Mistaken mistakes

The world is filled with tens of thousands of amateur sleuths these days trying to suss out whether articles, books, term papers or speeches were written by AI. One of the biggest clues: look for the little mistakes.

This is both terrible advice and terrifying for those of us who don’t now, nor would ever use AI in our writing. Those real, human-made writings are chock full of tiny mistakes. The literary equivalent of a sixth finger on a hand in an AI image. Forget Em-dashes, if finding little errors is considered evidence, then a whole lot of us are SCREWED.

And really, how could AI even generate a typo without actually typing anything? How long before we have a word for the AI equivalent?

Ok, I had to look. Google AI says (because that’s the top response in any Google search now is their AI Overview, a view I neither asked for nor wanted) that “An AI word mistake or fabricated fact is most commonly called an hallucination.”

Great. Now our computers are hallucinating.

It used to be that the tiny mistakes made writing MORE human. And not just the stubborn typo that makes it through six rounds of edits and print proofs and review copies into the final print and bound copy so that your aunt in Duluth can send you an email about how much they liked your books “Despite the line ‘She thought often of her time in pubic school’ on page 74.” What about that human-written line of code that made it past all the beta testers and soft launches of your new app to summarily crash the user’s system once they install the brilliant new program you wrote with your own fat, unreliable fingers? Would someone accuse AI of writing that small mistake?

Mistakes can be as personal as a signature to a real writer. If I had a nickel for every time I mistyped the word “form” when I meant “from” I’d have at least tripled my total royalties from all book sales across all 35 of my books. Allegedly AI would what? Learn from that after one time? Gross.

And isn’t the reason there are small mistakes in any writing generated by AI merely a result of it learning from the myriad of mistakes it ingested while gorging on a billion stolen words as companies illegally scraped every piece of public literature to feed the beast?

They’re mocking us with our own mistakes. That’s what writing teachers in MFA programs are for!

It has been said that those who cannot do, teach. Time to retire that phrase, it seems. Those who cannot do - use AI to do it for them.

Mistakes are the most human thing a human can do. We eat, breathe, make mistakes. Some more than others - husbands, for example. We make mistakes constantly. And so do writers. Writers who also happen to be married? Walking disaster zones. Mistakes everywhere. But would my wife trade me for an AI husband?

Don’t answer that.

I think one of the reasons we’re all resigned to the fact that “AI is coming” or even that “AI is here”, is that we accepted all those human mistakes years ago. We’ve made peace with it, as uncomfortable as it makes us. We’ve conditioned ourselves to the awkwardness, to the raw truth that we are imperfect.

And it also offers a cold comfort to know that even our chip-aided overlords are imperfect too.

Not that I should be surprised. Spellcheck betrays me all the time. And what is spellcheck but the Trojan Horse of AI? We all got used to it’s friendly reminders of our terrible spelling and grammar. We even let Clippy ™ into our homes and onto our computers to “suggest” ways to improve our writings. If we had started building the walls then, maybe we could have been able to fight off the coming invasion. But no, we took the free advice. We sheepishly admitted our shortcomings and ways our middle school English course had failed us, and invited in artificial assistance.

And then what does it do? It lets me write that “the desert was quiet tasty”. So instead of an appetizing after-dinner treat I have sandy silence. Is it still a sentence? Technically, I guess. But a jumble of words anything calling itself “intelligence” should catch, right?

This is why I never refer to myself as intelligent. I’m intelligent enough to know better.