A recent study published in Scientific Reports uncovered that the average reader couldn’t tell the difference between AI-generated poetry and human-written poetry. Most of these readers even preferred the AI works to the human ones.

In the study, researchers asked Open AI’s ChatGPT-3.5 to write poems in the style of famous poets such as Walt Whitman, T.S. Eliot, Sylvia Plath, Geoffrey Chaucer, Allen Ginsberg, Emily Dickinson, and William Shakespeare. With these poems in hand, the researchers asked study participants to read 10 poems (five by a real poet, five by a robot) and identify which were fake and which were real.

The participants guessed correctly only 46 percent of the time. ChatGPT was able to imitate the style of these poets in its compositions well enough to fool the participants—although we should keep in mind that the participants were non-experts on poetry. Cognitive scientist Keith Holyoak of the University of California, Los Angeles commented, “Essentially, ChatGPT has displayed its skill as a quasi-plagiarist.”

Plagiarism is a good description for what ChatGPT actually does. We need to set aside, from the outset, all the nonsense about “intelligence” or “understanding” or “consciousness” existing in AI chatbots like ChatGPT. Philosopher Edward Feser has done a good job explaining why that’s impossible. AI robots do not create poetry in the strict meaning of the term.

Rather, these programs have been trained on vast amounts of human text (including the works of the poets listed above) and developed certain word-association maps through that process. The program uses statistical data to predict what word a human author would likely place next in a sentence. The AI, basing itself entirely on the work of real humans, can simulate human thought by playing advanced number games with words and phrases. But the AI has no understanding of what the words mean—just as my word processor has no understanding of the words I’m typing into it right now. So AI is essentially derivative and unoriginal.

Interestingly, when participants were asked to rate the poems on qualities like rhythm and originality, they tended to give higher ratings to the AI poems (as long as they didn’t know whether the poem was by a human or a robot). Researchers theorized that the robot’s poems appealed to readers more because they were easier to understand. As Sarah Kuta wrote for Smithsonian Magazine, “A.I.-generated poems cannot match the complexity of human-authored verse,” and a non-expert reader will therefore prefer the easily understood (and arguably less interesting) AI works.

This doesn’t prove that AI poems are superior to human poems; if anything, it proves the opposite. The AI’s formulaic work might be more understandable to someone who isn’t accustomed to reading poetry, but that only reveals that the AI piece lacks the tension, complexity, and mystery of the greatest poetry, which can only come from the equally mysterious human consciousness and its encounter with a meaningful reality.

Researcher Brian Porter noted that “Emily Dickinson sometimes breaks the expected rhyme scheme on purpose. But the A.I.-generated poems generated in her style never did that once.” This is because the AI simply follows certain rules or patterns blindly because that’s what it’s programmed to do. AIs do not have the freedom or intentionality of a human poet who might choose to break the rules in order to better express the nuances of emotion and experience portrayed in the poem.

So really, the study reflects more on the readers than on the poetry. We live in a society where most people have not been taught to contend with or appreciate truly complex and elevated poetic works. This is why they prefer the simplistic AI verse.

But if people like the AI stuff better, doesn’t that mean that human poets have been superseded in poetry-writing, just as human chess players were superseded by computers in winning chess games? No. Art and chess are fundamentally different. Only if you narrowly consider the end of chess as “winning” would we say that AIs supersede humans. And the purpose of art is nothing like that—it’s not about “winning” according to a certain set of rules. It’s about the human soul, the human experience, and coming to understand the world itself.

As I recently argued in a piece for The Epoch Times, genuine art can only be created by a sensing, knowing, feeling, and willing being. It can only come from a soul. An “art-like object” (such as an AI poem) with no intended meaning or emotion behind it is not real art. Computers can simulate art in the same way they simulate the stock market, the weather, or a war, but it isn’t genuine art any more than the computer’s weather model is actually the weather.

In her write-up on the study for Smithsonian, Sarah Kuta claims that the study confirms our worst fears about AI one day replacing human artists and putting them out of work. Setting aside the fact that virtually no one makes a living as a poet these days, Kuta’s claim lacks justification because it contains a false assumption about the nature of art.

Only when art is viewed strictly from a commercial, transactional, and consumerist perspective could someone believe that robots could replace human artists. In that model, yes, a robot can produce an “art-product” that people will buy or consume as readily as something produced by a human—maybe more so.

But that’s only looking at the economic side of things, and it unduly weights the readers’ experience, locating the artistic occurrence completely in the onlooker, completely subjectively.

Objectively speaking, no artistic occurrence has taken place when someone reads an AI poem because there has not been any communication of meaning, emotion, or vision from one consciousness to another. It’s completely one-sided and, in a sense, self-generated by the human viewer or reader.

It is only because we long ago lost our understanding of what art—particularly literature—really is that we could fall for the AI ruse. When literary theory veered into absurdity in the second half of the 20th century, when postmodern academics undermined the very notion of words as expressions of truth from one mind to another, we opened the door to believing in AI “art.” We had already begun to locate meaning solely in the eye of the beholder, opening ourselves up to all sorts of deceptions, and paving the way for the adulation and admiration of blind, cold, heartless machines, auto-generating words that are, strictly speaking, gibberish.