Crud Meridian

I cannot really articulate how badly I do not want to write about AI. Or, at least, AI in its current newsworthy incarnation. I would love nothing more than to tell you about Jude Law’s delightful turn-of-the-millenium role as a polymer-sheened sex robot. But you don’t want to hear about Gigolo Joe, because you’re dead inside. You want a steroidal spellchecker that’s read all of Mein Kampf. “What if Microsoft Clippy dreamt of nothing but fire?” we asked, and so born was our answer. 

I can’t really articulate anything at all particularly well these days. You might be wondering why you haven’t read a brutally indulgent and cutting piece of satire in the last few months. Well, I’ve been busy lying on the floor. I’ve been busy staring at the wall. I’ve been busy watching House MD. My brain is full of fog and darkness the way only a real human being’s brain could be. “Connor,” you ask, “are you depressed?” to which I can only say “How did you get in here?” To which I can only say “How did you get anything?” 

I am at a loss about what to say, and I’m saying it. Large-Language Artificial Intelligence models are designed to create a facsimile of presence and intent by processing massive amounts of data about the world and then spitting out a statistically likely reply to any given prompt. When they answer “4” to “what is 2+2?” you must remember that they have not done the math. Underneath any given request to ChatGPT is the subtext of “What did the billion monkeys type?” and, in comparison, I am only one monkey. Otherwise, I am doing the same thing as ChatGPT – I am responding to the world with a series of expressions designed to convince you that I’m Here, that I Understand. That in my heart of hearts I can Know The World in a way that is substantially different from simply Delineating the Distance Between Its Parts. I am trying to convince myself of this in much the same way as I try to convince you. When you ask me “what’s the difference?” I can only offer that I am usefully maladapted in a way that ChatGPT can, by design, never be. It may have read every Jacobean Revenge Play in existence, but I’ve been saying “Increbidle” under my breath and quietly giggling for the past three months. ChatGPT cannot be stupid, and this is how we shall defeat it.

Here is the short of it: Reducing the majority of human expressions to their most likely versions is an intellectual heat-death, a flattening of all expression to the line of best fit. Language lives because we do it badly and weirdly, over and over and over and over and over. That I have remembered the horse_ebooks tweet “everything happens so much” since I read it over a decade ago is a testament not just to the strange saliency of the wording but to its profound mutancy. There – again. Mutancy isn’t a word. You’re welcome. The prescriptivists amongst us have underlined it in red, but these are the same perverts who would capitalize e.e. cummings and deny Dril his Nobel Prize. We can’t let them win, and their offer to write our emails is a crude little siren that does not, in fact, find me well. 

When you send me the ransom, I want to see the way your ink-stained fingers tore at each letter of the magazine. It’s how I’ll know you love me.