AI Detectors or Art Detestors? Are AI Checkers Reliable?

AI Detectors or Art Detestors? I’ve noticed a recent craze among Soupers: an increasing, almost reverent, reliance on AI detectors to judge the outcomes of poetry contests. The argument often made is that these detectors are among the best tools humans have developed to unmask identity thieves — those who supposedly use AI-generated poems to win laurels. To them, that amounts to dishonesty and open cheating. But what they fail to understand is this: AI detectors frequently flag poems as AI-generated not because they are fake, but because they are written with near-perfection. These tools are programmed to raise red flags on poems that heavily utilize poetic devices — metaphor, rhyme, alliteration, and other classical techniques — since such devices are common in the very same classic poems on which AI models were trained. I’ve personally experienced this many times. 

Poems I wrote 100% by myself have been classified by detectors as 100% AI-generated. I doubt I’m alone in this experience. It makes me believe that relying solely on AI detectors to distinguish human from machine may be — in the end — a spurious endeavor. Let’s be clear: there’s no way an artificial intelligence created by humans can become more intelligent than the humans who created it. That would be like saying humans are now wiser than the God who made them. This kind of thinking, if left unchecked, risks replacing human discernment with blind algorithmic judgment. If this trend isn’t curbed, we may soon see wishy-washy poems dominating winners’ lists — not just on PoetrySoup.com, but across other poetry platforms and magazines that rely blindly on these so-called “all-knowing” judges. 

This age may well go down in history as the era when substandard poems took precedence over superior ones. To those already hooked on this tech — tread with caution. To those considering joining the craze — look before you leap.