ChatGPT is an artificial intelligence site which provides good quality, free-of-charge answers to the sorts of questions teachers and university lecturers might ask. ChatGPT was created by OpenAI, an independent artificial intelligence research foundation co-founded by Elon Musk in 2015. Released last November, OpenAI’s chatbot is able to create stunningly human-like responses to a wide range of questions and various writing prompts. ChatGPT is trained on a large sample of text taken from the internet. It is quite hard to detect if someone has used ChatGPT for their answers.
I have given ChatGPT a few test questions and was stunned by how quick and clear it was.
So that is, or shortly will be, the end of non-supervised coursework. And the real reason is this – that many school-and-university age students cannot resist cheating. Believe me. Quite often they cheat because they know that rank orders matter and they fear the other students will cheat, pushing them down the ranks.
We can try to ban the use of AI, in the way New York State has now banned use of ChatGPT in schools. We can threaten students with punishment if they use AI for answers. But these are doomed to fail. Something you can do on your phone, undetected, cannot be policed.
All those people who deplore regular end-of-course exams which are designed to test knowledge are going to be very disappointed. Because the most obvious alternative to traditional exams is coursework. In the future, coursework will have to be supervised in a classroom, but that is horribly boring. as we found in schools before 2015 when there was lots of it about.
There are a number of other reasons why Ofqual took a scythe to GCSE coursework in 2015: there was widespread coursework ‘help’ given by schools and parents; the coursework was often dull and limiting; and far too many pupils did brilliantly in the coursework but terribly in the written exams, suggesting that something dodgy was going on.
Of course there are other forms of cheating too. University students who are allowed to write exam answers from home and simply exchange drafts with each other. Students who are given 12 hours to write a three-hour online paper because ‘they live in different time zones’ or ‘some of them have paid jobs in the day’.
But all this pales into insignificance compared to the impact of AI on coursework, at both school and university level.
So what this means is that educational assessment should focus on traditional, rigorous exams set under ‘exam conditions’ in properly invigilated exam halls. This is the fairest system, and fairness matters as much as anything in assessment.
Traditional exams have another great virtue – unless they are ‘open book’ exams (a crazy idea) they force students to memorise things. Medical students, for example, are forced to take traditional exams because, yes, we want doctors to know things. Knowing things is what proper education is about. Chatboxes and AI undermine the willingness of humans to memorise things and are, in this respect, very damaging.
By Professor Barnaby Lenon
Dean of Education, University of Buckingham
Is a test of memory really the best way to test someone’s understanding?
There has to be better ways!