We live in a silicon age of nonsense, the unnatural byproduct of computational creativity. As artificial intelligence churns its way toward writing like us, it casts off reams of gobbledygook. Is it any good?
To figure that out, we need a Turing test for twaddle. And the Alan Turing of nonsense, the closeted eccentric English genius who invented a whole field of thought, is Edward Lear (1812-1888). As he described himself in the poem How Pleasant to Know Mr. Lear!:
His mind is concrete and fastidious,
His nose is remarkably big,
His visage is more or less hideous,
His beard it resembles a wig.
Lear wrote The Owl and the Pussycat, introduced the world to the limerick, and developed a certain flavour of frippery that echoed through the next hundred years of culture. In 1964, poet Howard Moss explained that Lear’s nonsense is not senseless. Rather, it can “hold the plausible and the implausible in tension” and find in the absurd “an entertainment, a release, and a form of criticism.” Training your cat to walk across the keyboard won’t cut it.
Helpfully, Moss lists all the ingredients for premium nonsense. In your mise en place, you’ll need:
1. “The ability to think and feel like a child.”
2. “A highly developed sense of language.”
Do the machines measure up? It’s undeniable that their sense of language is highly developed, as thousands of software engineers have worked through terabytes of prose to develop it.
It’s at the ability to think and feel like a child that we plunge into the uncanny valley. Natural language processing may be in its infancy but that’s where the metaphor stops. It thinks like a machine and it doesn’t feel. Where we unconsciously expect a child-like sense of play, we get a robot. And that, as anyone who’s ever seen a horror movie with an evil kid in it knows, is unsettling.
The proof of this came when GPT-3, the most advanced neural network for free-form writing, set its unblinking red eye upon The Owl and the Pussycat. The original work, once voted Britain’s most popular childhood poem, ends with an avian-feline marriage party:
They dined on mince, and slices of quince,
Which they ate with a runcible spoon;
And hand in hand, on the edge of the sand,
They danced by the light of the moon,
They danced by the light of the moon.
When the technology savant behind Gwern.net asked for a fourth stanza, GPT-3 produced this:
And the Owl and the Pussy-Cat, they lived together
In perfect harmony,
Till the Owl, one cold winter’s night
Sank his talons in Pussy’s thigh
And then she screamed, and she shouted, and she cried,
“O’Fuffle! O’Fuffle! Fuffle-ee!”
But, alas! It was too late, for the Owl was dead;
And Pussy said, “Too bad,
You are the one who is too bad!”
Way to ruin a childhood classic, but top marks for form. The human formula for tiptop claptrap suggests that for the time being, we should focus the machines on sorting Japanese pastries, keep them away from our cats, and leave the whimsy to humans. Make them make sense; stop making nonsense.
Quick quips; lightning
“A nose that can see is worth two that sniff.”
— Eugene Ionesco
“In larger things we are convivial
What causes trouble is the trivial.”
— Richard Armour
“Here’s a good rule of thumb:
Too clever is dumb.”
— Ogden Nash, “Reflection on Ingenuity.”
Get Wit Quick No. 97 was generated by a runcible human brain clocking in at about 1.3 kilograms, though to be fair much of that is water weight. My book Elements of Wit: Mastering The Art of Being Interesting was concrete, fastidious, and wig-like. Ward off murderous digital owls by tapping the ❤️ below.