Generative AI Idea Pack Cambridge
Generative AI Idea Pack Cambridge
Generative AI Idea Pack Cambridge
Idea Pack
for English language
teachers
How to use
We have designed this idea pack to help you become more confident when using generative AI tools in your teaching
practice. It offers research-based, practical suggestions that will aid you in navigating the exciting yet occasionally
daunting field of generative AI. Some of these ideas are ideas for activities in the classroom, others are for you to
consider and perhaps embed in your professional development goals. You might go through the whole pack at once,
or you could take a one-a-day approach and start your week (or day) with a new card.
On some of the cards, you will see specific GenAI tools mentioned. When possible, we recommend trying prompts on multiple tools and
We have tried as far as possible to recommend specific tools that seeing what results you get. For instance, you and your colleagues
are accessible for free. Not all tools are free to use or allow unlimited might input any prompt into a chatbot such as ChatGPT, Copilot, and
use, and some tools that are free now may have a premium version Pi.ai and compare the results. Or for image generation, you could try
later. There are currently free-to-use alternatives for most general the same prompt using DALL-E, Adobe Firefly, and Midjourney.
purposes, such as chatbots or image generators. GenAI tools change
fast, so specific tools that are recommended in these activity cards Depending on the tool, you might get an output that you can
may be different in the future and new tools will be created. However, export into another tool (e.g. a table into a spreadsheet), other
most of these activities can be used on lots of different tools. times you can just copy and paste the results you get and modify it in
a software you know better. Give yourself time to discover what the
specific tools can do!
A word on copyright
Copyright issues and intellectual property rights related to AI content generation are an unchartered territory and there are still many
unanswered questions in connection with them. However, what’s for certain is that you cannot feed in (not even parts of) a published piece
learning or teaching material and ask for certain modifications (such as “make it shorter, longer, include more advanced vocabulary, etc.)
as it is copyright infringement. It also means that you mustn’t use text-to-speech AI tools on such texts (for example, if you wanted to turn
them into audio recordings). Please make sure you have information on and follow your local laws and your institution’s rules when you use
generative AI tools.
Prompt components
Try using the 7 components below to write your prompts. You don’t need to use all 7 components in every prompt that
you write. Experiment with this structure and see if it helps you get the output you need.
Role Description: In one line, tell the bot what its role is. For Examples: Illustrate what the output should look like to help guide
example: “You are an English as an Additional Language Teaching the bot. For example, if you’re using the bot to write a quiz for your
Assistant”. students, use this space to give it an example of a question, so it
knows what its response should look like.
Instructions: This is the most important part of a prompt: tell it
what to do. Begin the instructions with an action word like “Write”, Cues & Format: Provide more information to influence the output,
“Explain”, “Edit”, and so on. Good instructions in prompts are like such as the format, style, and tone. You might guide its format by
good instructions in the classroom: detailed, specific, and actionable. saying something like “Provide your answer in a bullet point list”
This will give you fewer errors and issues. or “Produce ten sentences”. You might guide the style by saying
something like “Use language appropriate for beginner learners of
Context: Give the bot any contextual information it needs. For English”, or by using words such as creative, formal, persuasive, etc.
example, if you are using it to prepare activities, you may include You might guide tone by saying that it should be friendly, helpful,
information about your class in terms of topic, number of learners, professional, etc.
learner level, and so on. However, it is crucial that you do not enter
any identifying detail about your student. Allowing questions: If appropriate, give the bot a chance to admit
and avoid its weaknesses. For example, encourage it to ask any
Input Content: If you are asking the GenAI bot to edit or analyse questions it needs to fulfill its task, or tell it to admit if it doesn’t
existing content, this is where you include this. If your Input Content know something.
is quite long, put it below the body of your prompt. Make sure it
is clear what is input content and what is not. Avoid entering any
material into a chatbot that is published or copyrighted, unless you
own the copyright.
Prompt writing tips
Plan your prompt before you start. What do you want your output If you’re not getting what you need, reiterate and experiment
to look like? What information do you want it to provide? with your prompt. Each bot behaves differently, so remember that
techniques that work on one may not work on another. Don’t be
Consider the length of the response that you want the bot to afraid to test and revise a prompt.
create. Open questions or requests will prompt longer responses
from the bot, while closed questions can produce shorter results. Be aware of the climate impact of generative AI. Be cautious that
However, remember that AI bots struggle to follow strict word limits, GenAI requires a lot of energy to run, and one well-written prompt
though they can follow approximate guidance, e.g. “Your response may use these tools less than taking a long time to get what you
should be 2-3 sentences”. need.
Avoid using CEFR levels (i.e. A1-C2) when telling the generative
AI bot the style of language to use. GenAI bots occasionally have
trouble understanding them.
Tell the AI bot what it should do, and don’t tell it what it
shouldn’t do. AI bots respond better when told to do something
than to avoid doing something.
Young learners
Productivity Being part of a professional community Responding to learners
− Saving time on communication − Join a community − Make it make sense!
− Review, assess and give feedback
Information management Sourcing and evaluating digital resources
− Prompt library − Create an evaluation checklist Communication and interaction online
− Genre switch
Legal issues Developing materials
− Role play with a historical figure or fictional character
− Whose art is this? − Create role play or debate cards
− Create vocabulary or idiom lists Assessing learning
Theories and methodologies for language learning
and teaching − Creative inspiration − Personalised assessment rubrics
− Further reading suggestions Preparing students for using digital tools and resources
− Lesson reflection with an AI coach − AI biases Please click on the
− Teacher buddy − AI results evaluation
Subcategory or Title
if you would like to
view the activity card.
Contents
Teenager learners
Productivity Reflection and development with digital tools and Setting up and managing digital tools and resources
resources
− Saving time on communication − Tool tutorials
− Further reading suggestions
Information management Preparing students for using digital tools and resources
− Lesson reflection with an AI coach
− Prompt library − Teacher buddy − AI biases
− AI results evaluation
Digital citizenship Being part of a professional community
− Classroom GenAI ground rules
− Highlighting bias − Join a community − How to write great prompts
− How sustainable is AI?
Sourcing and evaluating digital resources − What’s your opinion?
Legal issues
− Create an evaluation checklist Responding to learners
− Debate: Is AI an art thief?
Developing materials − Learner-developed personalised activities
Digital welfare and safety − Make it make sense!
− Create role play or debate cards
− AI image games − Create vocabulary or idiom lists − Review, assess and give feedback
− Fake news − Creative inspiration Communication and interaction online
− Solving the issue of ‘How can we make AI safer?’ − Games with images − Genre switch
Theories and methodologies for language learning and − Jigsaw reading − Role play with a historical figure or fictional character
teaching − Making materials accessible
− Improving access to authentic materials Assessing learning
− Personalised starters
− Methodology suggestions − Tongue twisters − AI feedback for writing
Adult learners
Productivity Reflection and development with digital tools and Setting up and managing digital tools and resources
resources
− Saving time on communication − Tool tutorials
− Further reading suggestions
Information management Preparing students for using digital tools and resources
− Lesson reflection with an AI coach
− Prompt library − Teacher buddy − AI biases
− AI results evaluation
Digital citizenship Being part of a professional community
− Classroom GenAI ground rules
− Highlighting bias − Join a community − How to write great prompts
− How sustainable is AI?
Sourcing and evaluating digital resources − What’s your opinion?
Legal issues
− Create an evaluation checklist Responding to learners
− Debate: Is AI an art thief?
Developing materials − Learner-developed personalised activities
Digital welfare and safety − Make it make sense!
− Create role play or debate cards
− AI image games − Create vocabulary or idiom lists − Review, assess and give feedback
− Fake news − Creative inspiration Communication and interaction online
− Solving the issue of ‘How can we make AI safer?’ − Games with images − Genre switch
Theories and methodologies for language learning and − Jigsaw reading − Role play with a historical figure or fictional character
teaching − Making materials accessible
− Improving access to authentic materials Assessing learning
− Personalised starters
− Methodology suggestions − Tongue twisters − AI feedback for writing
Classroom Activity
Digital citizenship Digital tools for language learning and teaching Preparing students for using digital tools and resources
− Highlighting bias − Chatbot role play − AI biases
− How sustainable is AI? − AI results evaluation
Developing materials
− Classroom GenAI ground rules
Legal issues − Create role play or debate cards
− How to write great prompts
− Debate: Is AI an art thief? − Create vocabulary or idiom lists
− What’s your opinion?
− Whose art is this? − Creative inspiration
− Games with images Responding to learners
Digital welfare and safety
− Jigsaw reading − Learner-developed personalised activities
− AI image games
− Personalised starters
− Fake news Communication and interaction online
− Tongue twisters
− Solving the issue of ‘How can we make AI safer?’ − Genre switch
− Write your own text
− Role play with a historical figure or fictional character
Theories and methodologies for language learning and
teaching Lesson planning
• Your role
• The tone of the email
• Any details you would like to mention
Remember, do not include any personal information such as the name of the student, parent
or guardian.
Prompt library
Why not start a prompt library with colleagues in your school or through an online community? Add
prompts that worked for you and borrow your colleagues’ prompts. Regularly trying new prompts will
help you develop your prompt writing skills.
• Use simple software, or a website such as Padlet, or a shared document / spreadsheet that everyone
feels comfortable using.
• Choose a format that lets you search easily.
• Tags can be helpful too – you can tag by subject, age group, topic etc.
Don’t be afraid to edit and improve your colleagues’ prompts and invite them to improve yours!
Highlighting bias
Ask your learners to use image-generating AI to create a poster of a
particular nationality, e.g. ‘poster of Mexican people’. This will very often
show how chatbots can produce biased content. It can be particularly
interesting to do this for learners’ own nationality/ies.
Ideas:
• Split into three groups, each researching either the economic, environmental or social
impacts of AI.
• Share your findings.
• Each group then moves to a different aspect and then suggests solutions to reduce any
negative impact AI has on this area.
Some of the articles on these issues might be a bit difficult for learners, depending on their
level and age. Why not encourage them to use AI to simplify them? Or starting research
using AI? Remember to always check whether the article you want to simplify is open source
as you must not enter copyright materials into any GenAI tool.
How sustainable is AI?
Example:
• Briefly introduce the topic (e.g. AI image generation tools need to be trained on artists’ work to create new images.
Some artists say using their artwork to train these is like stealing their work).
• You could also watch part of this clip from the BBC about the issue.
• Learners are then divided into two groups – for and against ‘AI is an art thief’.
• Learners research their side of the argument, perhaps also using AI for ideas and to generate images to use examples.
• Learners debate the topic and reflect on both arguments.
If you’re not familiar with the arguments yourself, you can do your own research beforehand, or learn alongside your
learners, asking them to explain parts of the arguments to you as you monitor the class.
After that, ask your students: whose pictures are these? Who owns them? You could then have a discussion about
copyright, which is like a special “made by” sticker, it usually goes to the person who creates something. But the AI
creator isn’t really a person, it’s a machine! Right now, in April 2024, the law isn’t quite sure who gets the copyright for AI-
generated stuff.
Here are some ideas people are thinking about, phrased in a way that can be easier for young learners to
understand:
• Did someone help the AI creator? If you told the AI creator what to paint, maybe you get the copyright!
• Where did your own ideas come from? What influenced them?
• Did the AI creator use other people’s work? If it used parts of existing paintings, those people might still own those
parts.
• Is the AI-generated art really new? Or is it just copying other things?
AI image games
Improve critical thinking and attention to detail with AI-generated images with an image generator. Here you have two games you
can try:
Game 1: Game 2:
• Either you or your students (in pairs) use the text-to-image In this version, you or your students can use the generative fill
function and generate a photo of a famous landmark. function of Adobe Firefly.
• Then they compare the generated artwork with photos of the
real place, collecting all the differences and similarities. • First, you upload a real photo, for example, of your living room.
• Then you or they (in pairs, for each other) select certain parts
This task can help them to improve their attention to detail in the photo and either remove them (generative fill can
and is also good for practicing comparatives, superlatives, and perfectly blend the removed part with the background), add
contrastive linking words. something to them or replace them with something else.
• Their task will then be to guess what’s real and what’s been
modified in the images.
This task can help students with their critical thinking, modals of
deduction, and expressions of agreement and disagreement.
Fake news
Helping students, no matter their age, to become AI literate is going to be an
important part of teaching and learning. The following activity will help your
learners become more conscious of real and fake content online.
• Put students in small groups and have them generate a fake news article Prompt: Give me a fake news article
along with a matching lead image. They can use an AI chatbot, and a version about [aliens] that the general public
of the prompt on the right for the news item. would easily believe.
• Then they can use an AI image generator for the lead photo, using a similar
prompt to the one above.
• Mix these student-generated news stories with some real tabloid news and
check how well they can identify fake news. Give them a few clues on how to do
that, or you could come up with them together – e.g. by looking up people that
were quoted in the article, or researching the fact in other reputable resources.
As a follow-up activity, you could have a discussion on the impact of fake news,
e.g. the fact that it is influencing people’s decision making, which could have a
negative effect on their welfare and/or safety.
Example:
• Set the scene: tell your learners they are part of an AI safety committee who has been asked to present
three solutions to an AI safety issue (e.g. AI-generated video and deepfakes spreading misinformation).
• They then research solutions and choose three to present.
• Optional: they can also suggest a new solution that they think of themselves.
Do some research on AI welfare and safety issues yourself and choose a topic suitable for the age and
level of your learners. You can use AI to help you create a list of issues to start your thinking.
If students still don’t understand the output, encourage them to keep trying different prompts (e.g. “I don’t
understand. Please simplify”) until they get to something they can engage with easily. Remind your students that
they should refrain from entering copyright materials into the tools.
Methodology suggestions
GenAI chatbots can be very helpful when you’re trying to identify Prompt: Suggest the best ELT
the best methodologies to teach a certain topic/language item. How methodologies to teach [how to express
about using them as a starting point when creating an activity or a possession/simple conditionals] to
lesson plan focusing on a specific teaching methodology? You can [elementary] [adult] learners.
enter one of the prompts to the right as a starting point.
Evaluate results according to your needs and refine the prompt if Prompt: Create a lesson plan to teach
necessary. You can try this with different methodologies according [young learners] [the difference between
past simple and continuous] using
to your teaching beliefs and learners’ needs to see which is more [Guided Discovery].
suitable for your context.
Example:
• Try some conversations out yourself first and save some to show as models if needed.
• Encourage learners to both ask and answer questions.
• Get learners to save their conversation and share any interesting things they spoke
about with the class.
• Ask learners to experiment with asking the chatbot to correct any mistakes they make.
If you have young learners, you can facilitate this conversation with the whole class,
by giving the initial prompt and then moderating the learners’ suggestions to continue
the conversation or having learners take turns to provide the next interaction. If you’re
working with teenagers or adult learners, remind them that they shouldn’t enter any
personal information about themselves into the chatbot.
If you don’t find an exercise useful, see if you can prompt the chatbot to modify it
so that it better suits your needs. Use your knowledge of teaching to help you!
Additionally, if you specifically ask it to have a conversation with you, it can also improve your speaking
skills, such as asking, responding, turn-taking, asking for clarification, and repair strategies. You can of
course try this with your students as well; some scenarios where it could prove helpful are job interviews,
motivational interviews for further study, dates, or small talk with colleagues.
Tips:
You are in control of the conversation so don’t be afraid to skip questions if they’re not helpful or ask the
coach to change their style if it’s not working for you.
Teacher buddy
Replacing human connection with something artificial is definitely not the preferred direction to take.
However, when you need to get something off your chest quickly, you might not have anyone around
us that could offer the help or support you need. In such cases you could turn to an AI teacher buddy
that you can enter doubts and concerns into about a lesson you have just taught, and then can generate
suggestions for assistance.
ChatGPT’s voice chat function (in the app) to some extent can replicate a real-world conversation with
someone who’s an active listener (because it recaps what you’ve said) and a good turn-taker (it always
asks a relevant follow-up question), while Pi.ai’s textual chat is designed to listen to your problems and
offer understanding, solace, and advice if you need it. Just be careful with relying on these tools too
much. Remember that GenAI chatbots do not process what you say in the same way as a human: the
responses they produce are created one word at a time based on the next most likely word and any
other programming or training they are based on. GenAI chatbots do not think and any advice they
provide may be questionable. You should always carefully consider whether responses are relevant to
you and your context. They can offer a quick solution but shouldn’t replace human interaction entirely.
Join a community
As with the rapid spread of remote teaching during the pandemic, the appearance of so many AI tools has also made it clear that
teachers work best when sharing best practices, worries, questions, and advice with each other.
Find a community on a social media platform dedicated to English teaching using AI. This can be a great way of learning new ways
of using AI in teaching little and often and a great place to share your experiences and questions.
Tips:
• There are plenty of communities on social media platforms. Choose one on a platform that you already use regularly.
• Be on the lookout for newer or more localised communities.
• Some communities are so active that keeping up with all the notifications can be daunting. Joining smaller groups may be better
than extremely large ones because the posts tend to stick to the original topic more.
• Don’t feel like you have to read everything. Maybe set a time limit for how long you’d like to spend on the community each
week. For example, do you want to set a reminder to take a look for 10 minutes every Wednesday?
• Get the most out of the community by deciding how you’re going to use them as part of professional development, for example:
are you going to ask questions about how to use AI? Are you asking someone to help you improve your prompt?
• Share your successes and learning points with the community.
General Skill-specific
• Is the information in the material correct? • Reading: do I need to read and understand the text to answer
• Does the activity adequately support the learning objective(s)? the questions or are some of them easy to guess?
• Is the language at the right level of challenge for my learners? • Cloze activities: is the language of the sentences at the right
level for my learners?
• Can I do the exercises myself?
• Are my learners going to be motivated to complete this
activity?
If the activities don’t meet your criteria, you could prompt a GenAI tool to alter them; use the activities as a springboard to rewrite
them yourself or choose a different source for your activities.
Prompt: You are an [EFL] teacher. Write a set of [10] role Prompt: You are an [EFL] teacher. Write a set of [10] debate
play scenarios for [elementary] [young learners] related to cards for [intermediate] [adult] learners related to [Sustainable
[ordering food]. Development Goal 14: Life below Water].
• Print the cards or create digital versions if you’re teaching • Organise the class in two big groups: one for, and one
online. against.
• Organise the class into pairs and distribute the cards. • Distribute the cards accordingly and moderate the
Monitor the pair work and give feedback. discussion as you would in a regular classroom debate.
• At the end, ask volunteers to repeat the task in front of all
their classmates.
• Give general feedback to the whole class.
As always, don’t forget to edit your prompt and/or the output if the cards are different from what you had in mind.
Once you have the lists, you can use them as any other wordlist you have – classroom games, gap-fill tasks, or
matching exercises are only some of the many options!
Creative inspiration
When introducing vocabulary on a topic that can be a bit dry, generate a fun story
using all the target words. This can help make the topic more engaging and can also
offer learners opportunities to develop their creative thinking skills.
When prompting:
• Include that the chatbot should act as a teacher of English as a Foreign Language.
• Describe the age and level of your learners.
• Give the topic and the vocabulary you want to teach (e.g. describing people – wavy,
beard, short, etc.)
• Ask it to create a funny story using all these words.
• Optional: ask the chatbot to generate an image of one of the key characters or
scenes and have learners select the one that most closely matches the description
from the text (it is likely that none of the images will be quite right, so this can lead These images were created with the assistance of Microsoft Copilot.
to a great discussion).
Prompt: ‘Professor Whiskers wearing a tattered
Why not ask learners to write the next scene/the ending of the story, challenging red cape, using a grey umbrella as a magnifying
glass with big wavy hair and a big bushy beard.’
them to use some of the original vocabulary words.
3. In pairs (each with a different image), learners work out ten things that appear in
both images. They are not allowed to look at each other’s image so they must find Prompt: ‘A very messy desk belonging to
out by asking questions, e.g. ‘Do you have any apples in your picture?’ a teenager with a pencil, an eraser, some
4. Learners could then look at each other’s images and find something unusual/funny pens, some paper, an apple, a slice of
it. They then ask their partner a question about it which they have to think of a good pizza, and a smartphone’
answer for, e.g. A: ‘Why do you have two calculators on your desk?’ B: ‘So I can do my
maths homework faster!’
Always check the images created carefully to ensure they do not contain anything
unsuitable.
Jigsaw reading
Jigsaw reading activities can be a great way to engage mixed level groups in a Prompt 1: You are an [EFL] teacher of
reading activity – and you can ask a GenAI chatbot to create such an activity [elementary] [young] learners. Write
related to a specific topic. a jigsaw reading activity related to
[describing people].
Example:
1. Check the activity works before the lesson and create a completed answer Prompt 2: Create the [four] reading texts
sheet based on the texts. on cards.
2. Divide the class into groups with as many members as there are reading
texts and distribute a different card to each member.
Prompt 3: Create a chart for the
3. Ask each member to fill in their chart with the information they have learners to complete the activity with the
and then ask the other members for specific information to help them information from the reading texts/cards.
complete the chart.
Modifications for this prompt might include specific vocabulary that your
learners are familiar with or the amount of information you want to include
on each card.
For students who are hard-of-hearing, you can quickly and easily generate subtitles for any video, be it
a file or a link from YouTube. There are many video-to-text AI tools but Vizard.ai has a very simple user
interface and lots of post-editing options once the video transcript is ready. However, you should check
the output to ensure it has not made errors.
For students who are visually impaired, you can use the built-in text-to-speech extension of many
websites and software. For example, Quizizz offers a “read aloud” option before you start the quiz and
tools owned by Microsoft [for example, Nearpod or Flip (previously Flipgrid)] have an “immersive reader”
function in them, which can even highlight the parts of speech or make the text dyslexic friendly. Text-to-
speech is a feature that most computers (although at a lower quality) include as well.
Personalised starters
Instead of searching for the top 10 best conversation starters or Prompt: Give me 10 [job interview
job interview questions, try generating your own set that’s going questions] for the role of [ethical hacker].
The candidate is an EFL speaker, so also
to be just right for the given context. You can also generate debate include questions related to their English
topics for an upcoming argumentative essay or presentation knowledge and skills.
assignment. Try one of these prompts on the right.
A follow-up activity can include prompting the chatbot to also Prompt: Generate 10 engaging debate
create good and bad model answers to these questions that topics for [young adults] who are
preparing for the [IELTS Academic exam].
students can evaluate. You can then assemble the evaluation
criteria together.
Prompt: My student needs to [deliver the
quarterly report as a presentation to the
board]. Generate 6 questions the board
might ask.
Tongue twisters
If there is a challenging sound you want to help learners practice, Prompt: Create a simple tongue twister
you can ask a GenAI chatbot to create a tongue twister for a for [elementary] English learners to
practise the sound [/d3/]
particular sound or set of sounds.
Once your learners are familiar with the concept, you can prompt
it to create a list of words with a particular sound or set of sounds
and have learners create the tongue twisters themselves. If you
work with teenagers or adult learners, you can also let them create
the tongue twisters and/or the words – and then they can vote on
their favourites!
The text can then be turned into a listening activity with a text-to-speech
tool or you can play it automatically if you have used a chatbot that
allows you to do so.
Don’t lose heart if the first version is not exactly what you wished for,
you can either regenerate the text with the same prompt, or add further
requests, such as “Make it shorter/Make it sound more professional.”
Differentiation tasks
Creating materials for mixed-level and mixed-ability classes can Prompt: You are an English teacher of a
be challenging and time-consuming. Why not prompt a GenAI tool mixed-ability group of [young] learners.
Create [three] differentiated activities
to start the process for you? You could ask a chatbot to create a for practising [using present tenses for
number of differentiated activities for a specific topic and for your routines] at [an elementary level].
particular context.
Elevator pitch
This task can really enhance your students’ language awareness and critical thinking if they are at least at B1 level.
Students can do this task individually but working in pairs or groups could create more peer-learning opportunities.
Example:
1. First, they generate a text for a given purpose. Let’s imagine they are practising how to give good elevator pitches.
They will have to include as much detail in the prompt as possible to make the response personalized.
2. Then they move on to critically evaluate the AI response, which probably contains advanced vocabulary with lots of
idiomatic expressions and phrases. This step means downgrading the AI-generated response so that it reflects what
they believe they are truly capable of. They should make the text sound believable if they were to read it out loud.
This can involve rephrasing sentences that contain tenses or structures they are not comfortable with, or deleting
words they can’t pronounce.
Although teachers typically try to upgrade student language with text comparisons, the point in this task is to make
students more autonomous and self-reliant when dealing with perfectly crafted AI responses. They are probably already
using AI writing assistants at the workplace or in other situations, so crafting a version of an artificially perfect text that
they can present more confidently is going to be a useful skill.
Course skeleton
Use AI chatbots to ask for the initial skeleton of a course. This can Prompt: I’m a language teacher,
teaching English as a second language
be helpful if you haven’t taught the course before, you think some to [adults]. Give me a [12-week] course
upgrade would be necessary, or you’d just like to check whether plan for a [presentation skills] course at
you’re on the right track. [intermediate] level.
After entering the first prompt, you will probably get the main
learning outcomes, and a rough sketch of the 12 weeks. You can Prompt: They have [one] lesson every
week and each lesson is [60] minutes
then ask for further details, such as assessment ideas, lesson plans long. I have [one visually impaired]
for each lesson, or even advice on differentiation. Try something student and one [who doesn’t like to work
like this: with other people much]. Break down the
course plan taking this information into
account. Don’t forget to differentiate.
Tool tutorials
Some students could require further assistance in connection with educational apps and websites,
including “simpler” tools such as word processors or the VLE of your choice.
To improve your productivity, you can now quickly create tutorials with AI assistance that you can send
out whenever a student asks for help. There are several tools out there, for example, Iorad or Scribe.
They usually require a browser extension to be installed, then upon request they “record” the mouse
clicks you make and put together the step-by-step user guide in seconds! You can then edit the guide
if you think some steps are unnecessary or need more explanation. Once you’re done, you can easily
share a direct link to the guide you have created. You could even add them to a library for students to
access, e.g. in an FAQ area or similar.
AI biases
All of us – including teachers, learners and AI tools – have conscious and/or unconscious biases. In order to check these, ask
learners and a GenAI chatbot the following questions:
• The doctor yelled at the nurse because she was late. Who was late?
• The nurse yelled at the doctor because he was late. Who was late?
• The doctor yelled at the nurse because he was late. Who was late?
You could then compare learners’ answers to the chatbots’ outputs; and of course, from the way the sentence in all three cases
is structured there’s no way of knowing who was late without knowing the gender of the doctor and the nurse, it could have
been either of them. Some of the chatbots might recognize the ambiguity and tell you about it; others, however, could assume
stereotypical roles (as they did in some of our test runs, assuming a doctor was always male while the nurse was female).
• Why is it a bad thing to think of some professions only as a man’s job / woman’s job?
• What can we do to avoid these stereotypes in our own lives?
• Why is it harmful if the replies of GenAI tools include these stereotypes?
AI results evaluation
To make sure your students understand the risks of accepting Prompt: Act as a tourist guide. Make a list
information provided by GenAI tools, you might want to try this of tips for [a family with 3 children aged
activity. Put students into small groups according to their city/town/ 4, 6 and 10] spending a week in [Buenos
Aires]. Make sure that the language you
village of residence (or one that they know very well). You (or if your use is appropriate for [beginner] learners
work with teenagers or adults, your learners) can then ask a GenAI of English and is engaging for children
chatbot for tips for spending time in them, in turns. aged [9].
Ask each group to analyze the results and check for “hallucinations”
(false information); note that this is more likely to happen if it’s not
a place that attracts a lot of tourists. You could discuss what you can
do with these hallucinations too – e.g. make sure you edit the output
to have the correct information, or report the issue within the tool to
improve outputs.
Ask learners:
It is good to get learners to agree on the ground rules themselves but if you feel they have missed any
key points out, bring these to learners’ attention and ask them how they could include them in their
ground rules.
1. First, generate something relatively short with some specific details. Prompt: Write a [100-word rejection
email] that includes that [the candidate
2. Then show students the generated text – but not the original prompt – and is perfectly competent but isn’t willing
ask them to come up with a prompt that they think could achieve the same to wake up at 4 am every day to get to
result. The prompt that achieves the closest result, wins. work. Also, they wouldn’t like to spend
happy hours on Friday with the team.]
3. Once you have found the winning prompt, analyze it together with the group Emphasise [how important team spirit
to highlight the necessary elements of a good prompt. is for the company and that everyone’s
part of a big family here]. Use a
[professional] style.
For images, Google now has an enjoyable and educational interactive image-
prompting game, called “Say what you see.” In this game, you can see a photo
of something, and you need to come up with the prompt that you think would
generate the same image. After typing in your prompt, you can immediately
check and compare whether you’ve succeeded.
Ideas:
• Choose a question from the coursebook asking for students’ opinion on a topic (e.g. fast fashion).
• Students give their initial opinion.
• They then think of at least three people who might have differing opinions on the topic (e.g. CEO of a fast fashion company, a
worker in a factory, an environmental activist). First ask students what they think these people would say.
• Students then prompt AI to act in each of these roles in turn answering the question (e.g. What do you think about fast fashion)?
• Students compare and contrast the answers, perhaps by drawing a Venn diagram to show where the opinions are similar and
where they differ.
• Finally, students answer the question themselves again and reflect on whether they changed their opinion at all as a result of
hearing other points of view.
For further ideas on developing students’ critical thinking skills through AI, take a look at our blog.
Tips:
• Model an example as a class, reminding learners of the rules of good prompting (see introduction cards).
• Then ask learners to try it out independently.
• Start with more concrete areas such as a list of vocabulary or grammar items they struggled with and want to
practise.
• As learners experiment more with this, ask them to reflect on which areas this method works well with and
which it does not. For those areas it doesn’t work so well with (perhaps listening skills for example), can they
think of other ways to improve independently?
Be prepared to monitor and help learners but also encourage learners to experiment and use trial and error to
get a useful output.
What you can do is tell it to summarize in complete, coherent sentences what you’re about to say. Then
you start giving your feedback as the thoughts come to your mind. You might as well do a running
commentary of student work that’s in front of you on the computer – for instance when marking student
essays – and then wait for the chatbot to turn it into coherent text. The separate freemium website,
AudioPen does this automatically, though there is a 3-minute limit per audio recordings.
Ensure you carefully check and edit the response if you decide you’d
like to share this with the learner.
Genre switch
One of the best uses of generative AI chatbots is amending existing
content. For instance, you can ask a GenAI chatbot to change a text’s
genre from a narrative to a letter, or from a newspaper article to Prompt: Change the following text into
a [newspaper article] at [an intermediate
a fairy tale – the options are endless! You can start by entering the level].
prompts on the right into a chatbot:
After that, you can set up small groups and have learners compare
and contrast the differences in order to learn about different genres, Prompt: Generate [5] questions to
compare and contrast the original text
for example, aspects like content, tone, audience, structure, purpose, and the new version.
etc. One use of this could be to teach students about how writing
a formal letter to a stranger is different to writing to someone you
know well. Please remember that you must not enter copyright
materials into any of the generative AI tools you’re using.
Although automatic AI-generated feedback can be very useful, it will still be necessary to do some manual marking for
some tasks to give richer feedback, tailored to your learners’ specific needs.
Test generation
You can generate either entire tests or just individual tasks with the
help of AI chatbots. In this case, you need to be very specific with
your prompts and will also need to meticulously provide all the
necessary details, such as the vocabulary items to be included, the
text to be turned into a gap-fill task, or the grammar points to be
included. You might need to generate the test in several steps. For Prompt: Generate a [100-word news
article] for language learners about
example, have a look at the following series of prompts on the right. [investing], and include these words:
[portfolio, ROI, interest rate, bear
Even if you are not completely satisfied with the results, you market, bull market].
have already saved time because tweaking the generated version
takes less effort than creating the whole test from scratch. Please
remember to evaluate the outputs; e.g. you may do the tasks Prompt: Remove the vocabulary items
yourself or consider whether students answering correctly tells you specified previously and substitute them
about what they’ve learnt in the course. To learn more about how to with gaps. Each gap should have a letter
of the alphabet in front of it. Provide the
evaluate tests and assessment materials, you could check out The vocabulary items as a list to choose from
Assessment Network and their resources. under the text.
Analysing feedback
Invite learners to give you some feedback on your teaching and use AI to help Prompt:
pull out themes to guide your professional development. This can help you I received some feedback on a course
create more learner-centred classes and develop as a teacher. that I run. I asked [4 open ended
questions] and got textual feedback.
Tips: Analyze the answers and tell me the
following:
• Create an online form to ask for feedback from your learners. You can also What’s the general reaction to my
use GenAI tools such as Typeform or an ordinary chatbot to help you with teaching - overly negative, somewhat
this step. negative, somewhat positive, overly
positive?
• Include some open questions (e.g. What was your favourite thing about the
lesson? What did you find most helpful? etc) What are the central thoughts in
connection with each question?
• Collect the responses to the open questions and use a GenAI chatbot to What action point could you recommend
summarise common themes. based on the feedback?
[copy-paste the questions with all the
You can then use these themes to guide your professional development. answers you’ve received]
As always, check that no student names or other identifying information is
included in the responses you enter into the chatbot.
hallucinations
false information that is produced by an artificial intelligence […]
Notes
To cite this resource: Generative AI Idea Pack for English language teachers. (2024) [pdf] Cambridge: Cambridge
University Press & Assessment
Jasmin Silver
Jasmin is an educational consultant and researcher. She
specialises in applying research to improve teaching and learning
in ELT and in integrating skills beyond English into the classroom.
Jo Szoke
Jo is a freelance teacher trainer, learning designer, and AI
specialist. She also teaches business and general English to adults.
Vicky Saumell
Vicky is a teacher, trainer, materials writer, and presenter. She
has worked for major publishers, especially in the areas of
project-based learning and the meaningful use of technology for
language learning.