Can artificial intelligence actually write genuine poetry, or is something essential always missing?

[Marnie]

As AI language tools have become increasingly capable of producing technically competent verse in almost any requested style, from Shakespearean sonnets to modern free verse, poetry has found itself at the centre of a genuinely interesting debate about what actually makes something count as art in the first place. Supporters of AI generated poetry point out that the technology can produce work that is often technically sound, sometimes genuinely moving on a purely surface level, and increasingly difficult for a casual reader to distinguish from human authored verse in a blind comparison, particularly when it comes to shorter, more formulaic pieces.

Skeptics argue that poetry has always been valued precisely because it emerges from genuine lived experience, real suffering, real joy and real specific memory, and that a piece of writing generated purely from statistical patterns in existing text, however polished the final result might sound, is fundamentally hollow in a way that no amount of technical competence can actually disguise once a reader knows the true origin of the piece.

The debate inevitably runs into the same difficult question that has surrounded art and authenticity for a very long time, whether the value of a poem lies primarily in the finished text itself, judged purely on its own merits, or in the specific human experience and intention that produced it, something an AI system by definition simply does not and cannot actually have. Where do you land on this genuinely thorny question?