Welcome! Let's start with an honest question
Have you ever had a colleague say, "Oh, just ask ChatGPT"? Maybe you nodded and thought, but what can it actually do for me?
That's where we start. No hype and no panic. We'll take an honest look at what AI assistants like ChatGPT, Gemini, Claude and Microsoft 365 Copilot are good at, where they struggle and what should always stay with you.
So, are teachers really using this stuff?
They are, and quite a lot of them now.
Teacher Tapp is a daily survey app that thousands of teachers in England answer. In November and December 2025 it asked around 8,000 to 10,000 teachers about AI. By that autumn, most of them said they used AI for school work every week. Only about 12% had never used it.
So what were they using it for? Among teachers using AI, the most common jobs were lesson planning (54%), creating quizzes (50%), test materials (45%), letters to parents (41%) and school reports (39%). Using AI during a live lesson was far rarer, at just 11%.
There's one detail I love. Only 14% had used AI to plan their most recent lesson. So for most teachers, AI is a helper they reach for now and then. It isn't running their planning.
What about the teachers who don't use it? Their top reason, given by 56% of them, was worry about reliability. Hold on to that word. We'll come back to it.
Does it actually save time?
Here's some proper evidence. In December 2024 the Education Endowment Foundation (EEF) published a trial with 259 science teachers in 68 secondary schools in England. Half of them used ChatGPT, along with a guide on how to use it well, to prepare their Year 7 and 8 lessons.
The result? The ChatGPT group spent about 56 minutes a week on that planning, against about 81 minutes for everyone else. That's a saving of roughly 25 minutes a week, or 31%. An expert panel didn't spot a drop in quality either, though the researchers said to treat that finding with some caution.
So yes, real time can be saved. But notice that those teachers had a guide, and they were still the ones in charge.
Why can AI sound so sure and still be wrong?
This is the big one, so let's go slowly.
An AI assistant doesn't look facts up the way you'd check a textbook. It has learned patterns from a huge amount of text. When you ask a question, it predicts the words most likely to come next. Most of the time that gives you something useful. But "sounds right" and "is right" aren't the same thing.
Think of a confident pupil who hasn't done the reading. Ask them about the causes of the First World War and they'll give you a fluent, well-structured answer. Some of it might be completely made up. They aren't lying on purpose. They're just filling the gap with something that sounds plausible.
AI does the same thing. In September 2025, OpenAI researchers explained that the way these models are trained and tested often rewards guessing over saying "I'm not sure". So the tool may invent a quote, a statistic or a book title, and deliver it in exactly the same calm, confident tone as a true fact. People call these made-up answers "hallucinations".
So here's the golden rule. Confidence is not evidence. You stay the expert in the room.
Sorting your day: help, check, or keep human
Let's make this practical. Picture your tasks sitting in three boxes.
Box 1: AI is a great help. First drafts and idea generation. Ten starter activities on fractions. A low-stakes retrieval quiz. Rewording instructions more simply. Turning your bullet points into a tidy parent newsletter. If it gets something slightly wrong here, you'll spot it and fix it fast.
Box 2: AI can help, but you check closely. Anything with facts, accuracy or a real person at the other end. Subject content and exam-style questions. Adapting a text for a pupil who finds reading hard. Drafting report comments or feedback. Here AI gives you a head start, but you read every line against what you know and against trusted sources.
Box 3: This stays human. The final judgement on a pupil's grade. Decisions about a child's wellbeing or safety. The conversation with a worried parent. Knowing that Jamie had a rough morning and needs a gentler start. AI doesn't know your pupils, and it can't take professional responsibility. You can.
Which box does marking a mock exam go in? Ask yourself who is accountable for the grade. That's you, so the final mark stays in Box 3, even if AI helped you draft some comments along the way.
A simple habit for every AI output
Whenever you use one of these tools, try this quick routine:
- Ask clearly for what you need.
- Read the output critically, as if a new trainee had written it.
- Check any facts, names or numbers against a source you trust.
- Adapt it with your own knowledge of your class before it goes anywhere near a pupil.
And one quick note. Your school may have rules about which tools you can use. We'll look at that properly in the next level.
Try it this week
Pick one Box 1 task, such as a five-question recap quiz. Ask an AI assistant to draft it, then run it through the four-step habit. Notice what you changed and why. That noticing is your professional judgement at work.
Key takeaways
- Most teachers in England were using AI weekly by autumn 2025 (Teacher Tapp, Nov–Dec 2025). Planning, quizzes and letters to parents were the top uses.
- The EEF's December 2024 trial found ChatGPT, used with a guide, cut science lesson planning time by about 31%.
- AI predicts likely words. It doesn't check facts, so it can sound confident and still be wrong.
- Sort your tasks into three boxes: AI helps, AI helps but you check closely, and stays human.
- Grades, wellbeing and safety decisions, and relationships stay with you.