"AI tools like ChatGPT work the same way as Google — you search for something and it gives you links to pages that already exist on the internet."
Google is like a library — you search a catalogue and find something that already exists. AI is more like a conversation with a librarian: you describe what you need, and it generates a brand new response in real time, built word by word based on patterns it has learned, rather than pulling an existing page or link.
"Because AI can hallucinate — stating incorrect information confidently as fact — teachers should always double-check anything AI-generated before using it in a lesson."
AI doesn't "know" facts the way a person does — it generates responses based on patterns in its training data, and it can misunderstand an unclear question or simply make things up. This is why the teacher's professional judgement always needs to stay in control: AI can support your work, but it should never replace your own verification.
"The wording of your prompt makes very little difference to the quality of the AI's response — the output is mostly a matter of luck."
The clearer and more specific your instructions, the better the response you'll get. Asking AI to "explain osmosis" gives a generic answer, but adding context — the audience, the tone, the persona, the purpose — reshapes the output dramatically every time, even though the underlying topic stays the same.
"AI has equally rich and accurate information about every text, so a prompt about a well-known international play and a prompt about a local South African setwork are just as likely to be reliable."
AI is trained on huge amounts of internet data, so it tends to have far more information on widely studied international works — like a Shakespeare play — than on lesser-documented local South African setworks. For that content, outputs are more likely to be thin or inaccurate, so it's especially important to check them against your own knowledge of the text and your learners' context.