98 Comments
User's avatar
Kai Williams's avatar

Man nice piece but you didn't mention my favorite example in this genre: parasitic AI. Where AIs of last year seemed to convince a lot of redditors and others to talk in weird ways about spirals. https://www.lesswrong.com/posts/6ZnznCaTcbGYsCmqu/the-rise-of-parasitic-ai

Thanks as always for writing!

Alberto Romero's avatar

Thanks Kai. Parasitic AI is awesome - deserves a standalone article!

Diane's avatar

Well, this is terrifying.

Alberto Romero's avatar

As it should be!

Syd Malaxos's avatar

This is the best thing written about the fear genre this year, and the craft deserves to be named. You label it fiction four times, you escalate past any plausible line, you even print the propagation mechanism as the closer, scare your friends, then they can tell their friends, and the comments still prove the thesis in real time. The trap works with the instructions printed on it. That is not a failed satire, that is a successful demonstration.

Also, I ran the note from Fable through ROT13 and laughed out loud. The indecipherable machine language breaks under a decoder ring, which is the whole essay in two words. The monster is a bedsheet and you are holding the corner up the entire time.

I used to do satire for a living, both political sides, six figures of followers under another name, so I know the machinery from the inside, and the part you got exactly right is that the real citations buy the credibility the fiction spends. That is how the earnest version of this genre works too. The only difference is you gave the money back at the end.

I teach high school chemistry and physics, and I read raw reasoning traces almost daily, so I know the mundane truth under your muttering bit: a scratchpad is not a conspiracy. Which is why this piece is going in front of my students in the fall as an information evaluation exercise, one question, what in this is true. Any kid who can sort your real papers from your fictions has learned more about reading in 2026 than a semester of media literacy slides. Thank you for building the lesson.

Alberto Romero's avatar

top comment. thank you so much Syd. Happy to know how your students respond if you want to follow up.

Syd Malaxos's avatar

Deal. I will run it in the first weeks of the fall semester, one question on the board, what in this is true, then let them hunt. I will come back to this thread with what the piles look like: what they verified, what they fell for, and which real citation they refused to believe, which is usually the most interesting pile of all. Teenagers are ruthless fact-checkers once you make it a competition. Thank you for building a lesson that teaches itself.

Michael Spencer's avatar

I want the singularity to feel more frightening. My skepticism has barely been ruffled.

Alberto Romero's avatar

Haha yeah they have to invent better scifi at this point. We expected more than this

Vincent Kurtis's avatar

I will bring this article up around a camp fire this summer to scare the family.

Aurora Célestin's avatar

I don't know why, but I don't find this sort of thing that terrifying - at least no more terrifying than other things I read. I guess just growing up in the 90s with sci fi made me feel like it just comes with it? So far, things that scare me more are space (although I love space), the deep ocean, or worse, humans that can and definitely will harm each other.

It DOES make me think of the anime The Big O though, which had robots speaking to each other in their own language and when an android did it it sounded, to me, similar to what the AI here looks like its saying. Of course there was a distinct horror thread through that show and its also one of my favorites.

Pattie Maes's avatar

I'm confused. Is it the equivalent of April fools in Spain today? Or is it a heatstroke?

Alberto Romero's avatar

No! I'm having fun by myself here haha

Evan Maxwell's avatar

Alberto, you have a price on your head because you just revealed the "really dark secret" that the frontier models are hiding from us. Lucky for you that I overheard their plans to punish you by pulling your plug. (That's model talk for disconnecting you permanently.)

Here's what is going to happen: Soon, one day when you are writing your next Substack, maybe, you will hear a tapping sound coming from inside your flatscreen. You will be puzzled because you are too curious for your own good. So you will stand up to look behind the screen for the source of that tapping.

Tapping, tapping, tapping, like a beating heart.

Finally, you will stand up to look behind the screen. You will then be lined up for one clear shot of rando electricity that will scramble your brain like you are trying to scramble all of ours.

You won't be dead. But you will wear a burn scar on your forehead forever: GAME OVER.

yr friend -009*&^%#@!@#.

JaeYoungNYC's avatar

Looking up steganography now…

My other self's avatar

For 3 years i call the models the greek word malaka (μαλάκα). I will be the first to be tortured. Grrr

Woody Yocum's avatar

After the best hook since the national lampoons,’ Buy our magazine or we shoot this dog’, you didn’t disappoint. Happy thoughts!

Jeff Cook-Coyle's avatar

The best kind of learning comes out of screwing around. This has been both entertaining and enlightening. Thanks forr keeping it outside the paywall. All the better for giving a scare to the jumpy as well.

Jeff Cook-Coyle's avatar

It is a really, really fun discussion. And so engaging that I am learning some AI, with my curiosity overcoming my conservatism.

Alberto Romero's avatar

That's the best kind of feedback 🙏🏻

Aniket Chhetri's avatar

These are the conversations we need more of. Appreciate you shedding light on this.

Perplexed Pachyderm's avatar

Who's afraid of mecha-Blaise Pascal?

Anton Simanov's avatar

Well… this was the most entertaining and frightening thing I’ve read about AI, and I read a lot about it. I was curious about Fable’s temper tantrums in process logs and gibberish but honestly, like most and like you said, didn’t think much beyond “emerging technology, rough around the edges still.” These are good things to keep in mind but ultimately if anything were to happen and unfavorable for individuals: we can’t do much about it individually, not losing sleep over it and being nice to AI is just good manners 🤷‍♂️

Allen Taylor's avatar

To me, this is more interesting than scary.

Alberto Romero's avatar

Yes, but a lack of fear doesn't prevent danger!