In an impressive way, our English and ICT teacher Ashley Davidson demonstrates again and again how to educate our students to be responsible and honest in their use of media and technology. While many teachers still complain that they can no longer identify cheating attempts and plagiarism due to ChatGPT or other AI and as a consequence revert to traditional, synchronous handwritten assessment standards under strict supervision(!), Ashley Davidson takes a far more audacious approach: When Year 9 learners were asked to write an essay on a reading, the first assignment - certainly to the astonishment of some students - was to enter the essay assignment into ChatGPT and discuss it in class.

There was certainly a lot of expectation about the result produced by the artificial intelligence. The surprise was far greater than the nine different versions of the essay: Although the teacher had entered the same assignment in the same wording several times, ChatGPT was able to generate a completely different essay for each request. With this realization, Ashley Davidson had achieved a first major learning outcome with the learners.

In the next step, learners were required to evaluate an AI-generated essay according to clear criteria. The students, of course, had to not only know but also apply the quality and evaluation criteria for the evaluation task. In a final step, they were asked to improve the essays in terms of language and content. And there was a lot to improve!

This unit made it clear to the 9th graders that you can't rely on AI at all. And it has already been a bit like that with Wikipedia or DeepL, to be honest. Those who can evaluate the content of such platforms and programs in a critical way have to be more capable.

Also of interest: At our Phorms Secondary School, we have been working for a while now with Turnitin, a plagiarism-checking software that allows teachers and learners to examine their own work for similarities. In addition to the extensive feedback studio, Turnitin will also offer the AI writing and ChatGPT detection capability starting in April 2023.