Jurassic Jaws Jones

Journal name:
Nature
Volume:
546,
Page:
696
Date published:
DOI:
doi:10.1038/546696a
Published online

Artistic licence.

Illustration by Jacey

So you want to be an actor, huh? Didn't your parents tell you acting is dead? Sorry if I spoilt that for you, kid. Sure, I've made some money in my time, but it's all over now. I haven't had a real job in years. If you don't know how it happened — boy, do I have a story for you.

First, they came for the artists with that art-melding neural network. Were you around when that was a thing? They take the style of one artist and feed it into the computer, which learns characteristics of the style and basically turns it into some web of numbers that's supposed to be like an artificial brain. Then you just put in another painting, and now you have Monet painting van Gogh's sunflowers. Funny, right? No, I don't know how it works exactly. If I did, I wouldn't be an actor. Anyways, they hooked their computers up to paintbrushes and automated mills and voilà, new Michelangelo frescos and sculptures, completely realistic. Turns out people would rather have Picasso paint their dog than commission any new art. Who wouldn't? Why, I gave my ex a Titian portrait of her for Christmas. No, that's not why we broke up.

Next, they came for the writers. At first, the networks could only generate silly texts that were like aliens trying to imitate Shakespeare. But somehow they figured out how to copy both the style and substance of stories just like they did with art, and then the floodgates opened. And yep, readers also don't want new things, they just want the same old things dressed up, and boy, are neural networks good at that. So now you can read David Copperfield set on Mars or Frankenstein written by Kafka on those ridiculous tablet readers that'll generate anything you damn well please.

You can probably guess that the musicians were next to become obsolete. We used to think that musical interpretation is something sublime and indescribable, but nope, neural networks can find all the parameters for them perfectly. No one wants to go to an unknown pianist's performance when they can listen to Rachmaninoff play the whole Chopin oeuvre. And don't think Chopin's oeuvre is sacred either — they just put works of any artist into the neural net, and it generates brand new pieces in seconds. Even musicologists couldn't distinguish between long-lost compositions and ones made up by the computer. They even learnt to render holograms of musicians playing. Did you see that abomination in New York? The grand premiere of Glenn Gould playing six new fugues by Bach. Heifetz playing Tchaikovsky's Violin Concerto No. 2. Just kill me.

Sorry, got a bit carried away there. I'm a sensitive soul, what can you say? If you don't know who they are, look them up. But make sure to look up the originals, not the neural-generated dreck. Anyways, what happened next? I called to my agent. I said, shouldn't we be concerned about this? He replied, sure, they can put Rachmaninoff's floating head on a fake body, but there's no way they can render an actor acting. Well guess what? One day we were laughing at a neural-network animation where the eyes moved into the nose, the next day people could watch Bette Davis in Star Wars in ultra-high definition, 100% realistic. It's insanity. Notice how there hasn't been a new actor or a new movie in a long time? Only endless remakes, redos, fusions, whatever crap they call them now. Titanic with the cast of Gone with the Wind. Studio Ghibli presents Blade Runner.

At first all the living actors decided we all had to band together, but people would rather watch Katherine Hepburn and Humphrey Bogart at their prime than any of us. So we all signed our faces away — I was one of the last holdouts, but I still had to do it in the end. People were forgetting who I was, even though I was the second-highest-paid actor a year ago. Is Jurassic Jaws Jones one of your favourite movies? Sorry to ruin it for you, but I didn't do anything — it was all neural-generated. And now they have those choose-your-own cast movies. I make $0.02 for each stream of The Godfather where they cast me as Michael Corleone. Actually, I'm probably Fredo most of the time.

If you want my advice, you gotta get ahead of the game. Be someone who makes neural networks, not someone who gets replaced by them. Now that they're done destroying the arts, they're already moving on to everyone else — teachers, politicians — you name it. And heck, they're probably training neural networks to make more neural networks. But at least if you learn them inside and out, maybe you can find something that they can't do, you know? Or, maybe you can figure out how us humans can be improved with implants or something. Don't be like me. I never learnt any maths or computer stuff, now I'm useless. All I'm good for is telling people crazy stories and signing autographs.

You're welcome. If you ever figure out a way to make me smart like in The Matrix, let me know before telling anyone else, okay? I'm counting on you, kid.

Author information

Affiliations

  1. Hal Y. Zhang enjoys thinking about future technologies.

Author details

Additional data