Geek House

An argument about AI, the six-fingered truth and a little extra...

“ … and they will read our diaries to find out what we were … ”

Joel and I had a fight the other day about AI. Not a debate: a fight.

I’m reading Bleak House at the moment — a novel about a protracted argument — so this seems appropriate. At least the Jarndyces were rowing over a pot of gold. We had a fight about AI. What a thing to fight about. What dorks.

I’m kidding, of course. Joel is my thought partner. With him, I think externally.

But the thing is we really do disagree about AI.

I don’t… like AI. I don’t trust it.¹ I have yet to be impressed with its output, apart from in language translation. Everything I read from ChatGPT sounds like plausible word-murals that run on and on, sentence after plausible sentence, without telling me anything. “Blah-di-bla is an important cultural phenomenon because… so-and-so is one of the most important historical figures of the 20th century… It was a time of great social change …”

Yawn af.

I am generic, benign narrator. Hear me roar whimper something bland and inoffensive into the conventional wisdom echo chamber.²

And the photo or video garbage? Joel showed me an Internet thing (believe the kids call them “memes”) of a slip-on fake finger above a six-fingered hand.

“Wear extra fingers,” it says. “Photo and video evidence will be inadmissible as it will appear to be AI-generated.”

Ha-fucking-ha.

It was supposed to be a joke. I tried to laugh. But it wasn’t funny. I could hardly breathe.

You can deep-fake me doing bad things. I can do bad things, whenever I want, and just claim it’s a deep-fake.

You’ll never know what’s real and what’s not. No one will. Maybe it won’t even matter, because the fake stuff will be more real to more people than the real stuff.

Now, Joel loves AI. ChatGPT is basically a third member of our relationship. He asks it everything: what to cook, where to eat, how to say “green beans” in French. We’ve talked about this before. We differ big time. He thinks, one day, AI will be conscious and function just like us. I don’t. Which of us is right?

Joel was raised in a homeschooling Christian family and is now the most dedicated atheist you’ll ever meet. He thinks we (humans) are nothing more than an intricate web of clicking machine parts. We perceive and think and exist and have sex and fart and blink and close our eyes to be transported elsewhere only because each atom in each cell is oiled just right and talks to all the other oily parts just so and they all click and whir in harmony, until one day — pfffft, bang — they don’t.

And we blink out, like a pixel on your flatscreen. Never to exist again.

I am a rational person. I can parse a sentence and flip a document like a pancake. But I can’t explain why this doesn’t sit right with me.

A collection of machine parts? Surely I am more than my machine parts.

I had laser eye surgery about 12 years ago. It was the kind of laser surgery where they tell you to lie back, keep you wide awake and do all the surgery on the surface of your eye, no cornea-cutting.

Numb to my, well, eyeballs, I watched what looked like panes of glass or translucent ice being pushed across the surface of my eye. It was exactly like lying at the bottom of a swimming pool, watching disturbances ruffle the surface.

Those machine parts of my eyeball had nothing to do with me. I was way behind-beneath-inside them.

Do I believe in God? No. Do I believe in “souls”? Probably not.

I believe when this run of consciousness is run, we’re done. Whatever we’ve put out there, whatever we’ve worked on, for better or worse gets added to humanity’s great sandpile.

That — right there, whatever’s left behind — is the something of me that sits apart from the oily machine.

“Take what it is you do. That’s how you’ll be remembered when your travelling days are through.”³

Do I believe ChatGPT can contribute to the sandpile? No.

That’s what we were arguing about. It was about art.

“Is ChatGPT art?” he asked. Not talking about its output, mind, but ChatGPT itself.

Joel said yes, thinking of the code that produces it.

I said no, ChatGPT is a tool. It may be well-crafted, the code even written in aesthetically pleasing ways — but it has been created with an optimisation metric in mind. It’s optimising, improving, helping us do something. That makes it, fundamentally, a tool, not art.

Shout me down. Go ahead. “Tools can be art. Craftspeople are artists, etc.”

Well, sure.

But a tool performs a function — art is the extra. Hell, it’s right there in the name: exTRA.

Art is the feeling, not the function: the emotion layered on top of the function.

ChatGPT can process all the novels ever written — and it can process all the secondary literature that’s ever been written about those novels. It can tell me who exactly Mr George is in Bleak House and why he seems to know Esther (no spoilers please, I’m neck deep).

But until the woman nursing the dead baby makes ChatGPT feel something, it has no capacity for art. It can’t add anything exTRA.

We could argue the point until the cows jump out of Guernica and come home, until Jarndyce v Jarndyce is resolved.

Where ChatGPT is a conventional wisdom echo chamber, art challenges conventional wisdom. They are diametrically opposed.

Art communicates a feeling, an emotion. It shakes you out of convention, makes you feel what you wouldn’t otherwise feel. Tools can do that, sure, but if you use them to cut wood, eat your breakfast or answer a question about green beans, they are — primarily, at root — tools, not art.

What about a novel? A novel is pure art. A (good) novel doesn’t pretend it has any job to do — it puts you behind the eyes of others and makes you feel. It is all exTRA: time-wasting, inefficient. Women’s books, goes the traditional perception: un-serious books without a serious purpose.

“As long as there’s a world we’ve got to sing those folks the truth.”

The only truth is everyone’s truth: not some condensed, consolidated bland approximation — but the noisy mess of everyone’s contradictory truths at once.

Plus a little exTRA on top.

Which is why ChatGPT will never be art. It’s why all of the essays telling me who so-and-so was and about places X, Y and Z with no individual perspective are bland as fuck and boring to read (not that I read them — straight to unsubscribe).

As ever,

Lauren Hough

said it already:

So, actually, maybe I do believe in souls after all.

Just don’t tell my atheist boyfriend.

On an entirely different note, here’s an awkward old offering from the annals you might have missed. On this day last year:

1. I am indebted to

Erik Hoel

and his recent piece (below) for triggering some of these thoughts — and my argument with Joel, thanks.

2. Ref: W. Dunn, “Trapped in the AI Echo Chamber”, The New Statesman, October 2023: “More popular among the AI community, however, is the idea that humans themselves are simply receiving inputs and generating outputs. Sam Altman, perhaps the most powerful person in AI, espouses this belief ... .”

3. This (and the other block quote) is from the song Same Old Train by Marty Stuart & Friends. I’ll cool it on the bluegrass next week, promise.

As for the pull quote at the top, that’s from the messiest and bloodiest novel I’ve read recently — Mantel’s APOGS. Mantel knew a thing or two about the mess of everyone’s contradictory truths — and how to be remembered when your travelling days are through.