Key to so many *Star Trek* plots and the great sci-fi films of my childhood, from *The Terminator *to *The Matrix* to *Bladerunner, *is the conceit of living in a world where it’s impossible to tell who is real and who is a simulation. While we *probably *don’t yet have humanoid intelligences that walk among us undetected, we do have chatbots; increasingly, they can pass themselves off as humans—and yes, even as poets. In April of 2023, I interviewed computer scientist and machine-learning expert Kristian Kime, who holds a PhD from Brandeis University. That [interview](https://www.youtube.com/watch?v=d_IZW4ejD7Y) is available on my YouTube channel and breaks down how large-language models work and what kind of thinking they can perform—or seem to perform. Since that time, staggering investments in AI have occurred, with Sung Cho of Goldman Sachs characterizing it as a [trillion-dollar tide](https://www.goldmansachs.com/insights/articles/will-the-1-trillion-of-generative-ai-investment-pay-off) of cash. As I write this, I am midway through releasing a follow-up [series of videos](https://www.youtube.com/playlist?list=PLzsfifIHOGSb2gYaklX2m2d2G3T9DqsON) surveying popular AI models and comparing their abilities at generating poetry. I look at blank verse, the villanelle, and free-verse prompts. Spoiler alert: the machines have gotten better, particularly at free verse. I do not believe that even an experienced poetry editor can reliably determine whether a poem is AI-generated or not. Stated another way, some submitters are certainly using AI, irrespective of editorial guidelines, and journals are likely getting hoodwinked. #@callout Whether our procedure of reading blind hinders or encourages machine-generated content is also unknown Here at *The Pierian,* we are discussing the implications and planning next steps. We have no policy at the moment on machine-generated content, and it seems counterproductive to promulgate unenforceable guidelines. Whether our procedure of reading blind hinders or encourages machine-generated content is also unknown. One might assume that elite journals perform additional checks based on the cover letter or simply discard those with dubious credentials. But preferring the credentials over the work has never been good policy. Career poets have a much greater incentive to cheat—their advancement is constantly in question, and the pressure to produce is unrelenting. Large language models are predictive: they generate words, based on their training, that their models indicate most commonly follow, word by word, left to right. In other words, they try to please. They incent us to choose their model over the competitors, thus pleasing their investors. There are many things these models cannot do, including determining what function poetry should play in society, if any. Only examining poetry’s past—and exploring its traditions—can tell us that. These models have no sense of the past, because they have no age, no day nor night, and no season. They are frozen in a single point-in-time of their training data. If generated poetry feels so threatening, it’s because it disrupts the contemporary role of poet-as-specialist, the guildsman who remains aloof from the rest of humanity and whose only obligation to it is to produce unique, if increasingly irrelevant, artifacts. One possible takeaway from *The Matrix* and its sci-fi brethren is that AI-generated hallucinations also confirm an ancient, Platonic fear that all poetry and art offer is deception.