings, and highlight safety steps.
Another group tries to follow the checklist in a dry run, without tools. They note where it was unclear.
Revise the checklists, then discuss which features made instructions easiest to follow.

Tools and lessons learnt

A school-approved AI tool or teacher-led use on the projector; printed instructions, paper or tablets.

Clear, step-by-step instructions help everyone.
AI can structure instructions, but people check whether they work in practice.
Safety steps must stay complete and visible in every simplified version.

Notes for teachers

Present the activity as a general communication task. Never single out learners who need support or enter information about individual learners into AI tools.

Peer testing takes place without tools.

Check what you have learnt!

Each question has one correct answer. Answer all seven questions, then view your result and the explanations.

1What should stay the same across differentiated task variants? Every sentence and the entire format of the task. The shared learning goal and essential technical content. The amount of help provided to learners.

2Which prompt describes a support need functionally? “Short instructions and visual cues are needed.” “Analyse this expert opinion and invent a diagnosis.” “Choose a level based on the learner’s name and health data.”

3What should be preserved when simplifying a workshop text? Only the length of the original text. Only easy words, without technical terms. The technical content, all safety rules and explained technical terms.

4What do we use AI for in this adaptation work? Diagnosing a learner’s condition. Drafting learning materials and task variants, with teacher checking. Automatically deciding grades and support arrangements.

5Which three criteria should be used to check every adaptation? Accuracy, appropriateness and usefulness. Text length, visual appeal and the speed of the AI response. The tool’s popularity, price and brand.

6Who should be offered adapted materials? Only learners who are publicly singled out. Only learners whose diagnoses have been uploaded to AI. Everyone who may benefit, without labelling.

7How should groups test the checklist in the “Make it clearer!” exercise? Immediately with tools, without checking safety steps. In a dry run without tools, noting unclear points. Only ask AI whether it is guaranteed to work.

References and further reading

References in the source document

UNESCO (2023): Guidance for generative AI in education and research. Paris: UNESCO.
European Parliament and Council Regulation (EU) 2024/1689, Article 5(1)(f) and Annex III.
Inclusion Europe (2009): Information for all – European standards for making information easy to read and understand.
CAST (2024): Universal Design for Learning Guidelines version 3.0.

Useful links

The legal dates and the exceptions to the prohibition on emotion recognition were clarified using European Commission guidance. Checked on 14 October 2026.

According to the source, images without a specified source were generated by Claude for Mac. External links open in a new tab and require an internet connection.

Credits and copyright

Author: Zsolt LengyeliTStudy Hungary Training and Research Centre
Reviewer: Mária HartyányiiTStudy Hungary Training and Research Centre
Project: 2025-1-LT01-KA220-VET-000362456
Learning material: M3-T3-MLC3

Attachments

Source material · PDFM3-T3-MLC3-TEXT-EN.pdf Practical exercises · PDFM3-T3-MLC3-EXC-EN.pdf

This work may be reused and adapted under the terms of the Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA 4.0).

Funded by the European Union. Views and opinions expressed are those of the author(s) only and do not necessarily reflect those of the European Union or the National Agency.