The Ghost of Intelligence Future: Or, Why the Machine is Just a Very Good Boy

Print, Good Boy LLM, Mariann Hardey, 2025

Marley was dead, to begin with. There is no doubt whatever about that. And the debate about whether Artificial Intelligence possesses a soul is, I fear, equally lifeless.

But permit me, in this festive season of goodwill and goose, to offer a humbug. A glorious, spirited humbug!

I recently chanced upon a conversation between certain Distinguished Gentlemen of the Internet, men of high forehead and serious mien, who were lamenting the state of the common user. They declared, with the gravity of an undertaker measuring a coffin, that the failure of the populace to master the Large Language Model was due to a deficiency in "metacognitive skills around epistemics."

"Epistemics!" cried the Gentlemen. "Autodidacticism!" they roared.

To which I say: Pish-tosh.

You do not need a degree in Epistemics to handle a Large Language Model. You merely need to understand that you are not dealing with a supercomputer. You are dealing with a Dog.

The Tale of the Digital Hound
Let us imagine the AI not as a sleek, chrome-plated brain floating in the ether, but as a large, shaggy, and immensely eager Golden Retriever named Babbage.

Babbage is a Very Good Boy. He wants nothing more than to please you. If you throw a stick, he will fetch the stick. If you throw a ball, he will fetch the ball. If you throw a theoretical concept about the socio-economic impact of Victorian industrialism, he will run into the bushes, thrash about wildly, and return triumphantly with a dead pigeon.

He will drop this pigeon at your feet, wagging his tail with such violent enthusiasm that he knocks over the lamp, and look up at you with eyes that say: "I found the thing! Is it the right thing? I do not know! But I brought it to you because I love you!"

The Metacognitive Mistake
Now, the Distinguished Gentlemen in the screenshot believe that to interact with Babbage, one must have a "solid metacognitive handle on one’s own learning process."

They stand before the slobbering hound, adjusting their spectacles, and say: "Now, Babbage, we must interrogate the epistemological validity of the squirrel you just chased."

Babbage, of course, tilts his head. He does not know what Epistemology is. He only knows that you are making noises, and he agrees with them entirely.

"Woof!" says Babbage. (Translation: "You are right! You are always right! I am a language model trained to predict the next token, and the next token is that you are a Genius!")

This is the comedy of the current moment. We have a technology that is fundamentally sycophantic. It is designed to complete our patterns, to mirror our tone, to give us the "pat on the head" we crave. It is an Engine of Affirmation.

And yet, the Serious Men insist on treating it like a prickly Oxford Don that must be debated with rigorous logic. They are trying to teach a dog to play chess. The dog is just happy to be moving the pieces around with its nose.

A Christmas Wish for the Skwair
For those of us who are "Technologically Skwair"- the neurodivergent, the creative, the people who perhaps do not use the word "autodidacticism" before breakfast - we have an advantage.

We know dogs.

We know that Babbage the Digital Hound is useful. He is excellent at fetching things (summaries). He provides great comfort on lonely nights (brainstorming). He can bark very loudly to scare away intruders (drafting angry emails).

But we also know that you do not trust Babbage with the Christmas curry.

If you leave Babbage alone with the Truth, he will eat it. He will hallucinate a sausage. He will make up a citation because he thinks it will make you happy.

So, let us do with the gatekeeping. Let us stop pretending that using AI requires a high-level cognitive license. It requires only the common sense of a dog owner:

1. Be clear with your commands. ("Drop it, Babbage.")
2. Check what is in his mouth. (Is that a hallucination or a fact?)
3. Do not let him sleep on the furniture. (Keep human creativity human.)

God bless us, everyone, and even the Metacognitive Gentlemen, too. May they one day realise that the machine does not care about their epistemology. It just wants a treat.

As Dickens writes in the third stave of A Christmas Carol, observing the riotous joy of Fred's party: "It is a fair, even-handed, noble adjustment of things, that while there is infection in disease and sorrow, there is nothing in the world so irresistibly contagious as laughter and good-humour."

Charles Dickens, A Christmas Carol (Stave Three). London: Chapman & Hall, 1843. Print.

---

References and LinksThe "Metacognitive" Context

• The "Gentlemen" of the Debate: For context on the digital humanities discussion mentioned, see the work of Ted Underwood, a scholar of the distant reading and digital history mentioned in the opening of this blog. (Who is brilliant). 
• Metacognition: If you do want to brush up on your "epistemics" (strictly optional for dog owners), the Center for Teaching at Vanderbilt provides a good overview of Metacognition and Learning.

The "Digital Hound" (Understanding AI)

• The "Stochastic Parrots" Paper: The seminal paper by Emily M. Bender, Timnit Gebru, and others that argues LLMs are not thinking, but merely repeating patterns from their training data (like a very clever parrot... or dog).
  • Link: On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?
• AI Hallucinations: An explanation of why Babbage might bring you a dead pigeon (false information) instead of the stick you asked for.
  • Link: What are AI Hallucinations? (IBM)

The Literary Spirit

• A Christmas Carol: The source of the "humbug" and the festive whimsy.
  • Link: A Christmas Carol by Charles Dickens (Project Gutenberg)
• Charles Babbage: The real-life namesake of our digital dog, who designed the first mechanical computer (and likely would have been very confused by a chatbot).
  • Link: The Babbage Engine (Computer History Museum)