Errors

The chatbots that rely on generative artificial intelligence have their uses, just as Wikipedia does. But relying on the details that either tool provides will land you in trouble eventually. In the case of AI chatbots, you’ll be in trouble sooner.

The common term for chatbot error is “hallucination,” and that’s already a dangerous term. A chatbot is technology, not a person or a character or a pet. It can’t hallucinate; it can present false claims confidently and with an air of authority. The most recent chatbot models continue to reply to queries with errors; this is not a problem only with earlier versions.

I have a laser focus that you’re not going to believe. I’ve had this superpower even before I became a writer. But non-writers think I’m exaggerating. If you’re a savant, it’s no big deal to you that you can count how many toothpicks fell out of a box onto the floor in just one glance. But if you’re not, it seems impossible.

Yesterday, I picked up a medical brochure in a doctor’s waiting room. I skimmed the front cover without really reading it, then I flipped the brochure over… and instantly saw the error: “indiviual.”

During the course of a career lasting over 20 years, would you care to guess how many times I have been given data from a software developer that later turned out not to be correct?

A maximum of three.

Once, another developer informed me that something I had written wasn’t true. I provided the developer with a copy of the e-mail in which a different developer had said it was true. (No, I didn’t rat out the original developer by providing their name.) The developer demonstrated to me that it could not possibly be true. So I changed the docs.