Two years ago, if you had told me that artificial intelligence would become one of the most powerful tools in my classroom, I would have been sceptical. Not dismissive but cautious. I work in a bilingual environment, teaching complex scientific ideas to students who are still developing academic English. My priority had always been clarity, structure, and inclusion. AI, at the time, felt like something distant, experimental, or even slightly gimmicky.
Then curiosity crept in.
What started as a simple “what if?” quickly turned into a professional reset. AI what I now jokingly call Absolutely Incredible forced me to rethink how I plan, how I give feedback, and how students learn. This isn’t a story about replacing teachers or automating education. It’s a story about upgrading our craft and reclaiming precision in teaching.
Prompts Are Your Best Friend
If there is one lesson, I learned early, it’s this: AI is only as good as the question you ask it. Prompts matter. A lot.
A vague prompt gives you a vague outcome. A well-thought-out prompt clear, purposeful, and rooted in classroom reality can save hours of planning and dramatically improve quality. Once I realised this, I started treating prompts the same way I treat lesson objectives: precise, intentional, and aligned to outcomes.
I’ll be honest when I saw how powerful the results were, I paid for the premium version of ChatGPT. But context matters. Teaching in China means access looks different. Tools like DeepSeek and DouBao are brilliant, free, and fast. Their limitation? They struggle to generate polished PDFs, Word documents, and Excel files. That’s where GPT still has the edge for now….
So I adapted — and this is the part that matters.
If you prefer, refuse or can’t afford the premium versions, that shouldn’t stop you. The key is to use the tools you do have intelligently.
Paid version or not I did the heavy thinking first. Planning, structuring, sequencing, and refining all on free platforms. I have taken old handwritten notes and digitised them by a simple picture and prompt such as “digitise these notes in the image – word for word”. This is where clarity is built. When you know you have a limited output, preparation becomes significant.
Once the thinking is done, you can paste the refined output from the free version into GPT for a final generation. Because the groundwork is solid, one clean prompt is enough. Paste, generate and suddenly you have a working classroom resource, not just a rough idea.
AI doesn’t replace professional thinking it amplifies it. It is intentional and an extension of your thought; a critical eye to unlock the cohesiveness of your thoughts. That is both scary and powerful take it as you wish.
Here’s an example of a prompt I regularly use for a KS3 bilingual class:
“Provide a set of bilingual English and Chinese key words with a brief description in a table.
Summarise the work – KEEP THE DETAIL IN BULLET POINTS – put key words or concepts in Chinese too (not all information).
Keep it in sequence. Ensure format keeps the superscript and subscript format.
Underline all titles.
Title size 32 font – body text size 24 font.
Activities to include:
– Fill in the blank exercise for definitions and/or solutions
– Sentence starter activity
– Matching key words to definitions in a table
Add solutions in the notes section.”
From there, you can tweak again on DeepSeek or DouBao before returning (or directly) to GPT for a final draft. Importantly, once the presentation is built, you are in control. So update, modify, and adapt.
I often add images manually from Google, it’s faster. You can be creative and use inbuilt tools like DALL·E, Copilot (built into PowerPoint), adobe firefly, GROK and Google Gemini. They’re useful, but access and practicality matter. The best tool is the one that works in your context. Experiment and decide.
Planning With ESL Learners in Mind
Planning for bilingual and ESL learners demands intentionality. You cannot jump straight to abstraction and expect success. Concrete first. Language scaffolding always. Cognitive load carefully managed.
This is where AI quietly excels.
With the right prompt, AI can generate multiple concrete models, analogies, and explanations tailored to ESL learners. It doesn’t replace professional judgement it supports it. I still decide what works. But instead of staring at a blank slide, I’m choosing from well-structured options.
Bilingual dictionaries have become a staple of my lessons. Students refer to them constantly. They support retrieval, reinforce subject-specific vocabulary, and reduce anxiety for learners who are unsure whether they’ve understood correctly. AI makes producing these dictionaries fast and consistent.
Here’s the real cheat code: most of what we teach already exists in specifications and textbooks. Instead of rewriting everything, I take photos of the relevant pages, upload them, and apply my prompt. The result? A lesson aligned to the specification, hitting the right learning objectives, and structured for my learners. Note the latter is relevant to ALL learners.
A word of caution, AI can be wrong. Confidently wrong. Every output needs a human eye and a subject specialist’s brain. Accuracy checking isn’t optional; it’s a professional responsibility.
Feedback: The Real Game Changer
This is where AI truly shines.
Feedback has always been one of the most time-consuming and emotionally loaded parts of teaching. Done poorly, it’s ignored. Done well, it transforms learning. AI gives us a way to scale good feedback without losing quality.
Students are already using AI whether we like it or not. Homework, in its traditional form, is largely obsolete. Students can take a photo of their work and receive step-by-step solutions instantly. You cannot police this effectively. Instead, you can design around it.
My approach is simple; I train the students on how to use it. As much as possible I control the narrative on how it’s used in my class and the expectations out of the class (I can hope). Students must write “AI” clearly on their work. My role then shifts from answer-checker to diagnostics. I question their reasoning. I probe understanding. I identify misconceptions. AI provides the steps; I assess the thinking.
Mark schemes are essential here. AI needs a reference point. When students revise, they upload the question and the mark scheme, allowing the AI to guide them step by step toward exam-standard responses.
So far, this approach is limited to GCSE and A-level students. The plan is to train teachers, parents, and younger students first before expanding access. Ethical and safe use must be taught, not assumed.
The Chatbot as a Learning Partner
Active learning requires retrieval, feedback, and reflection often without prompts. Traditionally, this has been hard to scale. AI changes that.
Students can now speak to a chatbot, explain their understanding, and receive immediate feedback. Visual learners get explanations. Verbal learners talk it through. It’s private, non-judgemental, and always available.
I recommend using it before bed: revisiting a concept, explaining it aloud, checking understanding. This strengthens neural connections between old knowledge and new learning.
I’ve used chatbots myself to break down complex ideas: thermodynamics, equilibrium, entropy into simple explanations. What emerges is something powerful: unlocked thought processes. By probing, questioning, and refining, students engage in something very close to the Feynman Technique explaining ideas simply to test understanding.
One rule I insist on: tell the chatbot to be honest. Clear. Direct. “Tell me if I’m wrong.” Learning depends on truth, not reassurance.
However, there is also a clear ethical concern. For some users, AI has become overly personal, leading to unhealthy dependence. In more serious cases, this raises safeguarding issues, particularly when students form emotional attachments to AI personas, such as so-called “AI girlfriends” or “AI boyfriends.” This is a phenomenon that many educators and families were not prepared for and did not fully understand when these tools first emerged
A Call to Upskill
AI is not a threat to teaching it’s a mirror. It exposes weak systems, inefficient habits, and outdated assumptions. But it also offers an opportunity: to upskill, reskill, and refine our practice.
If we don’t engage with this new reality, we risk becoming ill-equipped for the classrooms we already teach in. The upgrade is optional but the consequences of ignoring it are not.
Go and practice your craft. Learn the tools. Ask better questions. The teacher upgrade isn’t about technology. It’s about intent.
Sohaib Qureshi
MEd Transforming Practice
PS: As part of an acting member of the PeerSphere a community of thoughtful international educators. I would like to share a useful and growing resource on all things related to AI. Please feel free to explore the themes in the link below.
https://padlet.com/catherineellis3/ai-talking-points-1wv0l62e00go1wlg