Isaac Asimov’s fantastic short story ‘The Last Question’ has always struck me as vaguely implausible, not because of its depictions of the next trillion-or-so years of human evolution and civilisation, nor because of its wonderful twist, but simply because of what the ‘last question’, the hardest problem for the story’s artificial intelligence to solve, is. Tritely put, I thought that the ‘last question’ would be that of the meaning of life. Interestingly, Asimov disagrees with me. The last question was a scientific one, rather than a more philosophical one, because the thought that the latter would be entirely outside of the purview of artificial intelligence.
But current iterations of artificial intelligence are far from the masters of logic that Asimov imagined. ChatGPT can explain, in detail, how it is possible never to lose a game of noughts and crosses but, when asked to put this into practise, it plays with about as much skill as a toddler. What they are good at is usurping creativity and human thought with thoughtless knockoffs. This is an attack which should be resisted.
With this in mind, I admit to being baffled by reactions to the change in the format of finals examinations in modern languages. That the previous format of entirely open-book examinations is not practicable in the current age of artificial intelligence is obvious. There is, unfortunately, nothing stopping a student struggling in an exam from loading up ChatGPT and using it to plan or write an essay. Software which purports to detect AI-generated writing churns out far too many false positives to be reliable. And the academic arms race promoted by examinations means that any come edge, no matter how unscrupulous, will be taken by some. To allow this to go on harms both those who cheat and those who do not. Those who choose to cheat, by shouldering their preferred large language model with as much work as they can, surrender their thoughts to the mindless convulsions of an algorithm; they fail to develop the essential skills which a degree is supposed to foster. Those who choose not to will be at an undeniable disadvantage; their grades will suffer.
This raises an obvious question. If artificial intelligence really would improve people’s performance, should we not be teaching and encouraging students to use it in a productive manner? Plausibly. As long as one is not outsourcing one’s own thoughts to an artificial intelligence I see no real argument against its use, though, given its tendency to be confidently wrong, I have little faith in its research skills.
When it comes time for exams, however, the options on the table are closed book or open book. One protects essential and important skills whilst, admittedly, underpreparing students or the age of artificial intelligence. The other allows students to ignore and underdevelop these essential skills in favour of short-term gains in their marks. People who argue that this decision fails properly to prepare students for the future overlook the timeless skills that it is designed to protect and take their rightful place at the front of the queue of people ready to be replaced by computers. They are, as Milan Kundera put it, the allies of their own gravediggers.
I assume that, in many cases, the reactions are rationalised rather than rational. It is frustrating to get half, or three quarters, of the way through your degree only then to discover that you will not be able to flick your way through your notes if you forget a source or a quotation in the exam – or to learn that you are going to have to reacquaint yourself with the technology of a bygone era: the pen. The problem comes when such frustration is reimagined to be what it is not: a genuine critique of closed book exams. That in-person exams prioritise ‘outdated’ skills like memorisation is obviously a weak argument.
Memorisation is not outdated but nor is it the most important skill being protected by in- person exams. To risk sounding like an egghead, this is a strawman. I assume that what is secretly being said is ‘memorising material is such an unnecessary drag’. I sympathise. But this is not a principled stance and it should not be allowed to masquerade as one.
