Teachers: does this scenario sound familiar to you? You have a student who gets Cs and Ds, they haven’t taken much to creative writing, and their vocabulary is fairly small and limited. Suddenly this same student is handing in assignments that are longer than necessary, and using words like “voluminous” and “anti-authoritarian.” Clearly they are using chat-gpt, but how can you even call them out for this? Is there anyway to prove they aren’t doing the actual writing?

Programs that claim to check if writing comes from ChatGPT often don't work very well. They try to spot patterns in the text that might suggest it was generated by an AI, but human language is so complex that it's tough for these tools to be accurate. As AI writing improves and starts to mimic human styles more closely, it becomes even harder to tell the difference. So, while these tools can give some hints, they aren’t reliable for figuring out who really wrote the text.

In the above example you can see a popular "chat-gpt" checker actually reports the constitution of the USA as plagiarized from chat-gpt.

So what can actually be done? How can we stop students from using Chat GPT and ensure they are actually learning real writing skills. Well fortunately, there are two foolproof methods.

The first is to have your students write in-class. You can block Chat-GPT on school computers, and ensure they are actually writing. Simple enough!

The second, is to have your students write in Google Docs text documents, and submit you a link. The way Google Docs works is you have a history of what is being typed, as it is being typed - and it’s a simple matter to go through the revision history to see if any large chunks of text have been pasted into the document. Foolproof!