Does it rhyme? AI, Poetry, Scholarship and Schools
Dealing with complexity
I know very little about poetry, apart from my A-level studies focusing on some of the ‘greats’. John Donne still remains a favourite, and I can recite a few lines here and there to sprinkle something extra to an explanation given in a lesson. Yet, I recall asking someone, half-jokingly, who teaches and writes poetry, if their poems rhyme.* They smiled and gently explained how poetry takes different forms and that there were many brilliant poems that had been produced by contemporary writers. They were just not taught in schools.
I recalled this embarrassing and humbling conversation when I read a recent study that found that non-experts struggle to tell the difference between AI-generated poetry and human-written classics (which the AI had likely been trained on). Not only could they not tell the difference, they seemed to prefer the AI-generated versions. One conclusion drawn from the study is that it raises questions about creativity, but for me, it highlights an open problem in schools about how we teach students to value depth and complexity in art and other subject areas.
Jon Stone argues that the lack of familiarity with any artform severely limits our ability to get the most out of it. Moreover, it is not just a passing interaction with it, but sustained engagement over time. Yet in many schools, poetry has been reduced to a tick-box exercise of identifying metaphors and counting syllables. Students learn to spot technical devices but rarely engage with the historical context, cultural significance, or layered meanings that make poetry worth returning to again and again. That is not to lay the blame solely on poor teaching. Examination courses demand certain aspects to be studied and quantified.
As a result, we do not create expertise and then we ask non-experts to make judgements about what is ‘good’. This may sound familiar. Michael Gove, prior to the Brexit referendum, suggested that the people of the UK have had enough of experts. There is an understandable whiff of populism in this view because being an expert sets you apart from others and can be viewed as being snobbish. The discussion about Rupi Kaur’s work is one example of this. While expertise may not enjoy high public perception, we certainly find it useful when it matters to us. There’s a reason why parents move closer to schools they think will have the expertise to support their child or pay for tuition and school fees to guarantee the right help.
I don’t want to detract from the technical achievement of AI generating convincing poetry. We shouldn’t. It is impressive. However, I think there is fundamental question raised by the paper: how do we develop deeper appreciation for artistic complexity in our schools and society? One possible answer, which requires time, effort and human engagement, means working with actual poets, with difficult texts and with historical context. It means actively valuing and investing in scholarly expertise, a key aspect of what a good education should provide. The study, impressive as its findings about generative AI capabilities are, ultimately holds up a mirror to our own relationship with expertise and complexity in schools and society. I’m not sure we can truly be satisfied with the reflection.
*I ended up marrying them.