When DALL-E 3 launched last year, OpenAI put out a hype video of someone bringing their kid’s imagination to life. “My 5 year-old keeps talking about a ‘super-duper sunflower hedgehog’ — what does it look like?”
DALL-E comes up with a few options, one of which the kid decides to name Larry. More back and forth: We see Larry’s house. We learn what makes him “super-duper”: his kind-heartedness and his sunflower-petal quills. Eventually, ChatGPT turns Larry into a bedtime story, complete with illustrations.
Never mind the cloying schmaltz (“Word of Larry’s kind deeds had spread, and one by one, animals from all corners came to thank him, singing songs of gratitude and love”). Never mind the fact that no two images of Larry showed him looking the same. It was interesting anyway, because how cool is it to take what’s in your head and make a picture of it in seconds?
Reader, it is not cool. Not cool in any way. Because the inevitable evolution of this capability is bullshit.
Let me explain.
My Kindle’s lock screen is, of course, an ad platform. A few months ago, I noticed some unusual ads on it.
The first was for “Unlocking the Power of Mindfulness.” At a glance, it looked like any other cheesy self-help guide. A closer look showed something different. This “book” was written by “Whuk Sinte” and “Japey Addenrnn.” The subtitle was, “Sftgix of the pogrkess and veluing bokdj aidir them toward sellf-improvement and success.”
Next: “Sustainable Living: Simple Steps to a Greener Lifestyle.” The description: “Easy-to-implement tips for reducing waste and comccting energy, conservly, eco-water energy, eco-friendly shopping, and sustiainable cooking.”
Because who doesn’t want to comcct energy and conservly?
“The Enigma,” a “young adiult mystery adventure novel” written by “Collge Festers,” had the subhead “of wot hok gap treasiup novebr.” “The Art of Communication” — oh, the irony! — lets you master “convenications” in personal and professional life.
Before you tell me that there’s a way to turn off ads on the Kindle: I know. And that’s not the point. The point is that gen AI is flooding the zone with bullshit.
The seminal work on bullshit was published first as an essay and later as a book by Harry Frankfurt. Frankfurt defines bullshit as speech delivered without regard for truth. Where a liar intends to deceive, a bullshitter doesn’t care.
“[B]ullshit itself,” says Frankfurt, “is invariably produced in a careless or self-indulgent manner, …is never finely crafted, …in the making of it there is never [a] meticulously attentive concern with detail… Is the bullshitter by his very nature a mindless slob? Is his product necessarily messy or unrefined? The word shit does, to be sure, suggest this. Excrement is not designed or crafted at all; it is merely emitted, or dumped.” (Emphasis mine.)
These AI-generated books are not designed or crafted; they are merely emitted, or dumped. And the ease and speed of their emission create a massive problem for Amazon’s Kindle business.
Amazon is not a traditional publisher that reviews and selects books on merit. It is an aggregator, generating value by having zero transaction costs. “Self-publish easily. Publish print and digital formats in three simple steps, and see your book appear on Amazon stores around the world in 72 hours.”
When content is expensive to create, the filters can be low. When content itself has zero transactional costs, low filters are a recipe for disaster. No-cost stuff equals too much stuff.
Amazon, LinkedIn, Wikipedia, even scientific papers are getting flooded with AI bullshit. If we don’t act fast, we’ll drown in it.
Ngā mihi maioha / warm regards,
Kaila
