By Bernard Andrews, Philosophy Teacher
A spectre is haunting education—the spectre of artificial intelligence.
Will my students’ term essays be the fruits of genuine thinking or ChatGPT’s cold algorithms? Will AI ‘take over’, make us redundant? Every advance sees the collapse of another bastion of human uniqueness: chess, art, poetry, essays, music. What is education for, if computers can do it all better?
But whilst I don’t doubt that these tools are powerful and potentially dangerous, the belief that we are in competition with Artificial Intelligence is misplaced. We only believe we are in competition because the intelligence that our system reveres is artificial, the knowledge our system encourages us to teach is artificial.
ChatGPT is the, albeit painful, disinfecting sunlight on a diseased conception of the human condition. It illuminates what makes us truly unique and offers us an opportunity to reorient education as a distinctly human endeavour.
AI is the culmination of a seventeenth century project. Descartes’ obsession with automata, and Hobbes’ proclamation that ‘man is a machine’ kicked off a trend of modelling human nature in terms of the fanciest technology of the day: the human brain has been compared to a magnetic compass (1877), a church organ (1906), a radio (1937), and a telephone exchange (1943). Modern cognitive psychologists often describe the mind as an ‘information processor’.
Cognitive psychology was a postwar attempt to bring back mind and meaning to psychology after behaviourism’s ‘long cold winter’. But even Jerome Bruner, a founder of this movement, confessed that much cognitive psychology had failed, becoming an equally dehumanising computational psychology.
This mechanical view of the human condition reconceived knowledge as mere data and programming. So it’s no surprise that this is what our qualifications assess. And in a society drunk on competition, students find themselves in a frantic race to be most computer-like. They believe that to be educated is to be the passive recipient of data.
Consequently, exams have no inherent value, and the certificates are now far more valuable than the learning required to achieve them. A colleague asked a student if they would take an A* at A-Level without learning the subject. ‘Obviously!’ the student replied. The juice frankly isn’t worth the squeeze.
Knowledge is a uniquely human virtue
We must see the challenge of AI as an opportunity to reacquaint ourselves with knowledge —and remember why it is a uniquely human virtue.
Firstly, information and programming can never suffice for knowledge. A student may memorise the entirety of wikipedia, be able to recite it on demand, but still be knowledgeable only in a very narrow sense. No amount of data can guarantee that this information will be applied knowledgeably.
Secondly, knowledge can’t be reduced to stimulus-prompted algorithms because a knowledgeable action responds to the context in the right way, at the right time, for the right reasons and so on. This is the adverbial aspect of knowledge (so-called because it qualifies the action). It is precisely this that distinguishes the knowledgeable response from parrot-like repetition.
Thirdly, a knowledgeable action must be voluntary. If I accidentally give the right answer, I haven’t acted knowledgeably (even if I know the answer to be right). This separates us from mere information processors. It is illuminating that there is a legal defence of automatism that refers to an act committed without conscious volition, an automaton being a device that merely appears to operate spontaneously. A computer running AI may similarly simulate human decision-making, but it can only do what it is programmed to do and cannot refrain from doing so. Hence a computer can’t act voluntarily, can’t be culpable, and can’t be knowledgeable.
How do we cultivate intellectual virtues?
So how do our students learn when, why, where, and for what reasons to do things? How do they acquire this sensitivity to knowledge’s adverbial aspect? What else is needed besides information and rules? The answer is character. And not in a lofty Victorian pseudo-stoic sense, but simply in the sense that knowledge, along with intelligence, wisdom, understanding, and skill is an intellectual virtue, a character trait.
Our students can’t cultivate these virtues by being passive recipients of data and programming. They have to work for them. The horse must choose to drink. As Aristotle said, we ‘become builders by building and lyre players by playing the lyre; so too we become just by doing just acts, temperate by doing temperate acts, brave by doing brave acts.’ So teaching cannot consist of ‘just telling ‘em’. Our students must have opportunities to practise these virtues, to be wise, intelligent, skillful, understanding, and knowledgeable. Our job is not to make things easy, but appropriately hard – to challenge, debate, and discuss.
Standardisation vs particularisation
But this talk of character and virtues may seem dangerously subjective. Where we struggle to standardise, for example, in humanities subjects, we get wild variations in marks. And if we learnt anything from the Teacher Assessed Grades of Covid it was that teachers have their biases. So we must face up to that perennial tension in education between standardisation and particularisation.
The problem is, that in trying to standardise judgements, we invariably end up standardising responses instead. But a knowledgeable response is context-specific, so we end up lowering standards. (This is manifest in our degraded idea of examination. In schools, we think an examination is what a student does – which is odd. An examination is an investigation, a weighing up of evidence and so on —it is what the examiner does, not the examined.)
We must remind ourselves of what it means to examine and develop our skills. When we say a student is ‘good at history’, or ‘bad at philosophy’ what do we mean?
Firstly, because of the adverbial, context-dependent nature of knowledge and intelligence, we can only refine our judgements through acquaintance with cases, not through acquaintance with rubrics, frameworks, or maxims. We can only legislate for the minimum standards.
But these cases are not samples, either in the sense of a specimen for testing, nor in the sense of providing us with a general rule: ‘What a good one looks like’ (WAGOLL) is misleading in the sense that the example doesn’t tell us what’s habitually good (not least because if you thoughtlessly made a copy, it wouldn’t be a ‘good one’). A better title would be ‘this was a good one’.
Secondly, we can standardise by comparing cases and applying precedents. We can’t generalise at the level of individual acts or specimens, but we can at the level of dispositions and virtues. We sharpen our judgement of knowledge by seeing many and varied cases. (Hence, the importance of subject specialism can’t be overstated.)
Thirdly, explanations of our comparative judgement of cases can form the basis of formative assessment in the form of tips, and hints. (We must resist the temptation to systematise these or we’ll be back where we started.)
Finally, we can refine our abilities by testing ourselves. Our judgments of the students’ abilities are akin to hypotheses about how they will respond in certain situations —where they will flourish and where they will flounder. So we should see standardised tests (and for that matter assignments, questions in class and so on) not as assessments of the student’s ability, but as opportunities for us to hone our judging skills, to improve our understanding of our students.
We must shift our weight from exams and back to learning. We must make the learning inherently valuable so that using ChatGPT to write an essay would be like going to a gym and using a pallet stacker to lift weights.
Dialogue must take a central place in teaching, we acquire intellectual virtues by being intellectually virtuous. We must be our students’ sparring partners, exchanging jabs, practising combinations.
We need to renew our approach to examination and assessment in the classroom. We can keep the external and standardised assessments but they must be subservient to knowledge. The students will pass the exams because they are knowledgeable, but they will not be knowledgeable so that they can pass exams. We must refine our judgements through a large, diverse, and organic body of cases, and we must recognise that we can only judge our students if we understand them.
‘when we judge a particular action, we should consider many circumstances and the whole person who performed it, before we give the action a name’ —Michel de Montaigne