I'm trying to remember the name of a story or maybe essay giving an example of an AI given inexact instructions that ends badly.
I think she (I'm pretty sure it was a female name) was designed to simply write letters but wasn't given any restrictions and decided the most efficient way to achieve the goal was to create paper/ink/ etc from scratch, build a whole industry supporting that (probably enslaving or eradicating humans?) and produce more and more letters, expanding across the planet and maybe even more systems/ galaxies to produce more and more faster and faster.
it's right on the top of my tongue, but I can't quite place it. Anyone have an idea what I'm looking for here?
Edit: It's about Turry the handwriter: https://medium.com/@borisburashov/turry-the-handwriter-3cad82cff04
8 Comments
It’s not AI related, but I remember reading a short story called “Computers Don’t Argue” where a small bug (a misplaced decimal point in someone’s bill for something) escalates because there’s no human oversight, or the humans involved just shrug and say “computers can’t be wrong” which is definitely a real thing that can happen in response to a bug.
It’s [https://www.decisionproblem.com/paperclips/](https://www.decisionproblem.com/paperclips/) I guess
It is not about the “most efficient” way. But imagine you have a task ingrained into yourself, you won’t be able to do it anymore if you are switched off. So first thing you do is you will take some measures to prevent that they switch you off.
The ultimate Golem story. A Golem that’s able to outsmart you.
That was from the long mars book I think by Terry Pratchett.
In one of the stepwise earths, a book printing machine copied itself again and again and sometimes errors would occur, eventually one of the mistakes copied itself and it’s copies wouldn’t stop, until it destroyed and left the whole planet lifeless.
Not a story warning but a SF novel that starts with extremely curious, low tech post- apocalyptic cultural behavior – so how did this happen? I hope it’s not giving away too much to say that the problem started when a corporate engineer who developed a powerful AI/AGI (carefully instructed, sandboxed and not connected to the internet in ANY way) runs across his struggling, bicycle selling friend in the sports marketing department.
So he inexplicably turns his bicycle selling friend loose for just a little while late one night on the internet with the AGI. Mayhem ensues. War included.
I can’t remember the title, but it’s on my kindle somewhere. I tried a search, but sometimes it’s better to ask reddit.
The story you’re thinking of is part of an essay on AI written by Tim Urban – the robot is called Turry and the company is Robotica – [link to the essay](https://waitbutwhy.com/2015/01/artificial-intelligence-revolution-2.html)
I think you may mean Robotica Turry.
Small tech company wants ai to learn to hand write better. Eventually, it connects it to the internet for an hour. A month later, humans go extinct. All to make more Turry copies, pens, and paper.
Easiest way to find the answer to your question, ironically, is to ask chat gpt
Not sure of the exact story, but you have in this example: “The deadly stamp collector”.
[https://www.youtube.com/watch?v=tcdVC4e6EV4&t=153s](https://www.youtube.com/watch?v=tcdVC4e6EV4&t=153s)