Kamal Ghorui
Textpocalypse is near, and it has put authenticity in crisis. If any text can be generated by AI, how do we know what is authored ? Today, LLMs are becoming better and better at writing. Tomorrow, they may do something more. Right now, as Princeton researchers have established, AI systems mirror the human tendency to put a lower value on AI-generated content. One day, the machines might ditch this hierarchy. Humans might become agnostic about authorship. But that day is not here yet. Artificial hearts and knees are well and fine, but if someone’s caught passing off artificial words as natural ones, all hell can break loose.
As Mia Ballard has just found out. In what is being called the landmark, first AI scandal to hit the Big Five of publishing houses, Hachette has pulled her horror novel, Shy Girl . Because the internet said, I spy AI has written this, in lines both neat and algorithm-kissed. But by no means is such Sherlocking foolproof. All the chatbotting we do, is changing our cadences, after all. Not unlike how Coleridge wrote Kubla Khan , when he was, as is politely described, under the influence. There is no way of knowing that those who say they are being totally original, are actually being so. Shakespeare, some pundits say, wasn’t, but he also was. It’s complicated.
In his essay, Why AI will never supplant human novelists – or win the Booker Prize , Ian Leslie answers this, in terms of the fundamental reality of being alive in the world. It’s this that lets a novel about dragons feel deeply authentic. This is where authorship is more than a legal or commercial concern. We are engaging in a relationship with the consciousness behind the words. Strip that away, and doesn’t literature become just an elaborate ventriloquism act? We read the stories, for the voices. They can lie, but they should not be a lie. Dear reader, this, too, is not written by AI.
END OF ARTICLE
