On November 30, 2022, OpenAI released its artificial intelligence chatbot ChatGPT to the public. It was an immediate success and gained one million users in the first five days. At the time I was working as an ELT materials writer, and I put this new toy to use straight away. No more staring at a blank page for hours. No more hours spent crafting the perfect audio transcript. Just type a prompt, watch the output appear, and there you have it. When AI tools for educators first arrived, the promise was very simple: they would save us valuable time that could be used for other things. In reality what actually happened is we had to spend a great deal of time fixing these AI outputs.
As an experienced trainer and material writer, I was able to modify what I was getting out of the AI. I already knew what worked well in the classroom. I could spot a weak output at a glance and make the necessary changes. One could almost say I have internalised some good standards for level, for pedagogical structure, and for competencies. When AI gives me a first draft, I don’t outsource my pedagogical judgment.
But what if you’re early in your career, or you’re writing materials for the first time? How can you apply judgement to something that you’ve never learned to judge? The output looks great on the surface. It has headings, a numbering system, an answer key. It uses reasonable target vocabulary that you feel are probably good for the level. On the surface everything can look perfect, however there might be issues that are not immediately visible. For example: do the comprehension questions proceed from gist to more specific information gaps? Do the distractors make sense, or can the learners eliminate them without even reading the text? Is the vocabulary level really correct for the target group, or are some advanced terms going to derail the whole task?
If you’ve never written a worksheet from scratch, built an answer key, or assessed the level of an audio, you may have no internalised sense of what the right level of difficulty feels like on the page. You are not able to assess your worksheet because you have not yet developed a standard to do that. There is also very little published material available to help you develop these skills.
Then there is the risk of cognitive offloading – the idea that you will lose your judgement if you delegate everything to AI. This is a serious risk, but what if you have not even developed that sense of judgement? Since we have already observed that judgement is gained by writing your own worksheets, you will not gain that experience at all if all your work is delegated. In the end, there will be a whole cohort of trainers and material writers who do not have these skills, because they never engaged in the creative process themselves. Either material writers will:
Lose the skills they have acquired
Because of cognitive offloading onto AI tools.
Never gain these skills at all
Because of excessive task delegation from the outset.
AI can reduce the labour involved and increase your productivity, but it can affect the development of the material writer in a very negative way if it is allowed to. We get teachers who can produce materials faster than ever, but have no idea whether those materials are any good. The solution is to be more disciplined about using AI in the first place.
Workflows you can automate
There are different trainer workflows that can be automated. Some of these assist the materials writer but do not involve delegating the entire creative process – for example, stages 1, 2 and 4.
Ideation tasks
Data analysis, research, ideation/brainstorming, planning.
Planning content
Designing a syllabus or a lesson.
Creating content
Vocabulary lists, texts and dialogues, questions, exercises, role cards, scenarios, trainer notes, and answer keys.
Refining content
Refining, rewriting, editing, or changing the writing style.
Three tiers of material production
There are three tiers of material production: assisting, delegating and outsourcing. These involve different levels of AI use.
Materials written by the content writer
Generative AI is used during the ideation stages, including brainstorming, lesson planning, and research.
Generative AI used in the writing process
Prompts are designed to create complete activities. The material is then edited and reviewed by the writer, fact-checked, and reviewed for bias.
Generative AI does everything
Prompts are used to design courses, choose target vocabulary/grammar, create complete activities, and edit the final result.
To minimise the negative effects, it is wise to limit AI use to mainly assisting, and if you have to delegate writing tasks, do at least 20% of your workflow manually. For example, write one complete lesson from scratch per week. This is going to help you develop your pedagogical standards to an acceptable level. Compare your original work to AI-generated versions of the same task, so that you make your own criteria visible to yourself. Avoid outsourcing everything to AI, as that will lead to a loss of your critical judgement.
Before you generate anything, make sure you are clear about the purpose of the material, what output you expect, and what would cause you to reject it. These will become key criteria to assess the output. Keep a record of all materials you wrote entirely on your own. List all the defects you caught in AI output and why they mattered. Make a list of tasks you won’t delegate and the reasoning behind that. If you’re early in your career and a novice materials writer, find experienced colleagues who know how to write their own materials without AI. Ask them if they are willing to look at some AI-generated work and check if the criteria hold true. You’re not looking for someone to tell you whether AI is good or bad. You’re looking for someone who can help you build and test your own standards.
The debate about AI in education keeps asking: will AI make teachers worse at their jobs? That’s the wrong question. The right question is: under what conditions will teachers, especially novices, form the skills that make AI use professionally safe? Teacher education programmes need to design opportunities to develop standards explicitly, not hope they emerge by accident.
Use cases for generative AI in ELT
The use cases for generative AI in ELT are very extensive. The table below will give you an idea of how you can use AI to transform your everyday work.
| Use case | Example applications |
|---|---|
| Material creation | |
| Generating ideas for content | Brainstorming lesson topics, themes, and task types |
| Finding multiple perspectives | Exploring different cultural or professional viewpoints on a topic |
| Defining learner personas | Creating detailed profiles of target learners for materials design |
| Syllabus creation | Designing course outlines, learning objectives, and progression |
| Lesson planning | Structuring lessons, timing activities, sequencing tasks |
| Creating classroom and digital learning content | Writing texts, dialogues, exercises, interactive activities |
| Writing task instructions | Drafting clear, level-appropriate instructions for activities |
| Iterating, editing and improving content | Revising drafts, refining language, improving clarity |
| Analysing writing styles | Identifying features of genres, registers, or author styles |
| Changing writing styles | Adapting texts to different levels, tones, or genres |
| Creating titles and headlines | Writing engaging titles for lessons, units, or materials |
| Creating or selecting images for lessons | Generating image prompts or selecting appropriate visuals |
| Summarising complex information | Condensing articles, reports, or research for learners |
| Getting expert advice | Consulting AI as a subject-matter expert on content topics |
| Creating email templates | Drafting communications with learners, colleagues, or stakeholders |
| Creating reference material | Producing grammar guides, vocabulary lists, glossaries |
| Generating guidelines and checklists | Creating rubrics, self-assessment tools, quality criteria |
| Creating FAQs | Anticipating learner questions and drafting answers |
| Generating grading rubrics | Designing assessment criteria for tasks and projects |
| Generating feedback on learner output | Providing comments on writing, speaking, or task performance |
| Analysing and interpreting data | Reviewing learner performance data, survey results, or usage metrics |
| Writing answer keys | Producing correct answers and explanations for exercises |
| Creating role cards and scenarios | Designing role-play prompts, case studies, simulations |
| Adapting authentic texts | Simplifying or modifying real-world texts for specific levels |
| Creating audio scripts | Writing dialogues or monologues for listening activities |
| Designing assessment tasks | Writing test items, quizzes, or performance tasks |
| Marketing | |
| Market analysis | Researching trends, competitor offerings, learner needs |
| SEO optimisation | Improving visibility of course pages, blogs, or resources |
| Creating user personas | Defining target audiences for marketing campaigns |
| Analysing sales data | Interpreting enrollment trends, conversion rates, revenue |
| Identifying target groups | Segmenting audiences by level, profession, or goals |
| Generating keywords | Finding search terms for content and advertising |
| Creating meta descriptions | Writing summaries for web pages and course listings |
| Creating ideas for posts | Brainstorming social media and blog content |
| Generating LinkedIn posts | Drafting professional updates and thought leadership content |
| Creating marketing copy | Writing promotional text for courses, programmes, or services |
| Writing white papers | Producing in-depth reports on methodologies or outcomes |
| Creating email campaigns | Designing nurture sequences and promotional emails |
| Writing case studies | Documenting learner success stories and programme impact |
| Creating landing page content | Writing persuasive copy for course sign-up pages |
| Training | |
| Using AI as a subject-matter expert | Consulting AI for content knowledge in specialised domains |
| AI conversational partners | Practising speaking with AI chatbots or voice agents |
| Generating feedback on learner writing and speaking | Providing automated or semi-automated comments on output |
| Creating lesson reports | Summarising learner progress, attendance, or achievement |
| Creating trainer guides and notes | Writing teaching instructions, timing suggestions, and tips |
| Generating professional development content | Creating training materials for teachers on AI use |
| Simulating learner errors | Producing examples of common mistakes for analysis tasks |
| Creating differentiation options | Adapting tasks for mixed-level or special-needs groups |
| Generating extension activities | Producing additional tasks for fast finishers or homework |
| Supporting learner autonomy | Creating self-study guides, independent practice tasks |
| Facilitating telecollaboration | Designing tasks for virtual exchange or international projects |
| Creating reflective prompts | Writing questions for learner journals or self-assessment |
| Generating peer-review guidelines | Designing criteria and instructions for peer feedback |
| Supporting needs analysis | Creating surveys, interview questions, or diagnostic tasks |
| Creating progress trackers | Designing tools for learners to monitor their own development |
| Administrative and operational | |
| Scheduling and timetabling | Organising classes, sessions, or teacher assignments |
| Writing policy documents | Drafting institutional policies, guidelines, or procedures |
| Creating onboarding materials | Producing welcome packs, orientation content for learners or staff |
| Generating meeting agendas and minutes | Structuring and documenting team or department meetings |
| Writing grant proposals and reports | Drafting funding applications and progress reports |
| Creating surveys and questionnaires | Designing feedback forms, needs analysis tools |
| Translating and localising content | Adapting materials for different languages or cultural contexts |
| Managing learner records | Organising and summarising learner data and progress |
The image above is AI generated.
Comments are closed