In November a private software company, OpenAI, launched its new chatbot – ChatGPT3. It’s a blast. It can write limericks and poems, stories, and it can do multiple choice tests and write computer code. I asked it whether exams or assignments were better forms of assessments.

I was amazed at the result.

It sat on the fence! It says exams are good because they can cover the full range of material taught. They can help students to focus and prepare for a comprehensive test. It suggested that good exams can be more objective than assignments as they work to pre-determined answers and expectations. On the other hand, it pointed out that exams were short and that assignments allowed students more time to develop their arguments and helped students avoid pressures.

Some say that sitting on the fence isn’t easy as it ultimately becomes uncomfortable but I feel the chatbot is right. We need to retain a mix of exams and other forms of assessments. I’m bewildered by those who rail against exams – saying they are quaint, oldfashioned, inauthentic, or unfair, while overlooking the problems with take home assignments.

Here are a few personal comments that the chatbot didn’t unearth. To me the prospect of an exam always ensured that I would commit knowledge to memory – the whole or at least most of the syllabus. Some say that’s unnecessary these days. We are in an age where we can always look things up – but are we? Can you look things up during a job interview? Can you look things up when you are in a meeting trying to persuade your colleagues?

Can we look things up when we meet new people and form new friendships built on shared interests and knowledge? And can we look things up when we are daydreaming and skipping from one idea to the next as we develop our frameworks for understanding the world? Another thing I liked about exams was that they kept assessment separate from learning. I loved learning. I enjoyed lectures, tutorials, and laboratory classes. But I felt under pressure when continuous assessment intruded in the form of in-class quizzes or practicals that were marked on the spot. Assessments now more and more intrude into learning spaces and contaminate them. We had some fairly competitive students and the in-class assessments also introduced an element of competition that would have been better left till the end of term. Overall, it is hard to say whether exams were better or worse for mental health. I think it depended on the student.

As a lecturer I’ve set plenty of exams and assignments and have one observation. In the exams the students tended to get marks from zero to 100. But in the assignments, marks ranged from about 60 to 90. Why was that? I tried hard with rubrics and standard marking but for me the two processes delivered different results. Ultimately, I trusted the exam results more.

Post-pandemic there are reports that more cheating is being detected in online, remote assessments. You’ll hear that students also cheat on exams in halls but it isn’t at the same scale – sneaking off to the bathroom to check the formula ‘force equals mass by acceleration’ isn’t the same as outsourcing the whole exam to ChatGPT or to friends. No system is perfect but in hall or viva voce exams do provide a good deal of confidence and are important when we need to certify learning.

And we do need to certify learning. Not just for society but for the students themselves. To me perhaps the best thing about exams was that they marked the end of the year properly. I’m a great believer in building resilience and confidence in students. One has to support students but if you can set fair exams and help the students to prepare there is nothing better for their confidence than having gone through and done it, under some pressure, all by themselves.

As we move past the pandemic and reflect on all the innovations and opportunities, let’s set ourselves a test – to think deeply about what is worth adopting, what is worth retaining, and what is worth abandoning!